server.py

#!/usr/bin/env python
"""ViralLab — Engineer your influence. Turn noise into viral."""
import os
import subprocess
import sys
import threading
import time
import io
import zipfile
from pathlib import Path

# Load .env so BILIBILI_PROXY etc. work when running server directly
try:
    from dotenv import load_dotenv
    load_dotenv(Path(__file__).parent / ".env")
except ImportError:
    pass

import json
from typing import Optional

# Bypass proxy for news/video fetches (fixes "connection refused" when proxy is down)
_PROXY_VARS = ("HTTP_PROXY", "HTTPS_PROXY", "http_proxy", "https_proxy", "ALL_PROXY", "all_proxy")


def _ensure_macos_system_proxy_in_env(env: dict) -> dict:
    """If env has no HTTP(S)_proxy, copy macOS system proxy from scutil (QuickQ / gost / Clash)."""
    if env.get("HTTP_PROXY") or env.get("HTTPS_PROXY"):
        return env
    if sys.platform != "darwin":
        return env
    try:
        r = subprocess.run(
            ["scutil", "--proxy"],
            capture_output=True,
            text=True,
            timeout=2,
            cwd=Path(__file__).parent,
        )
        if r.returncode == 0 and r.stdout and ("HTTPEnable : 1" in r.stdout or "HTTPSEnable : 1" in r.stdout):
            host = port = None
            for line in r.stdout.splitlines():
                s = line.strip()
                if s.startswith("HTTPProxy :"):
                    host = s.split(":", 1)[1].strip()
                elif s.startswith("HTTPSProxy :"):
                    host = host or s.split(":", 1)[1].strip()
                elif s.startswith("HTTPPort :"):
                    port = s.split(":", 1)[1].strip()
                elif s.startswith("HTTPSPort :"):
                    port = port or s.split(":", 1)[1].strip()
            if host and port:
                env["HTTP_PROXY"] = env["HTTPS_PROXY"] = f"http://{host}:{port}"
    except Exception:
        pass
    return env


def _env_for_bilibili():
    """Env for B站 subprocess: use system proxy so Clash etc. can route bilibili.com to DIRECT (VPN on still works)."""
    env = os.environ.copy()
    env["PYTHONPATH"] = str(Path(__file__).parent)
    return _ensure_macos_system_proxy_in_env(env)


def _env_no_proxy():
    env = os.environ.copy()
    for v in _PROXY_VARS:
        env.pop(v, None)
    return env


def _env_for_topic_search_subprocess() -> dict:
    """Env for scripts/search_only.py: keep proxy + macOS scutil; zh multi-source can be slow."""
    env = _ensure_macos_system_proxy_in_env(os.environ.copy())
    env["PYTHONPATH"] = str(Path(__file__).parent)
    return env


def _topic_search_subprocess_timeout_sec() -> int:
    try:
        t = int(os.environ.get("NEWS_TOPIC_SEARCH_TIMEOUT_SEC", "180"))
    except ValueError:
        t = 180
    return max(60, min(t, 600))


from urllib.parse import quote
from flask import Flask, jsonify, redirect, request, send_from_directory, send_file, make_response

from src.parse_output import parse_file, parse_raw_news
from src.news_sources import raw_topic_file_stem
from src.video_tools import extract_youtube_id, score_berger
from src.content_angles import generate_angles
from src.minto_pyramid import structure_minto, minto_to_markdown
from src.run_history import add_lifecycle_to_items
from src.hot_trending import AUDIENCE_SEGMENTS, rerank_topics_by_segment
from src.media_transcribe import (
    download_youtube_video,
    has_videocaptioner,
    is_supported_external_media_url,
    test_douyin_session,
    transcribe_best_effort,
)
from src.transcription.whisperx_runner import has_whisperx
from src.image_ocr import ocr_image_bytes, ocr_job_for_url
from src.contagious_stepps import evaluate_contagious_article
from src.article_lab import build_viral_article_markdown
from src.publish_pipeline import (
    list_publish_items,
    load_markdown_from_output,
    publish_item_now,
    queue_publish_item,
    validate_publish_content,
)
from src.diagnostics import classify_error_message
from src.podcast_rss_sources import (
    REGION_TAGS,
    classify_focus_theme,
    classify_source_type,
    fetch_longform_and_podcasts,
    region_tags_for_lang,
)

app = Flask(__name__)
OUTPUT = Path(__file__).parent / "output"
_daily_stale_refresh_lock = threading.Lock()
_daily_last_stale_refresh_attempt = 0.0
NEWS_SEARCHES_FILE = OUTPUT / "news_searches.json"
LONGFORM_CACHE_TTL_SEC = 30 * 60
_LONGFORM_REFRESHING = {"en": False, "zh": False}
_virallab_apscheduler = None
_virallab_scheduler_lock = threading.Lock()

# Suggestions only — users can search any topic. Creator-relevant defaults per language zone.
DEFAULT_TOPIC_TIPS = [
    "AI agents", "creator economy", "climate tech", "no-code tools", "LLM trends",
    "creator tools", "indie hacking", "side projects", "AI explainers", "content strategy",
    "viral marketing", "social media trends", "YouTube algorithm", "newsletter monetization",
    "Substack trends", "podcast growth", "automation tools", "AI writing", "design tools",
    "remote work",
]
DEFAULT_TOPIC_TIPS_ZH = [
    "小红书爆款", "抖音运营", "短视频创作", "创作者经济", "直播带货", "私域流量",
    "电商直播", "内容营销", "AI写作", "B站UP主", "知乎好物", "微信公众号",
    "品牌出海", "小红书种草", "抖音算法", "剪映教程", "飞书文档", "即刻",
    "知识星球", "新榜",
]

# Viral Videos — creator-focused topic suggestions per language
DEFAULT_VIRAL_TIPS = [
    "AI explainer", "fashion haul", "viral marketing", "unboxing", "makeup tutorial",
    "cooking viral", "dance challenge", "prank", "product review", "day in my life",
]
DEFAULT_VIRAL_TIPS_ZH = [
    "热门", "鬼畜", "搞笑", "数码", "游戏", "知识", "时尚", "影视",
    "汽车", "日常", "穿搭",
]

# Audience segment topic tips (Tier 1: docs/AUDIENCE_INSIGHT_STRATEGY.md)
SEGMENT_TOPIC_TIPS = {
    "general": {"en": None, "zh": None},  # use DEFAULT_TOPIC_TIPS / DEFAULT_TOPIC_TIPS_ZH
    "tech": {
        "en": ["AI agents", "LLM trends", "developer tools", "open source", "SaaS", "product launch", "tech startup", "API", "automation"],
        "zh": ["AI", "科技", "产品", "效率工具", "开源", "开发者", "互联网", "数码", "算法"],
    },
    "entrepreneurs": {
        "en": ["indie hacking", "side project", "creator economy", "monetization", "growth", "launch", "startup", "marketing", "remote work"],
        "zh": ["创业", "副业", "出海", "变现", "个人品牌", "品牌出海", "运营", "私域流量", "增长"],
    },
    "creator": {
        "en": ["creator economy", "viral marketing", "content strategy", "YouTube algorithm", "newsletter", "Substack", "podcast growth"],
        "zh": ["小红书爆款", "抖音运营", "创作者经济", "直播带货", "内容营销", "B站UP主", "种草", "爆款"],
    },
}
SEGMENT_VIRAL_TIPS = {
    "general": {"en": None, "zh": None},
    "tech": {"en": ["AI explainer", "tech review", "developer", "startup"], "zh": ["知识", "数码", "科技"]},
    "entrepreneurs": {"en": ["indie hack", "launch", "growth", "side project"], "zh": ["创业", "出海", "变现"]},
    "creator": {"en": ["viral marketing", "content strategy", "tutorial"], "zh": ["爆款", "运营", "种草"]},
}


def _record_topic_search(topic: str) -> None:
    """Record a topic search with timestamp for time-based popular searches."""
    if not topic or len(topic) > 80:
        return
    from datetime import datetime, timezone
    data = {"searches": []}
    if NEWS_SEARCHES_FILE.exists():
        try:
            data = json.loads(NEWS_SEARCHES_FILE.read_text(encoding="utf-8"))
        except Exception:
            pass
    searches = data.get("searches", [])
    # Migrate old format: topics -> searches (no timestamp, treat as recent)
    if not searches and data.get("topics"):
        now = datetime.now(timezone.utc).isoformat()
        for t, c in data["topics"].items():
            searches.extend([{"topic": t, "ts": now}] * min(c, 10))
        data["topics"] = {}
    searches.append({"topic": topic.strip(), "ts": datetime.now(timezone.utc).isoformat()})
    # Keep last 30 days
    cutoff = datetime.now(timezone.utc).timestamp() - (30 * 24 * 3600)
    searches = [s for s in searches if _parse_ts(s.get("ts", "")) > cutoff]
    data["searches"] = searches[-5000:]
    OUTPUT.mkdir(exist_ok=True)
    NEWS_SEARCHES_FILE.write_text(json.dumps(data, ensure_ascii=False), encoding="utf-8")


def _parse_ts(s: str) -> float:
    from datetime import datetime
    try:
        return datetime.fromisoformat(s.replace("Z", "+00:00")[:26]).timestamp()
    except Exception:
        return 0


def _has_cjk(s: str) -> bool:
    """True if string contains CJK (Chinese/Japanese/Korean) characters."""
    for c in s:
        if "\u4e00" <= c <= "\u9fff" or "\u3000" <= c <= "\u303f" or "\uac00" <= c <= "\ud7af":
            return True
    return False


def _matches_lang(topic: str, lang: str) -> bool:
    """True if topic matches the language zone (no mixing in EN/zh views)."""
    has_cjk = _has_cjk(topic)
    return (lang == "zh" and has_cjk) or (lang == "en" and not has_cjk)


def _get_segment() -> str:
    """Audience segment from query or cookie. One of: general, tech, entrepreneurs, creator."""
    seg = (request.args.get("segment") or request.cookies.get("virallab_segment") or "general").lower()
    return seg if seg in ("general", "tech", "entrepreneurs", "creator") else "general"


def _get_platform_hot_topics(
    limit: int,
    lang: str,
    force_refresh: bool = False,
    segment: str = "general",
    *,
    allow_blocking_fetch: bool = True,
) -> tuple[list[str], bool, Optional[int]]:
    """Platform-wide hot topics. ZH: 微博、知乎、抖音、百度. EN: Hacker News, Reddit. Returns (topics, ok, cache_age_sec).
    If segment != general, topics are re-ranked by segment relevance. Never blocks >2s."""
    import threading
    cache_file = OUTPUT / "hot_trending.json"
    cache_ttl = 10 * 60  # 10 min
    key = "topics_zh" if lang == "zh" else "topics_en"
    now = __import__("time").time()

    def _run_refresh():
        try:
            script = Path(__file__).parent / "scripts" / "fetch_hot_trending.py"
            env = _env_no_proxy()
            env["PYTHONPATH"] = str(Path(__file__).parent)
            subprocess.run(
                [sys.executable, str(script)],
                capture_output=True,
                env=env,
                cwd=Path(__file__).parent,
                timeout=45,
            )
        except Exception:
            pass

    # Use fresh cache if available
    if not force_refresh and cache_file.exists():
        try:
            data = json.loads(cache_file.read_text(encoding="utf-8"))
            ts = data.get("ts", 0)
            topics = data.get(key, data.get("topics", []))
            if topics:
                if segment != "general":
                    topics = rerank_topics_by_segment(topics, lang, segment)
                if now - ts < cache_ttl:
                    return topics[:limit], True, int(now - ts)
                threading.Thread(target=_run_refresh, daemon=True).start()
                return topics[:limit], True, int(now - ts)
        except Exception:
            pass

    # Avoid blocking HTTP handlers (e.g. /news on cold deploy): fill cache in background only.
    if not allow_blocking_fetch and not force_refresh:
        threading.Thread(target=_run_refresh, daemon=True).start()
        return [], False, None

    # No usable cache or force_refresh: run sync fetch. User-initiated refresh gets longer timeout (~50s).
    sync_timeout = 55 if force_refresh else 8
    try:
        script = Path(__file__).parent / "scripts" / "fetch_hot_trending.py"
        env = _env_no_proxy()
        env["PYTHONPATH"] = str(Path(__file__).parent)
        subprocess.run(
            [sys.executable, str(script)],
            capture_output=True,
            env=env,
            cwd=Path(__file__).parent,
            timeout=sync_timeout,
        )
        if cache_file.exists():
            data = json.loads(cache_file.read_text(encoding="utf-8"))
            topics = data.get(key, data.get("topics", []))
            if topics:
                if segment != "general":
                    topics = rerank_topics_by_segment(topics, lang, segment)
                return topics[:limit], True, 0
    except Exception:
        pass
    threading.Thread(target=_run_refresh, daemon=True).start()
    return [], False, None


def _get_most_searched(
    limit: int = 10,
    time_range: str = "1d",
    lang: str = "en",
    force_hot_refresh: bool = False,
    segment: str = "general",
    *,
    allow_blocking_hot_fetch: bool = False,
) -> tuple[list[str], str, Optional[int]]:
    """Top topics. Returns (topics, source, hot_cache_age_sec). source: 'platform'|'app'|'default'. hot_cache_age only when platform."""
    from datetime import datetime, timezone
    seg_tips = SEGMENT_TOPIC_TIPS.get(segment, {})
    defaults = (seg_tips.get(lang) or (DEFAULT_TOPIC_TIPS_ZH if lang == "zh" else DEFAULT_TOPIC_TIPS))[:limit * 2]

    block_fetch = allow_blocking_hot_fetch or force_hot_refresh
    platform_topics, ok, cache_age = _get_platform_hot_topics(
        limit,
        lang,
        force_refresh=force_hot_refresh,
        segment=segment,
        allow_blocking_fetch=block_fetch,
    )
    if ok:
        return platform_topics[:limit], "platform", cache_age

    now = datetime.now(timezone.utc).timestamp()
    if time_range == "60m":
        cutoff = now - 3600
    elif time_range == "7d":
        cutoff = now - (7 * 24 * 3600)
    else:
        cutoff = now - (24 * 3600)
    if not NEWS_SEARCHES_FILE.exists():
        return defaults[:limit], "default", None
    try:
        data = json.loads(NEWS_SEARCHES_FILE.read_text(encoding="utf-8"))
        searches = data.get("searches", [])
        if not searches and data.get("topics"):
            sorted_topics = sorted(data["topics"].items(), key=lambda x: -x[1])
            result = [t[0] for t in sorted_topics[:limit] if _matches_lang(t[0], lang)]
            for d in defaults:
                if len(result) >= limit:
                    break
                if d not in result:
                    result.append(d)
            return result[:limit], "app" if data["topics"] else "default", None
        counts = {}
        for s in searches:
            ts = _parse_ts(s.get("ts", ""))
            if ts >= cutoff:
                t = s.get("topic", "").strip()
                if t and _matches_lang(t, lang):
                    counts[t] = counts.get(t, 0) + 1
        if not counts:
            return defaults[:limit], "default", None
        sorted_topics = sorted(counts.items(), key=lambda x: -x[1])
        result = [t[0] for t in sorted_topics[:limit]]
        for d in defaults:
            if len(result) >= limit:
                break
            if d not in result:
                result.append(d)
        return result[:limit], "app", None
    except Exception:
        return defaults[:limit], "default", None


def _news_refresh_max_topics() -> int:
    try:
        n = int(os.environ.get("NEWS_REFRESH_MAX_TOPICS", "15"))
    except ValueError:
        n = 15
    return max(5, min(n, 40))


def _topics_for_news_refresh() -> list[str]:
    """Queries for /news/refresh — platform hot tags across segments (ZH+EN), deduped by file stem."""
    cap = _news_refresh_max_topics()
    seen_stems: set[str] = set()
    out: list[str] = []
    for seg in ("general", "tech", "entrepreneurs", "creator"):
        for slang in ("zh", "en"):
            topics, _, _ = _get_most_searched(10, "1d", slang, False, seg)
            for t in topics or []:
                st = raw_topic_file_stem(t)
                if st in seen_stems:
                    continue
                seen_stems.add(st)
                out.append(t)
                if len(out) >= cap:
                    return out
    for slang in ("zh", "en"):
        defaults = DEFAULT_TOPIC_TIPS_ZH if slang == "zh" else DEFAULT_TOPIC_TIPS
        for d in defaults:
            st = raw_topic_file_stem(d)
            if st not in seen_stems:
                seen_stems.add(st)
                out.append(d)
            if len(out) >= cap:
                return out
    return out


def _news_card_rows(
    lang: str,
    time_range: str,
    segment: str,
    force_hot: bool,
    topic_filter: str,
) -> list[dict]:
    """One card block per live hot tag: stem, markdown file, display label, rescan query."""
    if topic_filter:
        stem = topic_filter.strip()
        disp = stem.replace("_", " ")
        return [{"stem": stem, "file": f"raw_{stem}.md", "display_topic": disp, "query_topic": disp}]
    tips, _, _ = _get_most_searched(
        limit=12,
        time_range=time_range,
        lang=lang,
        force_hot_refresh=force_hot,
        segment=segment,
    )
    rows: list[dict] = []
    seen: set[str] = set()
    for tag in tips:
        stem = raw_topic_file_stem(tag)
        if stem in seen:
            continue
        seen.add(stem)
        rows.append({"stem": stem, "file": f"raw_{stem}.md", "display_topic": tag, "query_topic": tag})
    if len(rows) < 6:
        defaults = DEFAULT_TOPIC_TIPS_ZH if lang == "zh" else DEFAULT_TOPIC_TIPS
        for d in defaults:
            if len(rows) >= 12:
                break
            st = raw_topic_file_stem(d)
            if st in seen:
                continue
            seen.add(st)
            rows.append({"stem": st, "file": f"raw_{st}.md", "display_topic": d, "query_topic": d})
    return rows


def _hot_tags_for_lang(lang: str, limit: int = 10) -> list[str]:
    """Unified hot tags references (EN and ZH kept separate)."""
    seg = _get_segment()
    tags, _source, _age = _get_most_searched(
        limit=max(limit, 10), time_range="1d", lang=lang, force_hot_refresh=False, segment=seg
    )
    if tags:
        return tags[:limit]
    return (DEFAULT_TOPIC_TIPS_ZH if lang == "zh" else DEFAULT_TOPIC_TIPS)[:limit]


def _longform_cache_path(lang: str) -> Path:
    return OUTPUT / f"longform_cache_{'zh' if lang == 'zh' else 'en'}.json"


def _read_longform_cache(lang: str) -> tuple[list[dict], list[str], float]:
    path = _longform_cache_path(lang)
    if not path.exists():
        return [], [], 0.0
    try:
        data = json.loads(path.read_text(encoding="utf-8"))
        items = data.get("items", [])
        sources = data.get("sources", [])
        ts = float(data.get("ts") or 0.0)
        if isinstance(items, list) and isinstance(sources, list) and ts > 0:
            return items, sources, ts
    except Exception:
        pass
    return [], [], 0.0


def _write_longform_cache(lang: str, items: list[dict], sources: list[str]) -> None:
    path = _longform_cache_path(lang)
    path.parent.mkdir(parents=True, exist_ok=True)
    payload = {
        "lang": "zh" if lang == "zh" else "en",
        "ts": time.time(),
        "items": items,
        "sources": sources,
    }
    path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")


def _refresh_longform_cache(lang: str, max_per_feed: int = 6) -> None:
    try:
        items, sources = fetch_longform_and_podcasts(lang=lang, max_per_feed=max_per_feed)
        if items:
            _write_longform_cache(lang, items, sources)
            print(f"[longform] refreshed {lang}: {len(items)} items from {len(sources)} feeds")
        else:
            print(f"[longform] refresh {lang}: 0 items (check network, proxy, or feedparser)")
    except Exception as e:
        print(f"[longform] refresh {lang} failed: {e}")
    finally:
        _LONGFORM_REFRESHING[lang] = False


def _kick_longform_refresh(lang: str, max_per_feed: int = 6) -> None:
    if _LONGFORM_REFRESHING.get(lang):
        return
    _LONGFORM_REFRESHING[lang] = True
    threading.Thread(
        target=_refresh_longform_cache,
        kwargs={"lang": lang, "max_per_feed": max_per_feed},
        daemon=True,
    ).start()


def _get_longform_feed(lang: str, max_per_feed: int = 6, force_refresh: bool = False) -> tuple[list[dict], list[str], bool, int]:
    now = time.time()
    cached_items, cached_sources, cache_ts = _read_longform_cache(lang)
    cache_age = int(now - cache_ts) if cache_ts else -1
    cache_ok = bool(cached_items)

    if cache_ok and not force_refresh:
        if cache_age <= LONGFORM_CACHE_TTL_SEC:
            return cached_items, cached_sources, True, cache_age
        _kick_longform_refresh(lang, max_per_feed=max_per_feed)
        return cached_items, cached_sources, True, cache_age

    try:
        items, sources = fetch_longform_and_podcasts(lang=lang, max_per_feed=max_per_feed)
        if items:
            _write_longform_cache(lang, items, sources)
            return items, sources, False, 0
    except Exception:
        pass

    if cache_ok:
        return cached_items, cached_sources, True, cache_age
    return [], [], False, -1


TAGLINE = "Engineer your influence. Turn noise into viral."
SPREAD_LINE = "ViralLab scores content by applying behavioral science - aka. STEPPS, design viral content - so easy"
TG_GROUP_URL = "https://t.me/virallab8"
TG_PERSONAL_URL = "https://t.me/miccakitt"
BLOG_URL = "https://funghi88.github.io/"
GITHUB_REPO_URL = "https://github.com/Funghi88/ViralLab"
X_URL = "https://x.com/miccakitt"


def _html_escape(s):
    return (
        str(s)
        .replace("&", "&amp;")
        .replace("<", "&lt;")
        .replace(">", "&gt;")
        .replace('"', "&quot;")
    )


def _sanitize_snippet(text: str) -> str:
    """Strip HTML tags and decode entities so snippets display as plain text, not raw HTML."""
    if not text or not isinstance(text, str):
        return ""
    import html
    import re
    s = html.unescape(text)
    s = re.sub(r"<[^>]*>?", "", s)  # strip tags including malformed/unclosed
    s = re.sub(r"\s+", " ", s).strip()
    return s[:500]


# Snippet placeholders: when snippet equals these, it's source attribution, not news content
_SNIPPET_PLACEHOLDERS = frozenset({"Hacker News front page", "HN", "Hacker News"})


def _snippet_is_placeholder(snippet: str) -> bool:
    """True when snippet is a source placeholder, not actual news content."""
    return (snippet or "").strip() in _SNIPPET_PLACEHOLDERS


def _source_display(source: str) -> str:
    """Source attribution for display. Use full name for respect."""
    if source == "Hacker News":
        return "Hacker News front page"
    return source or ""


def _get_source_for_item(item: dict) -> str:
    """Get source for display. When snippet is placeholder (e.g. HN), use it as source."""
    src = _source_display(item.get("source", "")) or _infer_source_from_url(item.get("url", ""))
    if not src and _snippet_is_placeholder(item.get("snippet", "")):
        snip = (item.get("snippet") or "").strip()
        if snip == "HN":
            return "Hacker News front page"
        return snip
    return src


def _source_type_for_item(item: dict) -> str:
    """Classify item source to support Daily filters."""
    return classify_source_type(_get_source_for_item(item))


def _infer_source_from_url(url: str) -> str:
    """Infer source from URL when source field is missing."""
    if not url:
        return ""
    u = (url or "").lower()
    if "news.ycombinator.com" in u:
        return "Hacker News front page"
    if "theverge.com" in u:
        return "The Verge"
    if "techcrunch.com" in u:
        return "TechCrunch"
    if "producthunt.com" in u:
        return "Product Hunt"
    return ""


def _get_lang():
    """Get user language from cookie or query param. Returns 'en' or 'zh'."""
    lang = request.cookies.get("virallab-lang") or request.args.get("lang", "en")
    return "zh" if lang == "zh" else "en"


def _get_region():
    """Get user region from URL param (for shared links) or cookie. Returns 'global', 'americas', 'europe', or 'asia'."""
    r = request.args.get("region") or request.cookies.get("virallab-region") or "global"
    return r if r in ("global", "americas", "europe", "asia") else "global"


def _t(key: str, lang: str) -> str:
    """Translation lookup. Key -> en/zh string."""
    T = {
        "daily_news_title": ("Daily News", "每日新闻"),
        "daily_news_desc": (
            "Top 3 = most talked about right now. News, long-form, and podcast sources in one feed.",
            "前 3 条为当下最热话题，并整合新闻、长文与播客来源",
        ),
        "daily_focus_label": ("Focus", "聚焦"),
        "daily_focus_all": ("All", "全部"),
        "daily_focus_news": ("News", "新闻"),
        "daily_focus_longform": ("Long-form", "长文"),
        "daily_focus_podcast": ("Podcast", "播客"),
        "daily_focus_hint": ("Filter cards by source type.", "按来源类型筛选卡片"),
        "mechanism": ("", ""),
        "refresh_now": ("Refresh now", "立即刷新"),
        "refresh_desc": (
            "fetch latest anytime. Server refreshes daily digest every 30 min by default; opening this page also triggers a background refresh if the digest is older than ~8h.",
            "随时拉取最新。服务器默认约每 30 分钟刷新每日新闻；若摘要超过约 8 小时未更新，打开本页也会在后台自动补跑一次。",
        ),
        "longform_refresh_desc": (
            "Cache-first for speed. Refreshes in background after 30 min, on a schedule, and when the server starts.",
            "优先读缓存；超过约 30 分钟、定时任务以及每次启动服务后都会在后台重新拉取。",
        ),
        "longform_cache_age": ("Cache age: {mins} min", "缓存时长：{mins} 分钟"),
        "last_updated": ("Last updated", "最后更新"),
        "sources": ("Sources", "来源"),
        "sources_curated": ("curated for content creators.", "为创作者精选"),
        "setup": ("SETUP", "设置"),
        "view_full_digest": ("View full digest (10 Top News · load more)", "查看完整摘要（10 条头条 · 加载更多）"),
        "full_digest_title": ("Full digest (10 Top News · load more)", "完整摘要（10 条头条 · 加载更多）"),
        "load_more": ("More", "更多"),
        "ranking_legend": ("Rising = new · Peaking = holding strong · Fading = declining", "上升 = 新话题 · 高峰 = 热度持稳 · 回落 = 热度下降"),
        "tooltip_rising": ("New this run — wasn't in the previous top list. A fresh topic gaining traction.", "本轮新上榜，上一轮未出现。新兴热门话题"),
        "tooltip_peaking": ("Same or better rank than before — the topic is at or near its peak attention.", "排名持平或上升，话题处于或接近关注高峰"),
        "tooltip_fading": ("Dropped in rank — was higher before, now declining in buzz.", "排名下降，热度回落"),
        "search_by_topic": ("Search by topic", "按主题搜索"),
        "whats_happening": ("what's happening around X", "了解 X 领域动态"),
        "no_news": ("No daily news yet. Run:", "尚无每日新闻，执行："),
        "source": ("Source", "来源"),
        "contagious_score": ("Contagious", "传播性"),
        "lifecycle_rising": ("rising", "上升"),
        "lifecycle_peaking": ("peaking", "高峰"),
        "lifecycle_fading": ("fading", "回落"),
        # Homepage & global
        "nav_home": ("Home", "首页"),
        "nav_home_desc": ("Landing & quick links", "首页与快速链接"),
        "nav_daily": ("Daily News", "每日新闻"),
        "nav_daily_desc": ("Top 3 topics today", "今日前 3 热门话题"),
        "nav_field": ("Your Field", "你的领域"),
        "nav_field_desc": ("Curated by industry", "按行业精选"),
        "nav_news": ("News by Topic", "按主题新闻"),
        "nav_news_desc": ("Search & browse news", "搜索与浏览新闻"),
        "nav_longform": ("Long-form & Podcasts", "长文与播客"),
        "nav_longform_desc": ("Deep articles, blogs, and podcasts", "深度文章、博客与播客"),
        "nav_viral": ("Viral Videos", "热门视频"),
        "nav_viral_desc": ("YouTube & Bilibili", "YouTube 与 Bilibili"),
        "nav_video2text": ("Video to Text", "视频转文字"),
        "nav_video2text_desc": ("Extract transcript from video", "视频提取逐字稿"),
        "nav_article_lab": ("Article Lab", "文章改写"),
        "nav_article_lab_desc": ("Upload markdown → viral rewrite", "上传 markdown → 结构化爆款改写"),
        "nav_publish": ("Publish Queue", "发布队列"),
        "nav_publish_desc": ("Approve and publish to channels", "审核并发布到渠道"),
        "nav_science": ("STEPPS & Structure", "评分与结构"),
        "nav_science_desc": ("Scoring + Minto Pyramid", "评分与金字塔结构"),
        "nav_china_content": ("China content", "中国内容"),
        "nav_china_content_desc": ("Fetch from Bilibili, XHS, Douyin, Zhihu, Shipinhao", "从B站、小红书、抖音、知乎、视频号拉取"),
        "hero_line1": ("ENGINEER YOUR", "打造你的"),
        "hero_line2": ("INFLUENCE.", "影响力"),
        "hero_subtitle": ("Your content agent.", "你的内容助手"),
        "hero_desc": ("Turn noise into viral. Gather trends, score with behavioral science, and design content that spreads.", "化杂讯为爆款。小红书、抖音、B站创作者必备 — 趋势、评分、内容角度"),
        "hero_cta": ("Explore Features", "探索功能"),
        "search_placeholder": ("Search viral videos (e.g. AI explainer)...", "搜索热门视频（如：AI 解说）..."),
        "how_we_score": ("How we score", "评分方式"),
        "tg_group": ("TG group", "TG 群组"),
        "home_title": ("Home", "首页"),
        "tagline": ("Engineer your influence. Turn noise into viral.", "打造你的影响力。化杂讯为爆款"),
        "spread_line": ("ViralLab scores content by applying behavioral science - aka. STEPPS, design viral content - so easy", "ViralLab 运用行为科学（STEPPS）评分内容，设计爆款内容更轻松"),
        "link_daily": ("Daily News", "每日新闻"),
        "link_daily_desc": ("Top 3 most talked about or key focus topics", "前 3 条最热门或重点话题"),
        "link_field": ("Your Field", "你的领域"),
        "link_field_desc": ("Curated trends and resources by industry", "按行业精选的趋势与资源"),
        "link_news": ("News by Topic", "按主题新闻"),
        "link_news_desc": ("Search and browse news by topic", "按主题搜索与浏览新闻"),
        "link_longform": ("Long-form & Podcasts", "长文与播客"),
        "link_longform_desc": ("Deep reads and episodes in one place", "深度阅读与播客聚合"),
        "link_viral": ("Viral Videos", "热门视频"),
        "link_viral_desc": ("YouTube & Bilibili, ranked by spread rate", "YouTube 与 Bilibili，按传播率排序"),
        "link_video2text": ("Video to Text", "视频转文字"),
        "link_video2text_desc": ("Paste URL → extract transcript", "粘贴链接 → 提取逐字稿"),
        "link_article_lab": ("Article Lab", "文章改写"),
        "link_article_lab_desc": ("Upload markdown and rewrite with STEPPS + Minto", "上传 markdown 并用 STEPPS + Minto 改写"),
        "link_publish": ("Publish Queue", "发布队列"),
        "link_publish_desc": ("Approve drafts and publish to X/blog", "审核草稿并发布到 X/博客"),
        "link_science": ("STEPPS & Structure", "评分与结构"),
        "link_science_desc": ("Scoring + Minto Pyramid", "评分与金字塔结构"),
        # Your Field page
        "field_title": ("Your field · Curated resources", "你的领域 · 精选资源"),
        "field_desc": ("Pick your niche — must-see trends and resources for Xiaohongshu, Douyin, Bilibili creators. Fashion-first, designer-built.", "选择你的领域 — 小红书、抖音、B站创作者必看的趋势与资源。时尚优先，设计师视角"),
        "field_trends": ("Trends", "趋势"),
        "field_color_forecasting": ("Color forecasting", "色彩预测"),
        "field_forecasting_sites": ("Forecasting sites", "预测网站"),
        "field_resources": ("Resources", "资源"),
        "field_select_above": ("Select a field above.", "请在上方选择领域"),
        "field_empty": ("No resources for this field.", "此领域暂无资源"),
        "field_region": ("Region", "地区"),
        "field_region_global": ("Global", "全球"),
        "field_region_americas": ("Americas", "美洲"),
        "field_region_europe": ("Europe", "欧洲"),
        "field_region_asia": ("Asia", "亚洲"),
        "field_share": ("Share", "分享"),
        "field_share_copied": ("Link copied!", "链接已复制！"),
        "field_share_to": ("Share to", "分享至"),
        "field_share_whatsapp": ("WhatsApp", "WhatsApp"),
        "field_share_telegram": ("Telegram", "Telegram"),
        "field_share_x": ("X", "X"),
        "field_share_linkedin": ("LinkedIn", "LinkedIn"),
        "field_share_copy": ("Copy link", "复制链接"),
        "field_share_text": ("Curated resources for content creators — ViralLab", "为创作者精选的趋势与资源 — ViralLab"),
        # News by topic — different content per language zone
        "news_title": ("News by topic", "按主题新闻"),
        "news_desc": (
            "Each section matches a live tag from above (hot list + segment). Refresh all only re-fetches those tags — not every old file on disk.",
            "每个板块对应上方热词（实时热榜 + 当前领域）。「刷新全部」只更新这些标签，不再遍历旧的 raw 主题文件。",
        ),
        "longform_title": ("Long-form & Podcasts", "长文与播客"),
        "longform_desc": ("Page sources:", "页面文章的来源："),
        "longform_scope": (
            "English mode: global English sources. Chinese mode: Chinese ecosystem, including Traditional Chinese, Singapore, Taiwan, and overseas Chinese communities.",
            "英文区：全球英文来源。中文区：中文生态来源（含繁体中文区、新加坡、台湾与海外华人社区）。",
        ),
        "longform_no_items": ("No long-form items yet. Refresh Daily News first.", "暂无长文内容，请先刷新每日新闻。"),
        "longform_view_daily": ("Open Daily News", "打开每日新闻"),
        "longform_theme_label": ("Focus themes", "主题聚焦"),
        "longform_theme_all": ("All", "全部"),
        "longform_theme_transition": ("Transition pain points", "转型痛点"),
        "longform_theme_one_person": ("One-person business", "一人公司"),
        "longform_theme_design_ai": ("Design & creativity in AI", "设计与创意 · AI 时代"),
        "news_placeholder": ("Search any topic (e.g. AI agents, creator economy)", "搜索主题（如：小红书爆款、抖音运营）"),
        "news_range_60m": ("60 mins", "60 分钟"),
        "news_range_1d": ("1 day", "1 天"),
        "news_range_7d": ("7 days", "7 天"),
        "news_tip_requested": ("Most requested in last {range}", "过去 {range} 最常搜索"),
        "audience_segment_label": ("Audience", "受众"),
        "news_tip_platform_hot": ("What users care about now — Weibo, Zhihu, Douyin, Baidu, 少数派", "用户正在关注 · 微博、知乎、抖音、百度、少数派 实时热搜"),
        "news_tip_platform_hot_en": ("What users care about now — HN, Reddit, Lobsters, Dev.to, GitHub", "用户正在关注 · HN、Reddit、Lobsters、Dev.to、GitHub 实时热门"),
        "news_tip_suggestions": ("Suggestions — try these", "建议 — 试试这些"),
        "hot_tags_title": ("Hot tags", "热门标签"),
        "hot_tags_desc": ("Reference tags from current ecosystem signals.", "来自当前生态信号的参考标签。"),
        "news_topic_label": ("Topic", "主题"),
        "news_no_results": ("Search a topic above to get curated news.", "搜索上方主题即可获取精选新闻"),
        "news_searching": ("Searching… Refresh in a few seconds.", "正在搜索… 几秒后刷新页面"),
        "news_english_topic_in_zh": ("This topic has English content. Please select a Chinese topic above.", "此主题为英文内容，请选择上方中文主题"),
        "news_chinese_topic_in_en": ("This topic has Chinese content. Switch to 中文 to view.", "此主题为中文内容，请切换至英文查看"),
        "news_show_all": ("Show all topics", "显示全部主题"),
        "news_search_failed": ("Search failed (proxy/network). Re-run when VPN is on.", "搜索失败（代理/网络）。请开启 VPN 后重试"),
        "news_parse_error": ("Could not render this topic file. Try refreshing or re-searching.", "该主题文件无法渲染，请刷新或重新搜索。"),
        "news_topic_no_hits": (
            "Last search returned no articles for this topic (network, VPN, or source limits). Re-search below.",
            "该主题上次搜索未返回文章（网络、VPN 或来源限制）。请在下方重新搜索。",
        ),
        "news_topic_unparsed": (
            "This topic file has no article list yet. Re-search below.",
            "该主题文件尚无文章列表，请在下方重新搜索。",
        ),
        "news_rescan_topic": ("Re-search this topic", "重新搜索此主题"),
        "news_refresh_all": ("Refresh all topics", "刷新全部主题"),
        "news_refresh_hot": ("Refresh hot list", "刷新热搜"),
        "news_refresh_interval": ("every 60 mins", "每 60 分钟"),
        "news_hot_updated": ("Updated {mins} min ago", "约 {mins} 分钟前更新"),
        "news_hot_just_now": ("Just updated", "刚刚更新"),
        "loading_searching": ("Searching topics…", "正在搜索主题…"),
        "loading_search_sub": ("This may take up to a minute. Please wait.", "可能需要一分钟，请稍候"),
        "loading_refreshing": ("Refreshing…", "正在刷新…"),
        "loading_refresh_sub": ("Please wait.", "请稍候"),
        # Viral Videos — different content per language zone
        "viral_title": ("Viral Videos", "热门视频"),
        "viral_subtitle": ("Ranked by spread rate", "按传播率排序"),
        "viral_desc": ("Search by topic and view videos ranked by spread rate.", "按主题搜索并查看按传播率排序的视频。"),
        "viral_placeholder": ("e.g. AI explainer, fashion haul", "如：AI 解说、穿搭测评、直播带货"),
        "viral_source_label": ("Source", "来源"),
        "viral_source_global": ("Global (YouTube)", "全球 (YouTube)"),
        "viral_source_china": ("China (Bilibili)", "中国 (Bilibili)"),
        "viral_hot_tags_hint": ("Use hot tags as quick query starters.", "可用热门标签作为快速搜索起点。"),
        "viral_search_btn": ("Search viral", "搜索热门"),
        "viral_what_we_serve": ("Global: YouTube videos. China: Bilibili videos. Every link opens the video on its platform.", "英文区：YouTube 视频。中文区：B站视频。点击即跳转至该平台观看。"),
        "viral_china_note": ("In China? YouTube needs VPN. Use China source for Bilibili (no VPN).", "在中国, YouTube 需魔法, 请选中国来源搜 B站"),
        "viral_china_fetch_title": ("Fetch content from platforms", "从平台拉取内容"),
        "viral_china_fetch_desc": ("Run fetch for selected platforms (Bilibili API + optional crawlers for XHS/Douyin/Zhihu/Shipinhao). Results merge into the list above", "对所选平台运行拉取（B站公开API + 小红书/抖音/知乎/视频号可选爬取），结果会合并到上方列表"),
        "china_crawl_duration": ("Crawl took {sec}s", "拉取耗时 {sec} 秒"),
        "viral_china_fetch_btn": ("Fetch", "拉取"),
        "viral_china_fetch_keywords": ("Keywords", "关键词"),
        "viral_china_platforms": ("Platforms", "平台"),
        "content_type_video": ("Video", "视频"),
        "content_type_post": ("Post", "帖子"),
        "content_type_article": ("Article", "文章"),
        "content_type_note": ("Note", "笔记"),
        "china_crawl_done": ("Crawl finished. Refresh the list to see results.", "拉取完成。请刷新上方列表查看。"),
        "china_crawl_no_items": ("No items fetched.", "未拉取到内容。"),
        "china_crawl_errors": ("({n} platform(s) failed — see below)", "（{n} 个平台失败 — 见下方）"),
        "china_content_page_title": ("China content", "中国内容"),
        "china_content_after_fetch": ("Fetched content appears below. You can also see it on the Viral page (source: China).", "拉取结果展示于本页下方；也可在「热门视频」选中国来源查看。"),
        "china_content_audience_label": ("Audience — add their keywords as tags below", "人群 — 点击后将该人群关心的关键词加入下方 tag"),
        "china_content_keyword_input_placeholder": ("Type a hot keyword (e.g. open claw), Enter or Add as tag", "输入你发现的热门关键词（如 open claw），回车或点击添加成 tag"),
        "china_content_add_tag": ("Add", "添加"),
        "china_content_tags_label": ("Keywords to crawl (click × to remove)", "将要爬取的关键词（点击 × 移除）"),
        "china_content_no_tags_yet": ("No keywords yet — pick an audience above or type your own", "暂无关键词 — 点击上方人群或输入自定义"),
        "china_content_try_tags": ("Select one or more keywords, then Fetch to crawl all together (max 10).", "勾选一个或多个关键词，点击拉取可一并爬取（最多10个）。"),
        "china_content_fetched_title": ("Fetched content", "拉取结果"),
        "china_content_no_items_yet": ("No content yet. Run Fetch above (requires Playwright in the same Python that runs the server).", "暂无内容。请在上方点击拉取（需在运行服务器的同一 Python 环境中安装 Playwright）。"),
        "china_content_eval_btn": ("Score & rewrite", "评估 · 改写"),
        "china_content_eval_score": ("Score", "评分"),
        "china_content_rewrite_directions": ("Rewrite directions", "改写方向"),
        "china_content_minto_title": ("Structure (Minto)", "结构（金字塔）"),
        "china_content_eval_loading": ("Evaluating…", "评估中…"),
        "china_content_eval_error": ("Evaluation failed", "评估失败"),
        "china_content_crawl_hint": ("Bilibili uses public API (no login). Xiaohongshu/Douyin often return no data without login.", "B站使用公开API（无需登录）；小红书/抖音未登录时多为空。"),
        "china_content_login_title": ("Login (optional)", "登录态（可选）"),
        "china_content_login_desc": ("Without login, XHS/Douyin often return no data. To use your session: run the command below on your machine; a browser will open — log in there once. Session is saved locally only; we never upload it.", "未登录时小红书/抖音等可能无法拉取。若需登录后拉取：请在本机终端运行下方命令，会打开浏览器，在页面中登录一次即可。会话仅保存在本机 config/，我们不会上传你的登录信息。"),
        "china_content_login_cmd": (".venv/bin/python -m tools.china_crawler login --platform xhs", ".venv/bin/python -m tools.china_crawler login --platform xhs"),
        "china_content_login_cmd_note": ("Replace xhs with douyin, zhihu, or shipinhao for other platforms. Bilibili does not need login.", "将 xhs 改为 douyin、zhihu 或 shipinhao 可保存其他平台登录态。B站无需登录。"),
        "china_content_login_risk": ("Using your account to crawl may be detected by the platform (rate limit or ban). We recommend a separate account; we do not post or delete your content. See docs/CHINA_CRAWLER_LOGIN.md.", "使用账号登录爬取可能被平台检测、限流或封号，建议使用小号。本工具不会代你发帖/删帖，不会删除你账号内容；详见 docs/CHINA_CRAWLER_LOGIN.md。"),
        "china_content_free_compliant_title": ("Free & compliant", "免费且合规"),
        "china_content_hot_topics_label": ("Today's hot topics (click to add as keyword)", "今日最热话题（点击加入为关键词）"),
        "china_content_free_compliant_desc": ("No cost, no scraping, no login: use Bilibili (public API) here, and open the links below to search on each platform in your browser. Fully compliant.", "不付费、不爬取、不登录：B站用本站公开 API；其余平台点击链接在浏览器中搜索，合规使用。"),
        "china_content_free_bilibili": ("Bilibili (in-app)", "B站（站内）"),
        "china_content_free_search_links": ("Search in browser", "在浏览器中搜索"),
        "china_content_load_error": ("China content page failed to load.", "中国内容页暂时无法加载。"),
        "viral_china_content_link": ("Fetch from Bilibili, 小红书, 抖音, 知乎, 视频号", "从B站、小红书、抖音、知乎、视频号拉取内容"),
        "viral_suggestions": ("Suggestions — try these", "建议 — 试试这些"),
        "viral_lang_hint": ("This topic is in English. Try China source for Chinese content.", "此主题为英文，建议选择中国来源搜索中文内容"),
        "viral_berger_unproven": ("—", "—"),
        "viral_berger_methodology": ("Score: title+description only (Berger keywords). Not AI. 0 views = unproven.", "评分：仅标题与描述（Berger 关键字）。非 AI。0 播放 = 未验证"),
        "viral_loading": ("Finding what's spreading…", "正在加载…"),
        "viral_empty_title": ("Nothing here yet.", "暂无结果"),
        "viral_empty_body": ("Try one of these searches to see viral content scored in real time:", "试试这些搜索，实时查看爆款评分："),
        "viral_error": ("Error", "加载失败"),
        "viral_china_unavailable": ("Bilibili may be unreachable from your network. Try switching to Global (YouTube) source above.", "B站 API 可能无法访问（需在中国网络）。请切换至上方「全球 (YouTube)」来源重试。"),
        "viral_china_proxy_placeholder": (
            "e.g. http://127.0.0.1:18080 (gost / same as Cursor)",
            "本机 HTTP 前端，如 http://127.0.0.1:18080（gost，与 Cursor 一致）",
        ),
        "viral_china_retry_with_proxy": ("Retry", "重试"),
        "viral_timeout": ("Request timeout", "请求超时"),
        "content_angles": ("3 ways to ride this trend", "3 种内容角度"),
        # Video to Text — different content per language zone
        "video2text_title": ("Video to Text", "影片转文字"),
        "video2text_tagline": ("Paste a video. Get the transcript.", "粘贴视频。获取逐字稿"),
        "video2text_accuracy_note": (
            "**Douyin:** save the clip to disk and paste **local video/audio path** here (recommended — URL fetch breaks often without heavy cookie setup). YouTube prefers existing captions first. Other URLs use yt-dlp / VideoCaptioner; optional WhisperX for speech-only.",
            "**抖音（推荐）：** 先把视频**保存到本机**，在下面粘贴 **`/路径/文件名.mp4`**（或拖拽得到的路径）；不要在抖音依赖网页链接。**YouTube：** 优先现成字幕。**其它：** yt-dlp / VideoCaptioner，可选 WhisperX 语音转文字。",
        ),
        "video2text_placeholder": (
            "Paste local path e.g. /Users/you/Movies/clip.mp4 or a YouTube / Bilibili / XHS URL",
            "可粘贴本地文件路径（如 `/Users/you/Movies/抖音.mp4`）或 YouTube/B站/小红书等链接（抖音不推荐用链接）",
        ),
        "video2text_btn": ("Get transcript →", "获取逐字稿 →"),
        "douyin_session_test_btn": ("Douyin URL test (optional)", "可选：抖音链接自检"),
        "douyin_session_testing": ("Testing Douyin URL...", "正在检测抖音链接…"),
        "video2text_hint": (
            "Local file path recommended for Douyin. Also: YouTube, Bilibili, Xiaohongshu, Shipinhao, Xiaoyuzhou, podcast links · e.g. ",
            "抖音建议用**本地视频路径**；也支持 YouTube、B站、小红书、视频号、小宇宙、播客链接 · 例如 ",
        ),
        "video2text_note_minto": ("After you get a transcript, use the <strong>Structure (Minto)</strong> tab above the transcript to see conclusion → key points → evidence.", "获取逐字稿后，用上方的<strong>结构（金字塔）</strong>标签查看结论 → 要点 → 论据。"),
        "video2text_tab_structure": ("Structure (Minto)", "结构（金字塔）"),
        "video2text_export_mode": ("Export", "导出"),
        "video2text_export_raw": ("Original transcript", "原文版"),
        "video2text_export_minto": ("Minto version", "金字塔版"),
        "video2text_goto_article_lab": ("Rewrite this transcript →", "用这份逐字稿去改写 →"),
        "article_lab_prefilled_from": ("Prefilled from saved file", "已从保存的逐字稿载入"),
        "article_lab_prefill_missing": (
            "Could not load that file from output — paste markdown or upload below.",
            "未能从 output 载入该文件，请粘贴正文或上传。",
        ),
        "video2text_processing": ("Processing your transcript…", "正在处理逐字稿…"),
        "video2text_processing_sub": ("This can take a few seconds for captions and longer for ASR. Please keep this page open.", "字幕模式通常几秒，ASR 模式会更久。请保持页面开启。"),
        "video2text_err_douyin": (
            "Douyin URL fetch failed. Prefer a **local downloaded file**: paste `/path/to/video.mp4` here and retry. Cookie-based URL fetching is brittle.",
            "抖音**链接**拉取失败。**推荐**改贴**本地下载的视频路径**（如 `/Movies/片段.mp4`）再获取逐字稿；依赖 Cookie 在线拉取很不稳定。",
        ),
        "video2text_err_xhs": (
            "Xiaohongshu: the note must include video (not image-only). If it still fails, open the note in Chrome while logged in and retry, or set YTDLP_COOKIES_BROWSER / YTDLP_COOKIES_PROFILE. Align terminal HTTP_PROXY with Cursor: http://127.0.0.1:18080 (gost); avoid stale QuickQ ports. Short links retry without a broken proxy.",
            "小红书：需为视频笔记（纯图文无法转写）。若仍失败，请在已登录的 Chrome 中打开该笔记后重试，或配置 YTDLP_COOKIES_BROWSER / YTDLP_COOKIES_PROFILE。终端 HTTP_PROXY 建议与 Cursor 一致使用 http://127.0.0.1:18080（gost），勿沿用已失效的 QuickQ 口；短链在代理不可连时会尝试直连。",
        ),
        "video2text_err_generic": (
            "Transcript unavailable for this link right now.",
            "暂时无法获取该链接的逐字稿。",
        ),
        "video2text_your_transcripts": ("Your transcripts", "你的逐字稿"),
        "video2text_none": ("None yet", "尚无"),
        "video2text_download_btn": ("Download video", "下载视频"),
        "video2text_download_progress": ("Downloading video…", "正在下载视频…"),
        "video2text_download_hint": (
            "YouTube only: uses yt-dlp (<code>pip install yt-dlp</code>). Respect copyright and YouTube's terms.",
            "仅 YouTube：使用 yt-dlp（<code>pip install yt-dlp</code>）。请遵守版权与 YouTube 条款。",
        ),
        "youtube_dl_missing_url": ("Paste a URL first.", "请先粘贴链接。"),
        "youtube_dl_need_ytdlp": (
            "yt-dlp is not installed. Run: pip install yt-dlp",
            "未检测到 yt-dlp。请运行：pip install yt-dlp",
        ),
        "youtube_dl_not_youtube": (
            "Only YouTube watch or youtu.be links are supported for download.",
            "下载仅支持 YouTube 或 youtu.be 链接。",
        ),
        "youtube_dl_failed": ("Could not download this video.", "无法下载该视频。"),
        "article_lab_title": ("Article Lab", "文章改写"),
        "article_lab_desc": ("Upload or paste markdown. We output a structured contagious version in markdown.", "上传或粘贴 markdown，输出结构化传播版本（markdown）。"),
        "article_lab_intro": (
            "Rewrite your markdown into a structured, contagious version with Minto + STEPPS.",
            "将 markdown 改写为更有结构、更可传播的版本（Minto + STEPPS）。",
        ),
        "article_lab_input_label": ("Input markdown", "输入 markdown"),
        "article_lab_upload_multi_label": ("Upload multiple .md files (batch)", "批量上传多个 .md 文件"),
        "article_lab_btn": ("Rewrite article", "改写文章"),
        "article_lab_btn_batch": ("Batch rewrite", "批量改写"),
        "article_lab_download": ("Download rewritten markdown", "下载改写后的 markdown"),
        "article_lab_queue_publish": ("Queue to publish", "加入发布队列"),
        "publish_title": ("Publish Queue", "发布队列"),
        "publish_desc": ("Queue rewritten content, approve, then publish to X or blog.", "将改写内容加入队列，审核后发布到 X 或博客。"),
        "publish_handoff_title": ("WeChat MP & Xiaohongshu (Cursor)", "微信公众号与小红书（Cursor）"),
        "publish_handoff_body": (
            "Finish markdown here, then in Cursor ask the Agent to run global skills **wechat-mp-auto** or **xiaohongshu-auto** (browser → mp.weixin.qq.com / creator.xiaohongshu.com). Copy from below or open a saved file under /view/…. Skills live in ~/cursor-skills-source after link-to-cursor.sh.",
            "在此定稿 markdown 后，到 Cursor 让 Agent 执行全局技能 **wechat-mp-auto** 或 **xiaohongshu-auto**（浏览器打开 mp.weixin.qq.com / creator.xiaohongshu.com）。从下方复制正文，或通过 /view/文件名 打开 output 中的 .md。技能在 ~/cursor-skills-source，需已执行 link-to-cursor.sh。",
        ),
        "publish_form_title": ("Create publish draft", "创建发布草稿"),
        "publish_platform": ("Platform", "平台"),
        "publish_to_x": ("X", "X"),
        "publish_to_blog": ("Blog", "博客"),
        "publish_schedule": ("Schedule (optional)", "定时（可选）"),
        "publish_add_btn": ("Add to queue", "加入队列"),
        "publish_now_btn": ("Publish now", "立即发布"),
        "publish_empty": ("No publish items yet.", "暂无发布项。"),
        "publish_status_queued": ("Queued", "待发布"),
        "publish_status_published": ("Published", "已发布"),
        "publish_status_failed": ("Failed", "失败"),
        # Campaign page
        "campaign_eyebrow": ("ViralLab · Free · No API key needed", "ViralLab · 免费 · 无需 API 密钥"),
        "campaign_h1_line1": ("Stop guessing", "停止猜测"),
        "campaign_h1_line2": ("why content ", "为什么内容会"),
        "campaign_h1_em": ("spreads.", "爆红"),
        "campaign_h1_cta": ("Now you know.", "现在你知道了"),
        "campaign_hero_sub1": ("You're one score away", "一个分数就能"),
        "campaign_hero_sub2": (" from understanding any viral video, trend, or idea. ViralLab gives you the same framework Fortune 500 brands use — powered by Jonah Berger's peer-reviewed science — ", "让你读懂任何爆款视频、趋势或创意。ViralLab 给你 Fortune 500 品牌用的同一套框架 — 基于 Jonah Berger 的同行评审科学 — "),
        "campaign_hero_sub3": ("free, in seconds.", "免费，几秒搞定"),
        "campaign_cta_score": ("Score a video now", "立即评分视频"),
        "campaign_cta_trends": ("See today's trends", "看今日热门"),
        "campaign_proof_books": ("Based on 4 NYT bestsellers", "基于 4 本纽约时报畅销书"),
        "campaign_proof_studies": ("80+ peer-reviewed studies", "80+ 篇同行评审研究"),
        "campaign_proof_brands": ("Used by Apple, Google, Nike", "Apple、Google、Nike 都在用"),
        "campaign_proof_free": ("Free — no keys, no signup", "免费 — 无密钥、免注册"),
        "campaign_tag_story": ("The story behind this tool", "这个工具背后的故事"),
        "campaign_story_before": ("Before", "从前"),
        "campaign_story_before_h": ("You publish.<br>You hope.", "你发布。<br>你祈祷"),
        "campaign_story_before_b": ("You watch other creators blow up on the same topics you covered. You guess. You copy. You wonder why yours never lands the same way.", "你看着别人用同样的主题爆红。你猜。你模仿。你不懂为什么你的永远差一点。"),
        "campaign_story_disc": ("Discovery", "发现"),
        "campaign_story_disc_h": ("Virality isn't<br>luck. It's science.", "爆红不是<br>运气。是科学"),
        "campaign_story_disc_b": ("Professor Jonah Berger spent 30 years studying why people share things. His STEPPS framework reveals the exact psychological triggers behind every piece of content that spreads.", "Jonah Berger 教授花了 30 年研究人们为什么分享。他的 STEPPS 框架揭示了每则爆款内容背后的心理触发点。"),
        "campaign_story_now": ("Now", "现在"),
        "campaign_story_now_h": ("You create with<br>an unfair edge.", "你创作时<br>自带不公平优势"),
        "campaign_story_now_b": ("ViralLab scores any video or trend against all six STEPPS signals — and tells you exactly which lever to pull to make your content spread further.", "ViralLab 用六个 STEPPS 信号评分任何视频或趋势 — 并告诉你该拉哪根杆让内容传得更远。"),
        "campaign_tag_who": ("Who this is for", "适合谁用"),
        "campaign_who_h": ("You're already ahead.<br>This makes it obvious.", "你已经领先了。<br>这让它一目了然"),
        "campaign_who_desc": ("The creators who understand <em>why</em> content spreads — not just what's trending — are the ones building audiences that last. You're about to become one of them.", "懂<em>为什么</em>内容会爆 — 不只是什么在红 — 的创作者，才能建立持久的受众。你即将成为其中一员。"),
        "campaign_audience_yt": ("YouTubers & Video Creators", "YouTuber 与视频创作者"),
        "campaign_audience_yt_d": ("Score trending videos before you script yours. Know which Berger signals made them pop — then engineer those signals into your own content.", "在写脚本前先评分热门视频。知道哪些 Berger 信号让它们爆红 — 再把这些信号设计进你的内容。"),
        "campaign_audience_news": ("Newsletter & Blog Writers", "电子报与博客作者"),
        "campaign_audience_news_d": ("Discover what's spreading in your niche today. The Daily Digest surfaces the top 3 stories — with content angles you can write up in hours.", "发现你领域今天在传什么。每日精选呈现前 3 则话题 — 附带几小时内就能写出的内容角度。"),
        "campaign_audience_social": ("Social Media Managers", "社交媒体经理"),
        "campaign_audience_social_d": ("Stop presenting gut-feel content calendars. Walk into every brief with a Berger score and a clear explanation of <em>why</em> this trend is worth riding.", "别再拿感觉派内容日历简报。带着 Berger 评分和<em>为什么</em>值得跟风的清楚说明，走进每个 briefing。"),
        "campaign_audience_china": ("China-Market Creators", "中国市场创作者"),
        "campaign_audience_china_d": ("We pull from Bilibili, Douyin, and Xiaohongshu natively. Discover what's spreading in Chinese content — and score it with the same framework.", "我们原生支持 B站、抖音、小红书。发现中文内容在传什么 — 用同一套框架评分。"),
        "campaign_tag_how": ("How it solves your real problems", "如何解决你的真实痛点"),
        "campaign_how_h": ("Every time you sit down to create,<br>ViralLab has the answer.", "每次坐下来创作时，<br>ViralLab 都有答案"),
        "campaign_trigger": ("Every time you're stuck on what to post — open ViralLab first.", "每次卡在「不知道发什么」时 — 先打开 ViralLab"),
        "campaign_mech1_pain": ("I don't know what's trending right now", "我不知道现在什么在红"),
        "campaign_mech1_sol": ("Daily Digest — top 3 topics, refreshed daily", "每日精选 — 前 3 话题，每日更新"),
        "campaign_mech1_detail": ("Curated for your field. Fresh every morning.", "依你的领域精选。每天早晨新鲜出炉"),
        "campaign_mech1_cta": ("Open Daily →", "每日精选 →"),
        "campaign_mech2_pain": ("I don't know why that video went viral", "我不知道那支视频为什么爆红"),
        "campaign_mech2_sol": ("Berger Score — 6 signals, one number", "Berger 评分 — 6 信号，一个数字"),
        "campaign_mech2_detail": ("See exactly which STEPPS principles fired — and which didn't.", "清楚看到哪些 STEPPS 原则触发了 — 哪些没有"),
        "campaign_mech2_cta": ("Score a video →", "评分视频 →"),
        "campaign_mech3_pain": ("I want to repurpose this video into content", "我想把这支视频转成内容"),
        "campaign_mech3_sol": ("Video-to-Text — transcript + full Berger analysis", "视频转文字 — 逐字稿 + 完整 Berger 分析"),
        "campaign_mech3_detail": ("Paste a YouTube URL. Get transcript and score in seconds.", "贴上 YouTube 链接。几秒获取逐字稿与评分"),
        "campaign_mech3_cta": ("Try it →", "试试 →"),
        "campaign_mech4_pain": ("Generic trends don't apply to my niche", "通用趋势不适用我的领域"),
        "campaign_mech4_sol": ("Your Field — curated resources by creator niche", "你的领域 — 依创作者领域精选资源"),
        "campaign_mech4_detail": ("Fashion, tech, food, beauty, education, and more.", "时尚、科技、美食、美妆、教育等"),
        "campaign_mech4_cta": ("Pick your field →", "选择领域 →"),
        "campaign_mech5_pain": ("I need more source variety, not just YouTube", "我需要更多来源，不只 YouTube"),
        "campaign_mech5_sol": ("News Search — DuckDuckGo + Bilibili, no API key", "新闻搜索 — DuckDuckGo + B站，无需 API 密钥"),
        "campaign_mech5_detail": ("Search any topic. Global and China sources covered.", "搜索任一主题。全球与中国来源皆涵盖"),
        "campaign_mech5_cta": ("Search now →", "立即搜索 →"),
        "campaign_tag_score": ("What a Berger score looks like", "Berger 评分长这样"),
        "campaign_score_h": ("Not a black box. A breakdown.", "不是黑盒。是拆解"),
        "campaign_score_label": ("/100 · Strong Signal", "/100 · 强信号"),
        "campaign_score_title": ("Strong Emotion + Social Currency. Weak on Triggers.", "强情绪 + 强社交货币。触发弱"),
        "campaign_score_body": ("This video scored high because it opens with a surprising claim (Social Currency) and builds genuine excitement (Emotion). It loses points on Triggers — there's no recurring context that would make viewers think of it again naturally.", "这支视频高分因为开场有惊人主张（社交货币）并营造真实兴奋感（情绪）。触发失分 — 没有让观众自然想起的 recurring 情境。"),
        "campaign_score_fix": ("Fix →", "修正 →"),
        "campaign_score_fix_detail": ("Add one line anchoring this to a recurring moment: ", "加一句把这个锚定到 recurring 情境："),
        "campaign_score_fix_example": ("\"Every time you open your feed and see this type of content…\"", "「每次打开动态看到这类内容时…」"),
        "campaign_cred_tag": ("The science behind the score", "评分背后的科学"),
        "campaign_cred_quote": ("\"The key question isn't 'how do I make something go viral?' — it's 'why do people talk and share things in the first place?'\"", "「关键问题不是『怎么让东西爆红？』— 而是『人们为什么会谈论和分享？』」"),
        "campaign_cred_byline": ("— Jonah Berger, Wharton School Professor &amp; author of <em>Contagious</em>", "— Jonah Berger，沃顿商学院教授、《疯潮行销》作者"),
        "campaign_cred_books": ("NYT Bestselling books", "纽约时报畅销书"),
        "campaign_cred_studies": ("Peer-reviewed studies", "同行评审研究"),
        "campaign_cred_tenure": ("Wharton research tenure", "沃顿研究资历"),
        "campaign_cred_f500": ("Apple · Google · Nike · Gates", "Apple · Google · Nike · Gates"),
        "campaign_tag_steps": ("Get started in 3 easy steps", "3 步骤轻松开始"),
        "campaign_steps_h": ("Your first score in under 2 minutes.", "2 分钟内搞定第一个评分"),
        "campaign_steps_desc": ("No account. No API key. Just paste and discover.", "免账号。免 API 密钥。贴上就能发现"),
        "campaign_step1_title": ("Pick your field", "选择你的领域"),
        "campaign_step1_body": ("Tell ViralLab your niche — tech, fashion, food, education, or more. Every result, trend, and score gets calibrated to your audience.", "告诉 ViralLab 你的领域 — 科技、时尚、美食、教育等。每个结果、趋势、评分都会依你的受众调整。"),
        "campaign_step1_time": ("⏱ 30 seconds", "⏱ 30 秒"),
        "campaign_step2_title": ("Find what's spreading", "发现正在传播的"),
        "campaign_step2_body": ("Open Daily Digest for today's top 3 topics, or search any trend in News or Viral. See what's gaining traction right now.", "打开每日精选看今日前 3 话题，或在新闻或热门搜索任一趋势。看看现在什么在涨。"),
        "campaign_step2_time": ("⏱ 60 seconds", "⏱ 60 秒"),
        "campaign_step3_title": ("Score it & act", "评分并行动"),
        "campaign_step3_body": ("Paste a YouTube URL into Video-to-Text. Get a full Berger score, the STEPPS breakdown, and one concrete fix to make your version spread further.", "把 YouTube 链接贴到视频转文字。获取完整 Berger 评分、STEPPS 拆解，以及一个具体修正让你的版本传得更远。"),
        "campaign_step3_time": ("⏱ 10 seconds", "⏱ 10 秒"),
        "campaign_cta_h": ("Your next viral piece of content<br>starts with a <em>score.</em>", "你的下一则爆款内容<br>从一个<em>评分</em>开始"),
        "campaign_cta_sub": ("Free. No signup. No API keys. Just paste a YouTube URL and discover exactly why content spreads — and how to engineer it into yours.", "免费。免注册。免 API 密钥。贴上 YouTube 链接就能发现内容为什么会爆 — 以及如何把它设计进你的"),
        "campaign_cta_btn": ("Score your first video — free →", "免费评分你的第一支视频 →"),
        "campaign_cta_nudge": ("or <a href='/daily' style='color:rgba(255,255,255,0.5); text-decoration:underline;'>browse today's trending topics</a>", "或 <a href='/daily' style='color:rgba(255,255,255,0.5); text-decoration:underline;'>浏览今日热门话题</a>"),
        "campaign_tag_social": ("Join the conversation", "加入讨论"),
        "campaign_social_h": ("Creators using ViralLab<br>don't create alone.", "用 ViralLab 的创作者<br>不孤单创作"),
        "campaign_social_desc": ("Follow for weekly breakdowns of viral content, new feature drops, and a community of creators who think before they post.", "追踪每周爆款拆解、新功能预告，以及会先想再发的创作者社群"),
        "campaign_social_tg_name": ("ViralLab Community", "ViralLab 社群"),
        "campaign_social_tg_promise": ("Weekly STEPPS breakdowns. Early feature drops. Creator tips. This is where the conversation lives.", "每周 STEPPS 拆解、新功能抢先、创作者技巧。讨论都在这里"),
        "campaign_social_tg_join": ("Join group →", "加入群组 →"),
        "campaign_social_x_name": ("X / Twitter", "X / Twitter"),
        "campaign_social_x_promise": ("Daily observations on what's spreading and why. The shorter version of the science.", "每日观察什么在传、为什么。科学的简短版"),
        "campaign_social_x_join": ("Follow →", "追踪 →"),
        "campaign_social_blog_name": ("The Blog", "博客"),
        "campaign_social_blog_promise": ("Long-form breakdowns of viral campaigns, Berger deep-dives, and creator case studies.", "爆款案例长文拆解、Berger 深度解析、创作者个案研究"),
        "campaign_social_blog_join": ("Read →", "阅读 →"),
        "campaign_social_dm_name": ("Direct Channel", "直达频道"),
        "campaign_social_dm_promise": ("Personal updates, behind-the-scenes on ViralLab's development, and direct access.", "个人动态、ViralLab 开发幕后、直达联系"),
        "campaign_social_dm_join": ("Follow →", "追踪 →"),
        "campaign_title": ("Engineer Your Influence", "打造你的影响力"),
        # Campaign v2 (simplified landing)
        "camp2_greeting": ("Hey, welcome to ViralLab", "嗨，欢迎来到 ViralLab"),
        "camp2_h1": ("A tool for understanding why content spreads.", "帮你理解内容为什么会传播的工具"),
        "camp2_p": ("You paste a video or search a topic — we tell you <strong>which psychological triggers</strong> are making it travel. Not a black box. A real breakdown you can actually use.", "你粘贴视频或搜索话题 — 我们告诉你<strong>哪些心理触发点</strong>让它传播。不是黑盒。是你可以真正使用的拆解。"),
        "camp2_btn_primary": ("Try it with a video →", "用视频试试 →"),
        "camp2_btn_ghost": ("See today's trends", "看今日趋势"),
        "camp2_whats_in": ("What's in here", "这里有什么"),
        "camp2_tool_daily": ("Daily News", "每日精选"),
        "camp2_tool_daily_desc": ("Top 3 topics spreading today, curated for your field.", "今日传播中的前 3 话题，按你的领域精选"),
        "camp2_tool_viral": ("Viral Videos", "热播视频"),
        "camp2_tool_viral_desc": ("Search YouTube & Bilibili. See scores for each video.", "搜索 YouTube 和 B站。查看每支视频的评分"),
        "camp2_tool_v2t": ("Video to Text", "视频转文字"),
        "camp2_tool_v2t_desc": ("Paste a URL, get the transcript + a full score breakdown.", "粘贴链接，获取逐字稿 + 完整评分拆解"),
        "camp2_tool_field": ("Your Field", "你的领域"),
        "camp2_tool_field_desc": ("Resources and trends filtered to your niche.", "按你的领域筛选的资源和趋势"),
        "camp2_tool_news": ("News by Topic", "按话题搜索新闻"),
        "camp2_tool_news_desc": ("Search any topic. Global and China sources.", "搜索任意话题。全球和中国来源"),
        "camp2_how": ("How the score works", "评分如何运作"),
        "camp2_step1": ("You give it a video or some text.", "你提供视频或文本"),
        "camp2_step2": ("It checks for six things that research shows make people share: <strong>social currency, triggers, emotion, visibility, usefulness, and story.</strong>", "它会检查六个研究证明会让人们分享的因素：<strong>社交货币、触发点、情绪、可见性、有用性、故事。</strong>"),
        "camp2_step3": ("You get a score, a breakdown of what's strong, what's weak, and a one-line suggestion for the easiest fix.", "你会得到评分、强弱项的拆解，以及一条最易改进的建议"),
        "camp2_science": ("The scoring is based on <strong>Jonah Berger's STEPPS framework</strong> — a model built from years of research into why people share things. It's not the only way to think about virality, but it's one of the most grounded ones out there.", "评分基于 Jonah Berger 的 STEPPS 框架 — 多年研究人们为何分享的模型。它不是思考爆款的唯一方式，却是最 grounded 的之一。"),
        "camp2_science_link": ("Learn more →", "了解更多 →"),
        "camp2_community": ("Stay in the loop", "保持联系"),
        "camp2_community_p": ("If you find this useful, come say hi. There's a Telegram group where we share breakdowns and new features as they ship.", "如果觉得有用，来打个招呼。有个 Telegram 群组，我们会在那里分享拆解和新功能。"),
        "camp2_comm_tg": ("Telegram group", "Telegram 群组"),
        "camp2_comm_x": ("𝕏 @miccakitt", "𝕏 @miccakitt"),
        "camp2_comm_blog": ("Blog", "博客"),
        "camp2_comm_direct": ("Direct updates", "直达更新"),
        "camp2_footer": ("No signup, no API key needed. Just open a tool and use it.", "无需注册，无需 API 密钥。打开工具就能用。"),
        "camp2_footer_built": ("Built by", "由"),
        "camp2_footer_welcome": ("— feedback always welcome.", "— 欢迎反馈"),
        # Science page (STEPPS)
        "science_hero_h2": ("Not guesswork—<em>peer-reviewed</em> virality.", "非猜测 — <em>同行评审</em> 的爆款科学"),
        "science_hero_tagline": ("Every ViralLab score runs on Jonah Berger's STEPPS framework — 30 years of behavioral research condensed into a single, auditable algorithm. The same science Fortune 500 brands use to engineer word-of-mouth, now for your content.", "每个 ViralLab 评分都基于 Jonah Berger 的 STEPPS 框架 — 30 年行为研究浓缩成一套可稽核的算法。Fortune 500 品牌用来设计口碑的科学，现在为你的内容服务"),
        "science_hero_cta": ("See how we score", "看我们如何评分"),
        "science_proof_4": ("NYT bestsellers on contagiousness", "纽约时报畅销书：传播学"),
        "science_proof_4_sub": ("Verified", "已验证"),
        "science_proof_80": ("Peer-reviewed studies underpinning the model", "同行评审研究支撑模型"),
        "science_proof_80_sub": ("Academic citations", "学术引用"),
        "science_proof_30yr": ("Of word-of-mouth research at Wharton", "沃顿商学院口碑研究资历"),
        "science_proof_30yr_sub": ("Research tenure", "研究资历"),
        "science_proof_f500": ("Brands including Apple, Google & Nike", "Apple、Google、Nike 等品牌"),
        "science_proof_f500_sub": ("Applied clients", "应用客户"),
        "science_proof_cta": ("See it in action →", "看实际评分 →"),
        "science_berger_name": ("Jonah Berger · Professor, Wharton School", "Jonah Berger · 沃顿商学院教授"),
        "science_berger_quote": ("\"The key question isn't 'how do I make something go viral?' — it's 'why do people talk and share things in the first place?'\"", "「关键问题不是『怎么让东西爆红？』— 而是『人们为什么会谈论和分享？』」"),
        "science_cred_1": ("4 New York Times bestselling books", "4 本纽约时报畅销书"),
        "science_cred_2": ("Wharton School Professor since 1996", "1996 年起任沃顿商学院教授"),
        "science_cred_3": ("80+ peer-reviewed studies published", "80+ 篇同行评审研究"),
        "science_cred_4": ("Consulted for Apple, Google, Nike, Gates Foundation", "为 Apple、Google、Nike、盖兹基金会提供顾问"),
        "science_cred_5": ("Cited 10,000+ times in academic literature", "学术文献引用 10,000+ 次"),
        "science_why_holds": ("Why Berger's work holds up", "Berger 的研究为何站得住脚"),
        "science_why_holds_desc": ("Most viral frameworks are post-hoc pattern matching. Berger's work starts with <em>why humans share</em>, then derives patterns from behavioral science and controlled experiments.", "多数爆款框架是事后归纳。Berger 从<em>人类为什么分享</em>出发，再从行为科学与对照实验推导模式"),
        "science_4keys_title": ("The 4 Keys to Viral Content", "爆款内容的 4 把钥匙"),
        "science_4keys_desc": ("Each book unlocks a different lever. ViralLab applies all of them to your content.", "每本书解锁不同杠杆。ViralLab 将它们全部应用于你的内容"),
        "science_stepps_title": ("STEPPS — What We Look For", "STEPPS — 我们看什么"),
        "science_stepps_desc": ("Hover each principle to see the behavioral question our algorithm answers.", "悬停每个原则查看我们算法回答的行为问题"),
        "science_keywords": ("Keywords", "关键字"),
        "science_how_score_h": ("How We Score", "我们如何评分"),
        "science_how_score_p": ("We scan content for <strong>STEPPS</strong> (Social Currency, Triggers, Emotion, Public, Practical, Stories) and <strong>Magic Words</strong>—the language that research shows boosts sharing. We also score <strong>hook strength</strong> (first 15 words: question, stat, surprising claim, conflict) and <strong>narrative arc</strong> (setup→conflict→resolution). Content with 3+ STEPPS principles gets a bonus. Transparent formula, no black box.", "我们扫描内容的 <strong>STEPPS</strong>（社交货币、触发、情绪、公开、实用、故事）与 <strong>Magic Words</strong>—研究证实能提升分享的语言。我们也评分<strong>开场强度</strong>（前 15 字：问句、数据、惊人主张、冲突）与<strong>叙事弧</strong>（铺陈→冲突→解决）。含 3+ STEPPS 原则的内容有加分。透明公式，非黑盒"),
        "science_formula": ("STEPPS: up to 20 pts per principle. Magic words: +5 pts each. Hook: up to 15 pts. Narrative arc: +10 pts. 3+ principles: +10 pts. Total capped at 100.", "STEPPS：每原则最高 20 分。Magic Words：每个 +5 分。开场：最高 15 分。叙事弧：+10 分。3+ 原则：+10 分。总分上限 100"),
        "science_lab_title": ("From Berger's lab to your score", "从 Berger 实验室到你的评分"),
        "science_lab_desc": ("We don't match keywords. We detect the underlying behavioral patterns Berger's research proved actually drive sharing.", "我们不匹配关键字。我们侦测 Berger 研究证实真正驱动分享的行为模式"),
        "science_magic_title": ("Magic Words", "Magic Words"),
        "science_magic_desc": ("Words that boost engagement", "提升参与的词汇"),
        "science_trust_h": ("The same framework Fortune 500 brands use. <em>Now for your content.</em>", "Fortune 500 品牌用的同一套框架。<em>现在为你的内容服务。</em>"),
        "science_trust_sub": ("Thirty years of peer-reviewed research. One transparent score. Know exactly why your content spreads — or doesn't.", "30 年同行评审研究。一个透明评分。清楚知道你的内容为什么会爆 — 或不会"),
        "science_cta_score": ("Score your content →", "评分你的内容 →"),
        "science_cta_book": ("Read Contagious by Berger", "阅读 Berger《疯潮行销》"),
        "science_translate_1_t": ("Signal, not syntax", "信号，非语法"),
        "science_translate_1_b": ("Emotion is scored by pacing, narrative tension, and vocal dynamics — not the word 'amazing'. Social Currency is detected by insight density and identity framing — not the word 'exclusive'.", "情绪由节奏、叙事张力、语调评分 — 非『amazing』一字。社交货币由洞察密度与身份框架侦测 — 非『exclusive』一字"),
        "science_translate_2_t": ("Hook scored separately", "开场独立评分"),
        "science_translate_2_b": ("The first 5 seconds are the single highest predictor of shareability. We score your hook independently before aggregating the full signal.", "前 5 秒是分享率的最强预测。我们在汇总前独立评分你的开场"),
        "science_translate_3_t": ("Transparent, not black-box", "透明，非黑盒"),
        "science_translate_3_b": ("Every score breaks down by principle. You see exactly which STEPPS signals fired, which were weak, and what to change. No mysterious confidence scores.", "每个评分都按原则拆解。你清楚看到哪些 STEPPS 信号触发、哪些弱、该改什么。没有神秘信心分数"),
        "science_translate_4_t": ("Platform-calibrated weights", "平台校准权重"),
        "science_translate_4_b": ("Emotion matters more on TikTok. Practical Value matters more on YouTube tutorials. Weights adjust to the platform context of your content.", "情绪在 TikTok 较重要。实用价值在 YouTube 教学较重要。权重依平台情境调整"),
        # Science v2 (How we score)
        "science_v2_title": ("📊 How we score", "📊 我们如何评分"),
        "science_v2_page_title": ("How We Score", "我们如何评分"),
        "science_v2_intro1": ("ViralLab scoring is based on <strong>Jonah Berger's STEPPS</strong> framework.", "ViralLab 评分基于 <strong>Jonah Berger 的 STEPPS</strong> 框架。"),
        "science_v2_intro2": ("Use it as a practical lens to improve hooks, structure, and shareability.", "把它当作优化开场、结构与传播性的实用视角。"),
        "science_v2_stepps_title": ("The six signals we score (STEPPS)", "我们评分的六个信号 (STEPPS)"),
        "science_v2_stepps_intro": ("Berger identified six things that consistently make content more shareable:", "Berger 发现六个让内容更易分享的因素："),
        "science_v2_stepps_S_name": ("Social Currency", "社交货币"),
        "science_v2_stepps_S_desc": ("People share things that make them look good or knowledgeable. Content that feels exclusive or insightful gets passed on.", "人们分享让自己显得有见识的内容。感觉独家或洞察力强的会被转发。"),
        "science_v2_stepps_T_name": ("Triggers", "触发"),
        "science_v2_stepps_T_desc": ("Things that link to everyday moments get recalled naturally. \"Every time I see X, I think of Y.\"", "与日常时刻连结的内容会被自然想起。「每次看到 X，我就想到 Y。」"),
        "science_v2_stepps_E_name": ("Emotion", "情绪"),
        "science_v2_stepps_E_desc": ("High-arousal emotions — awe, excitement, anger — drive sharing. Low-arousal ones (contentment, sadness) don't as much.", "高唤醒情绪 — 敬畏、兴奋、愤怒 — 驱动分享。低唤醒（满足、悲伤）较少。"),
        "science_v2_stepps_P1_name": ("Public", "公开"),
        "science_v2_stepps_P1_desc": ("Visible things get copied. If people can see others doing or sharing something, they're more likely to do it too.", "可见的事物会被模仿。若人们看到别人在做或分享，更可能跟进。"),
        "science_v2_stepps_P2_name": ("Practical Value", "实用价值"),
        "science_v2_stepps_P2_desc": ("Useful content gets shared because people want to help others. \"You have to see this tip.\"", "实用内容被分享因为人们想帮助他人。「你一定要看这个技巧。」"),
        "science_v2_stepps_S2_name": ("Stories", "故事"),
        "science_v2_stepps_S2_desc": ("Information travels better inside a story. People retell narratives, not facts.", "信息在故事里传播更好。人们复述的是叙事，不是事实。"),
        "science_v2_formula_title": ("How the number is calculated", "分数如何计算"),
        "science_v2_formula_intro": ("We scan the transcript for signals in each category and add up the points:", "我们扫描逐字稿中各类别信号并累加分数："),
        "science_v2_formula_r1": ("<strong>Each STEPPS signal</strong> — keywords, framing, structure", "<strong>每个 STEPPS 信号</strong> — 关键词、框架、结构"),
        "science_v2_formula_r2": ("<strong>Magic words</strong> — language shown to boost engagement", "<strong>Magic words</strong> — 研究证实提升参与的语言"),
        "science_v2_formula_r3": ("<strong>Hook strength</strong> — first 15 words: question, stat, surprise", "<strong>开场强度</strong> — 前 15 字：问句、数据、惊喜"),
        "science_v2_formula_r4": ("<strong>Narrative arc</strong> — setup → conflict → resolution", "<strong>叙事弧</strong> — 铺陈 → 冲突 → 解决"),
        "science_v2_formula_r5": ("<strong>3+ signals active</strong> — content hits multiple principles", "<strong>3+ 信号激活</strong> — 内容触及多个原则"),
        "science_v2_formula_r6": ("<strong>Total</strong>", "<strong>总分</strong>"),
        "science_v2_formula_pts1": ("up to 20 pts", "最高 20 分"),
        "science_v2_formula_pts2": ("+5 pts each", "每个 +5 分"),
        "science_v2_formula_pts3": ("up to 15 pts", "最高 15 分"),
        "science_v2_formula_pts4": ("+10 pts", "+10 分"),
        "science_v2_formula_pts5": ("+10 pts", "+10 分"),
        "science_v2_formula_pts6": ("capped at 100", "上限 100"),
        "science_v2_honest": ("<strong>A note on accuracy:</strong> This score is based on text analysis — keywords, structure, and framing. It doesn't account for delivery, visuals, timing, or luck, which all matter. Think of it as one useful lens, not a definitive verdict. A low score doesn't mean the content is bad. A high score doesn't guarantee it spreads.", "<strong>关于准确性：</strong>此评分基于文字分析 — 关键词、结构、框架。未考虑表达、画面、时机或运气，这些都很重要。把它当作一个有用视角，非最终定论。低分不代表内容差。高分也不保证会传播。"),
        "science_v2_link_score": ("Score a video →", "评分视频 →"),
        "science_v2_link_transcript": ("Transcript + score", "逐字稿 + 评分"),
        "science_minto_title": ("Minto Pyramid — Structure", "Minto 金字塔 — 结构"),
        "science_minto_desc": ("Lead with the answer, then key points, then evidence. In social media, users decide in seconds whether to keep watching, so this structure improves clarity, retention, and repurposing speed across formats. ViralLab extracts this structure from transcripts on Video to Text so you can turn one script into hooks, threads, and long-form drafts with less guesswork.", "结论先行，再要点，再论据。社交媒体里用户几秒内就决定是否继续看，这种结构能显著提升清晰度、完播和多平台改写效率。ViralLab 会在「视频转文字」里自动提取该结构，让你用同一份逐字稿更快产出开场、线程和长文草稿。"),
        "science_minto_cta": ("Try it on Video to Text →", "在视频转文字中试试 →"),
    }
    en, zh = T.get(key, (key, key))
    return zh if lang == "zh" else en


def _lifecycle_tooltip_html(lang: str, life: str) -> str:
    """Tooltip content for one lifecycle: show only the definition for that rank."""
    key = f"tooltip_{life}" if life in ("rising", "peaking", "fading") else "tooltip_peaking"
    return _html_escape(_t(key, lang))


# Function cards for Features dropdown (Buffer-style)
FUNCTION_CARDS = [
    {"href": "/daily", "icon": "📅", "title": "Daily News", "desc": "Top 3 most talked about topics today"},
    {"href": "/longform", "icon": "📚", "title": "Long-form & Podcasts", "desc": "Deep reads and episodes in one place"},
    {"href": "/news", "icon": "📰", "title": "News by Topic", "desc": "Search by topic and read signals"},
    {"href": "/viral", "icon": "▶️", "title": "Viral Videos", "desc": "Search viral videos ranked by spread rate"},
    {"href": "/china-content", "icon": "📥", "title": "China content", "desc": "Fetch from Bilibili, XHS, Douyin, Zhihu"},
    {"href": "/video-to-text", "icon": "📝", "title": "Video to Text", "desc": "Paste URL → transcript + Berger score"},
    {"href": "/article-lab", "icon": "✍️", "title": "Article Lab", "desc": "Rewrite with Minto + STEPPS"},
    {"href": "/publish", "icon": "🚀", "title": "Publish Queue", "desc": "Approve drafts and publish to channels"},
    {"href": "/science", "icon": "📊", "title": "STEPPS & Structure", "desc": "Scoring and writing framework"},
]


def _sidebar_html(active="", lang="en"):
    """Aegis-style sidebar navigation. lang: en or zh for toggle."""
    nav_items = [
        ("/", "🏠", "nav_home", "nav_home_desc"),
        ("/daily", "📅", "nav_daily", "nav_daily_desc"),
        ("/field", "🎯", "nav_field", "nav_field_desc"),
        ("/news", "📰", "nav_news", "nav_news_desc"),
        ("/longform", "📚", "nav_longform", "nav_longform_desc"),
        ("/viral", "▶️", "nav_viral", "nav_viral_desc"),
        ("/china-content", "📥", "nav_china_content", "nav_china_content_desc"),
        ("/video-to-text", "📝", "nav_video2text", "nav_video2text_desc"),
        ("/article-lab", "✍️", "nav_article_lab", "nav_article_lab_desc"),
        ("/publish", "🚀", "nav_publish", "nav_publish_desc"),
        ("/science", "📊", "nav_science", "nav_science_desc"),
    ]
    links = ""
    for href, icon, title_key, desc_key in nav_items:
        title, desc = _t(title_key, lang), _t(desc_key, lang)
        is_active = " active" if active == href else ""
        links += f'<a href="{href}" class="sidebar-link{is_active}"><span class="sidebar-icon">{icon}</span><span class="sidebar-text"><strong>{_html_escape(title)}</strong><small>{_html_escape(desc)}</small></span></a>'
    next_path = request.path or "/"
    toggle_label = "EN" if lang == "zh" else "中文"
    toggle_lang = "en" if lang == "zh" else "zh"
    toggle_href = f"/set-lang?lang={toggle_lang}&next={_html_escape(next_path)}"
    subtitle = _t("hero_subtitle", lang)
    return f'''<aside class="sidebar">
        <a href="/" class="sidebar-brand">
            <span class="logo-avatar"><img src="/static/logo-avatar.png" alt=""></span>
            <span class="logo-text-wrap">
                <span class="logo-text">ViralLab</span>
                <span class="sidebar-subtitle">{_html_escape(subtitle)}</span>
            </span>
        </a>
        <nav class="sidebar-nav">{links}</nav>
        <div class="sidebar-footer">
            <a href="{toggle_href}" class="btn-lang" id="langToggle">{toggle_label} ▾</a>
        </div>
    </aside>'''


def _header_html(active="", lang="en"):
    """Compact top bar with logo, search and community links."""
    placeholder = _t("search_placeholder", lang)
    return f'''<header class="top-bar">
        <div class="header-search-wrap">
            <span class="search-icon">⌕</span>
            <input type="text" class="header-search" placeholder="{_html_escape(placeholder)}" id="headerSearch">
        </div>
        <div class="top-bar-links-wrap">
            <div class="top-bar-links">
                <a href="{TG_GROUP_URL}" target="_blank" rel="noopener">💬 {_html_escape(_t("camp2_comm_tg", lang))}</a>
                <a href="{X_URL}" target="_blank" rel="noopener">{_html_escape(_t("camp2_comm_x", lang))}</a>
                <a href="{BLOG_URL}" target="_blank" rel="noopener">📖 {_html_escape(_t("camp2_comm_blog", lang))}</a>
                <a href="{GITHUB_REPO_URL}" target="_blank" rel="noopener">✈️ {_html_escape(_t("camp2_comm_direct", lang))}</a>
            </div>
            <div class="top-bar-hover-card">{_html_escape(_t("camp2_community_p", lang))}</div>
        </div>
    </header>'''


def _hero_html(lang="en"):
    """Hero: left text, right thumbnail image."""
    greeting = _t("camp2_greeting", lang)
    line1 = _t("hero_line1", lang)
    line2 = _t("hero_line2", lang)
    subtitle = _t("hero_subtitle", lang)
    desc = _t("hero_desc", lang)
    cta = _t("hero_cta", lang)
    return f'''<section class="hero hero-bg">
        <div class="hero-left">
            <p class="hero-greeting">👋 {_html_escape(greeting)}</p>
            <h1 class="hero-headline"><span>{_html_escape(line1)}</span><span>{_html_escape(line2)}</span></h1>
            <p class="hero-subtitle">{_html_escape(subtitle)}</p>
            <p class="hero-desc">{_html_escape(desc)}</p>
            <a href="/daily" class="hero-cta">{_html_escape(cta)}</a>
        </div>
        <div class="hero-visual">
            <div class="hero-thumbnail-wrap">
                <img src="/static/hero-thumbnail.png" alt="ViralLab" class="hero-thumbnail">
            </div>
        </div>
    </section>'''


def _base(title, body, active="", hero="", lang=None, og_title=None, og_desc=None, og_image=None):
    lang = lang or _get_lang()
    base_url = request.host_url.rstrip("/")
    og_title = og_title or f"{title} — ViralLab"
    og_desc = og_desc or "Your content agent. Engineer your influence. Curated trends, daily news, and resources for content creators."
    og_image = og_image or f"{base_url}/static/og-image.png"
    canonical = request.url
    return f"""<!DOCTYPE html>
<html lang="{lang}">
<head>
    <meta charset="utf-8">
    <meta name="viewport" content="width=device-width, initial-scale=1">
    <link rel="icon" href="/static/favicon-32.png" type="image/png" sizes="32x32">
    <title>{_html_escape(title)} — ViralLab</title>
    <meta property="og:type" content="website">
    <meta property="og:title" content="{_html_escape(og_title)}">
    <meta property="og:description" content="{_html_escape(og_desc)}">
    <meta property="og:image" content="{_html_escape(og_image)}">
    <meta property="og:url" content="{_html_escape(canonical)}">
    <meta property="og:site_name" content="ViralLab">
    <meta name="twitter:card" content="summary_large_image">
    <meta name="twitter:title" content="{_html_escape(og_title)}">
    <meta name="twitter:description" content="{_html_escape(og_desc)}">
    <meta name="twitter:image" content="{_html_escape(og_image)}">
    <link rel="preconnect" href="https://fonts.googleapis.com">
    <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
    <link href="https://fonts.googleapis.com/css2?family=Poppins:ital,wght@0,300;0,400;0,500;0,600;0,700;0,800;0,900;1,300;1,400&display=swap" rel="stylesheet">
    <style>
        :root {{ --bg: #faf8f5; --card: #fff; --text: #2d4a2b; --muted: #5a6b58; --accent: #2d4a2b; --accent-hover: #1e3520; --border: #e8e4dc; --warm: #f5f2ed; --cta: #e8a54d; --cta-border: #2d4a2b; --cta-text: #2d4a2b; --header-height: 4.25rem; --footer-height: 80px; --accent-mid: #3d6b3a; --accent-light: #6a9e67; --cta-dark: #c8883a; --cta-pale: #fdf3dc; --cream-dark: #ede8de; --muted-light: #8fa88c; --ink: #1a2419; --radius: 4px; --shadow: 0 2px 12px rgba(45,74,43,0.08); --shadow-lg: 0 8px 32px rgba(45,74,43,0.13); }}
        * {{ box-sizing: border-box; }}
        html {{ scroll-behavior: smooth; overflow-x: hidden; }}
        :root {{ --ease: cubic-bezier(0.25, 0.1, 0.25, 1); --ease-out: cubic-bezier(0.33, 1, 0.68, 1); --dur: 0.25s; }}
        body {{ font-family: "Poppins", -apple-system, BlinkMacSystemFont, sans-serif; background: var(--bg); color: var(--text); margin: 0; padding: 0; line-height: 1.55; overflow-x: hidden; }}
        body::before {{ content: ""; position: fixed; top: 0; left: 0; right: 0; height: 3px; background: var(--accent); z-index: 999; }}
        html[lang="zh"] .sidebar-subtitle, html[lang="zh"] .hero-subtitle, html[lang="zh"] .hero-desc, html[lang="zh"] .hero-cta, html[lang="zh"] .tagline, html[lang="zh"] .sidebar-text strong, html[lang="zh"] .sidebar-text small {{ letter-spacing: 0.12em; }}
        .app-layout {{ display: flex; min-height: 100vh; overflow-x: hidden; }}
        .sidebar {{ width: 220px; flex-shrink: 0; background: var(--bg); border-right: 1px solid var(--border); padding: 0; position: fixed; left: 0; top: 0; bottom: 0; display: flex; flex-direction: column; z-index: 40; }}
        .sidebar-brand {{ display: flex; align-items: center; gap: 0.5rem; padding: 0 1.25rem; height: var(--header-height); box-sizing: border-box; text-decoration: none; color: var(--text); border-bottom: 1px solid var(--border); margin-bottom: 1rem; }}
        .sidebar-brand:hover {{ color: var(--accent); }}
        .sidebar-nav {{ display: flex; flex-direction: column; gap: 0.25rem; flex: 1; min-height: 0; overflow-y: auto; padding: 0.5rem 0; }}
        .sidebar-link {{ display: flex; align-items: flex-start; gap: 0.6rem; padding: 0.6rem 1.25rem; text-decoration: none; color: var(--muted); font-size: 0.9rem; transition: background var(--dur) var(--ease-out), color var(--dur) var(--ease-out); }}
        .sidebar-link:hover {{ background: var(--warm); color: var(--accent); }}
        .sidebar-link.active {{ background: var(--warm); color: var(--accent); font-weight: 500; }}
        .sidebar-icon {{ font-size: 1.1rem; flex-shrink: 0; }}
        .sidebar-text {{ display: flex; flex-direction: column; gap: 0.1rem; }}
        .sidebar-text strong {{ font-size: 0.9rem; color: inherit; }}
        .sidebar-text small {{ font-size: 0.75rem; opacity: 0.85; }}
        .sidebar-footer {{ margin-top: auto; height: var(--footer-height); padding: 0 1.25rem; border-top: 1px solid var(--border); display: flex; align-items: center; font-size: 0.85rem; flex-shrink: 0; box-sizing: border-box; }}
        .sidebar-footer a {{ color: var(--text); text-decoration: none; }}
        .sidebar-footer a:hover {{ color: var(--accent); }}
        .main-content {{ margin-left: 220px; flex: 1; display: flex; flex-direction: column; min-width: 0; overflow-x: hidden; padding: 0 78px calc(2rem + var(--footer-height)) 48px; }}
        .top-bar {{ position: sticky; top: 0; z-index: 50; height: var(--header-height); padding: 0 78px 0 48px; margin-bottom: 1rem; margin-left: -48px; margin-right: -78px; background: var(--bg); overflow: visible; display: flex; align-items: center; justify-content: space-between; gap: 1.5rem; box-sizing: border-box; }}
        .top-bar-links-wrap {{ position: relative; margin-left: auto; }}
        .top-bar-links {{ display: flex; align-items: center; gap: 1rem; }}
        .top-bar-links a {{ color: var(--muted); text-decoration: none; font-size: 0.75rem; font-weight: 300; transition: color var(--dur) var(--ease-out); }}
        .top-bar-links a:hover {{ color: var(--accent); }}
        .top-bar-hover-card {{ position: absolute; top: 100%; right: 0; margin-top: 0.5rem; padding: 1rem 1.25rem; max-width: 240px; font-size: 0.6rem; font-weight: 400; line-height: 1.5; color: var(--card); background: var(--accent); border-radius: var(--radius); box-shadow: var(--shadow-lg); opacity: 0; pointer-events: none; transition: opacity var(--dur) var(--ease-out); z-index: 100; }}
        .top-bar-links-wrap:hover .top-bar-hover-card {{ opacity: 1; }}
        .top-bar::after {{ content: ""; position: absolute; left: 0; bottom: 0; width: calc(100vw - 220px); height: 1px; background: var(--border); pointer-events: none; }}
        .wrap {{ max-width: 1100px; margin: 0 auto; padding: 0 2rem 2rem; }}
        .page-header {{ margin-bottom: 2.5rem; position: relative; padding-top: 0.5rem; }}
        .page-header::before {{ content: ''; position: absolute; top: 0; left: 0; width: 48px; height: 3px; background: var(--cta); border-radius: 1px; }}
        h1 {{ font-size: 1.75rem; font-weight: 800; margin: 0 0 0.5rem 0; letter-spacing: -0.02em; line-height: 1.25; color: var(--ink); }}
        .tagline {{ color: var(--muted); font-size: 1rem; margin: 0; line-height: 1.6; font-weight: 400; max-width: 42ch; }}
        .stats {{ display: grid; grid-template-columns: repeat(3, 1fr); gap: 1.5rem; margin: 2.5rem 0; }}
        .stat {{ background: var(--card); padding: 1.25rem 1.5rem; border-radius: var(--radius); border: 1px solid var(--border); box-shadow: var(--shadow); text-align: left; transition: transform var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out); }}
        .stat:hover {{ transform: translateY(-2px); box-shadow: var(--shadow-lg); border-color: var(--accent-light); }}
        .stat-num {{ font-size: 1.75rem; font-weight: 700; color: var(--accent); letter-spacing: -0.02em; }}
        .stat-label {{ font-size: 0.8rem; color: var(--muted); margin-top: 0.25rem; }}
        section {{ margin-bottom: 3rem; }}
        section:first-of-type {{ margin-top: 0; }}
        .section-title {{ font-size: 1.5rem; font-weight: 800; color: var(--ink); letter-spacing: -0.01em; margin-bottom: 0.5rem; line-height: 1.3; }}
        .section-desc {{ font-size: 0.9375rem; color: var(--muted); margin-bottom: 1.25rem; max-width: 56ch; line-height: 1.6; }}
        .cards {{ display: grid; gap: 1.25rem; grid-template-columns: repeat(auto-fill, minmax(280px, 1fr)); align-items: stretch; }}
        .daily-cards-grid {{ grid-template-columns: repeat(3, 1fr); gap: 1.75rem; }}
        .card {{ background: var(--card); border-radius: var(--radius); padding: 1.25rem 1.5rem; border: 1px solid var(--border); box-shadow: var(--shadow); transition: transform var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out); display: flex; flex-direction: column; gap: 0.5rem; min-height: 100%; overflow: visible; }}
        .card:hover {{ transform: translateY(-2px); box-shadow: var(--shadow-lg); border-color: var(--accent-light); }}
        .card h3 {{ font-size: 1.05rem; font-weight: 600; margin: 0; line-height: 1.45; color: var(--text); letter-spacing: -0.01em; flex: 1; min-width: 0; word-break: break-word; }}
        .card h3 a {{ color: inherit; text-decoration: none; }}
        .card h3 a:hover {{ color: var(--accent); }}
        .card p {{ font-size: 0.9rem; color: var(--muted); margin: 0; font-weight: 400; line-height: 1.55; }}
        .card-snippet {{ font-size: 0.9rem; color: var(--muted); line-height: 1.55; margin: 0; word-break: break-word; overflow-wrap: break-word; display: -webkit-box; -webkit-line-clamp: 4; -webkit-box-orient: vertical; overflow: hidden; flex: 1; min-height: 0; }}
        .card-source {{ font-size: 0.68rem; color: var(--muted); opacity: 0.75; margin: 0; }}
        .card-header {{ display: flex; justify-content: space-between; align-items: flex-start; gap: 0.75rem; overflow: visible; }}
        .card-header h3 {{ flex: 1; min-width: 0; }}
        .card-header .badges {{ flex-shrink: 0; display: flex; flex-direction: column; align-items: flex-end; gap: 0.35rem; }}
        .berger {{ font-size: 0.75rem; font-weight: 700; padding: 0.25rem 0.6rem; border-radius: var(--radius); white-space: nowrap; }}
        .berger-high {{ background: rgba(45,74,43,0.2); color: var(--accent); }}
        .berger-mid {{ background: rgba(232,165,77,0.35); color: var(--cta-text); }}
        .berger-low {{ background: rgba(45,74,43,0.1); color: var(--muted); }}
        .meta {{ font-size: 0.78rem !important; color: var(--muted); opacity: 0.9; }}
        .card-error {{ border-left: 4px solid #dc2626; }}
        .muted {{ color: var(--muted); font-size: 0.9rem; line-height: 1.55; }}
        .content-meta {{ display: flex; flex-direction: column; gap: 0.35rem; margin-bottom: 1.25rem; }}
        .content-meta .sources-label {{ margin: 0; }}
        .sources-label {{ color: var(--muted); font-size: 0.8rem; font-weight: 400; line-height: 1.5; opacity: 0.92; }}
        .sources-label a {{ color: var(--text); text-decoration: none; }}
        .sources-label a:hover {{ color: var(--accent); }}
        .browse-more {{ color: var(--muted); font-size: 0.85rem; font-weight: 400; margin: 1.75rem 0 1rem 0; opacity: 0.92; line-height: 1.5; }}
        .browse-more a {{ color: var(--text); text-decoration: none; }}
        .browse-more a:hover {{ color: var(--accent); }}
        .refresh-link {{ color: var(--muted) !important; text-decoration: none !important; }}
        .refresh-link:hover {{ color: var(--accent) !important; }}
        .tip-range-wrap .refresh-link {{ color: var(--muted) !important; text-decoration: none !important; }}
        .content-meta a {{ color: var(--muted) !important; text-decoration: none !important; }}
        .content-meta a:hover {{ color: var(--accent) !important; }}
        .browse-more a:hover {{ color: var(--accent-hover); }}
        .digest-cards {{ display: flex; flex-direction: column; gap: 1.5rem; margin-top: 1.75rem; }}
        .digest-card.load-more-hidden {{ display: none; }}
        .btn-load-more {{ display: block; margin: 2rem auto; padding: 0.75rem 2rem; border: 1px solid var(--border); border-radius: var(--radius); background: var(--card); color: var(--text); font-size: 0.9375rem; font-weight: 600; cursor: pointer; font-family: inherit; transition: border-color var(--dur), color var(--dur); }}
        .btn-load-more:hover {{ border-color: var(--accent); color: var(--accent); }}
        .digest-card {{ background: var(--card); border-radius: var(--radius); padding: 1.25rem 1.5rem; border: 1px solid var(--border); box-shadow: var(--shadow); transition: transform var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out); overflow: visible; }}
        .digest-card:hover {{ transform: translateY(-2px); box-shadow: var(--shadow-lg); border-color: var(--accent-light); }}
        .digest-card-header {{ display: flex; align-items: flex-start; gap: 0.75rem; margin-bottom: 0.6rem; overflow: visible; }}
        .digest-card-badges {{ display: flex; flex-direction: column; align-items: flex-end; gap: 0.35rem; flex-shrink: 0; }}
        .digest-card-num {{ font-size: 0.85rem; font-weight: 700; color: var(--accent); min-width: 1.75rem; opacity: 0.9; }}
        .digest-card-title {{ flex: 1; font-size: 1.05rem; font-weight: 600; margin: 0; line-height: 1.45; color: var(--text); letter-spacing: -0.01em; }}
        .digest-card-title a {{ color: inherit; text-decoration: none; }}
        .digest-card-title a:hover {{ color: var(--accent); }}
        .digest-card-snippet {{ font-size: 0.9rem; color: var(--muted); margin: 0 0 0.5rem 0; line-height: 1.6; font-weight: 400; }}
        .digest-card-source {{ font-size: 0.68rem; color: var(--muted); opacity: 0.75; margin: 0.15rem 0 0.85rem 0; }}
        .digest-card-pills {{ display: flex; flex-wrap: wrap; gap: 0.5rem; margin-top: 0.25rem; }}
        .pill-tag {{ font-size: 0.72rem; padding: 0.28rem 0.65rem; border-radius: 999px; background: rgba(45,74,43,0.08); color: var(--muted); font-weight: 500; letter-spacing: 0.02em; }}
        .digest-top-section {{ margin-bottom: 0.5rem; display: flex; flex-direction: column; gap: 0.5rem; align-items: flex-start; }}
        .digest-top-section .browse-more {{ margin: 0; }}
        pre {{ max-width: 100%; overflow-x: auto; white-space: pre-wrap; word-break: break-all; box-sizing: border-box; }}
        code {{ background: var(--warm); padding: 0.2rem 0.5rem; border-radius: 6px; font-size: 0.85em; color: var(--text); }}
        footer {{ position: fixed; bottom: 0; left: 220px; right: 0; height: var(--footer-height); padding: 0 2rem; font-size: 0.8rem; color: var(--muted); background: var(--bg); z-index: 50; box-sizing: border-box; display: flex; flex-direction: column; justify-content: center; border-top: 1px solid var(--border); }}
        footer a {{ color: var(--accent); text-decoration: none; }}
        footer a:hover {{ text-decoration: underline; }}
        .spread {{ font-size: 0.75rem; margin-bottom: 0.35rem; }}
        .form {{ display: flex; gap: 0.5rem; margin-bottom: 1.25rem; max-width: 560px; flex-wrap: wrap; align-items: center; }}
        .form input {{ flex: 1; padding: 0.75rem 1rem; border-radius: var(--radius); border: 1.5px solid var(--border); background: var(--bg); color: var(--text); font-size: 0.9375rem; font-family: inherit; transition: border-color var(--dur) var(--ease-out); }}
        .tip-card {{ background: var(--card); border: 1px solid var(--border); border-radius: var(--radius); padding: 1.25rem 1.5rem; margin-bottom: 1.25rem; box-shadow: var(--shadow); transition: border-color var(--dur) var(--ease-out); }}
        .tip-card:hover {{ border-color: var(--accent-light); }}
        .china-result-box {{ overflow-wrap: break-word; word-break: break-word; max-width: 100%; min-width: 0; overflow-x: auto; white-space: pre-wrap; font-size: 0.8rem; line-height: 1.4; padding: 0.5rem 0; box-sizing: border-box; }}
        .tip-header {{ display: flex; flex-wrap: wrap; align-items: center; justify-content: space-between; gap: 0.75rem; margin-bottom: 0.5rem; }}
        .tip-label {{ font-size: 0.625rem; font-weight: 700; color: var(--muted); text-transform: uppercase; letter-spacing: 0.12em; }}
        .tip-range-wrap {{ display: flex; gap: 0.5rem; font-size: 0.8rem; }}
        .tip-range {{ color: var(--muted); text-decoration: none; padding: 0.2rem 0.5rem; border-radius: 4px; transition: color var(--dur), background var(--dur); }}
        .tip-range:hover {{ color: var(--accent); }}
        .tip-range.active {{ color: var(--accent); font-weight: 600; background: rgba(45,74,43,0.08); }}
        .viral-source-wrap {{ display: flex; align-items: center; gap: 0.4rem; }}
        .viral-source-label {{ font-size: 0.75rem; font-weight: 600; color: var(--muted); }}
        .viral-source-select {{ padding: 0.25rem 0.5rem; font-size: 0.8rem; }}
        .segment-pills-wrap {{ display: flex; flex-wrap: wrap; gap: 0.35rem; align-items: center; }}
        .segment-pill {{ padding: 0.3rem 0.6rem; border-radius: var(--radius); font-size: 0.75rem; font-weight: 500; text-decoration: none; color: var(--muted); background: var(--cream-dark); transition: background var(--dur), color var(--dur); }}
        .segment-pill:hover {{ color: var(--accent); background: rgba(45,74,43,0.1); }}
        .segment-pill.active {{ color: var(--accent); background: rgba(45,74,43,0.12); font-weight: 600; }}
        .audience-inference-note {{ font-size: 0.7rem; color: var(--muted); margin: 0.6rem 0 0; line-height: 1.4; }}
        .tip-chips {{ display: flex; flex-wrap: wrap; gap: 0.4rem; align-items: center; }}
        .tip-chip-form {{ display: inline-flex; margin: 0; }}
        .tip-chip {{ padding: 0.4rem 0.75rem; border-radius: var(--radius); border: 1px solid var(--border); background: var(--card); color: var(--text); font-size: 0.8125rem; font-weight: 500; cursor: pointer; transition: background var(--dur), border-color var(--dur), color var(--dur); white-space: nowrap; user-select: none; -webkit-tap-highlight-color: transparent; }}
        .tip-chip:hover {{ background: var(--accent); border-color: var(--accent); color: white; }}
        .tip-chip-active {{ background: var(--accent); border-color: var(--accent); color: white; }}
        .china-content-wrap {{ max-width: 100%; margin: 0; padding: 0 0 100px; }}
        .china-layout-grid {{ display: grid; grid-template-columns: minmax(0, 1.4fr) minmax(0, 1fr); gap: 1rem; align-items: start; }}
        .china-stack {{ display: grid; gap: 1rem; align-content: start; }}
        .china-panel {{ margin: 0; width: 100%; }}
        .china-panel .section-desc {{ max-width: none; }}
        .china-wide {{ grid-column: 1 / -1; }}
        .china-title-desc {{ max-width: none; white-space: normal; overflow-wrap: anywhere; margin-bottom: 0.85rem; }}
        .china-platform-row {{ display: flex; flex-wrap: wrap; gap: 0.65rem 0.9rem; }}
        .china-platform-row label {{ font-size: 0.875rem; color: var(--text); }}
        .china-actions {{ display: flex; flex-wrap: wrap; align-items: center; gap: 0.65rem; }}
        .china-help-text {{ font-size: 0.82rem; color: var(--muted); margin: 0; }}
        @media (max-width: 1180px) {{ .china-layout-grid {{ grid-template-columns: 1fr; }} }}
        .topic-label {{ font-size: 0.625rem; font-weight: 700; color: var(--muted); text-transform: uppercase; letter-spacing: 0.12em; margin: 1.5rem 0 0.5rem 0; padding-bottom: 0.35rem; border-bottom: 1px solid var(--border); grid-column: 1 / -1; }}
        .topic-label:first-child {{ margin-top: 0; }}
        .form input:focus {{ outline: none; border-color: var(--accent-light); }}
        .form button {{ padding: 0.75rem 1.5rem; border-radius: var(--radius); border: none; background: var(--cta); color: var(--cta-text); cursor: pointer; font-weight: 700; font-size: 0.9375rem; font-family: inherit; box-shadow: 0 2px 10px rgba(232,165,77,0.3); transition: transform var(--dur) var(--ease-out), background var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out); }}
        .form button:hover {{ transform: translateY(-2px); background: #e0963f; box-shadow: 0 6px 18px rgba(232,165,77,0.4); }}
        .stepps-grid {{ display: grid; gap: 1.25rem; grid-template-columns: repeat(auto-fill, minmax(260px, 1fr)); }}
        .stepp {{ background: var(--card); padding: 1.25rem 1.5rem; border-radius: var(--radius); border: 1px solid var(--border); box-shadow: var(--shadow); transition: transform var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out); }}
        .stepp:hover {{ transform: translateY(-2px); box-shadow: var(--shadow-lg); border-color: var(--accent-light); }}
        .stepp h3 {{ margin: 0 0 0.75rem 0; font-size: 0.625rem; font-weight: 700; color: var(--muted); text-transform: uppercase; letter-spacing: 0.12em; }}
        .stepp p {{ margin: 0; font-size: 0.9rem; color: var(--muted); line-height: 1.55; }}
        .stepp .kw {{ font-size: 0.85rem; color: var(--accent); margin-top: 0.5rem; font-weight: 500; }}
        .science-hero {{ margin-bottom: 2.5rem; }}
        .science-hero h2 {{ font-size: 1.85rem; font-weight: 700; color: var(--accent); margin: 0 0 0.5rem 0; letter-spacing: -0.02em; }}
        .science-hero h2 em {{ font-style: italic; color: var(--accent); }}
        .science-hero .tagline {{ font-size: 1.1rem; color: var(--muted); line-height: 1.6; margin: 0; max-width: 42ch; }}
        .science-proof {{ display: flex; flex-wrap: wrap; gap: 1.5rem 2.5rem; align-items: center; padding: 1.25rem 1.5rem; background: var(--card); border-radius: var(--radius); border: 1px solid var(--border); margin-bottom: 2.5rem; box-shadow: var(--shadow); }}
        .proof-item {{ font-size: 0.95rem; color: var(--text); display: flex; align-items: center; gap: 0.4rem; }}
        .proof-num {{ font-weight: 700; color: var(--accent); font-size: 1.1rem; }}
        .proof-cta {{ margin-left: auto; padding: 0.5rem 1rem; background: var(--cta); color: var(--cta-text); font-weight: 600; font-size: 0.9rem; text-decoration: none; border-radius: 8px; white-space: nowrap; transition: background var(--dur), transform var(--dur); }}
        .proof-cta:hover {{ background: #e0963f; transform: translateX(2px); }}
        .books-grid {{ display: grid; grid-template-columns: repeat(4, 1fr); gap: 1.25rem; margin-bottom: 2.5rem; }}
        .book-card {{ background: var(--card); border-radius: var(--radius); padding: 1.25rem; border: 1px solid var(--border); box-shadow: var(--shadow); transition: transform var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out); text-decoration: none; color: var(--text); display: flex; flex-direction: column; align-items: center; text-align: center; min-height: 280px; }}
        .book-card:hover {{ transform: translateY(-2px); box-shadow: var(--shadow-lg); border-color: var(--accent-light); }}
        .book-cover {{ width: 100%; max-width: 120px; aspect-ratio: 2/3; object-fit: cover; border-radius: 8px; box-shadow: 0 4px 12px rgba(0,0,0,0.15); margin-bottom: 0.85rem; }}
        .book-cover-placeholder {{ width: 100%; max-width: 120px; aspect-ratio: 2/3; background: var(--warm); border-radius: 8px; margin-bottom: 0.85rem; display: none; align-items: center; justify-content: center; font-size: 1.5rem; font-weight: 600; color: var(--muted); }}
        .book-title {{ font-size: 0.95rem; font-weight: 600; line-height: 1.35; margin: 0 0 0.4rem 0; color: var(--accent); }}
        .book-unlocks {{ font-size: 0.8rem; color: var(--muted); line-height: 1.45; }}
        .mechanism-box {{ background: linear-gradient(135deg, rgba(45,74,43,0.06) 0%, rgba(232,165,77,0.08) 100%); border-radius: var(--radius); padding: 1.75rem 2rem; border: 1px solid var(--border); margin-bottom: 2.5rem; }}
        .mechanism-box h3 {{ font-size: 1rem; font-weight: 600; color: var(--accent); margin: 0 0 1rem 0; letter-spacing: -0.01em; }}
        .mechanism-box p {{ font-size: 0.95rem; color: var(--text); line-height: 1.65; margin: 0 0 0.75rem 0; }}
        .mechanism-box p:last-child {{ margin-bottom: 0; }}
        .mechanism-box strong {{ color: var(--accent); }}
        .score-formula {{ font-size: 0.9rem; background: var(--card); padding: 1rem 1.25rem; border-radius: var(--radius); border: 1px solid var(--border); margin-top: 1rem; font-family: ui-monospace, monospace; color: var(--muted); }}
        @media (max-width: 900px) {{ .books-grid {{ grid-template-columns: repeat(2, 1fr); }} }}
        @media (max-width: 500px) {{ .books-grid {{ grid-template-columns: 1fr; }} }}
        @media (max-width: 640px) {{ .science-proof {{ flex-direction: column; align-items: flex-start; }} .proof-cta {{ margin-left: 0; margin-top: 0.5rem; }} }}
        .science-hero .hero-cta {{ display: inline-flex; align-items: center; gap: 0.5rem; padding: 0.75rem 1.5rem; background: var(--accent); color: var(--card); font-weight: 600; text-decoration: none; border-radius: 10px; font-size: 0.95rem; transition: background var(--dur), transform var(--dur); }}
        .science-hero .hero-cta:hover {{ background: var(--accent-hover); transform: translateY(-2px); }}
        .science-wrap {{ max-width: 100%; margin: 0; padding: 0 0 100px; }}
        .science-wrap .section-title {{ font-size: 0.8125rem; font-weight: 700; letter-spacing: 0.08em; text-transform: uppercase; color: var(--muted-light); margin-bottom: 1.25rem; }}
        .science-v2-h1 {{ font-size: 1.375rem; font-weight: 700; letter-spacing: -0.01em; margin-bottom: 1rem; color: var(--ink); }}
        .science-v2-p {{ font-size: 0.9375rem; color: var(--muted); line-height: 1.75; margin-bottom: 1rem; }}
        .science-v2-p strong {{ color: var(--ink); font-weight: 600; }}
        .science-v2-p a {{ color: var(--accent-light); text-decoration: none; }}
        .science-v2-p a:hover {{ text-decoration: underline; }}
        .science-hr {{ border: none; border-top: 1px solid var(--border); margin: 2.25rem 0; }}
        .berger-block {{ background: var(--cream-dark); border-radius: var(--radius); padding: 1.25rem 1.375rem; margin-bottom: 1.5rem; }}
        .berger-name {{ font-size: 0.9375rem; font-weight: 700; color: var(--ink); margin-bottom: 0.2rem; }}
        .berger-role {{ font-size: 0.8125rem; color: var(--muted); margin-bottom: 0.875rem; }}
        .berger-facts {{ display: flex; flex-direction: column; gap: 0.45rem; }}
        .berger-fact {{ display: grid; grid-template-columns: 18px 1fr; gap: 0.625rem; font-size: 0.8125rem; color: var(--muted); line-height: 1.5; }}
        .stepps-list {{ display: flex; flex-direction: column; gap: 0; border: 1px solid var(--border); border-radius: var(--radius); overflow: hidden; }}
        .stepps-item {{ padding: 0.875rem 1.125rem; border-bottom: 1px solid var(--border); display: grid; grid-template-columns: 28px 1fr; gap: 0.875rem; align-items: start; }}
        .stepps-item:last-child {{ border-bottom: none; }}
        .stepps-letter {{ width: 28px; height: 28px; background: var(--cream-dark); border-radius: 50%; display: flex; align-items: center; justify-content: center; font-size: 0.75rem; font-weight: 800; color: var(--accent); flex-shrink: 0; margin-top: 1px; }}
        .stepps-name {{ font-size: 0.875rem; font-weight: 600; color: var(--ink); margin-bottom: 0.2rem; }}
        .stepps-desc {{ font-size: 0.8125rem; color: var(--muted); line-height: 1.55; }}
        .formula-block {{ background: var(--card); border: 1px solid var(--border); border-radius: var(--radius); padding: 1.125rem 1.25rem; }}
        .formula-row {{ display: grid; grid-template-columns: 1fr auto; align-items: center; gap: 0.75rem; padding: 0.5rem 0; border-bottom: 1px solid var(--border); font-size: 0.8125rem; color: var(--muted); }}
        .formula-row:last-child {{ border-bottom: none; }}
        .formula-row strong {{ color: var(--ink); font-weight: 600; }}
        .formula-pts {{ font-size: 0.75rem; font-weight: 600; color: var(--accent); white-space: nowrap; }}
        .honest-note {{ background: var(--cta-pale); border-left: 3px solid var(--cta); border-radius: 0 var(--radius) var(--radius) 0; padding: 0.875rem 1rem; font-size: 0.8125rem; color: var(--muted); line-height: 1.65; }}
        .honest-note strong {{ color: var(--ink); font-weight: 600; }}
        .bottom-links {{ margin-top: 2.25rem; display: flex; gap: 1rem; font-size: 0.8125rem; }}
        .bottom-links a {{ color: var(--accent-light); text-decoration: none; }}
        .bottom-links a:hover {{ text-decoration: underline; }}
        .science-proof-grid {{ display: grid; grid-template-columns: repeat(4, 1fr); gap: 1px; background: var(--border); border-radius: 12px; overflow: hidden; padding: 0; }}
        .science-proof-grid .science-proof-item {{ display: flex; flex-direction: column; gap: 0.35rem; padding: 1.5rem 1.25rem; background: var(--card); }}
        .science-proof-grid .science-proof-label {{ font-size: 0.9rem; color: var(--text); line-height: 1.4; }}
        .science-proof-grid .proof-sublabel {{ font-size: 0.7rem; text-transform: uppercase; letter-spacing: 0.08em; color: var(--muted); margin-top: 0.25rem; }}
        .science-proof-grid .science-proof-cta {{ grid-column: 1 / -1; margin: 0; padding: 1rem 1.5rem; text-align: center; background: var(--warm); border-radius: 0; }}
        .science-proof-grid .science-proof-cta:hover {{ background: var(--cta); }}
        .science-berger-card .science-quote {{ margin: 0 0 1.25rem 0; padding-left: 1rem; border-left: 3px solid var(--accent); font-style: italic; color: var(--text); font-size: 1rem; line-height: 1.6; }}
        .science-creds {{ display: flex; flex-direction: column; gap: 0.5rem; }}
        .science-creds-item {{ display: flex; align-items: center; gap: 0.5rem; font-size: 0.9rem; color: var(--muted); }}
        .science-creds-dot {{ width: 6px; height: 6px; background: var(--cta); border-radius: 50%; flex-shrink: 0; }}
        .science-timeline {{ display: flex; flex-direction: column; gap: 0; margin-top: 1.5rem; }}
        .science-timeline-item {{ display: grid; grid-template-columns: 4rem 1fr; gap: 1.25rem; padding-bottom: 1.5rem; position: relative; }}
        .science-timeline-item:not(:last-child)::before {{ content: ''; position: absolute; left: 1.4rem; top: 2.2rem; bottom: 0; width: 1px; background: var(--border); }}
        .science-timeline-node {{ display: flex; flex-direction: column; align-items: center; }}
        .science-timeline-year {{ font-size: 0.8rem; color: var(--muted); letter-spacing: 0.05em; }}
        .science-timeline-dot {{ width: 10px; height: 10px; background: var(--accent); border-radius: 50%; margin-top: 0.35rem; }}
        .science-timeline-tag {{ display: inline-block; font-size: 0.7rem; letter-spacing: 0.08em; text-transform: uppercase; background: rgba(232,165,77,0.25); color: var(--cta-text); padding: 0.2rem 0.5rem; border-radius: 4px; margin-bottom: 0.4rem; }}
        .science-timeline-title {{ font-size: 1rem; font-weight: 600; color: var(--text); margin-bottom: 0.35rem; }}
        .science-timeline-body {{ font-size: 0.9rem; color: var(--muted); line-height: 1.55; margin: 0; }}
        .science-translate-grid {{ display: grid; grid-template-columns: 1fr 1fr; gap: 2rem; margin-top: 1.5rem; align-items: start; }}
        .science-algo-block {{ margin: 0; }}
        .science-algo-code {{ margin: 0; font-size: 0.8rem; line-height: 1.7; color: var(--muted); white-space: pre-wrap; overflow-x: auto; }}
        .science-algo-code .algo-kw {{ color: var(--accent); }}
        .science-translate-list {{ display: flex; flex-direction: column; gap: 1.25rem; }}
        .science-translate-item {{ display: grid; grid-template-columns: 2.5rem 1fr; gap: 1rem; align-items: start; }}
        .science-translate-num {{ font-size: 1.5rem; font-weight: 700; color: var(--border); line-height: 1; }}
        .science-translate-title {{ font-size: 1rem; font-weight: 600; color: var(--text); margin-bottom: 0.35rem; }}
        .science-translate-body {{ font-size: 0.9rem; color: var(--muted); line-height: 1.55; margin: 0; }}
        .science-trust-closer {{ background: linear-gradient(135deg, rgba(45,74,43,0.08) 0%, rgba(232,165,77,0.06) 100%); border-radius: 16px; padding: 2.5rem 2rem; border: 1px solid var(--border); margin-top: 2rem; }}
        .science-trust-headline {{ font-size: 1.5rem; font-weight: 700; color: var(--text); margin: 0 0 0.75rem 0; line-height: 1.35; }}
        .science-trust-headline em {{ font-style: italic; color: var(--accent); }}
        .science-trust-sub {{ font-size: 1rem; color: var(--muted); line-height: 1.6; margin: 0 0 1.5rem 0; max-width: 36ch; }}
        .science-trust-ctas {{ display: flex; flex-wrap: wrap; gap: 0.75rem; align-items: center; }}
        .science-cta-ghost {{ padding: 0.6rem 1.25rem; border: 1px solid var(--border); border-radius: 8px; background: var(--card); color: var(--text); font-size: 0.9rem; text-decoration: none; transition: border-color var(--dur), color var(--dur); }}
        .science-cta-ghost:hover {{ border-color: var(--accent); color: var(--accent); }}
        @media (max-width: 900px) {{ .science-proof-grid {{ grid-template-columns: repeat(2, 1fr); }} .science-translate-grid {{ grid-template-columns: 1fr; }} }}
        @media (max-width: 560px) {{ .science-proof-grid {{ grid-template-columns: 1fr; }} }}
        .resource-list {{ margin: 0.5rem 0 0 0; padding-left: 1.25rem; font-size: 0.9rem; color: var(--muted); line-height: 1.6; }}
        .resource-list li {{ margin-bottom: 0.4rem; }}
        .resource-list a {{ color: var(--accent); text-decoration: none; }}
        .resource-list a:hover {{ text-decoration: underline; }}
        .field-select {{ padding: 0.6rem 1rem; border-radius: var(--radius); border: 1px solid var(--border); background: var(--card); font-size: 0.9375rem; max-width: 220px; font-family: inherit; }}
        .field-controls {{ display: flex; flex-wrap: wrap; align-items: flex-end; gap: 1rem; margin-bottom: 1.5rem; }}
        .field-control-group {{ display: flex; flex-direction: column; gap: 0.35rem; }}
        .field-control-label {{ font-size: 0.625rem; font-weight: 700; color: var(--muted); text-transform: uppercase; letter-spacing: 0.12em; }}
        .field-share-wrap {{ margin-left: auto; display: flex; align-items: center; gap: 0.5rem; }}
        .share-dropdown {{ position: relative; }}
        .btn-share {{ padding: 0.5rem 1rem; border-radius: var(--radius); border: 1px solid var(--border); background: var(--card); color: var(--text); font-size: 0.9rem; font-weight: 600; cursor: pointer; font-family: inherit; transition: border-color 0.2s, color 0.2s; }}
        .btn-share:hover {{ border-color: var(--accent); color: var(--accent); }}
        .share-panel {{ display: none; position: absolute; top: 100%; right: 0; margin-top: 0.5rem; background: var(--card); border-radius: var(--radius); box-shadow: var(--shadow-lg); border: 1px solid var(--border); padding: 0.75rem; min-width: 140px; z-index: 200; }}
        .share-panel.share-panel-open {{ display: flex; flex-direction: column; gap: 0.35rem; }}
        .share-panel-label {{ font-size: 0.7rem; font-weight: 600; color: var(--muted); text-transform: uppercase; letter-spacing: 0.05em; }}
        .share-link {{ font-size: 0.9rem; color: var(--text); text-decoration: none; padding: 0.4rem 0.5rem; border-radius: 6px; transition: background 0.15s; }}
        .share-link:hover {{ background: var(--warm); color: var(--accent); }}
        .share-link.share-copy {{ background: none; border: none; cursor: pointer; font-family: inherit; text-align: left; width: 100%; }}
        .share-feedback {{ font-size: 0.8rem; color: var(--accent); }}
        .source-select {{ padding: 0.6rem 1rem; border-radius: var(--radius); border: 1px solid var(--border); background: var(--card); font-size: 0.9rem; font-family: inherit; }}
        .angles-wrap {{ margin-top: 0.5rem; padding-top: 0.75rem; border-top: 1px solid var(--border); }}
        .angles-label {{ font-size: 0.625rem; font-weight: 700; color: var(--muted); text-transform: uppercase; letter-spacing: 0.1em; opacity: 0.88; display: block; margin-bottom: 0.35rem; }}
        .angles-list {{ margin: 0; padding-left: 1.25rem; font-size: 0.85rem; color: var(--accent); line-height: 1.5; }}
        .angles-list li {{ margin-bottom: 0.2rem; }}
        .loader-text {{ font-style: italic; }}
        .loader-spinner {{ display: inline-block; width: 20px; height: 20px; border: 2px solid var(--border); border-top-color: var(--accent); border-radius: 50%; animation: spin 0.8s linear infinite; }}
        .empty-state {{ padding: 2.5rem 1.5rem; text-align: center; }}
        .empty-state-icon {{ font-size: 2rem; margin-bottom: 1rem; opacity: 0.6; }}
        .empty-state-title {{ font-size: 1.15rem; font-weight: 600; color: var(--text); margin-bottom: 0.5rem; }}
        .empty-state-body {{ font-size: 0.95rem; color: var(--muted); margin-bottom: 1.25rem; max-width: 32ch; margin-left: auto; margin-right: auto; }}
        .empty-state-chips {{ display: flex; flex-wrap: wrap; gap: 0.5rem; justify-content: center; }}
        .empty-state-loading {{ text-align: center; padding: 2rem; }}
        .loading-overlay {{ position: fixed; inset: 0; background: rgba(250,248,245,0.92); z-index: 9999; display: flex; align-items: center; justify-content: center; }}
        .loading-overlay-content {{ text-align: center; padding: 2rem; }}
        .loading-overlay-content .loader-spinner {{ margin-bottom: 1rem; }}
        .loading-overlay-content .loading-title {{ font-size: 1.1rem; font-weight: 600; color: var(--text); margin: 0 0 0.35rem 0; }}
        .loading-overlay-content .loading-sub {{ font-size: 0.9rem; color: var(--muted); margin: 0; }}

        @keyframes spin {{ to {{ transform: rotate(360deg); }} }}
        .lifecycle {{ font-size: 0.7rem; font-weight: 600; padding: 0.2rem 0.5rem; border-radius: 4px; text-transform: uppercase; white-space: nowrap; }}
        .lifecycle-rising {{ background: rgba(34,197,94,0.2); color: #15803d; }}
        .lifecycle-peaking {{ background: rgba(232,165,77,0.35); color: var(--cta-text); }}
        .lifecycle-fading {{ background: rgba(107,114,128,0.2); color: #6b7280; }}
        .lifecycle-wrap {{ position: relative; display: inline-block; }}
        .lifecycle-wrap .lifecycle-tooltip {{ position: absolute; left: 50%; transform: translateX(-50%) translateY(6px); bottom: 100%; margin-bottom: 0.5rem; width: 240px; padding: 0.6rem 0.75rem; background: #1e3520; color: #fff; border-radius: 10px; box-shadow: var(--shadow-lg); font-size: 0.62rem; font-weight: 400; text-transform: none; opacity: 0; visibility: hidden; pointer-events: none; transition: opacity 0.2s, transform 0.2s, visibility 0.2s; z-index: 1000; line-height: 1.4; text-align: left; white-space: normal; }}
        .lifecycle-wrap:hover .lifecycle-tooltip {{ opacity: 1; visibility: visible; transform: translateX(-50%) translateY(0); }}
        .badges {{ display: flex; gap: 0.5rem; align-items: center; flex-wrap: wrap; }}
        .site-header {{ display: flex; flex-direction: column; width: 100vw; max-width: 100vw; margin-left: calc(50% - 50vw); margin-right: calc(50% - 50vw); margin-bottom: 1.5rem; background: var(--bg); }}
        .header-divider {{ height: 1px; background: var(--border); width: 100%; flex-shrink: 0; }}
        .header-row {{ display: flex; align-items: center; justify-content: space-between; padding: 0.9rem 2rem; gap: 2rem; max-width: 1100px; margin: 0 auto; width: 100%; box-sizing: border-box; }}
        .header-top-left, .header-top-right, .header-bottom-right {{ flex: 0 0 auto; }}
        .header-top-center {{ flex: 1; display: flex; justify-content: center; }}
        .header-top-right, .header-bottom-right {{ display: flex; align-items: center; gap: 1.25rem; }}
        .btn-lang {{ background: none; border: none; color: var(--text); font-size: 0.9rem; cursor: pointer; font-family: inherit; display: flex; align-items: center; gap: 0.2rem; padding: 0; }}
        .btn-lang:hover {{ color: var(--accent); }}
        .logo-row {{ display: flex; align-items: center; gap: 0.5rem; text-decoration: none; }}
        .logo-row:hover .logo-text {{ color: var(--accent); }}
        .logo-text-wrap {{ display: flex; flex-direction: column; gap: 0.1rem; }}
        .logo-avatar {{ width: 36px; height: 36px; border-radius: 50%; overflow: hidden; flex-shrink: 0; display: flex; align-items: center; justify-content: center; }}
        .logo-text {{ font-size: 1.2rem; font-weight: 700; color: var(--text); letter-spacing: -0.02em; }}
        .sidebar-subtitle {{ font-size: 0.6rem; font-weight: 300; color: var(--muted); line-height: 1.2; display: block; }}
        .logo-avatar img {{ width: 100%; height: 100%; object-fit: cover; }}
        .header-icon {{ color: var(--text); text-decoration: none; font-size: 0.9rem; }}
        .header-icon:hover {{ color: var(--accent); }}
        .header-search-wrap {{ display: flex; align-items: center; flex: 1; max-width: 320px; border: 1px solid var(--border); border-radius: 10px; padding: 0.4rem 0.75rem; gap: 0.5rem; background: var(--card); box-shadow: 0 1px 4px rgba(0,0,0,0.04); }}
        .search-icon {{ color: var(--accent); font-size: 0.9rem; }}
        .header-search {{ flex: 1; border: none; background: none; color: var(--text); font-size: 0.9rem; font-family: inherit; }}
        .header-search::placeholder {{ color: var(--muted); }}
        .header-search:focus {{ outline: none; }}
        .nav-main {{ display: flex; align-items: center; gap: 1.5rem; flex: 1; justify-content: center; }}
        .nav-trigger {{ background: var(--warm); border: 1px solid var(--border); border-radius: 8px; color: var(--text); font-size: 0.9rem; cursor: pointer; font-family: inherit; padding: 0.4rem 0.75rem; display: flex; align-items: center; gap: 0.25rem; }}
        .nav-trigger:hover {{ background: var(--border); }}
        .nav-link {{ color: var(--muted); text-decoration: none; font-size: 0.95rem; }}
        .nav-link:hover, .nav-link.active {{ color: var(--accent); font-weight: 500; }}
        .nav-dropdown {{ position: relative; }}
        .nav-dropdown-panel {{ display: none; position: fixed; top: auto; left: 220px; margin-top: 0.5rem; background: var(--card); border-radius: 12px; box-shadow: 0 10px 40px rgba(0,0,0,0.12); border: 1px solid var(--border); padding: 1rem; min-width: 320px; max-width: calc(100vw - 240px); z-index: 100; }}
        .nav-dropdown:hover .nav-dropdown-panel {{ display: block; }}
        .func-cards {{ display: grid; grid-template-columns: 1fr 1fr; gap: 0.75rem; }}
        .func-card {{ display: block; padding: 1rem; border-radius: 10px; text-decoration: none; color: var(--text); transition: background 0.15s; }}
        .func-card:hover {{ background: var(--bg); }}
        .func-icon {{ font-size: 1.5rem; display: block; margin-bottom: 0.35rem; }}
        .func-title {{ font-weight: 600; font-size: 0.95rem; display: block; margin-bottom: 0.2rem; }}
        .func-desc {{ font-size: 0.8rem; color: var(--muted); line-height: 1.4; }}
        .header-utils {{ display: flex; align-items: center; gap: 0.5rem; margin-left: auto; }}
        .btn-util {{ padding: 0.4rem 0.75rem; border-radius: 8px; font-size: 0.85rem; cursor: pointer; font-family: inherit; }}
        .btn-outline {{ border: 1px solid var(--border); background: var(--card); color: var(--text); }}
        .btn-outline:hover {{ border-color: var(--accent); color: var(--accent); }}
        .chevron {{ font-size: 0.7em; opacity: 0.8; }}
        .hero {{ display: grid; grid-template-columns: 1fr 1fr; gap: 3rem; align-items: center; margin-bottom: 3rem; padding: 2.5rem 0; min-height: 300px; position: relative; border-radius: 24px; }}
        .hero::before {{ content: ""; position: absolute; inset: 0; background: repeating-linear-gradient(transparent, transparent 29px, rgba(45,74,43,0.08) 29px, rgba(45,74,43,0.08) 30px); opacity: 0; transition: opacity 0.4s var(--ease-out); pointer-events: none; border-radius: 24px; }}
        .hero:hover::before {{ opacity: 1; }}
        .hero-left {{ max-width: 420px; position: relative; z-index: 1; }}
        .hero-greeting {{ font-size: 0.75rem; font-weight: 300; color: var(--muted); margin: 0 0 0.6rem 0; }}
        .hero-headline {{ font-size: 2.25rem; font-weight: 700; color: var(--accent); line-height: 1.2; margin: 0 0 0.5rem 0; letter-spacing: -0.02em; }}
        .hero-headline span {{ display: block; }}
        .hero-subtitle {{ font-size: 0.95rem; font-weight: 600; color: var(--accent); letter-spacing: 0.04em; text-transform: uppercase; margin: 0 0 1rem 0; opacity: 0.85; }}
        .page-header h1 {{ color: var(--ink); font-size: 1.5rem; font-weight: 800; }}
        .hero-desc {{ color: var(--muted); font-size: 1.05rem; line-height: 1.65; margin-bottom: 1.5rem; font-weight: 400; }}
        .hero-cta {{ display: inline-flex; align-items: center; gap: 0.5rem; padding: 0.85rem 1.75rem; border-radius: var(--radius); background: var(--cta); color: var(--cta-text); font-weight: 700; text-decoration: none; font-size: 0.95rem; font-family: inherit; border: none; box-shadow: 0 4px 14px rgba(232,165,77,0.35); transition: transform var(--dur) var(--ease-out), background var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out); }}
        .hero-cta:hover {{ transform: translateY(-2px); background: #e0963f; box-shadow: 0 8px 22px rgba(232,165,77,0.4); }}
        .page-links {{ display: grid; grid-template-columns: repeat(auto-fill, minmax(200px, 1fr)); gap: 1.25rem; margin-top: 2.5rem; }}
        .page-link-card {{ display: flex; flex-direction: column; gap: 0.35rem; padding: 1.25rem 1.5rem; background: var(--card); border-radius: var(--radius); border: 1px solid var(--border); text-decoration: none; color: var(--text); transition: transform var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out); box-shadow: var(--shadow); }}
        .page-link-card:hover {{ transform: translateY(-2px); border-color: var(--accent-light); box-shadow: var(--shadow-lg); }}
        .page-link-icon {{ font-size: 1.75rem; }}
        .page-link-card strong {{ font-size: 1.05rem; font-weight: 600; letter-spacing: -0.01em; line-height: 1.35; }}
        .page-link-card small {{ font-size: 0.82rem; color: var(--muted); font-weight: 400; line-height: 1.45; }}
        .hero-visual {{ display: flex; align-items: center; justify-content: center; position: relative; z-index: 1; padding: 2rem; }}
        .hero-thumbnail-wrap {{ background: var(--accent); border-radius: 45% 55% 52% 48% / 55% 45% 55% 45%; overflow: hidden; display: flex; align-items: center; justify-content: center; width: 300px; height: 300px; transition: transform 0.4s var(--ease-out), box-shadow 0.4s var(--ease-out); }}
        .hero:hover .hero-thumbnail-wrap {{ transform: scale(1.05); box-shadow: 0 12px 40px rgba(45,74,43,0.2); }}
        .hero-thumbnail {{ width: 100%; height: 100%; object-fit: cover; transition: transform 0.5s var(--ease-out); }}
        .hero:hover .hero-thumbnail {{ transform: scale(1.08); }}
        .hero-bg {{ background: transparent; }}
        .science-hero {{ margin-bottom: 2.5rem; }}
        .science-hero h2 {{ font-size: 1.85rem; font-weight: 700; color: var(--accent); margin: 0 0 0.5rem 0; letter-spacing: -0.02em; }}
        .science-hero h2 em {{ font-style: italic; color: var(--accent); }}
        .science-hero .tagline {{ font-size: 1.1rem; color: var(--muted); line-height: 1.6; margin: 0; max-width: 42ch; }}
        .berger-bio {{ background: var(--card); border-radius: var(--radius); padding: 1.75rem 2rem; border: 1px solid var(--border); box-shadow: var(--shadow); margin-bottom: 2.5rem; }}
        .berger-bio h3 {{ font-size: 1rem; font-weight: 600; color: var(--muted); text-transform: uppercase; letter-spacing: 0.05em; margin: 0 0 0.75rem 0; }}
        .berger-bio p {{ font-size: 0.95rem; color: var(--text); line-height: 1.65; margin: 0; }}
        .berger-bio a {{ color: var(--accent); text-decoration: none; }}
        .berger-bio a:hover {{ text-decoration: underline; }}
        .books-grid {{ display: grid; grid-template-columns: repeat(4, 1fr); gap: 1.25rem; margin-bottom: 2.5rem; }}
        .book-card {{ background: var(--card); border-radius: var(--radius); padding: 1.25rem; border: 1px solid var(--border); box-shadow: var(--shadow); transition: transform var(--dur) var(--ease-out), box-shadow var(--dur) var(--ease-out), border-color var(--dur) var(--ease-out); text-decoration: none; color: var(--text); display: flex; flex-direction: column; align-items: center; text-align: center; }}
        .book-card:hover {{ transform: translateY(-2px); box-shadow: var(--shadow-lg); border-color: var(--accent-light); }}
        .book-cover {{ width: 100%; max-width: 120px; aspect-ratio: 2/3; object-fit: cover; border-radius: 8px; box-shadow: 0 4px 12px rgba(0,0,0,0.15); margin-bottom: 0.85rem; }}
        .book-cover-placeholder {{ width: 100%; max-width: 120px; aspect-ratio: 2/3; background: var(--warm); border-radius: 8px; margin-bottom: 0.85rem; display: none; align-items: center; justify-content: center; font-size: 1.5rem; font-weight: 600; color: var(--muted); }}
        .book-title {{ font-size: 0.9rem; font-weight: 600; line-height: 1.35; margin: 0 0 0.25rem 0; }}
        .book-year {{ font-size: 0.75rem; color: var(--muted); }}
        .mechanism-box {{ background: linear-gradient(135deg, rgba(45,74,43,0.06) 0%, rgba(232,165,77,0.08) 100%); border-radius: var(--radius); padding: 1.75rem 2rem; border: 1px solid var(--border); margin-bottom: 2.5rem; }}
        .mechanism-box h3 {{ font-size: 1rem; font-weight: 600; color: var(--accent); margin: 0 0 1rem 0; letter-spacing: -0.01em; }}
        .mechanism-box p {{ font-size: 0.95rem; color: var(--text); line-height: 1.65; margin: 0 0 0.75rem 0; }}
        .mechanism-box p:last-child {{ margin-bottom: 0; }}
        .mechanism-box strong {{ color: var(--accent); }}
        .score-formula {{ font-size: 0.9rem; background: var(--card); padding: 1rem 1.25rem; border-radius: var(--radius); border: 1px solid var(--border); margin-top: 1rem; font-family: ui-monospace, monospace; color: var(--muted); }}
        @media (max-width: 900px) {{ .books-grid {{ grid-template-columns: repeat(2, 1fr); }} }}
        @media (max-width: 500px) {{ .books-grid {{ grid-template-columns: 1fr; }} }}
        .onboarding-backdrop {{ position: fixed; inset: 0; background: rgba(45,74,43,0.75); backdrop-filter: blur(4px); z-index: 9999; display: flex; align-items: center; justify-content: center; padding: 1.5rem; }}
        .onboarding-modal {{ background: var(--card); max-width: 480px; width: 100%; border-radius: 16px; overflow: hidden; box-shadow: 0 24px 48px rgba(0,0,0,0.2); }}
        .onboarding-progress {{ height: 3px; background: var(--border); }}
        .onboarding-progress-fill {{ height: 100%; background: var(--accent); transition: width 0.3s ease; width: 33.33%; }}
        .onboarding-body {{ padding: 2rem 2rem 1.5rem; }}
        .onboarding-step-label {{ font-size: 0.7rem; letter-spacing: 0.1em; text-transform: uppercase; color: var(--muted); margin-bottom: 0.5rem; }}
        .onboarding-title {{ font-size: 1.35rem; font-weight: 700; color: var(--text); margin-bottom: 0.75rem; }}
        .onboarding-desc {{ font-size: 0.95rem; color: var(--muted); line-height: 1.6; margin: 0; }}
        .onboarding-footer {{ display: flex; justify-content: space-between; align-items: center; padding: 1rem 2rem 1.5rem; border-top: 1px solid var(--border); }}
        .btn-onboard-skip {{ background: none; border: none; color: var(--muted); font-size: 0.9rem; cursor: pointer; }}
        .btn-onboard-skip:hover {{ color: var(--text); }}
        .btn-onboard-next {{ background: var(--accent); color: var(--card); border: none; padding: 0.6rem 1.25rem; border-radius: 8px; font-weight: 600; font-size: 0.9rem; cursor: pointer; }}
        .btn-onboard-next:hover {{ background: var(--accent-hover); }}
        .reveal {{ opacity: 1; transform: none; transition: opacity 0.65s ease, transform 0.65s ease; }}
        .reveal.visible {{ opacity: 1; transform: none; }}
        @media (max-width: 900px) {{ .daily-cards-grid {{ grid-template-columns: repeat(2, 1fr); }} }}
        @media (max-width: 768px) {{ .app-layout {{ flex-direction: column; }} .sidebar {{ width: 100%; height: auto; position: relative; flex-direction: row; flex-wrap: wrap; }} .sidebar-nav {{ flex-direction: row; flex-wrap: wrap; overflow: visible; }} .sidebar-link {{ flex: 1 1 auto; }} .main-content {{ margin-left: 0; }} .stats {{ grid-template-columns: 1fr; }} .cards {{ grid-template-columns: 1fr; }} .daily-cards-grid {{ grid-template-columns: 1fr; }} footer {{ left: 0; }} }}
        @media (max-width: 640px) {{ .hero {{ grid-template-columns: 1fr; }} .hero-visual {{ order: -1; }} .hero-thumbnail-wrap {{ width: 240px; height: 240px; }} .form {{ flex-wrap: wrap; }} }}
    </style>
</head>
<body>
    <div class="app-layout">
        {_sidebar_html(active, lang)}
        <main class="main-content">
        {_header_html(active, lang)}
        {hero if hero else f'<header class="page-header"><h1>ViralLab</h1><p class="tagline">{_html_escape(_t("tagline", lang))}</p></header>'}
        {body}
        <!-- Footer: position:fixed at bottom of viewport, aligned with sidebar-footer. Do not change. -->
        <footer>
            <p class="spread">{_html_escape(_t("camp2_footer", lang))}</p>
            <p>{_html_escape(_t("camp2_footer_built", lang))} <a href="{TG_PERSONAL_URL}" target="_blank" rel="noopener">@miccakitt</a> {_html_escape(_t("camp2_footer_welcome", lang))}</p>
        </footer>
        </main>
    </div>
    <div id="loadingOverlay" class="loading-overlay" style="display:none" data-msg-search="{_html_escape(_t("loading_searching", lang))}" data-msg-search-sub="{_html_escape(_t("loading_search_sub", lang))}" data-msg-refresh="{_html_escape(_t("loading_refreshing", lang))}" data-msg-refresh-sub="{_html_escape(_t("loading_refresh_sub", lang))}">
        <div class="loading-overlay-content">
            <span class="loader-spinner"></span>
            <p class="loading-title" id="loadingTitle"></p>
            <p class="loading-sub" id="loadingSub"></p>
        </div>
    </div>
        <script>
        (function loadingOverlay() {{
            const el = document.getElementById('loadingOverlay');
            if (!el) return;
            const title = document.getElementById('loadingTitle');
            const sub = el.querySelector('.loading-sub');
            function show(mode) {{
                if (mode === 'search') {{ title.textContent = el.getAttribute('data-msg-search'); sub.textContent = el.getAttribute('data-msg-search-sub'); }}
                else {{ title.textContent = el.getAttribute('data-msg-refresh'); sub.textContent = el.getAttribute('data-msg-refresh-sub'); }}
                el.style.display = 'flex';
            }}
            document.querySelectorAll('form[action*="search-news"]').forEach(function(f) {{ f.addEventListener('submit', function() {{ show('search'); }}); }});
            document.querySelectorAll('a.refresh-link').forEach(function(a) {{
                const href = a.getAttribute('href') || '';
                if (!/\\/(news|daily)\\/refresh/.test(href) && !/refresh_hot=1/.test(href)) return;
                a.addEventListener('click', function() {{ show('refresh'); }});
            }});
        }})();
        document.getElementById('headerSearch')?.addEventListener('keydown', function(e) {{
            if (e.key === 'Enter') {{
                e.preventDefault();
                const q = this.value.trim() || 'viral trending';
                const viralInput = document.getElementById('viralQuery');
                const viralForm = document.getElementById('viralSearchForm');
                if (viralInput && viralForm) {{ viralInput.value = q; viralForm.dispatchEvent(new Event('submit')); }}
                else {{ window.location = '/viral?q=' + encodeURIComponent(q); }}
            }}
        }});
        (function scrollReveal() {{
            const reveals = document.querySelectorAll('.reveal');
            const obs = new IntersectionObserver(function(entries) {{
                entries.forEach(function(e) {{
                    if (e.isIntersecting) {{ e.target.classList.add('visible'); obs.unobserve(e.target); }}
                }});
            }}, {{ threshold: 0.1 }});
            reveals.forEach(function(el) {{ obs.observe(el); }});
        }})();
        (function() {{
            const d = new Date();
            const y = d.getFullYear(), m = String(d.getMonth()+1).padStart(2,'0'), day = String(d.getDate()).padStart(2,'0');
            const locale = document.documentElement.lang === 'zh' ? 'zh-TW' : 'en-US';
            const wd = d.toLocaleDateString(locale, {{ weekday: 'short' }});
            const el = document.getElementById('dailyNewsDate');
            if (el) el.textContent = y + '-' + m + '-' + day + ' ' + wd;
        }})();
        (function onboarding() {{
            if (localStorage.getItem('virallab-onboarding-done')) return;
            const steps = [
                {{ title: 'Pick your field.', desc: 'ViralLab tailors trends and scores to your niche. Tell us who you create for.' }},
                {{ title: 'Find what\'s spreading.', desc: 'The Viral page surfaces top videos in your niche — scored with Berger\'s STEPPS. See why each piece spreads.' }},
                {{ title: 'Score your content.', desc: 'Paste any YouTube URL into Video-to-Text. Get transcript + Berger score + fix suggestions.' }}
            ];
            let step = 0;
            const overlay = document.createElement('div');
            overlay.id = 'onboardingOverlay';
            overlay.innerHTML = '<div class="onboarding-backdrop"><div class="onboarding-modal"><div class="onboarding-progress"><div class="onboarding-progress-fill"></div></div><div class="onboarding-body"><div class="onboarding-step-label">Step ' + (step+1) + ' of 3</div><div class="onboarding-title"></div><p class="onboarding-desc"></p></div><div class="onboarding-footer"><button type="button" class="btn-onboard-skip">Skip</button><button type="button" class="btn-onboard-next">Next →</button></div></div></div>';
            overlay.querySelector('.onboarding-title').textContent = steps[0].title;
            overlay.querySelector('.onboarding-desc').textContent = steps[0].desc;
            const progressFill = overlay.querySelector('.onboarding-progress-fill');
            const titleEl = overlay.querySelector('.onboarding-title');
            const descEl = overlay.querySelector('.onboarding-desc');
            const stepLabel = overlay.querySelector('.onboarding-step-label');
            const nextBtn = overlay.querySelector('.btn-onboard-next');
            function updateStep() {{
                stepLabel.textContent = 'Step ' + (step+1) + ' of 3';
                titleEl.textContent = steps[step].title;
                descEl.textContent = steps[step].desc;
                progressFill.style.width = ((step+1)/3*100) + '%';
                nextBtn.textContent = step === 2 ? 'Start scoring →' : 'Next →';
            }}
            function done() {{ overlay.remove(); localStorage.setItem('virallab-onboarding-done', '1'); }}
            overlay.querySelector('.btn-onboard-skip').onclick = done;
            nextBtn.onclick = function() {{ if (step < 2) {{ step++; updateStep(); }} else done(); }};
            overlay.querySelector('.onboarding-backdrop').onclick = function(e) {{ if (e.target === this) done(); }};
            document.body.appendChild(overlay);
        }})();
        </script>
    </div>
<script>
/* Dev auto-refresh: only active on 127.0.0.1 */
if (location.hostname === '127.0.0.1') {{
    var _devDown = false;
    setInterval(function() {{
        fetch(location.href, {{cache: 'no-store', method: 'HEAD'}})
            .then(function() {{ if (_devDown) {{ _devDown = false; location.reload(); }} }})
            .catch(function() {{ _devDown = true; }});
    }}, 1500);
}}
</script>
</body>
</html>"""


def _list_digests():
    if not OUTPUT.exists():
        return []
    files = []
    for f in OUTPUT.glob("*.md"):
        name = f.stem
        if name == "daily_news":
            files.append({"type": "daily_news", "file": f.name})
        elif name == "daily_news_zh":
            files.append({"type": "daily_news_zh", "file": f.name})
        elif name.startswith("digest_"):
            files.append({"type": "digest", "topic": name[7:], "file": f.name})
        elif name.startswith("raw_"):
            files.append({"type": "raw", "topic": name[4:], "file": f.name})
        elif name.startswith("videos_"):
            files.append({"type": "videos", "query": name[7:], "file": f.name})
        elif name.startswith("transcript_"):
            files.append({"type": "transcript", "video_id": name[10:], "file": f.name})
    return sorted(files, key=lambda x: x["file"])


@app.route("/api/digests")
def list_digests():
    return jsonify(_list_digests())


@app.route("/api/digests/<path:name>")
def get_digest(name):
    if ".." in name or "/" in name:
        return jsonify({"error": "Invalid path"}), 400
    path = OUTPUT / name
    if not path.exists():
        return jsonify({"error": "Not found"}), 404
    return send_from_directory(OUTPUT, name, mimetype="text/markdown")


@app.route("/api/export/<path:name>")
def api_export(name):
    """Export digest as Markdown, Notion, or Obsidian format."""
    if ".." in name or "/" in name:
        return jsonify({"error": "Invalid path"}), 400
    path = OUTPUT / name
    if not path.exists():
        return jsonify({"error": "Not found"}), 404
    fmt = request.args.get("format", "markdown").lower()
    content = path.read_text(encoding="utf-8", errors="replace")
    if fmt == "obsidian":
        content = "---\ntags: [virallab, trends]\n---\n\n" + content
    elif fmt == "notion":
        content = f"# ViralLab export\n\n*Exported for Notion*\n\n---\n\n" + content
    filename = path.stem + ".md"
    from flask import Response
    return Response(content, mimetype="text/markdown", headers={
        "Content-Disposition": f"attachment; filename={filename}",
    })


@app.route("/setup")
def view_setup():
    """Serve SETUP.md for API configuration instructions."""
    path = Path(__file__).parent / "SETUP.md"
    if not path.exists():
        return "SETUP.md not found", 404
    body = path.read_text(encoding="utf-8", errors="replace")
    try:
        from markdown import markdown
        html_body = markdown(body, extensions=["fenced_code", "tables"])
    except ImportError:
        html_body = f"<pre>{_html_escape(body)}</pre>"
    return f"""<!DOCTYPE html>
<html><head><meta charset="utf-8"><title>ViralLab Setup</title>
<link href="https://fonts.googleapis.com/css2?family=Poppins:wght@400;500;600&display=swap" rel="stylesheet">
<style>body{{font-family:Poppins,sans-serif;max-width:720px;margin:2rem auto;padding:1.5rem;line-height:1.6;color:#2d4a2b}}
a{{color:#2d4a2b}} code{{background:#f5f2ed;padding:0.2rem 0.4rem;border-radius:4px;font-size:0.9em}}
pre{{background:#f5f2ed;padding:1rem;border-radius:8px;overflow-x:auto;max-width:100%;box-sizing:border-box;white-space:pre-wrap;word-break:break-all}}</style>
</head><body>{html_body}</body></html>""", 200, {"Content-Type": "text/html; charset=utf-8"}


def _render_full_digest(filename: str):
    """Render daily_news.md or daily_news_zh.md as styled Buffer-style cards in main app layout."""
    _maybe_trigger_daily_refresh_if_stale()
    lang = "zh" if "zh" in filename else "en"
    path = OUTPUT / filename
    if not path.exists():
        return None
    _, parsed = parse_file(path)
    run_type = "daily_news_zh" if lang == "zh" else "daily_news"
    parsed = add_lifecycle_to_items(run_type, parsed[:60])

    updated_at = ""
    if (OUTPUT / "daily_news_updated.txt").exists():
        updated_at = (OUTPUT / "daily_news_updated.txt").read_text(encoding="utf-8").strip()

    INITIAL_SHOW = 10
    cards_html = ""
    for i, item in enumerate(parsed, 1):
        snip = item.get("snippet", "") or ""
        c_score = evaluate_contagious_article(item.get("title", ""), snip, lang=lang)
        c_total = c_score.get("total", 0)
        c_top = " + ".join(c_score.get("top_signals", []))
        c_badge = (
            f'<span class="berger {_score_badge_class(c_total)}" title="{_html_escape(c_top)}">{_html_escape(_t("contagious_score", lang))} {c_total}</span>'
            if c_total > 0
            else ""
        )
        angles = generate_angles("", item.get("title", ""), snip, count=2, lang=lang)
        pills = "".join(f'<span class="pill-tag">{_html_escape(a)}</span>' for a in angles)
        life = item.get("lifecycle", "peaking")
        life_label = _t(f"lifecycle_{life}", lang) if life in ("rising", "peaking", "fading") else life
        life_tt = _lifecycle_tooltip_html(lang, life)
        hidden_class = " load-more-hidden" if i > INITIAL_SHOW else ""
        src = _get_source_for_item(item)
        src_type = _source_type_for_item(item)
        snip_html = f'<p class="digest-card-snippet">{_html_escape(_sanitize_snippet(snip))}</p>' if snip else ""
        src_html = f'<p class="digest-card-source">Source: {_html_escape(src)}</p>' if src else ""
        cards_html += f'''<div class="digest-card{hidden_class}" data-digest-index="{i}" data-source-type="{_html_escape(src_type)}">
            <div class="digest-card-header">
                <span class="digest-card-num">{i}</span>
                <h3 class="digest-card-title"><a href="{_html_escape(item["url"])}" target="_blank" rel="noopener">{_html_escape(item["title"])}</a></h3>
                <div class="digest-card-badges">
                    <span class="lifecycle-wrap"><span class="lifecycle lifecycle-{life}">{_html_escape(life_label)}</span><span class="lifecycle-tooltip">{life_tt}</span></span>
                    {c_badge}
                </div>
            </div>
            {snip_html}
            {src_html}
            <div class="digest-card-pills">{pills}</div>
        </div>'''

    load_more_btn = ""
    if len(parsed) > INITIAL_SHOW:
        load_more_btn = f'''
        <button type="button" class="btn-load-more" id="digestLoadMore">{_t("load_more", lang)}</button>
        <script>
        (function() {{
            var btn = document.getElementById("digestLoadMore");
            var hidden = document.querySelectorAll(".digest-card.load-more-hidden");
            if (!btn || hidden.length === 0) return;
            btn.addEventListener("click", function() {{
                var toShow = 50;
                for (var i = 0; i < toShow && hidden.length > 0; i++) {{
                    hidden[0].classList.remove("load-more-hidden");
                    hidden = document.querySelectorAll(".digest-card.load-more-hidden");
                }}
                if (hidden.length === 0) btn.style.display = "none";
            }});
        }})();
        </script>'''

    title_t = _t("daily_news_title", lang)
    back_link = f'<p class="browse-more"><a href="/daily">← {_t("nav_daily", lang)}</a></p>'
    export_links = f'<p class="sources-label">Export: <a href="/api/export/{filename}?format=markdown">Markdown</a> · <a href="/api/export/{filename}?format=notion">Notion</a> · <a href="/api/export/{filename}?format=obsidian">Obsidian</a></p>'
    refresh_url = f"/daily/refresh?next=/view/{filename}"
    refresh_options = f'<p class="sources-label"><a href="{refresh_url}" class="refresh-link">{_t("refresh_now", lang)}</a> — {_t("refresh_desc", lang)}</p>'
    filter_bar = f'''
    <div class="tip-card" style="margin-top:0.75rem;">
        <div class="tip-header">
            <span class="tip-label">{_html_escape(_t("daily_focus_label", lang))}</span>
            <span class="tip-range-wrap">{_html_escape(_t("daily_focus_hint", lang))}</span>
        </div>
        <div class="tip-chips">
            <button type="button" class="tip-chip tip-chip-active digest-filter-btn" data-filter="all">{_html_escape(_t("daily_focus_all", lang))}</button>
            <button type="button" class="tip-chip digest-filter-btn" data-filter="news">{_html_escape(_t("daily_focus_news", lang))}</button>
            <button type="button" class="tip-chip digest-filter-btn" data-filter="longform">{_html_escape(_t("daily_focus_longform", lang))}</button>
            <button type="button" class="tip-chip digest-filter-btn" data-filter="podcast">{_html_escape(_t("daily_focus_podcast", lang))}</button>
        </div>
    </div>'''

    body = f"""
    <section>
        <div class="section-title">📅 {title_t} · {_t("full_digest_title", lang)}</div>
        <div class="content-meta">
            {refresh_options}
            <p class="sources-label">{_t("last_updated", lang)}: {_html_escape(updated_at)}</p>
            {export_links}
        </div>
        {filter_bar}
        <div class="digest-top-section">
            {back_link}
        </div>
        <div class="digest-cards">{cards_html}</div>
        {load_more_btn}
    </section>
    <script>
    (function() {{
        var buttons = document.querySelectorAll(".digest-filter-btn");
        var cards = document.querySelectorAll(".digest-card");
        if (!buttons.length || !cards.length) return;
        function applyFilter(filter) {{
            cards.forEach(function(card) {{
                var t = card.getAttribute("data-source-type") || "news";
                card.style.display = (filter === "all" || t === filter) ? "" : "none";
            }});
            buttons.forEach(function(btn) {{
                btn.classList.toggle("tip-chip-active", btn.getAttribute("data-filter") === filter);
            }});
        }}
        buttons.forEach(function(btn) {{
            btn.addEventListener("click", function() {{
                applyFilter(btn.getAttribute("data-filter") || "all");
            }});
        }});
    }})();
    </script>
    """
    return _base(title_t, body, "/daily", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/view/<path:name>")
def view_digest(name):
    if ".." in name or "/" in name:
        return "Invalid path", 400
    path = OUTPUT / name
    if not path.exists():
        return "Not found", 404

    # Styled full digest for daily news
    if name in ("daily_news.md", "daily_news_zh.md"):
        result = _render_full_digest(name)
        if result:
            return result

    body = path.read_text(encoding="utf-8", errors="replace")
    escaped = _html_escape(body)
    export_links = f'''
    <p style="margin-bottom:1rem;font-size:0.9rem;">
        Export: <a href="/api/export/{_html_escape(name)}?format=markdown">Markdown</a> ·
        <a href="/api/export/{_html_escape(name)}?format=notion">Notion</a> ·
        <a href="/api/export/{_html_escape(name)}?format=obsidian">Obsidian</a>
    </p>
    '''
    html = f"""<!DOCTYPE html>
<html><head><meta charset="utf-8"><title>{_html_escape(name)}</title>
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600&display=swap" rel="stylesheet">
</head>
<body style="font-family:Inter,sans-serif;max-width:720px;margin:2rem auto;padding:1.5rem;white-space:pre-wrap;background:#fafafa;color:#1a1a1a;line-height:1.6">
{export_links}
{escaped}
</body></html>"""
    return html, 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/api/video-to-text", methods=["POST"])
def api_video_to_text():
    req_json = request.get_json(silent=True) or {}
    url = (request.form.get("url") or req_json.get("url") or "").strip()
    if not url:
        return jsonify({"error": "Missing url"}), 400
    transcript, caption_source, fail_detail = _fetch_transcript_from_input(url, _get_lang())
    if not transcript:
        hint = (
            "ASR unavailable: install `videocaptioner` and/or optional WhisperX "
            "(`pip install -r requirements-asr.txt`; see README)."
        )
        payload = {"error": "Transcript unavailable", "hint": hint}
        if fail_detail:
            payload["detail"] = fail_detail[:24000]
        return jsonify(payload), 400

    score = score_berger(transcript)
    lang = _get_lang()
    stepps = _score_to_stepps(score.get("breakdown", {}), lang)
    title = (transcript[:80] + "…") if len(transcript) > 80 else transcript
    md = _build_markdown_export(
        url=url,
        title=title,
        score=score["total"],
        stepps=stepps,
        transcript=transcript,
        magic_words=score.get("magic_words_found", []),
        lang=lang,
        include_minto=True,
    )
    OUTPUT.mkdir(exist_ok=True)
    out_id = str(abs(hash(url)))[:12]
    out_file = OUTPUT / f"transcript_{out_id}.md"
    out_file.write_text(md, encoding="utf-8")
    return jsonify({
        "file": out_file.name,
        "url": f"/view/{out_file.name}",
        "score": score["total"],
        "caption_source": caption_source,
    })


@app.route("/api/image-ocr", methods=["POST"])
def api_image_ocr():
    """OCR images from URLs (direct image or public HTML page) or multipart uploads.

    JSON body: ``url`` (single), ``urls`` (list), optional ``mode`` (auto | image | page).
    Form: same fields plus file field ``image`` / ``images`` / ``file``.
    """
    req_json = request.get_json(silent=True) or {}
    mode = (req_json.get("mode") or request.form.get("mode") or "auto").strip()
    single = (req_json.get("url") or request.form.get("url") or "").strip()
    urls_raw = req_json.get("urls")
    urls: list[str] = []
    if isinstance(urls_raw, list):
        urls = [str(u).strip() for u in urls_raw if str(u).strip()]
    elif isinstance(urls_raw, str) and urls_raw.strip():
        urls = [p.strip() for p in urls_raw.replace(",", " ").split() if p.strip()]
    if single:
        urls.insert(0, single)

    files = list(request.files.getlist("image")) + list(request.files.getlist("images"))
    f_one = request.files.get("file")
    if f_one and f_one.filename:
        files.append(f_one)

    if not urls and not files:
        return jsonify({"ok": False, "error": "Provide url, urls, or an image file"}), 400

    results: list[dict] = []
    text_chunks: list[str] = []

    for u in urls:
        job = ocr_job_for_url(u, mode=mode)
        results.append({"input": u, **job})
        if job.get("mode") == "page" and job.get("results"):
            for sub in job["results"]:
                t = (sub or {}).get("text") or ""
                if t:
                    text_chunks.append(t)
        elif job.get("results"):
            for sub in job["results"]:
                t = (sub or {}).get("text") or ""
                if t:
                    text_chunks.append(t)

    for f in files:
        if not f.filename:
            continue
        data = f.read()
        lines, text = ocr_image_bytes(data)
        ok = bool(text or lines)
        results.append({
            "input": f.filename,
            "ok": ok,
            "mode": "upload",
            "text": text,
            "lines": lines,
            "error": None if ok else "no_text_or_unreadable",
        })
        if text:
            text_chunks.append(text)

    return jsonify({
        "ok": True,
        "results": results,
        "text_combined": "\n\n".join(text_chunks).strip(),
    })


def _youtube_download_filename_ok(name: str) -> bool:
    """Allow only files we create: ``youtube_{11-char id}.{ext}``."""
    if not name or "/" in name or ".." in name:
        return False
    if not name.startswith("youtube_") or "." not in name:
        return False
    rest = name[8:]  # after "youtube_"
    vid_part, ext = rest.rsplit(".", 1)
    if len(vid_part) != 11 or not ext.isalnum() or not (2 <= len(ext) <= 8):
        return False
    return all(c.isalnum() or c in "_-" for c in vid_part)


@app.route("/api/youtube-media/<name>")
def api_youtube_media_download(name):
    if not _youtube_download_filename_ok(name):
        return jsonify({"error": "Invalid file"}), 400
    try:
        out_dir = OUTPUT.resolve()
        path = (OUTPUT / name).resolve()
    except OSError:
        return jsonify({"error": "Invalid path"}), 400
    if path.parent != out_dir:
        return jsonify({"error": "Invalid path"}), 400
    if not path.is_file():
        return jsonify({"error": "Not found"}), 404
    return send_file(path, as_attachment=True, download_name=name)


@app.route("/api/youtube-download", methods=["POST"])
def api_youtube_download():
    req_json = request.get_json(silent=True) or {}
    url = (request.form.get("url") or req_json.get("url") or "").strip()
    lang = _get_lang()
    if not url:
        return jsonify({"error": _t("youtube_dl_missing_url", lang)}), 400
    OUTPUT.mkdir(exist_ok=True)
    # Werkzeug access-log lines appear after the response; yt-dlp can run minutes.
    print(f"[youtube-download] start {url[:200]!r}", flush=True)
    path, err = download_youtube_video(url, OUTPUT)
    if not path:
        if err == "no_ytdlp":
            return jsonify({"error": _t("youtube_dl_need_ytdlp", lang)}), 400
        if err == "not_youtube":
            return jsonify({"error": _t("youtube_dl_not_youtube", lang)}), 400
        if err == "missing_url":
            return jsonify({"error": _t("youtube_dl_missing_url", lang)}), 400
        detail = err[:800] if err else ""
        print(f"[youtube-download] failed {err[:120]!r}", flush=True)
        return jsonify({"error": _t("youtube_dl_failed", lang), "detail": detail}), 502
    print(f"[youtube-download] ok {path.name}", flush=True)
    return jsonify({"file": path.name, "url": f"/api/youtube-media/{path.name}"})


@app.route("/china-access")
def china_access():
    """China access guide (VPN, sources)."""
    path = Path(__file__).parent / "CHINA_ACCESS.md"
    if not path.exists():
        return "China access guide not found", 404
    body = path.read_text(encoding="utf-8", errors="replace")
    escaped = _html_escape(body)
    content = f'<section><pre style="white-space:pre-wrap;font-size:0.95rem;">{escaped}</pre></section>'
    return _base("China Access", content, ""), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/api/china-platforms")
def api_china_platforms():
    """China platform search links."""
    from src.china_sources import CHINA_PLATFORM_LINKS, get_china_search_url
    query = request.args.get("q", "viral trending").strip()
    return jsonify({
        "query": query,
        "platforms": [
            {"id": k, "name": v["name"], "url": get_china_search_url(k, query), "desc": v["desc"]}
            for k, v in CHINA_PLATFORM_LINKS.items()
        ],
    })


@app.route("/api")
def api_info():
    return jsonify({
        "name": "ViralLab API",
        "version": "1.0",
        "tagline": TAGLINE,
        "endpoints": {
            "digests": "/api/digests",
            "get": "/api/digests/<filename>",
            "export": "/api/export/<filename>?format=markdown|notion|obsidian",
            "video_to_text": "POST /api/video-to-text",
            "youtube_download": "POST /api/youtube-download (JSON or form: url)",
            "viral_videos": "GET /api/viral-videos?q=<query>&source=global|china",
            "health": "/api/health",
            "refresh_daily": "GET /api/refresh-daily (for cron)",
            "refresh_videos": "GET /api/refresh-videos (for cron)",
        },
    })


@app.route("/api/health")
def health():
    return jsonify({"status": "ok", "project": "virallab"})


def _run_refresh_daily():
    """Run daily news refresh (subprocess). Used by API and scheduler."""
    script = Path(__file__).parent / "scripts" / "daily_news.py"
    env = _env_no_proxy()
    env["PYTHONPATH"] = str(Path(__file__).parent)
    # Daily EN+ZH source aggregation can exceed 60s when multiple sources are slow.
    timeout_sec = int(os.environ.get("DAILY_REFRESH_TIMEOUT_SEC", "420"))
    proc = subprocess.Popen(
        [sys.executable, str(script)],
        stdout=subprocess.PIPE,
        stderr=subprocess.PIPE,
        env=env,
        cwd=Path(__file__).parent,
    )
    try:
        stdout, stderr = proc.communicate(timeout=timeout_sec)
    except subprocess.TimeoutExpired:
        proc.kill()
        print(f"[daily_refresh] timed out after {timeout_sec}s")
        raise
    if proc.returncode != 0:
        err = (stderr or b"").decode("utf-8", errors="ignore").strip()
        out = (stdout or b"").decode("utf-8", errors="ignore").strip()
        if err:
            print(f"[daily_refresh] failed: {err}")
        elif out:
            print(f"[daily_refresh] failed: {out}")
        else:
            print(f"[daily_refresh] failed with code {proc.returncode}")
        return False
    return True


def _daily_digest_age_hours() -> float | None:
    """Hours since ``daily_news_updated.txt`` was written, or None if missing/unparseable."""
    from datetime import datetime, timezone

    p = OUTPUT / "daily_news_updated.txt"
    if not p.exists():
        return None
    s = p.read_text(encoding="utf-8").strip()
    if not s:
        return None
    s = s.replace(" UTC", "").strip()
    try:
        dt = datetime.strptime(s, "%Y-%m-%d %H:%M").replace(tzinfo=timezone.utc)
        return (datetime.now(timezone.utc) - dt).total_seconds() / 3600.0
    except ValueError:
        return None


def _maybe_trigger_daily_refresh_if_stale() -> None:
    """When digest is older than threshold, start a debounced background refresh (visitors + long-running server)."""
    try:
        max_age_h = float(os.environ.get("DAILY_AUTO_REFRESH_IF_OLDER_THAN_HOURS", "8"))
    except ValueError:
        max_age_h = 8.0
    if max_age_h <= 0:
        return
    try:
        debounce_sec = int(os.environ.get("DAILY_AUTO_REFRESH_DEBOUNCE_SEC", "2700"))
    except ValueError:
        debounce_sec = 2700
    debounce_sec = max(300, debounce_sec)

    age = _daily_digest_age_hours()
    if age is not None and age < max_age_h:
        return

    global _daily_last_stale_refresh_attempt
    with _daily_stale_refresh_lock:
        now = time.time()
        if now - _daily_last_stale_refresh_attempt < debounce_sec:
            return
        _daily_last_stale_refresh_attempt = now

    def _bg():
        try:
            _run_refresh_daily()
        except Exception:
            pass

    threading.Thread(target=_bg, daemon=True).start()


@app.route("/api/refresh-daily")
def api_refresh_daily():
    """Refresh daily news. Manual or cron. Optional: ?key=CRON_SECRET for cron."""
    import os
    key = request.args.get("key", "")
    secret = os.environ.get("CRON_SECRET", "")
    if secret and key != secret:
        return jsonify({"error": "Unauthorized"}), 401
    try:
        ok = _run_refresh_daily()
        if not ok:
            return jsonify({"error": "Daily news refresh failed"}), 500
        return jsonify({"status": "ok", "message": "Daily news refreshed"})
    except subprocess.TimeoutExpired:
        return jsonify({"error": "Timeout"}), 504
    except Exception as e:
        return jsonify({"error": str(e)}), 500


@app.route("/api/refresh-videos")
def api_refresh_videos():
    """Refresh viral videos (videos_trending_viral.md). Optional: ?key=CRON_SECRET for cron."""
    import os
    key = request.args.get("key", "")
    secret = os.environ.get("CRON_SECRET", "")
    if secret and key != secret:
        return jsonify({"error": "Unauthorized"}), 401
    try:
        _run_refresh_videos()
        return jsonify({"status": "ok", "message": "Viral videos refreshed"})
    except subprocess.TimeoutExpired:
        return jsonify({"error": "Timeout"}), 504
    except Exception as e:
        return jsonify({"error": str(e)}), 500


@app.route("/api/search-news", methods=["POST"])
def api_search_news():
    """Search news by topic. Runs search in background, redirects immediately (avoids timeout on Render)."""
    req_json = request.get_json(silent=True) or {}
    json_mode = (request.args.get("format") == "json") or request.is_json
    topic = (request.form.get("topic") or request.args.get("topic") or req_json.get("topic") or "").strip()
    if not topic:
        if json_mode:
            return jsonify({"error": "Missing topic", "diagnostic": classify_error_message("missing topic", context="news")}), 400
        return redirect("/news")
    if json_mode:
        from src.news_sources import fetch_all_topic_sources
        try:
            max_results = int(req_json.get("max_results", request.args.get("max_results", 12)) or 12)
        except (TypeError, ValueError):
            max_results = 12
        try:
            items, sources = fetch_all_topic_sources(topic, target_total=max(1, min(max_results, 50)))
            if not items:
                diag = {
                    "code": "no_results",
                    "category": "data",
                    "message": f"No results for topic: {topic}",
                    "hint": "Try a broader keyword or another language.",
                    "retryable": True,
                }
                return jsonify({"query": topic, "items": [], "sources": sources, "diagnostic": diag}), 200
            return jsonify({"query": topic, "items": items, "sources": sources, "diagnostic": None}), 200
        except Exception as e:
            diag = classify_error_message(str(e), context="news")
            return jsonify({"error": str(e), "diagnostic": diag}), 500
    _record_topic_search(topic)
    safe_topic = raw_topic_file_stem(topic)
    range_param = request.form.get("range") or request.args.get("range") or "1d"
    if range_param not in ("60m", "1d", "7d"):
        range_param = "1d"
    redirect_url = f"/news?topic={quote(safe_topic, safe='')}&range={range_param}"

    def _run_search():
        try:
            script = Path(__file__).parent / "scripts" / "search_only.py"
            subprocess.run(
                [sys.executable, str(script), topic],
                check=True,
                env=_env_for_topic_search_subprocess(),
                cwd=Path(__file__).parent,
                timeout=_topic_search_subprocess_timeout_sec(),
            )
        except Exception as e:
            print(f"[search-news] failed for {topic!r}: {e}")

    import threading
    threading.Thread(target=_run_search, daemon=True).start()
    return redirect(redirect_url)


STEPPS_ORDER = ["social_currency", "triggers", "emotion", "public", "practical", "stories"]
STEPPS_LABELS = {"social_currency": "Social Currency", "triggers": "Triggers", "emotion": "Emotion", "public": "Public", "practical": "Practical", "stories": "Stories"}
STEPPS_LABELS_ZH = {"social_currency": "社交货币", "triggers": "触发", "emotion": "情绪", "public": "公开", "practical": "实用", "stories": "故事"}
FIX_SUGGESTIONS = {
    "triggers": "Add a time/context anchor — 'Next time you open your feed…' — to boost Triggers.",
    "stories": "Wrap your core insight in a 3-part arc: Setup → Tension → Resolution.",
    "public": "Show social proof visually — crowd, screenshot, or 'join X people' framing.",
    "social_currency": "Add insider framing — 'most people don't know…' or exclusive angle.",
    "emotion": "Raise arousal: awe, excitement, or surprise in the hook.",
    "practical": "Add actionable steps — '3 ways to…' or specific how-to.",
}
FIX_SUGGESTIONS_ZH = {
    "triggers": "加时间/情境锚 —「下次打开动态时…」— 提升触发。",
    "stories": "用三段弧包装核心洞察：铺陈 → 张力 → 解决。",
    "public": "可视化社会证明 — 人群、截图或「加入 X 人」框架。",
    "social_currency": "加内行框架 —「多数人不知道…」或独家角度。",
    "emotion": "提高唤醒：开场用敬畏、兴奋或惊喜。",
    "practical": "加可执行步骤 —「3 种方法…」或具体 how-to。",
}


def _berger_interpretation(score: dict, lang: str = "en") -> str:
    """One-line interpretation: top 2 signals, weak 1, optional fix. For tooltip."""
    labels = STEPPS_LABELS_ZH if lang == "zh" else STEPPS_LABELS
    fix_map = FIX_SUGGESTIONS_ZH if lang == "zh" else FIX_SUGGESTIONS
    bd = score.get("breakdown", {})
    if not bd:
        return ""
    sorted_sigs = sorted(((k, v) for k, v in bd.items() if v > 0), key=lambda x: -x[1])
    top = [labels.get(k, k.replace("_", " ").title()) for k, _ in sorted_sigs[:2]]
    weak_key = min((k for k in STEPPS_ORDER if bd.get(k, 0) < 15), key=lambda k: bd.get(k, 20), default=None)
    weak = labels.get(weak_key, weak_key.replace("_", " ").title()) if weak_key else None
    fix = fix_map.get(weak_key, "") if weak_key else ""
    if lang == "zh":
        if top and weak:
            return f"强 {top[0]}" + (f" + {top[1]}" if len(top) > 1 else "") + (f"。弱于 {weak}。{fix}" if fix else f"。弱于 {weak}。")
        if top:
            return f"强 {top[0]}" + (f" + {top[1]}" if len(top) > 1 else "") + "。"
        return ""
    if top and weak:
        return f"Strong {top[0]}" + (f" + {top[1]}" if len(top) > 1 else "") + (f". Weak on {weak}. {fix}" if fix else f". Weak on {weak}.")
    if top:
        return f"Strong {top[0]}" + (f" + {top[1]}" if len(top) > 1 else "") + "."
    return ""


def _berger_breakdown_list(breakdown: dict, lang: str = "en") -> list:
    """Ordered list of {letter, name, score} for STEPPS bars."""
    labels = STEPPS_LABELS_ZH if lang == "zh" else STEPPS_LABELS
    letters = ["S", "T", "E", "P", "P", "S"]
    return [{"letter": letters[i], "name": labels.get(k, k.replace("_", " ").title()), "score": breakdown.get(k, 0)} for i, k in enumerate(STEPPS_ORDER)]


def _score_badge_class(score: int) -> str:
    if score >= 60:
        return "berger-high"
    if score >= 35:
        return "berger-mid"
    return "berger-low"


@app.route("/api/viral-videos", methods=["GET"])
def api_viral_videos():
    """Fetch viral videos. Query: q (search term). Source: global (default) or china."""
    lang = _get_lang()
    query = request.args.get("q", "viral trending").strip() or "viral trending"
    source = request.args.get("source", "global").lower()
    err_diag: dict | None = None
    try:
        if source == "china":
            from src.video_tools import score_berger
            from src.china_sources import get_crawler_china_results

            videos = []
            script = Path(__file__).parent / "scripts" / "fetch_bilibili_json.py"
            args = [sys.executable, str(script), query, "20"]
            kw = dict(capture_output=True, text=True, cwd=Path(__file__).parent, timeout=25)
            # 1) Try with system proxy (VPN+Clash 时走代理，Clash 把 bilibili 直连)
            result = subprocess.run(args, env=_env_for_bilibili(), **kw)
            if result.returncode != 0 or not (result.stdout or "").strip():
                # 2) Fallback: no proxy，直连 B 站（国内未开代理时）
                env_direct = _env_no_proxy()
                env_direct["PYTHONPATH"] = str(Path(__file__).parent)
                result = subprocess.run(args, env=env_direct, **kw)
            if result.returncode != 0:
                err = result.stderr.strip() or result.stdout.strip()
                try:
                    data = json.loads(err)
                    err = data.get("error", err)
                    err_diag = data.get("diagnostic")
                except json.JSONDecodeError:
                    err_diag = classify_error_message(err, context="china_video")
                raise RuntimeError(err)
            try:
                videos = json.loads(result.stdout) if result.stdout.strip() else []
            except (json.JSONDecodeError, ValueError):
                videos = []
            try:
                platforms_param = request.args.get("platforms", "").strip()
                platform_filter = [p.strip().lower() for p in platforms_param.split(",") if p.strip()] or None
                crawler = get_crawler_china_results(platform_filter=platform_filter)
                seen_urls = {v.get("url") for v in videos if isinstance(v, dict) and v.get("url")}
                for c in crawler:
                    if isinstance(c, dict):
                        u = c.get("url")
                        if u and u not in seen_urls:
                            seen_urls.add(u)
                            videos.append(c)
            except Exception:
                pass
        else:
            from src.video_tools import score_berger, extract_youtube_id
            max_fetch = 24 if (lang == "en" and source == "global") else 12
            script = Path(__file__).parent / "scripts" / "fetch_videos_json.py"
            env = _env_no_proxy()
            env["PYTHONPATH"] = str(Path(__file__).parent)
            result = subprocess.run(
                [sys.executable, str(script), query, str(max_fetch)],
                capture_output=True,
                text=True,
                env=env,
                cwd=Path(__file__).parent,
                timeout=25,
            )
            if result.returncode != 0:
                err = result.stderr.strip() or result.stdout.strip()
                try:
                    data = json.loads(err)
                    err = data.get("error", err)
                    err_diag = data.get("diagnostic")
                except json.JSONDecodeError:
                    err_diag = classify_error_message(err, context="video")
                raise RuntimeError(err)
            videos = json.loads(result.stdout)
        out = []
        _china_platform_display = {"bilibili": "Bilibili", "xiaohongshu": "Xiaohongshu", "douyin": "Douyin", "shipinhao": "Shipinhao", "zhihu": "Zhihu"}
        for v in videos:
            # English channels: only show English content (no CJK in title/desc)
            if lang == "en" and source == "global":
                title = v.get("title", "")
                desc = v.get("description", "") or ""
                if _has_cjk(title) or _has_cjk(desc):
                    continue
            text = f"{v.get('title', '')} {v.get('description', '')}"
            score = score_berger(text)
            title = v.get("title", "")
            desc = (v.get("description") or "")[:200]
            angles = generate_angles(query, title, desc, count=3, lang=lang)
            if source == "china":
                platform = _china_platform_display.get((v.get("platform") or "bilibili").lower(), "Bilibili")
            else:
                platform = "YouTube"
            views_int = v.get("views_int", 0) or 0
            berger_val = score["total"]
            interpretation = _berger_interpretation(score)
            breakdown_list = _berger_breakdown_list(score.get("breakdown", {}))
            out.append({
                "title": title,
                "url": v.get("url"),
                "platform": platform,
                "content_type": v.get("content_type", "video"),
                "views": v.get("views") or str(views_int) or "0",
                "berger": berger_val,
                "magic": ", ".join(score["magic_words_found"]) or "—",
                "desc": (v.get("description") or "")[:150],
                "angles": angles,
                "interpretation": interpretation,
                "breakdown": breakdown_list,
            })
            if len(out) >= 12:
                break
        # Lang hint when Chinese UI user searches English topic on global source
        lang_hint = None
        if lang == "zh" and source == "global" and not _has_cjk(query):
            lang_hint = _t("viral_lang_hint", lang)

        return jsonify({"query": query, "source": source, "videos": out, "lang_hint": lang_hint})
    except subprocess.TimeoutExpired:
        timeout_msg = _t("viral_timeout", lang)
        return jsonify({"error": timeout_msg, "diagnostic": classify_error_message(timeout_msg, context="video")}), 504
    except Exception as e:
        err_msg = str(e)
        diag = err_diag or classify_error_message(err_msg, context="china_video" if source == "china" else "video")
        if source == "china":
            return jsonify({
                "error": err_msg,
                "diagnostic": diag,
                "china_unavailable": True,
                "china_hint": _t("viral_china_unavailable", lang),
            }), 500
        return jsonify({"error": err_msg, "diagnostic": diag}), 500


@app.route("/api/china-crawl", methods=["POST"])
def api_china_crawl():
    """Run China crawler for given platforms and keywords. Writes to output/china_crawler_{platform}.json."""
    import time
    from urllib.parse import quote

    t0 = time.perf_counter()
    deadline = t0 + 55.0  # Keep request responsive; return partial results when budget exceeded.
    time_budget_hit = False
    lang = _get_lang()
    data = request.get_json(silent=True) or {}
    platforms_raw = data.get("platforms", request.form.get("platforms", ""))
    kw_raw = data.get("keywords", request.form.get("keywords", ""))
    if isinstance(kw_raw, list):
        keywords_list = [str(k).strip() for k in kw_raw if str(k).strip()][:10]
    else:
        parts = [p.strip() for p in str(kw_raw or "").split(",") if p.strip()][:10]
        keywords_list = parts if parts else ["热门"]
    if not keywords_list:
        keywords_list = ["热门"]
    valid_platforms = ("xhs", "douyin", "shipinhao", "zhihu", "bilibili", "tiktok", "x", "instagram")
    if isinstance(platforms_raw, list):
        platforms = [p for p in platforms_raw if p and str(p).lower() in valid_platforms]
    else:
        platforms = [p.strip().lower() for p in str(platforms_raw).split(",") if p.strip() and p.strip().lower() in valid_platforms]
    if not platforms:
        return jsonify({"ok": False, "error": "No valid platforms (use bilibili, xhs, douyin, shipinhao, zhihu, tiktok, x, instagram)"}), 400

    from src.china_sources import get_china_search_url

    def _fallback_search_url(platform: str, keyword: str) -> str:
        p = (platform or "").lower()
        if p == "zhihu":
            return "https://www.zhihu.com/search?type=content&q=" + quote(keyword)
        if p == "xhs":
            return get_china_search_url("xiaohongshu", keyword)
        return get_china_search_url(p, keyword)

    def _classify_status(platform: str, fetched_count: int, kw_attempted: int, kw_with_items: int, reasons: list[str]) -> str:
        if fetched_count <= 0:
            reason_text = " ".join(reasons).lower()
            if any(k in reason_text for k in ("登录", "login", "cookie", "cookies", "fresh cookies", "扫码", "verify_login")):
                return "blocked_login"
            if any(k in reason_text for k in ("风控", "captcha", "forbidden", "403", "429", "blocked", "rate limit")):
                return "blocked_risk"
            return "network_error"
        if kw_attempted > 1 and kw_with_items < kw_attempted:
            return "partial"
        if reasons:
            return "partial"
        return "ok"

    root = Path(__file__).parent
    out_dir = root / "output"
    out_dir.mkdir(parents=True, exist_ok=True)
    env = {**_env_no_proxy(), "PYTHONPATH": str(root), "CHINA_CRAWLER_OUTPUT_DIR": str(out_dir)}
    fetched = {}
    errors = []
    all_items = []
    seen_urls = set()
    status_map = {
        p: {
            "platform": p,
            "status": "network_error",
            "fetched": 0,
            "keywords": keywords_list,
            "keywords_with_items": 0,
            "source": "crawler",
            "reasons": [],
            "fallback_url": _fallback_search_url(p, keywords_list[0]),
        }
        for p in platforms
    }

    try:
        from src.third_party_china_api import fetch_platform_via_third_party
    except ImportError:
        fetch_platform_via_third_party = lambda p, k, max_results=20: []

    for keyword in keywords_list:
        if time.perf_counter() > deadline:
            time_budget_hit = True
            break
        for p in platforms:
            if time.perf_counter() > deadline:
                time_budget_hit = True
                break
            try:
                # Prefer third-party API (Just One API / Apify) when token is set
                third_party = fetch_platform_via_third_party(p, keyword, max_results=15)
                if third_party:
                    status_map[p]["source"] = "third_party_api"
                    fetched[p] = fetched.get(p, 0) + len(third_party)
                    status_map[p]["fetched"] = fetched[p]
                    status_map[p]["keywords_with_items"] += 1
                    for i in third_party:
                        if isinstance(i, dict) and i.get("url") and i["url"] not in seen_urls:
                            seen_urls.add(i["url"])
                            all_items.append(i)
                    continue
                # Fallback: run Playwright crawler
                r = subprocess.run(
                    [sys.executable, "-m", "tools.china_crawler", "run", "--platform", p, "--type", "search", "--keywords", keyword, "--max", "15"],
                    capture_output=True,
                    text=True,
                    env=env,
                    cwd=root,
                    timeout=25,
                )
                if r.returncode == 0:
                    try:
                        path = out_dir / f"china_crawler_{p}.json"
                        file_data = json.loads(path.read_text(encoding="utf-8")) if path.exists() else {}
                        items = file_data.get("items") or []
                        fetched[p] = fetched.get(p, 0) + len(items)
                        status_map[p]["fetched"] = fetched[p]
                        if items:
                            status_map[p]["keywords_with_items"] += 1
                        for i in items:
                            if isinstance(i, dict) and i.get("url") and i["url"] not in seen_urls:
                                seen_urls.add(i["url"])
                                all_items.append(i)
                    except Exception:
                        pass
                else:
                    err = (r.stderr or r.stdout or "").strip()
                    if err:
                        status_map[p]["reasons"].append((err.split("\n")[0] or err).strip()[:160])
                        if "No module named" in err and "playwright" in err.lower():
                            errors.append("Playwright not installed. Run: source .venv/bin/activate && pip install playwright && python -m playwright install chromium")
                        else:
                            first_line = (err.split("\n")[0] or err).strip()[:80]
                            errors.append(f"{p}: {first_line}")
            except subprocess.TimeoutExpired:
                status_map[p]["reasons"].append("timeout")
                errors.append(f"{p}({keyword[:8]}): timeout")
            except Exception as e:
                status_map[p]["reasons"].append(str(e)[:160])
                errors.append(f"{p}({keyword[:8]}): {str(e)[:100]}")

    for p in platforms:
        ps = status_map[p]
        if p == "douyin" and int(ps.get("fetched") or 0) <= 0 and not (root / "config" / "china_crawler_douyin_state.json").exists():
            ps["reasons"].append("Douyin may require local crawler login state. Run: .venv/bin/python -m tools.china_crawler login --platform douyin")
        ps["status"] = _classify_status(
            platform=p,
            fetched_count=int(ps.get("fetched") or 0),
            kw_attempted=len(keywords_list),
            kw_with_items=int(ps.get("keywords_with_items") or 0),
            reasons=ps.get("reasons") or [],
        )
        ps["reasons"] = list(dict.fromkeys(ps.get("reasons") or []))[:2]

    fallback_links = []
    for p in platforms:
        ps = status_map[p]
        if ps["status"] in ("blocked_login", "blocked_risk", "network_error", "partial"):
            fallback_links.append({
                "platform": p,
                "status": ps["status"],
                "keyword": keywords_list[0],
                "url": ps.get("fallback_url") or "",
            })

    errors = list(dict.fromkeys(errors))[:3]
    total = len(all_items)
    duration_sec = round(time.perf_counter() - t0, 1)
    message = _t("china_crawl_done", lang) if total > 0 else _t("china_crawl_no_items", lang)
    if time_budget_hit:
        errors.append("Time budget reached; returned partial results.")
        message = ("Partial crawl complete." if lang != "zh" else "拉取时间达到上限，已返回部分结果。")
    if errors:
        message += " " + (errors[0] if len(errors) == 1 else _t("china_crawl_errors", lang).format(n=len(errors)))
    return jsonify({
        "ok": True,
        "fetched": fetched,
        "message": message,
        "errors": errors,
        "items": all_items,
        "duration_sec": duration_sec,
        "platform_status": status_map,
        "fallback_links": fallback_links,
    })


@app.route("/api/china-content-evaluate", methods=["POST"])
def api_china_content_evaluate():
    """Evaluate text with STEPPS + Minto; return score, structure, and rewrite directions. Own framework only."""
    from src.rewrite_direction import rewrite_directions

    data = request.get_json(silent=True) or {}
    text = (data.get("text") or "").strip()
    lang = (data.get("lang") or _get_lang() or "zh").lower()[:2]
    if lang != "en":
        lang = "zh"
    if not text:
        return jsonify({"ok": False, "error": "Missing text"}), 400
    try:
        score_data = score_berger(text)
        breakdown = score_data.get("breakdown") or {}
        stepps = _berger_breakdown_list(breakdown, lang)
        minto = {}
        if len(text.strip()) > 200:
            minto = structure_minto(text, lang=lang)
        directions = rewrite_directions(score_data, minto or None, lang)
        return jsonify({
            "ok": True,
            "score": score_data.get("total", 0),
            "breakdown": breakdown,
            "stepps": stepps,
            "minto": minto,
            "rewrite_directions": directions,
        })
    except Exception as e:
        return jsonify({"ok": False, "error": str(e)}), 500


def _build_china_item_markdown(item: dict, lang: str = "zh") -> str:
    """Build AI-friendly markdown for one China content item."""
    title = str(item.get("title") or "").strip() or "Untitled"
    url = str(item.get("url") or "").strip() or "-"
    platform = str(item.get("platform") or "").strip() or "-"
    content_type = str(item.get("content_type") or "").strip() or "-"
    views = str(item.get("views") or "").strip() or "-"
    desc = str(item.get("description") or "").strip()
    if not desc:
        desc = str(item.get("snippet") or "").strip()
    transcript_text = f"{title}\n\n{desc}".strip()

    if (lang or "zh").lower().startswith("zh"):
        prompts = "\n".join([
            '- "基于逐字稿写 3 个短视频开场钩子（15字内）"',
            '- "按小红书风格改写成种草笔记，保留核心信息"',
            '- "按抖音口播节奏改写成 30 秒脚本：开场-冲突-结论"',
            '- "用金字塔原理重写：结论先行 + 3 个要点 + 对应论据"',
        ])
        remix_title = "AI 改写提示词"
        transcript_title = "逐字稿"
        source_title = "来源信息"
    else:
        prompts = "\n".join([
            '- "Rewrite into 3 short-form video hooks (under 12 words each)"',
            '- "Rewrite for Xiaohongshu note style while preserving key facts"',
            '- "Rewrite as a 30-second Douyin voiceover script: hook-conflict-payoff"',
            '- "Restructure with Minto Pyramid: conclusion first, then 3 key points with evidence"',
        ])
        remix_title = "AI Remix Prompts"
        transcript_title = "Transcript"
        source_title = "Source"

    return f"""# China Content Transcript Export

## {source_title}

- Title: {title}
- Platform: {platform}
- Type: {content_type}
- Views: {views}
- URL: {url}

---

## {transcript_title}

{transcript_text}

---

## {remix_title}

{prompts}
"""


@app.route("/api/china-content-export-md", methods=["POST"])
def api_china_content_export_md():
    """Return markdown text for one China content item (client decides whether to download)."""
    data = request.get_json(silent=True) or {}
    item = data.get("item") or {}
    if not isinstance(item, dict):
        return jsonify({"ok": False, "error": "Invalid item"}), 400
    lang = (data.get("lang") or _get_lang() or "zh").lower()[:2]
    markdown_text = _build_china_item_markdown(item, lang=lang)
    title = str(item.get("title") or "").strip() or "china-content"
    safe = "".join(c for c in title if c.isalnum() or c in "-_").strip("-_")
    if not safe:
        safe = "china-content"
    filename = f"transcript_{safe[:40]}.md"
    return jsonify({"ok": True, "markdown": markdown_text, "filename": filename})


@app.route("/api/china-content-rewrite", methods=["POST"])
def api_china_content_rewrite():
    """Proxy transcript text to a user-provided rewrite model endpoint."""
    from urllib import request as urlrequest
    from urllib.error import URLError, HTTPError

    data = request.get_json(silent=True) or {}
    text = (data.get("text") or "").strip()
    if not text:
        return jsonify({"ok": False, "error": "Missing text"}), 400
    lang = (data.get("lang") or _get_lang() or "zh").lower()[:2]
    style = (data.get("style") or "").strip()
    rewrite_api_url = (data.get("rewrite_api_url") or os.getenv("CHINA_REWRITE_API_URL") or "").strip()
    if not rewrite_api_url:
        return jsonify({
            "ok": False,
            "error": "Missing rewrite API URL. Fill it in the page input or set CHINA_REWRITE_API_URL in .env.",
        }), 400

    prompt = (
        "请改写为更适合中文社交媒体发布的版本，保留事实，不要编造。输出中文，结构清晰，先给标题再给正文。"
        if lang == "zh"
        else "Rewrite this for social media publication with high clarity. Keep facts accurate and avoid fabrication."
    )
    if style:
        prompt += f" Style requirement: {style}"

    payload = {
        "prompt": prompt,
        "text": text,
        "lang": lang,
        "source": "virallab-china-content",
    }
    body = json.dumps(payload).encode("utf-8")
    req = urlrequest.Request(
        rewrite_api_url,
        data=body,
        headers={"Content-Type": "application/json"},
        method="POST",
    )

    try:
        with urlrequest.urlopen(req, timeout=45) as resp:
            raw = resp.read().decode("utf-8", errors="replace")
    except HTTPError as e:
        err_body = ""
        try:
            err_body = e.read().decode("utf-8", errors="replace")[:300]
        except Exception:
            pass
        return jsonify({"ok": False, "error": f"Rewrite API HTTP {e.code}. {err_body}".strip()}), 502
    except URLError as e:
        return jsonify({"ok": False, "error": f"Rewrite API unreachable: {e.reason}"}), 502
    except Exception as e:
        return jsonify({"ok": False, "error": f"Rewrite API failed: {e}"}), 502

    rewritten = ""
    try:
        data_resp = json.loads(raw)
        if isinstance(data_resp, dict):
            rewritten = (
                data_resp.get("rewrite")
                or data_resp.get("text")
                or data_resp.get("output")
                or data_resp.get("answer")
                or data_resp.get("content")
                or ""
            )
            if not rewritten and isinstance(data_resp.get("choices"), list) and data_resp["choices"]:
                c0 = data_resp["choices"][0] or {}
                msg = c0.get("message") if isinstance(c0, dict) else None
                if isinstance(msg, dict):
                    rewritten = msg.get("content") or ""
        elif isinstance(data_resp, list) and data_resp:
            rewritten = str(data_resp[0])
    except Exception:
        rewritten = raw

    rewritten = (rewritten or "").strip()
    if not rewritten:
        return jsonify({"ok": False, "error": "Rewrite API returned empty content."}), 502
    return jsonify({"ok": True, "rewritten_text": rewritten[:20000]})


STEPPS_DESCRIPTIONS = {
    "social_currency": (
        ("People share things that make them look good, smart, and ahead of the curve.", "Does sharing this make me look knowledgeable?"),
        ("人们分享让自己看起来更聪明、更领先的内容。", "分享这个会让我显得有见识吗？"),
    ),
    "triggers": (
        ("Content that connects to people at the right time and context, so they think about it when it's relevant.", "Will I remember this at the right moment?"),
        ("内容在对的时间与情境连结人们，让他们在相关时想起。", "我会在对的时刻想起这个吗？"),
    ),
    "emotion": (
        ("When we care, we share. Awe, excitement, and other emotions drive sharing.", "Does this make me feel something shareable?"),
        ("我们在乎就会分享。敬畏、兴奋等情绪驱动分享。", "这会让我产生想分享的感受吗？"),
    ),
    "public": (
        ("Visible content gets copied. Public proof of use increases adoption.", "Can others see me sharing this?"),
        ("可见的内容会被模仿。公开使用证明提升采用。", "别人能看到我在分享这个吗？"),
    ),
    "practical": (
        ("People share things they find helpful and believe will benefit others.", "Will this genuinely help someone I know?"),
        ("人们分享觉得有用、相信能帮到别人的内容。", "这会真正帮到我认识的人吗？"),
    ),
    "stories": (
        ("Content wrapped in compelling stories—information travels as idle chatter.", "Is there a story I'd naturally retell?"),
        ("包装在引人入胜故事中的内容 — 信息随闲聊传播。", "有我会自然转述的故事吗？"),
    ),
}

# Berger research timeline for "Why it works"
BERGER_TIMELINE = [
    {"year": "1996", "tag": "Foundation", "title": "Wharton Chair appointment", "body": "Berger joins Wharton School of Business, University of Pennsylvania — beginning 25+ years of word-of-mouth research."},
    {"year": "2013", "tag": "Breakthrough", "title": "Contagious published", "body": "The STEPPS framework reaches #1 on NYT Bestseller list. Apple, Google, Nike begin applying the model internally."},
    {"year": "2016", "tag": "Expansion", "title": "Invisible Influence", "body": "Berger extends the model: social conformity and hidden behavioral triggers documented across 80+ peer-reviewed studies."},
    {"year": "2020", "tag": "Algorithm", "title": "The Catalyst", "body": "Barrier-removal mechanics formalized. Resistance is shown to be a stronger predictor of inaction than lack of motivation."},
    {"year": "2023", "tag": "Language", "title": "Magic Words", "body": "Linguistic analysis at scale: specific words shown to increase compliance, sharing, and persuasion by measurable margins."},
]


# Open Library cover IDs (cover_i) — reliable image URLs
BERGER_BOOKS = [
    {"title": "Contagious", "subtitle": "Why Things Catch On", "year": "2013", "cover_id": "12130640", "url": "https://jonahberger.com/books/contagious/", "unlocks": "The 6 principles that make ideas spread—used by Fortune 500 brands."},
    {"title": "Invisible Influence", "subtitle": "The Hidden Forces That Shape Behavior", "year": "2016", "cover_id": "8869123", "url": "https://jonahberger.com/books/invisible-influence/", "unlocks": "Why people copy, conform, and share—the hidden triggers."},
    {"title": "The Catalyst", "subtitle": "How to Change Anyone's Mind", "year": "2020", "cover_id": "9295723", "url": "https://jonahberger.com/books/the-catalyst/", "unlocks": "Remove resistance. Lower friction. Make change happen."},
    {"title": "Magic Words", "subtitle": "What to Say to Get Your Way", "year": "2023", "cover_id": "13170768", "url": "https://jonahberger.com/magic-words/", "unlocks": "One word can boost compliance 50%. We surface them."},
]


@app.route("/science")
def science():
    """Jonah Berger STEPPS explainer — How we score (v2)."""
    lang = _get_lang()
    stepps_data = [
        ("S", "science_v2_stepps_S_name", "science_v2_stepps_S_desc"),
        ("T", "science_v2_stepps_T_name", "science_v2_stepps_T_desc"),
        ("E", "science_v2_stepps_E_name", "science_v2_stepps_E_desc"),
        ("P", "science_v2_stepps_P1_name", "science_v2_stepps_P1_desc"),
        ("P", "science_v2_stepps_P2_name", "science_v2_stepps_P2_desc"),
        ("S", "science_v2_stepps_S2_name", "science_v2_stepps_S2_desc"),
    ]
    stepps_html = "".join(
        f'<div class="stepps-item"><div class="stepps-letter">{letter}</div><div><div class="stepps-name">{_html_escape(_t(nk, lang))}</div><div class="stepps-desc">{_html_escape(_t(dk, lang))}</div></div></div>'
        for letter, nk, dk in stepps_data
    )
    formula_rows = [
        ("science_v2_formula_r1", "science_v2_formula_pts1"),
        ("science_v2_formula_r2", "science_v2_formula_pts2"),
        ("science_v2_formula_r3", "science_v2_formula_pts3"),
        ("science_v2_formula_r4", "science_v2_formula_pts4"),
        ("science_v2_formula_r5", "science_v2_formula_pts5"),
        ("science_v2_formula_r6", "science_v2_formula_pts6"),
    ]
    formula_html = "".join(
        f'<div class="formula-row"><span>{_t(rk, lang)}</span><span class="formula-pts">{_t(pk, lang)}</span></div>'
        for rk, pk in formula_rows
    )
    hero = f'<header class="page-header"><h1 class="science-v2-h1">{_t("science_v2_title", lang)}</h1></header>'
    body = f"""
    <div class="science-wrap">
      <p class="science-v2-p">{_t("science_v2_intro1", lang)}</p>
      <p class="science-v2-p">{_t("science_v2_intro2", lang)}</p>

      <hr class="science-hr" />

      <div class="section-title">{_t("science_v2_stepps_title", lang)}</div>
      <p class="science-v2-p">{_t("science_v2_stepps_intro", lang)}</p>
      <div class="stepps-list">{stepps_html}</div>

      <hr class="science-hr" />

      <div class="section-title">{_t("science_minto_title", lang)}</div>
      <p class="science-v2-p">{_t("science_minto_desc", lang)}</p>
      <p class="science-v2-p"><a href="/video-to-text">{_html_escape(_t("science_minto_cta", lang))}</a></p>

      <hr class="science-hr" />

      <div class="section-title">{_t("science_v2_formula_title", lang)}</div>
      <p class="science-v2-p">{_t("science_v2_formula_intro", lang)}</p>
      <div class="formula-block">{formula_html}</div>

      <hr class="science-hr" />

      <div class="honest-note">{_t("science_v2_honest", lang)}</div>

      <div class="bottom-links">
        <a href="/viral">{_html_escape(_t("science_v2_link_score", lang))}</a>
        <a href="/video-to-text">{_html_escape(_t("science_v2_link_transcript", lang))}</a>
      </div>
    </div>
    """
    return _base(_t("science_v2_page_title", lang), body, "/science", hero=hero, lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


STEPPS_ORDER_TUPLES = [
    ("social_currency", "S", "Social Currency", "社交货币"),
    ("triggers", "T", "Triggers", "触发"),
    ("emotion", "E", "Emotion", "情绪"),
    ("public", "P", "Public", "公开"),
    ("practical", "P", "Practical", "实用"),
    ("stories", "S", "Stories", "故事"),
]


def _get_score_tier(score: int) -> str:
    if score >= 85:
        return "Highly Viral"
    if score >= 70:
        return "Strong Signal"
    if score >= 50:
        return "Moderate Potential"
    return "Weak Signal"


def _score_to_stepps(breakdown: dict, lang: str = "en") -> list[dict]:
    """Map score_berger breakdown to STEPPS array with letter, name, score, color."""
    out = []
    for key, letter, name_en, name_zh in STEPPS_ORDER_TUPLES:
        s = breakdown.get(key, 0)
        if s >= 15:
            color = "#6a9e67"
        elif s >= 8:
            color = "#e8a54d"
        else:
            color = "#c0392b"
        name = name_zh if lang == "zh" else name_en
        out.append({"letter": letter, "name": name, "score": s, "color": color})
    return out


def _build_markdown_export(
    url: str,
    title: str,
    score: int,
    stepps: list[dict],
    transcript: str,
    magic_words: list[str],
    lang: str = "en",
    include_minto: bool = True,
) -> str:
    """Build AI-ready markdown string. Includes Minto Pyramid structure when transcript is usable."""
    sorted_stepps = sorted(stepps, key=lambda s: s["score"], reverse=True)
    top2 = sorted_stepps[:2]
    weak = sorted_stepps[-1]
    top_names = " + ".join(s["name"] for s in top2)
    table_rows = "\n".join(
        f"| {s['name']} | {s['score']} | {round((s['score'] / 20) * 100)}% |"
        for s in stepps
    )
    remix_prompts = [
        f'- "Rewrite as a Twitter/X thread using the top signals: {top_names}"',
        '"Extract the 5 most shareable moments from this transcript"',
        '"Write a YouTube title using Social Currency and Emotion signals"',
        '"Turn the hook into a 30-second Reels/Shorts script"',
        '"Restructure this script with the Minto Pyramid: one conclusion first, then 3–5 key points, then supporting evidence."',
    ]
    remix_prompts_str = "\n".join(remix_prompts)
    mw = ", ".join(magic_words) if magic_words else "None detected"

    minto_section = ""
    if include_minto and len(transcript.strip()) > 60:
        minto = structure_minto(transcript, lang=lang)
        if minto.get("conclusion") or minto.get("key_points"):
            minto_section = f"\n---\n\n## Minto Pyramid (structure)\n\n{minto_to_markdown(minto, lang)}\n"

    return f"""# ViralLab Transcript Analysis

**Source:** {url}
**Title:** {title}
**Analysed by:** ViralLab · Berger STEPPS Framework

---

## Berger Score: {score}/100

| Signal | Score | /20 |
|--------|-------|-----|
{table_rows}

**Top signals:** {top_names}
**Weakest signal:** {weak["name"]}

### Fix suggestion
Add a recurring context anchor to boost {weak["name"]}:
*"Every time you sit down to plan content — think about this..."*

---

## Magic Words Detected
{mw}
{minto_section}
---

## Transcript

{transcript}

---

## AI Remix Prompts

Use this transcript with any AI assistant:

{remix_prompts_str}

---
*Generated by ViralLab · Berger STEPPS + Minto Pyramid*"""


def _fetch_transcript_prefer_manual(video_id: str, lang: str) -> tuple[str, str]:
    """Fetch transcript, preferring manual captions. Returns (transcript, caption_source)."""
    try:
        from youtube_transcript_api import YouTubeTranscriptApi
        api = YouTubeTranscriptApi()
        lang_codes = ["en", "zh", "zh-Hans", "zh-Hant"] if lang == "zh" else ["en", "en-US", "en-GB"]
        try:
            transcript_list = api.list(video_id)
            for code in lang_codes:
                try:
                    t = transcript_list.find_manually_created_transcript([code])
                    fetched = t.fetch()
                    text = " ".join(s.text for s in fetched.snippets)
                    return text, "creator-uploaded (manual)"
                except Exception:
                    pass
            for code in lang_codes:
                try:
                    t = transcript_list.find_generated_transcript([code])
                    fetched = t.fetch()
                    text = " ".join(s.text for s in fetched.snippets)
                    return text, "auto-generated"
                except Exception:
                    pass
        except Exception:
            pass
        fetched = api.fetch(video_id)
        if fetched and fetched.snippets:
            text = " ".join(s.text for s in fetched.snippets)
            return text, "unknown"
    except Exception:
        pass
    return "", ""


@app.route("/api/test-douyin-session", methods=["POST"])
def api_test_douyin_session():
    data = request.get_json(silent=True) or {}
    url = (data.get("url") or "").strip()
    result = test_douyin_session(url)
    return jsonify(result)


def _fetch_transcript_from_input(input_ref: str, lang: str) -> tuple[str, str, str]:
    """Fetch transcript from YouTube captions first, then ASR fallback.

    Returns ``(text, caption_source, failure_detail)``. ``failure_detail`` is non-empty when
    text is empty and ASR failed after download attempts (often yt-dlp/VideoCaptioner stderr).
    """
    ref = (input_ref or "").strip()
    if not ref:
        return "", "", ""

    vid = extract_youtube_id(ref)
    if vid:
        transcript, caption_source = _fetch_transcript_prefer_manual(vid, lang)
        if transcript:
            return transcript, caption_source, ""

    if (ref.startswith("http://") or ref.startswith("https://")) and not is_supported_external_media_url(ref):
        # Still try ASR fallback for generic podcast/article pages.
        pass
    out = transcribe_best_effort(ref, lang=lang)
    if out.text:
        return out.text, out.source, ""
    return "", "", (out.detail or "").strip()


@app.route("/video-to-text", methods=["GET", "POST"])
def video_to_text():
    lang = _get_lang()
    msg = ""
    fail_detail = ""
    result = None
    if request.method == "POST":
        url = (request.form.get("url") or "").strip()
        if url:
            try:
                transcript, caption_source, fail_detail = _fetch_transcript_from_input(url, lang)
                if not transcript:
                    if has_videocaptioner() or has_whisperx():
                        ul = (url or "").lower()
                        if "douyin.com" in ul or "v.douyin.com" in ul:
                            msg = _t("video2text_err_douyin", lang)
                        elif "xiaohongshu.com" in ul or "xhslink.com" in ul:
                            msg = _t("video2text_err_xhs", lang)
                        else:
                            msg = _t("video2text_err_generic", lang)
                    else:
                        msg = (
                            "No captions found. Install ASR: "
                            "`pip install videocaptioner` and/or optional WhisperX "
                            "(`pip install -r requirements-asr.txt`; see README)."
                        )
                else:
                    score_data = score_berger(transcript)
                    stepps = _score_to_stepps(score_data.get("breakdown", {}), lang)
                    sorted_stepps = sorted(stepps, key=lambda s: s["score"], reverse=True)
                    top2 = sorted_stepps[:2]
                    weak = sorted_stepps[-1]
                    top_names = [s["name"] for s in top2]
                    title = (transcript[:80] + "…") if len(transcript) > 80 else transcript
                    fix_suggestion = (
                        f'Add a recurring context anchor to boost {weak["name"]}: '
                        f'<em>"Every time you sit down to plan content — think about this..."</em>'
                    )
                    markdown_raw = _build_markdown_export(
                        url=url,
                        title=title,
                        score=score_data["total"],
                        stepps=stepps,
                        transcript=transcript,
                        magic_words=score_data.get("magic_words_found", []),
                        lang=lang,
                        include_minto=False,
                    )
                    markdown_minto = _build_markdown_export(
                        url=url,
                        title=title,
                        score=score_data["total"],
                        stepps=stepps,
                        transcript=transcript,
                        magic_words=score_data.get("magic_words_found", []),
                        lang=lang,
                        include_minto=True,
                    )
                    minto = structure_minto(transcript, lang=lang) if transcript.strip() else {}
                    result = {
                        "url": url,
                        "title": title,
                        "transcript": transcript,
                        "minto": minto,
                        "score": score_data["total"],
                        "tier": _get_score_tier(score_data["total"]),
                        "stepps": stepps,
                        "top_names": top_names,
                        "weak_name": weak["name"],
                        "fix": fix_suggestion,
                        "magic_words": score_data.get("magic_words_found", []),
                        "markdown": markdown_minto,
                        "markdown_raw": markdown_raw,
                        "markdown_minto": markdown_minto,
                        "caption_source": caption_source,
                    }
                    OUTPUT.mkdir(exist_ok=True)
                    out_id = str(abs(hash(url)))[:12]
                    out_file = OUTPUT / f"transcript_{out_id}.md"
                    out_file.write_text(markdown_minto, encoding="utf-8")
                    result["transcript_file"] = out_file.name
            except Exception as e:
                msg = f"Error: {e}"
        else:
            msg = "Enter a URL"

    transcript_files = [x for x in _list_digests() if x["type"] == "transcript"]
    transcript_list = ""
    for x in transcript_files:
        transcript_list += f'<li><a href="/view/{_html_escape(x["file"])}">{_html_escape(x["file"])}</a></li>'

    v2t_styles = """
    .v2t-wrap{max-width:100%;margin:0;padding:0 0 100px;}
    .v2t-page-header{margin-bottom:32px;}
    .v2t-page-header h1{font-size:22px;font-weight:700;letter-spacing:-0.01em;margin-bottom:8px;color:var(--ink);}
    .v2t-page-header p{font-size:14px;color:var(--muted);line-height:1.65;max-width:480px;}
    .v2t-accuracy-note{font-size:13px;color:var(--muted);line-height:1.65;padding:12px 16px;background:var(--cream-dark);border-radius:var(--radius);margin-bottom:28px;}
    .v2t-accuracy-note strong{color:var(--ink);font-weight:600;}
    .v2t-form-block{margin-bottom:12px;}
    .v2t-input-row{display:flex;gap:8px;}
    .v2t-url-input{flex:1;font-family:inherit;font-size:14px;padding:11px 14px;border:1.5px solid var(--border);border-radius:var(--radius);background:var(--card);outline:none;}
    .v2t-url-input:focus{border-color:var(--accent-light);box-shadow:0 0 0 3px rgba(106,158,103,0.1);}
    .v2t-submit-btn{font-family:inherit;font-size:14px;font-weight:600;background:var(--cta);color:var(--cta-text);border:none;border-radius:var(--radius);padding:11px 22px;cursor:pointer;white-space:nowrap;}
    .v2t-submit-btn:hover{background:#f5b960;transform:translateY(-1px);}
    .v2t-submit-btn[disabled]{opacity:.75;cursor:not-allowed;transform:none;}
    .v2t-download-btn{font-family:inherit;font-size:14px;font-weight:600;background:transparent;color:var(--accent);border:1.5px solid var(--accent-light);border-radius:var(--radius);padding:11px 18px;cursor:pointer;white-space:nowrap;}
    .v2t-download-btn:hover{background:var(--cream-dark);}
    .v2t-download-btn[disabled]{opacity:.6;cursor:not-allowed;}
    .v2t-form-hint{font-size:12px;color:var(--muted-light);margin-top:8px;}
    .v2t-form-hint code{font-family:monospace;font-size:11px;background:var(--cream-dark);padding:1px 5px;border-radius:3px;color:var(--muted);}
    .v2t-processing{display:none;margin:10px 0 14px;padding:10px 12px;border:1px dashed var(--accent-light);border-radius:var(--radius);background:var(--cream-dark);}
    .v2t-processing.visible{display:block;}
    .v2t-processing strong{display:block;font-size:12px;color:var(--ink);}
    .v2t-processing small{display:block;font-size:11px;color:var(--muted);margin-top:3px;line-height:1.45;}
    .v2t-result-section{display:none;margin-top:28px;}
    .v2t-result-section.visible{display:block;}
    .v2t-score-summary{display:flex;align-items:center;gap:16px;padding:16px 20px;background:var(--card);border:1px solid var(--border);border-radius:var(--radius);margin-bottom:16px;}
    .v2t-score-num{font-size:36px;font-weight:800;color:var(--cta-dark);line-height:1;flex-shrink:0;}
    .v2t-score-meta{flex:1;}
    .v2t-score-tier{font-size:14px;font-weight:600;color:var(--ink);margin-bottom:3px;}
    .v2t-score-detail{font-size:13px;color:var(--muted);line-height:1.5;}
    .v2t-score-fix{font-size:12px;color:var(--muted);margin-top:6px;padding:8px 12px;background:var(--cta-pale);border-left:2px solid var(--cta);border-radius:0 var(--radius) var(--radius) 0;}
    .v2t-score-fix strong{color:var(--cta-dark);}
    .v2t-stepps-mini{display:flex;gap:3px;margin-top:10px;}
    .v2t-sm-wrap{flex:1;display:flex;flex-direction:column;align-items:center;gap:4px;}
    .v2t-sm-bar{width:100%;height:28px;background:var(--cream-dark);border-radius:2px;position:relative;overflow:hidden;}
    .v2t-sm-fill{position:absolute;bottom:0;left:0;right:0;border-radius:2px;animation:v2tUp 0.9s cubic-bezier(.4,0,.2,1) forwards;animation-delay:var(--d,0s);}
    @keyframes v2tUp{from{height:0}to{height:var(--h)}}
    .v2t-sm-ltr{font-size:9px;font-weight:700;color:var(--muted-light);}
    .v2t-tabs-label{font-size:12px;color:var(--muted);margin:0 0 6px;}
    .v2t-tabs-hint{font-weight:500;color:var(--ink);}
    .v2t-result-tabs{display:flex;gap:2px;border-bottom:1px solid var(--border);margin-bottom:0;}
    .v2t-r-tab{font-size:13px;font-weight:500;color:var(--muted);padding:9px 16px;cursor:pointer;border-bottom:2px solid transparent;margin-bottom:-1px;}
    .v2t-r-tab.active{color:var(--accent);border-bottom-color:var(--accent);font-weight:600;}
    .v2t-r-content{display:none;}
    .v2t-r-content.active{display:block;}
    .v2t-transcript-box{background:var(--card);border:1px solid var(--border);border-top:none;border-radius:0 0 var(--radius) var(--radius);padding:18px 20px;max-height:380px;overflow-y:auto;font-size:14px;line-height:1.8;}
    .v2t-transcript-box p{margin-bottom:6px;}
    .v2t-minto-box .v2t-minto-wrap{padding:0;}
    .v2t-minto-label{font-size:11px;text-transform:uppercase;letter-spacing:0.05em;color:var(--muted);margin:14px 0 6px;}
    .v2t-minto-label:first-child{margin-top:0;}
    .v2t-minto-conclusion{font-weight:500;color:var(--ink);margin:0 0 8px;}
    .v2t-minto-list{margin:0 0 8px;padding-left:20px;}
    .v2t-minto-empty{color:var(--muted);font-size:13px;}
    .v2t-markdown-box{background:#1e2a1e;border:1px solid #2d4a2b;border-top:none;border-radius:0 0 var(--radius) var(--radius);padding:18px 20px;max-height:380px;overflow-y:auto;font-family:monospace;font-size:12.5px;line-height:1.8;color:#b8d4b5;white-space:pre-wrap;}
    .v2t-action-row{display:flex;justify-content:flex-end;gap:8px;padding:8px 12px;background:var(--cream-dark);border:1px solid var(--border);border-top:none;border-radius:0 0 var(--radius) var(--radius);}
    .v2t-act-btn{font-family:inherit;font-size:12px;font-weight:600;padding:6px 14px;border-radius:var(--radius);cursor:pointer;}
    .v2t-act-btn.solid{background:var(--accent);color:#fff;border:none;}
    .v2t-act-btn.solid:hover{background:var(--accent-mid);}
    .v2t-act-btn.outline{background:transparent;color:var(--muted);border:1px solid var(--border);}
    .v2t-act-btn.outline:hover{color:var(--accent);border-color:var(--accent-light);}
    .v2t-small-notes{margin-top:32px;border-top:1px solid var(--border);padding-top:20px;display:flex;flex-direction:column;gap:10px;}
    .v2t-note-item{display:grid;grid-template-columns:18px 1fr;gap:10px;font-size:13px;color:var(--muted);line-height:1.55;}
    .v2t-note-item strong{color:var(--ink);font-weight:600;}
    .v2t-bottom-links{margin-top:28px;display:flex;gap:16px;font-size:13px;}
    .v2t-bottom-links a{color:var(--accent-light);text-decoration:none;}
    .v2t-bottom-links a:hover{text-decoration:underline;}
    @media (max-width:480px){.v2t-input-row{flex-direction:column;}}
    """

    if result:
        stepps_html = "".join(
            f'<div class="v2t-sm-wrap" title="{_html_escape(s["name"])}">'
            f'<div class="v2t-sm-bar"><div class="v2t-sm-fill" style="--h:{(s["score"]/20)*100}%;--d:{i*0.08}s;background:{s["color"]}"></div></div>'
            f'<div class="v2t-sm-ltr">{s["letter"]}</div></div>'
            for i, s in enumerate(result["stepps"])
        )
        top_name = result["top_names"][0] if result["top_names"] else ""
        weak_name = result["weak_name"]
        if len(result["top_names"]) >= 2:
            detail_text = f"{top_name} + {result['top_names'][1]} firing. {weak_name} is weakest."
        else:
            detail_text = f"{top_name} is strongest · {weak_name} is weakest."
        transcript = result["transcript"]
        transcript_paras = "".join(
            f"<p>{_html_escape(p)}</p>"
            for p in transcript.split("\n") if p.strip()
        ) if "\n" in transcript else f"<p>{_html_escape(transcript)}</p>"
        markdown_escaped = _html_escape(result["markdown"])
        markdown_raw_escaped = _html_escape(result.get("markdown_raw") or result["markdown"])
        download_title = (result["title"] or "transcript").replace(" ", "-")[:50]
        download_title = "".join(c for c in download_title if c.isalnum() or c in "-_") or "transcript"
        fix_html = f'<strong>Quick fix:</strong> {result["fix"]}'
        transcript_file = result.get("transcript_file") or ""
        article_lab_href = (
            f"/article-lab?prefill_file={quote(transcript_file)}" if transcript_file else "/article-lab"
        )
        minto = result.get("minto") or {}
        has_minto = bool(minto.get("conclusion") or minto.get("key_points"))
        if has_minto:
            minto_conclusion = _html_escape((minto.get("conclusion") or "").strip())
            minto_points = "".join(
                f'<li>{_html_escape(p)}</li>' for p in (minto.get("key_points") or [])
            )
            minto_groups = minto.get("groups") or []
            minto_html = f'<div class="v2t-minto-wrap"><p class="v2t-minto-label">Conclusion (answer first)</p><p class="v2t-minto-conclusion">{minto_conclusion}</p>'
            if minto_points:
                minto_html += f'<p class="v2t-minto-label">Key points</p><ol class="v2t-minto-list">{minto_points}</ol>'
            if minto_groups:
                minto_html += '<p class="v2t-minto-label">Supporting evidence</p>'
                for idx, g in enumerate(minto_groups, 1):
                    ev_items = g.get("evidence") or []
                    ev_html = "".join(f'<li>{_html_escape(e)}</li>' for e in ev_items[:3])
                    if not ev_html:
                        ev_html = '<li>(No direct evidence extracted)</li>'
                    minto_html += f'<p class="v2t-minto-label" style="margin-top:8px;">For point {idx}</p><ul class="v2t-minto-list">{ev_html}</ul>'
            else:
                minto_evidence = "".join(
                    f'<li>{_html_escape(e)}</li>' for e in (minto.get("evidence") or [])[:5]
                )
                if minto_evidence:
                    minto_html += f'<p class="v2t-minto-label">Supporting evidence</p><ul class="v2t-minto-list">{minto_evidence}</ul>'
            minto_html += "</div>"
        else:
            minto_html = '<p class="v2t-minto-empty">Could not extract a stable Minto structure from this transcript yet. Try a cleaner transcript or longer sample.</p>'
        result_html = f"""
        <div class="v2t-result-section visible" id="v2tResult">
            <div class="v2t-score-summary">
                <div class="v2t-score-num">{result["score"]}</div>
                <div class="v2t-score-meta">
                    <div class="v2t-score-tier">{_html_escape(result["tier"])}</div>
                    <div class="v2t-score-detail">{_html_escape(detail_text)}</div>
                    <div class="v2t-stepps-mini">{stepps_html}</div>
                </div>
            </div>
            <div class="v2t-score-fix">{fix_html}</div>
            <div class="v2t-rewrite-cta" style="margin-top:14px;">
                <a class="v2t-submit-btn" href="{_html_escape(article_lab_href)}" style="display:inline-block;text-decoration:none;">{_html_escape(_t("video2text_goto_article_lab", lang))}</a>
            </div>
            <div style="margin-top:16px;">
                <p class="v2t-tabs-label">View: <span class="v2t-tabs-hint">Transcript · Structure (Minto) · Markdown</span></p>
                <div class="v2t-result-tabs">
                    <div class="v2t-r-tab active" data-tab="transcript" onclick="v2tSwitchTab('transcript')">Transcript</div>
                    <div class="v2t-r-tab" data-tab="structure" onclick="v2tSwitchTab('structure')">{_t("video2text_tab_structure", lang)}</div>
                    <div class="v2t-r-tab" data-tab="markdown" onclick="v2tSwitchTab('markdown')">Markdown</div>
                </div>
                <div class="v2t-r-content active" id="v2tTabTranscript">
                    <div class="v2t-transcript-box" id="v2tTranscriptContent">{transcript_paras}</div>
                    <div class="v2t-action-row"><button class="v2t-act-btn outline" onclick="v2tCopy('v2tTranscriptContent')">Copy</button></div>
                </div>
                <div class="v2t-r-content" id="v2tTabStructure">
                    <div class="v2t-transcript-box v2t-minto-box">{minto_html}</div>
                </div>
                <div class="v2t-r-content" id="v2tTabMarkdown">
                    <div class="v2t-markdown-box" id="v2tMarkdownContent">{markdown_escaped}</div>
                    <div class="v2t-action-row">
                        <label for="v2tExportMode" style="font-size:12px;color:var(--muted);align-self:center;">{_html_escape(_t("video2text_export_mode", lang))}:</label>
                        <select id="v2tExportMode" class="v2t-act-btn outline" style="padding:6px 10px;">
                            <option value="minto">{_html_escape(_t("video2text_export_minto", lang))}</option>
                            <option value="raw">{_html_escape(_t("video2text_export_raw", lang))}</option>
                        </select>
                        <button class="v2t-act-btn outline" onclick="v2tCopyMarkdown()">Copy</button>
                        <button class="v2t-act-btn solid" onclick="v2tDownload()">Download .md</button>
                    </div>
                </div>
            </div>
            <textarea id="v2tMarkdownData" style="display:none">{_html_escape(result["markdown"])}</textarea>
            <textarea id="v2tMarkdownDataRaw" style="display:none">{markdown_raw_escaped}</textarea>
            <script>window.v2tDownloadTitle={json.dumps(download_title)};</script>
        </div>"""
    else:
        result_html = '<div class="v2t-result-section" id="v2tResult"></div>'

    msg_parts: list[str] = []
    if msg:
        msg_parts.append(f'<p class="muted" style="margin-bottom:0.65rem;">{_html_escape(msg)}</p>')
    if fail_detail:
        msg_parts.append(
            '<pre style="white-space:pre-wrap;word-break:break-word;font-size:11px;line-height:1.45;'
            "max-height:340px;overflow:auto;padding:10px 12px;background:var(--cream-dark);"
            'border:1px solid var(--border);border-radius:var(--radius);margin-bottom:1rem;">'
            f'{_html_escape(fail_detail[:14000])}</pre>'
        )
    msg_combo = "".join(msg_parts)

    body = f"""
    <style>{v2t_styles}</style>
    <div class="v2t-wrap">
        <div class="v2t-page-header">
            <h1>📝 {_html_escape(_t("video2text_title", lang))}</h1>
            <p>{_html_escape(_t("video2text_tagline", lang))}</p>
        </div>
        <div class="v2t-accuracy-note">{_t("video2text_accuracy_note", lang)}</div>
        <div class="v2t-form-block">
            <form method="post" id="v2tForm">
                <div class="v2t-input-row">
                    <input type="text" name="url" class="v2t-url-input" id="v2tUrlInput" placeholder="{_html_escape(_t("video2text_placeholder", lang))}" value="{_html_escape(result["url"]) if result else ""}" required>
                    <button type="submit" class="v2t-submit-btn">{_html_escape(_t("video2text_btn", lang))}</button>
                    <button type="button" class="v2t-download-btn" id="v2tDownloadVideoBtn">{_html_escape(_t("video2text_download_btn", lang))}</button>
                    <button type="button" class="v2t-act-btn outline" id="v2tTestDouyinBtn">{_html_escape(_t("douyin_session_test_btn", lang))}</button>
                </div>
                <div class="v2t-form-hint">{_html_escape(_t("video2text_hint", lang))}<code>https://www.youtube.com/watch?v=dQw4w9WgXcQ</code> · <code>https://www.xiaoyuzhoufm.com/episode/...</code></div>
                <div class="v2t-form-hint">{_t("video2text_download_hint", lang)}</div>
                <div id="v2tDownloadStatus" class="v2t-form-hint" style="color:var(--ink);"></div>
                <div id="v2tDouyinSessionResult" class="v2t-form-hint"></div>
            </form>
            <div id="v2tProcessing" class="v2t-processing">
                <strong>{_html_escape(_t("video2text_processing", lang))}</strong>
                <small>{_html_escape(_t("video2text_processing_sub", lang))}</small>
            </div>
        </div>
        {msg_combo}
        {result_html}
        {f'<p class="section-desc muted" style="margin-top:0.85rem;">{_t("video2text_note_minto", lang)}</p>' if result else ''}
    </div>
    <script>
    function v2tSwitchTab(name){{
        document.querySelectorAll('.v2t-r-tab').forEach(function(t){{ t.classList.toggle('active', t.dataset.tab===name); }});
        document.querySelectorAll('.v2t-r-content').forEach(function(c){{ c.classList.toggle('active', c.id==='v2tTab'+name.charAt(0).toUpperCase()+name.slice(1)); }});
    }}
    function v2tCopy(id){{ navigator.clipboard.writeText(document.getElementById(id).innerText); }}
    function v2tSelectedMarkdown(){{
        var mode = document.getElementById('v2tExportMode')?.value || 'minto';
        if(mode === 'raw') return document.getElementById('v2tMarkdownDataRaw');
        return document.getElementById('v2tMarkdownData');
    }}
    function v2tCopyMarkdown(){{ var ta=v2tSelectedMarkdown(); if(ta) navigator.clipboard.writeText(ta.value); }}
    function v2tDownload(){{
        var ta=v2tSelectedMarkdown();
        if(!ta) return;
        var blob=new Blob([ta.value],{{type:'text/markdown'}});
        var a=document.createElement('a'); a.href=URL.createObjectURL(blob);
        var mode = document.getElementById('v2tExportMode')?.value || 'minto';
        var suffix = mode === 'raw' ? '-raw' : '-minto';
        a.download=(window.v2tDownloadTitle||'transcript')+suffix+'.md';
        a.click();
    }}
    (function(){{
        var form = document.getElementById('v2tForm');
        if (!form) return;
        var testBtn = document.getElementById('v2tTestDouyinBtn');
        var hintEl = document.getElementById('v2tDouyinSessionResult');
        form.addEventListener('submit', function(){{
            var box = document.getElementById('v2tProcessing');
            var btn = form.querySelector('.v2t-submit-btn');
            if (box) box.classList.add('visible');
            if (btn) {{
                btn.disabled = true;
                btn.textContent = {json.dumps(_t("video2text_processing", lang))};
            }}
        }});
        var dlBtn = document.getElementById('v2tDownloadVideoBtn');
        var dlStatus = document.getElementById('v2tDownloadStatus');
        if (dlBtn) {{
            dlBtn.addEventListener('click', async function(){{
                var input = document.getElementById('v2tUrlInput') || form.querySelector('input[name="url"]');
                var u = input ? input.value.trim() : '';
                if (!u) {{ if (dlStatus) dlStatus.textContent = {json.dumps(_t("youtube_dl_missing_url", lang))}; return; }}
                var old = dlBtn.textContent;
                dlBtn.disabled = true;
                if (dlStatus) dlStatus.textContent = {json.dumps(_t("video2text_download_progress", lang))};
                try {{
                    var r = await fetch('/api/youtube-download', {{
                        method: 'POST',
                        headers: {{ 'Content-Type': 'application/json' }},
                        body: JSON.stringify({{ url: u }})
                    }});
                    var d = await r.json();
                    if (!r.ok) {{
                        if (dlStatus) dlStatus.textContent = (d.error || 'Download failed') + (d.detail ? (' — ' + String(d.detail).slice(0, 400)) : '');
                        return;
                    }}
                    if (dlStatus) dlStatus.textContent = '';
                    var href = d.url && d.url.indexOf('http') === 0 ? d.url : (window.location.origin + (d.url || ''));
                    var a = document.createElement('a');
                    a.href = href;
                    if (d.file) a.download = d.file;
                    a.rel = 'noopener';
                    document.body.appendChild(a);
                    a.click();
                    document.body.removeChild(a);
                }} catch (e) {{
                    if (dlStatus) dlStatus.textContent = (e && e.message) ? e.message : 'Download failed';
                }} finally {{
                    dlBtn.disabled = false;
                    dlBtn.textContent = old;
                }}
            }});
        }}
        if (testBtn) {{
            testBtn.addEventListener('click', async function(){{
                var input = form.querySelector('input[name="url"]');
                var old = testBtn.textContent;
                testBtn.disabled = true;
                testBtn.textContent = {json.dumps(_t("douyin_session_testing", lang))};
                if (hintEl) hintEl.textContent = "";
                try {{
                    var r = await fetch('/api/test-douyin-session', {{
                        method: 'POST',
                        headers: {{ 'Content-Type': 'application/json' }},
                        body: JSON.stringify({{ url: input ? input.value : '' }})
                    }});
                    var d = await r.json();
                    if (hintEl) {{
                        var extra = (d.reasons && d.reasons.length) ? (' · ' + d.reasons.join(' | ')) : '';
                        hintEl.textContent = (d.message || 'Session check finished.') + extra;
                    }}
                }} catch (e) {{
                    if (hintEl) hintEl.textContent = (e && e.message) ? e.message : 'Session check failed';
                }} finally {{
                    testBtn.disabled = false;
                    testBtn.textContent = old;
                }}
            }});
        }}
    }})();
    var r=document.getElementById('v2tResult');
    if(r&&r.classList.contains('visible')) r.scrollIntoView({{behavior:'smooth',block:'start'}});
    </script>
    """
    return _base(_t("video2text_title", lang), body, "/video-to-text", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


def _load_creator_fields(lang="en", region="global"):
    base = "creator_fields_zh" if lang == "zh" else "creator_fields"
    if region != "global":
        fname = f"{base}_{region}.json"
        path = Path(__file__).parent / fname
        if path.exists():
            return json.loads(path.read_text(encoding="utf-8"))
    fname = f"{base}.json"
    path = Path(__file__).parent / fname
    if path.exists():
        return json.loads(path.read_text(encoding="utf-8"))
    if lang == "zh":
        path = Path(__file__).parent / "creator_fields.json"
        if path.exists():
            return json.loads(path.read_text(encoding="utf-8"))
    return {}


def _render_home():
    """Landing page with hero and links to sections."""
    lang = _get_lang()
    cards = [
        ("/daily", "📅", "link_daily", "link_daily_desc"),
        ("/news", "📰", "link_news", "link_news_desc"),
        ("/longform", "📚", "link_longform", "link_longform_desc"),
        ("/viral", "▶️", "link_viral", "link_viral_desc"),
        ("/china-content", "📥", "nav_china_content", "nav_china_content_desc"),
        ("/video-to-text", "📝", "link_video2text", "link_video2text_desc"),
        ("/article-lab", "✍️", "link_article_lab", "link_article_lab_desc"),
        ("/publish", "🚀", "link_publish", "link_publish_desc"),
        ("/science", "📊", "link_science", "link_science_desc"),
    ]
    cards_html = "".join(
        f'<a href="{href}" class="page-link-card"><span class="page-link-icon">{icon}</span>'
        f'<strong>{_html_escape(_t(title_key, lang))}</strong>'
        f'<small>{_html_escape(_t(desc_key, lang))}</small></a>'
        for href, icon, title_key, desc_key in cards
    )
    body = f'<section class="page-links">{cards_html}</section>'
    return _base(_t("home_title", lang), body, "/", hero=_hero_html(lang), lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/article-lab", methods=["GET", "POST"])
def article_lab():
    lang = _get_lang()
    msg = ""
    result = None
    results = []
    input_md = ""
    prefill_file_arg = (request.args.get("prefill_file") or "").strip()
    prefill_notice = ""
    if request.method == "GET" and prefill_file_arg:
        _pf_title, pre_content = load_markdown_from_output(prefill_file_arg)
        if pre_content.strip():
            input_md = pre_content
            prefill_notice = (
                f'<p class="muted" style="font-size:0.86rem;margin:0 0 0.6rem 0;">'
                f'{_html_escape(_t("article_lab_prefilled_from", lang))} '
                f'<code>{_html_escape(Path(prefill_file_arg).name)}</code></p>'
            )
        else:
            prefill_notice = (
                f'<p class="muted" style="font-size:0.86rem;margin:0 0 0.6rem 0;">'
                f'{_html_escape(_t("article_lab_prefill_missing", lang))} '
                f'<code>{_html_escape(Path(prefill_file_arg).name)}</code></p>'
            )
    if request.method == "POST":
        input_md = (request.form.get("markdown") or "").strip()
        file_items = [f for f in request.files.getlist("md_files") if f and f.filename]

        inputs = []
        if input_md:
            inputs.append(("pasted_markdown", input_md))
        for f in file_items:
            try:
                content = f.read().decode("utf-8", errors="replace")
            except Exception:
                content = ""
            if content.strip():
                inputs.append((f.filename or "uploaded_markdown", content))

        if not inputs:
            msg = "Please provide markdown text or upload a .md file."
        else:
            for source_name, content in inputs:
                try:
                    rewritten, meta = build_viral_article_markdown(content, lang=lang)
                    OUTPUT.mkdir(exist_ok=True)
                    out_id = str(abs(hash(source_name + rewritten)))[:12]
                    out_file = OUTPUT / f"article_rewrite_{out_id}.md"
                    out_file.write_text(rewritten, encoding="utf-8")
                    results.append(
                        {
                            "markdown": rewritten,
                            "file": out_file.name,
                            "title": meta.get("title", ""),
                            "berger": meta.get("berger_total", 0),
                            "contagious": meta.get("contagious_total", 0),
                            "source_name": source_name,
                        }
                    )
                except Exception as e:
                    msg = f"Error: {e}"
                    break
            if len(results) == 1:
                result = results[0]

    result_html = ""
    if results:
        cards = []
        download_files = []
        for idx, r in enumerate(results, 1):
            md_id = f"articleLabMarkdown{idx}"
            md_esc = _html_escape(r["markdown"])
            download_files.append(r["file"])
            cards.append(
                f"""
                <div class="tip-card">
                    <div class="tip-header">
                        <span class="tip-label">{_html_escape(r["title"])} · {_html_escape(r["source_name"])}</span>
                        <span class="tip-range-wrap">Berger {r["berger"]} · {_html_escape(_t("contagious_score", lang))} {r["contagious"]}</span>
                    </div>
                    <div class="v2t-markdown-box" id="{md_id}">{md_esc}</div>
                    <div class="v2t-action-row">
                        <button class="v2t-act-btn outline" onclick="navigator.clipboard.writeText(document.getElementById('{md_id}').innerText)">Copy</button>
                        <a class="v2t-act-btn outline" href="/publish?prefill_file={_html_escape(r['file'])}">{_html_escape(_t("article_lab_queue_publish", lang))}</a>
                        <a class="v2t-act-btn solid" href="/view/{_html_escape(r['file'])}" target="_blank" rel="noopener">{_html_escape(_t("article_lab_download", lang))}</a>
                    </div>
                </div>
                """
            )
        zip_link = ""
        if len(download_files) > 1:
            qs = "&".join(f"file={quote(x)}" for x in download_files)
            zip_link = f'<div style="margin:0.4rem 0 0.8rem 0;"><a class="v2t-act-btn solid" href="/api/article-lab/download-all?{qs}">Download all (.zip)</a></div>'
        result_html = zip_link + "\n".join(cards)

    body = f"""
    <section>
        <div class="section-title">✍️ {_html_escape(_t("article_lab_title", lang))}</div>
        <p class="section-desc muted">{_html_escape(_t("article_lab_desc", lang))}</p>
        {prefill_notice}
        <div class="tip-card" style="margin-bottom:0.9rem;">
            <p class="sources-label" style="font-size:0.86rem; line-height:1.6; margin:0;">{_html_escape(_t("article_lab_intro", lang))}</p>
        </div>
        <form method="post" enctype="multipart/form-data" class="tip-card">
            <div class="tip-header">
                <span class="tip-label">{_html_escape(_t("article_lab_upload_multi_label", lang))}</span>
            </div>
            <input type="file" name="md_files" multiple accept=".md,text/markdown" class="v2t-url-input" style="margin-bottom:0.8rem;">
            <div class="tip-header">
                <span class="tip-label">{_html_escape(_t("article_lab_input_label", lang))}</span>
            </div>
            <textarea name="markdown" class="v2t-url-input" style="min-height:220px; width:100%;">{_html_escape(input_md)}</textarea>
            <div style="margin-top:0.8rem;">
                <button type="submit" class="v2t-submit-btn" id="articleLabSubmit">{_html_escape(_t("article_lab_btn", lang))}</button>
                <span class="muted" style="margin-left:0.6rem; font-size:0.78rem;">{_html_escape(_t("article_lab_btn_batch", lang))}</span>
            </div>
            <div id="articleLabProcessing" class="v2t-processing" style="margin-top:0.55rem;">
                <strong>{_html_escape(_t("video2text_processing", lang))}</strong>
                <small>{_html_escape(_t("video2text_processing_sub", lang))}</small>
            </div>
        </form>
        {f'<p class="muted">{_html_escape(msg)}</p>' if msg else ''}
        {result_html}
    </section>
    <script>
    (function() {{
      var form = document.querySelector('form[enctype="multipart/form-data"]');
      if (!form) return;
      form.addEventListener('submit', function() {{
        var box = document.getElementById('articleLabProcessing');
        var btn = document.getElementById('articleLabSubmit');
        if (box) box.classList.add('visible');
        if (btn) {{
          btn.disabled = true;
          btn.textContent = {json.dumps(_t("video2text_processing", lang))};
        }}
      }});
    }})();
    </script>
    """
    return _base(_t("article_lab_title", lang), body, "/article-lab", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/publish", methods=["GET", "POST"])
def publish_lab():
    lang = _get_lang()
    msg = ""
    prefill_title = ""
    prefill_content = ""
    prefill_platform = request.args.get("platform", "x")
    prefill_file = (request.args.get("prefill_file") or "").strip()
    quality_warnings: list[str] = []
    x_ready = bool(os.environ.get("X_PUBLISH_WEBHOOK", "").strip())
    blog_ready = bool(os.environ.get("BLOG_PUBLISH_WEBHOOK", "").strip())
    if prefill_file:
        t, c = load_markdown_from_output(prefill_file)
        prefill_title = t
        prefill_content = c

    if request.method == "POST":
        action = (request.form.get("action") or "queue").strip()
        if action == "publish_now":
            item_id = (request.form.get("item_id") or "").strip()
            ok, out_msg, _item = publish_item_now(item_id)
            msg = out_msg if ok else out_msg
        else:
            title = (request.form.get("title") or "").strip()
            content = (request.form.get("content") or "").strip()
            platform = (request.form.get("platform") or "x").strip()
            schedule_at = (request.form.get("schedule_at") or "").strip()
            if not content:
                msg = "Please input content first."
            else:
                quality_warnings = validate_publish_content(content)
                item = queue_publish_item(
                    title=title or "Untitled post",
                    content=content,
                    platform=platform,
                    source_file=(request.form.get("source_file") or "").strip(),
                    schedule_at=schedule_at,
                )
                if item.get("duplicate"):
                    msg = f"Duplicate detected. Reusing existing item · {item.get('id')}"
                else:
                    msg = f"{_t('publish_status_queued', lang)} · {item.get('id')}"
                    prefill_title = ""
                    prefill_content = ""

    items = list_publish_items(limit=50)
    status_label = {
        "queued": _t("publish_status_queued", lang),
        "published": _t("publish_status_published", lang),
        "failed": _t("publish_status_failed", lang),
    }
    cards = []
    for it in items:
        title = _html_escape(it.get("title", "Untitled"))
        content = _html_escape((it.get("content", "") or "")[:360])
        platform = _html_escape(it.get("platform", "x"))
        status = it.get("status", "queued")
        status_text = _html_escape(status_label.get(status, status))
        created = _html_escape((it.get("created_at", "") or "").replace("T", " ")[:16])
        source_file = _html_escape(it.get("source_file", ""))
        extra = f" · {source_file}" if source_file else ""
        publish_btn = ""
        if status in ("queued", "failed"):
            publish_btn = f"""
            <form method="post" style="display:inline;">
                <input type="hidden" name="action" value="publish_now">
                <input type="hidden" name="item_id" value="{_html_escape(it.get("id", ""))}">
                <button type="submit" class="v2t-act-btn solid">{_html_escape(_t("publish_now_btn", lang))}</button>
            </form>
            """
        cards.append(
            f"""
            <div class="tip-card">
                <div class="tip-header">
                    <span class="tip-label">{title}</span>
                    <span class="tip-range-wrap">{status_text} · {platform} · {created}</span>
                </div>
                <p class="section-desc muted" style="margin:0.2rem 0 0.55rem 0;">{_html_escape(it.get("id", ""))}{extra}</p>
                <div class="v2t-markdown-box" style="max-height:220px;">{content}</div>
                <div class="v2t-action-row">{publish_btn}</div>
            </div>
            """
        )
    queue_html = "\n".join(cards) if cards else f'<p class="muted">{_html_escape(_t("publish_empty", lang))}</p>'
    warnings_html = ""
    if quality_warnings:
        warnings_html = "<div class='tip-card' style='margin-top:0.7rem;'><p class='section-desc muted' style='margin:0;'>" + " ".join(
            _html_escape(w) for w in quality_warnings
        ) + "</p></div>"
    readiness_html = (
        "<div class='tip-card' style='margin:0.75rem 0;'>"
        "<p class='section-desc muted' style='margin:0;'>"
        f"X webhook: {'configured' if x_ready else 'not set'} · "
        f"Blog webhook: {'configured' if blog_ready else 'not set'}. "
        "Without webhook, publish is logged locally only."
        "</p></div>"
    )

    body = f"""
    <section>
        <div class="section-title">🚀 {_html_escape(_t("publish_title", lang))}</div>
        <p class="section-desc muted">{_html_escape(_t("publish_desc", lang))}</p>
        <div class="tip-card" style="margin:0.75rem 0;">
            <div class="tip-header"><span class="tip-label">{_html_escape(_t("publish_handoff_title", lang))}</span></div>
            <p class="section-desc muted" style="margin:0;line-height:1.6;">{_html_escape(_t("publish_handoff_body", lang))}</p>
        </div>
        {readiness_html}
        <form method="post" class="tip-card">
            <div class="tip-header"><span class="tip-label">{_html_escape(_t("publish_form_title", lang))}</span></div>
            <input type="text" name="title" value="{_html_escape(prefill_title)}" placeholder="Post title" class="v2t-url-input" style="margin-bottom:0.6rem;">
            <textarea name="content" class="v2t-url-input" style="min-height:180px; width:100%; margin-bottom:0.6rem;" placeholder="Write content or prefill from rewritten markdown...">{_html_escape(prefill_content)}</textarea>
            <div class="tip-header" style="gap:0.55rem; flex-wrap:wrap;">
                <span class="tip-label">{_html_escape(_t("publish_platform", lang))}</span>
                <select name="platform" class="v2t-act-btn outline" style="padding:6px 10px;">
                    <option value="x"{" selected" if prefill_platform != "blog" else ""}>{_html_escape(_t("publish_to_x", lang))}</option>
                    <option value="blog"{" selected" if prefill_platform == "blog" else ""}>{_html_escape(_t("publish_to_blog", lang))}</option>
                </select>
                <label style="font-size:12px;color:var(--muted);">{_html_escape(_t("publish_schedule", lang))}</label>
                <input type="text" name="schedule_at" placeholder="2026-03-25 09:00 (optional)" class="v2t-url-input" style="max-width:230px;">
                <button type="submit" class="v2t-submit-btn">{_html_escape(_t("publish_add_btn", lang))}</button>
            </div>
            <input type="hidden" name="source_file" value="{_html_escape(prefill_file)}">
        </form>
        {warnings_html}
        {f'<p class="muted" style="margin-top:0.7rem;">{_html_escape(msg)}</p>' if msg else ""}
        <div class="cards" style="margin-top:0.9rem;">{queue_html}</div>
    </section>
    """
    return _base(_t("publish_title", lang), body, "/publish", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/api/article-lab/download-all")
def article_lab_download_all():
    files = request.args.getlist("file")
    safe_files = []
    for f in files:
        name = Path(f or "").name
        if name.endswith(".md"):
            safe_files.append(name)
    if not safe_files:
        return jsonify({"ok": False, "error": "No files selected."}), 400

    buf = io.BytesIO()
    with zipfile.ZipFile(buf, "w", compression=zipfile.ZIP_DEFLATED) as zf:
        for name in safe_files[:50]:
            p = OUTPUT / name
            if p.exists() and p.is_file():
                zf.writestr(name, p.read_text(encoding="utf-8", errors="replace"))
    buf.seek(0)
    return send_file(
        buf,
        mimetype="application/zip",
        as_attachment=True,
        download_name="article-lab-rewrites.zip",
    )


def _render_daily():
    """Daily News page only. Uses lang for UI and content (EN vs ZH). Fallback to EN when ZH empty."""
    _maybe_trigger_daily_refresh_if_stale()
    lang = _get_lang()
    items = _list_digests()
    news_type = "daily_news_zh" if lang == "zh" else "daily_news"
    daily_news_file = next((x for x in items if x["type"] == news_type), None)
    # When zh has no content, fallback to EN so page loads with something
    if lang == "zh" and daily_news_file:
        _, pre_parsed = parse_file(OUTPUT / daily_news_file["file"])
        if not pre_parsed:
            daily_news_file = next((x for x in items if x["type"] == "daily_news"), None)
            news_type = "daily_news" if daily_news_file else "daily_news_zh"
    elif lang == "zh" and not daily_news_file:
        daily_news_file = next((x for x in items if x["type"] == "daily_news"), None)
        news_type = "daily_news" if daily_news_file else "daily_news_zh"
    daily_cards = []
    if daily_news_file:
        path = OUTPUT / daily_news_file["file"]
        _, parsed = parse_file(path)
        run_type = "daily_news_zh" if lang == "zh" else "daily_news"
        parsed = add_lifecycle_to_items(run_type, parsed[:3])
        for item in parsed:
            snip = item.get("snippet", "") or ""
            c_score = evaluate_contagious_article(item.get("title", ""), snip, lang=lang)
            c_total = c_score.get("total", 0)
            c_top = " + ".join(c_score.get("top_signals", []))
            c_badge = (
                f'<span class="berger {_score_badge_class(c_total)}" title="{_html_escape(c_top)}">{_html_escape(_t("contagious_score", lang))} {c_total}</span>'
                if c_total > 0
                else ""
            )
            angles = generate_angles("", item.get("title", ""), snip, count=2, lang=lang)
            angles_html = "".join(f'<li>{_html_escape(a)}</li>' for a in angles)
            life = item.get("lifecycle", "peaking")
            life_label = _t(f"lifecycle_{life}", lang) if life in ("rising", "peaking", "fading") else life
            life_tt = _lifecycle_tooltip_html(lang, life)
            src = _get_source_for_item(item)
            src_type = _source_type_for_item(item)
            snip_html = f'<p class="card-snippet">{_html_escape(_sanitize_snippet(snip))}</p>' if snip else ""
            src_html = f'<p class="card-source">Source: {_html_escape(src)}</p>' if src else ""
            daily_cards.append(f'''<div class="card daily-card" data-source-type="{_html_escape(src_type)}">
                <div class="card-header"><h3><a href="{_html_escape(item["url"])}" target="_blank" rel="noopener">{_html_escape(item["title"])}</a></h3><span class="badges"><span class="lifecycle-wrap"><span class="lifecycle lifecycle-{life}">{_html_escape(life_label)}</span><span class="lifecycle-tooltip">{life_tt}</span></span>{c_badge}</span></div>
                {snip_html}
                {src_html}
                <div class="angles-wrap"><span class="angles-label">{_html_escape(_t("content_angles", lang))}:</span><ul class="angles-list">{angles_html}</ul></div>
            </div>''')
    no_news_msg = f'<div class="card"><p class="muted">{_t("no_news", lang)} <code>python main.py --daily-news</code></p></div>'
    daily_html = "\n".join(daily_cards) if daily_cards else no_news_msg
    view_file = "daily_news_zh.md" if lang == "zh" else "daily_news.md"
    news_url = "/news?lang=zh" if lang == "zh" else "/news"
    browse_more = f'<p class="browse-more"><a href="/view/{view_file}">{_t("view_full_digest", lang)}</a> · <a href="{news_url}">{_t("search_by_topic", lang)}</a> — {_t("whats_happening", lang)}</p>' if daily_cards else ""
    sources_label = ""
    sources_fname = "daily_news_sources_zh.txt" if lang == "zh" else "daily_news_sources.txt"
    sources_file = OUTPUT / sources_fname
    if sources_file.exists():
        sources_text = sources_file.read_text(encoding="utf-8").strip()
        if sources_text:
            sources_label = f'<p class="sources-label">{_t("sources", lang)}: {_html_escape(sources_text)} — {_t("sources_curated", lang)}</p>'

    updated_label = ""
    updated_file = OUTPUT / "daily_news_updated.txt"
    if updated_file.exists():
        updated_label = f'<p class="sources-label">{_t("last_updated", lang)}: {_html_escape(updated_file.read_text(encoding="utf-8").strip())}</p>'

    refresh_options = f'<p class="sources-label"><a href="/daily/refresh" class="refresh-link">{_t("refresh_now", lang)}</a> — {_t("refresh_desc", lang)}</p>'
    filter_bar = f'''
    <div class="tip-card" style="margin-top:0.75rem;">
        <div class="tip-header">
            <span class="tip-label">{_html_escape(_t("daily_focus_label", lang))}</span>
            <span class="tip-range-wrap">{_html_escape(_t("daily_focus_hint", lang))}</span>
        </div>
        <div class="tip-chips">
            <button type="button" class="tip-chip tip-chip-active daily-filter-btn" data-filter="all">{_html_escape(_t("daily_focus_all", lang))}</button>
            <button type="button" class="tip-chip daily-filter-btn" data-filter="news">{_html_escape(_t("daily_focus_news", lang))}</button>
            <button type="button" class="tip-chip daily-filter-btn" data-filter="longform">{_html_escape(_t("daily_focus_longform", lang))}</button>
            <button type="button" class="tip-chip daily-filter-btn" data-filter="podcast">{_html_escape(_t("daily_focus_podcast", lang))}</button>
        </div>
    </div>'''

    title_t = _t("daily_news_title", lang)
    body = f"""
    <section>
        <div class="section-title">📅 {title_t}</div>
        <p class="section-desc muted">{_t("daily_news_desc", lang)}</p>
        <div class="content-meta">
            {refresh_options}
            {updated_label}
            {sources_label}
        </div>
        {filter_bar}
        <div class="cards daily-cards-grid">{daily_html}</div>
        {browse_more}
    </section>
    <script>
    (function() {{
        var buttons = document.querySelectorAll(".daily-filter-btn");
        var cards = document.querySelectorAll(".daily-card");
        if (!buttons.length || !cards.length) return;
        function applyFilter(filter) {{
            cards.forEach(function(card) {{
                var t = card.getAttribute("data-source-type") || "news";
                card.style.display = (filter === "all" || t === filter) ? "" : "none";
            }});
            buttons.forEach(function(btn) {{
                btn.classList.toggle("tip-chip-active", btn.getAttribute("data-filter") === filter);
            }});
        }}
        buttons.forEach(function(btn) {{
            btn.addEventListener("click", function() {{
                applyFilter(btn.getAttribute("data-filter") || "all");
            }});
        }});
    }})();
    </script>
    """
    return _base(title_t, body, "/daily", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


def _render_field():
    """Your Field page only."""
    lang = _get_lang()
    region = _get_region()
    field_param = request.args.get("f", "").strip()
    creator_fields = _load_creator_fields(lang, region)
    cat_labels = [
        ("trends", _t("field_trends", lang)),
        ("color_forecasting", _t("field_color_forecasting", lang)),
        ("forecasting_sites", _t("field_forecasting_sites", lang)),
        ("resources", _t("field_resources", lang)),
    ]
    field_options = "".join(f'<option value="{k}"' + (' selected' if k == field_param else '') + f'>{_html_escape(v["name"])}</option>' for k, v in creator_fields.items())
    field_cards_html = ""
    if creator_fields:
        first_id = field_param if field_param in creator_fields else next(iter(creator_fields))
        first = creator_fields[first_id]
        for cat, label in cat_labels:
            items_list = first.get(cat, [])
            if items_list:
                field_cards_html += f'<div class="stepp"><h3>{_html_escape(label)}</h3><ul class="resource-list">'
                for r in items_list[:5]:
                    field_cards_html += f'<li><a href="{_html_escape(r["url"])}" target="_blank" rel="noopener">{_html_escape(r["name"])}</a> — {_html_escape(r.get("desc", ""))}</li>'
                field_cards_html += "</ul></div>"
        field_cards_html = field_cards_html or f'<p class="muted">{_t("field_select_above", lang)}</p>'

    cat_labels_js = json.dumps([[c, l] for c, l in cat_labels])
    empty_msg = _html_escape(_t("field_empty", lang))
    region_opts = [
        ("global", _t("field_region_global", lang)),
        ("americas", _t("field_region_americas", lang)),
        ("europe", _t("field_region_europe", lang)),
        ("asia", _t("field_region_asia", lang)),
    ]
    region_select = "".join(
        f'<option value="{r}"' + (' selected' if r == region else '') + f'>{_html_escape(l)}</option>'
        for r, l in region_opts
    )
    share_label = _html_escape(_t("field_share", lang))
    share_copied = _html_escape(_t("field_share_copied", lang))
    body = f"""
    <section>
        <div class="section-title">{_html_escape(_t("field_title", lang))}</div>
        <p class="section-desc muted">{_html_escape(_t("field_desc", lang))}</p>
        <div class="field-controls">
            <div class="field-control-group">
                <label class="field-control-label">{_html_escape(_t("field_region", lang))}</label>
                <select class="field-select" id="regionSelect" onchange="setRegion(this.value)">
                    {region_select}
                </select>
            </div>
            <div class="field-control-group">
                <label class="field-control-label">&nbsp;</label>
                <select class="field-select" id="fieldSelect" onchange="updateField(this.value); updateShareUrl();">
                    {field_options}
                </select>
            </div>
            <div class="field-control-group field-share-wrap">
                <div class="share-dropdown">
                    <button type="button" class="btn-share" id="shareBtn" title="{share_label}">↗ {share_label}</button>
                    <div class="share-panel" id="sharePanel">
                        <span class="share-panel-label">{_html_escape(_t("field_share_to", lang))}</span>
                        <a class="share-link" href="#" data-platform="whatsapp" target="_blank" rel="noopener">{_html_escape(_t("field_share_whatsapp", lang))}</a>
                        <a class="share-link" href="#" data-platform="telegram" target="_blank" rel="noopener">{_html_escape(_t("field_share_telegram", lang))}</a>
                        <a class="share-link" href="#" data-platform="x" target="_blank" rel="noopener">{_html_escape(_t("field_share_x", lang))}</a>
                        <a class="share-link" href="#" data-platform="linkedin" target="_blank" rel="noopener">{_html_escape(_t("field_share_linkedin", lang))}</a>
                        <button type="button" class="share-link share-copy" data-platform="copy">{_html_escape(_t("field_share_copy", lang))}</button>
                    </div>
                </div>
                <span class="share-feedback" id="shareFeedback"></span>
            </div>
        </div>
        <div id="fieldResources" class="stepps-grid">{field_cards_html}</div>
    </section>
    <script>
    const fields = {json.dumps(creator_fields)};
    const catLabels = {cat_labels_js};
    const emptyMsg = {json.dumps(empty_msg)};
    const shareCopied = {json.dumps(share_copied)};
    const baseUrl = window.location.origin + '/field';
    function updateShareUrl() {{
        const f = document.getElementById('fieldSelect').value;
        const r = document.getElementById('regionSelect').value;
        let url = baseUrl;
        const params = [];
        if (f) params.push('f=' + encodeURIComponent(f));
        if (r && r !== 'global') params.push('region=' + encodeURIComponent(r));
        if (params.length) url += '?' + params.join('&');
        return url;
    }}
    function updateField(id) {{
        const f = fields[id];
        if (!f) return;
        let html = '';
        for (const [cat, label] of catLabels) {{
            const items = f[cat] || [];
            if (items.length) {{
                html += '<div class="stepp"><h3>' + label + '</h3><ul class="resource-list">';
                items.slice(0,5).forEach(r => html += '<li><a href="' + r.url + '" target="_blank">' + r.name + '</a> — ' + (r.desc||'') + '</li>');
                html += '</ul></div>';
            }}
        }}
        document.getElementById('fieldResources').innerHTML = html || '<p class="muted">' + emptyMsg + '</p>';
    }}
    function setRegion(val) {{
        const f = document.getElementById('fieldSelect').value;
        let next = '/field';
        if (f) next += '?f=' + encodeURIComponent(f);
        if (val !== 'global') next += (next.indexOf('?') >= 0 ? '&' : '?') + 'region=' + encodeURIComponent(val);
        window.location = '/set-region?region=' + encodeURIComponent(val) + '&next=' + encodeURIComponent(next);
    }}
    const shareText = {json.dumps(_t("field_share_text", lang))};
    document.getElementById('shareBtn').addEventListener('click', function(e) {{
        e.stopPropagation();
        document.getElementById('sharePanel').classList.toggle('share-panel-open');
    }});
    document.addEventListener('click', function() {{ document.getElementById('sharePanel').classList.remove('share-panel-open'); }});
    document.getElementById('sharePanel').addEventListener('click', function(e) {{ e.stopPropagation(); }});
    document.querySelectorAll('.share-link').forEach(function(el) {{
        el.addEventListener('click', function(e) {{
            e.preventDefault();
            const url = updateShareUrl();
            const platform = el.getAttribute('data-platform');
            const text = shareText + ' ' + url;
            let shareUrl = '';
            if (platform === 'whatsapp') shareUrl = 'https://wa.me/?text=' + encodeURIComponent(text);
            else if (platform === 'telegram') shareUrl = 'https://t.me/share/url?url=' + encodeURIComponent(url) + '&text=' + encodeURIComponent(shareText);
            else if (platform === 'x') shareUrl = 'https://twitter.com/intent/tweet?url=' + encodeURIComponent(url) + '&text=' + encodeURIComponent(shareText);
            else if (platform === 'linkedin') shareUrl = 'https://www.linkedin.com/sharing/share-offsite/?url=' + encodeURIComponent(url);
            else if (platform === 'copy') {{
                navigator.clipboard.writeText(url).then(function() {{
                    document.getElementById('shareFeedback').textContent = shareCopied;
                    setTimeout(function() {{ document.getElementById('shareFeedback').textContent = ''; }}, 2000);
                }});
                document.getElementById('sharePanel').classList.remove('share-panel-open');
                return;
            }}
            if (shareUrl) {{ window.open(shareUrl, '_blank', 'noopener,noreferrer,width=600,height=500'); }}
            document.getElementById('sharePanel').classList.remove('share-panel-open');
        }});
    }});
    </script>
    """
    og_title = _t("field_title", lang)
    og_desc = _t("field_desc", lang)
    return _base(_t("nav_field", lang), body, "/field", lang=lang, og_title=og_title, og_desc=og_desc), 200, {"Content-Type": "text/html; charset=utf-8"}


def _render_news():
    """News by topic page only."""
    lang = _get_lang()
    topic_filter = (request.args.get("topic") or "").strip().replace(" ", "_")
    if lang == "zh":
        news_html_override = bool(topic_filter and not _has_cjk(topic_filter))
    else:
        news_html_override = bool(topic_filter and _has_cjk(topic_filter))
    news_cards = []
    time_range = request.args.get("range", "1d")
    if time_range not in ("60m", "1d", "7d"):
        time_range = "1d"
    range_labels = {"60m": _t("news_range_60m", lang), "1d": _t("news_range_1d", lang), "7d": _t("news_range_7d", lang)}
    search_form = f'''
        <form method="post" action="/api/search-news" class="form">
            <input type="hidden" name="range" value="{_html_escape(time_range)}">
            <input type="text" name="topic" placeholder="{_html_escape(_t("news_placeholder", lang))}" required>
            <button type="submit">{_t("search_by_topic", lang)}</button>
        </form>'''
    hot_tags = _hot_tags_for_lang(lang, limit=12)
    hot_tag_chips = "".join(
        f'<form method="post" action="/api/search-news" class="tip-chip-form"><input type="hidden" name="topic" value="{_html_escape(t)}"><input type="hidden" name="range" value="{_html_escape(time_range)}"><button type="submit" class="tip-chip">{_html_escape(t)}</button></form>'
        for t in hot_tags
    )
    hot_tags_card = f'''
        <div class="tip-card" style="margin-top:0.75rem;">
            <div class="tip-header">
                <span class="tip-label">{_html_escape(_t("hot_tags_title", lang))}</span>
                <span class="tip-range-wrap">{_html_escape(_t("hot_tags_desc", lang))}</span>
            </div>
            <div class="tip-chips">{hot_tag_chips}</div>
        </div>'''
    force_hot = request.args.get("refresh_hot") == "1"
    segment = _get_segment()
    tip_topics, tip_source, hot_cache_age = _get_most_searched(limit=12, time_range=time_range, lang=lang, force_hot_refresh=force_hot, segment=segment)
    segment_query = f"segment={{0}}&range={time_range}"
    if lang == "zh":
        segment_query += "&lang=zh"
    segment_pills = "".join(
        f'<a href="/news?{segment_query.format(seg_id)}" class="segment-pill{" active" if segment == seg_id else ""}">{AUDIENCE_SEGMENTS[seg_id]["name_zh"] if lang == "zh" else AUDIENCE_SEGMENTS[seg_id]["name_en"]}</a>'
        for seg_id in ("general", "tech", "entrepreneurs", "creator")
    )
    tip_chips = "".join(
        f'<form method="post" action="/api/search-news" class="tip-chip-form"><input type="hidden" name="topic" value="{_html_escape(t)}"><input type="hidden" name="range" value="{_html_escape(time_range)}"><button type="submit" class="tip-chip{" tip-chip-active" if topic_filter and raw_topic_file_stem(t) == topic_filter else ""}">{_html_escape(t)}</button></form>'
        for t in tip_topics[:12]
    )
    range_links = "".join(
        f'<a href="/news?range={r}" class="tip-range{" active" if r == time_range else ""}">{_html_escape(l)}</a>'
        for r, l in range_labels.items()
    )
    if tip_source == "platform":
        tip_label = _t("news_tip_platform_hot_en", lang) if lang == "en" else _t("news_tip_platform_hot", lang)
        hot_meta = ""
        if hot_cache_age is not None:
            refresh_url = "/news?refresh_hot=1" + ("&lang=en" if lang == "en" else "")
            if hot_cache_age < 60:
                hot_meta = f'<a href="{refresh_url}" class="refresh-link" title="{_t("news_refresh_hot", lang)}">↻ {_t("news_hot_just_now", lang)}</a>'
            else:
                mins = hot_cache_age // 60
                hot_meta = f'<a href="{refresh_url}" class="refresh-link" title="{_t("news_refresh_hot", lang)}">↻ {_t("news_hot_updated", lang).format(mins=mins)}</a>'
    elif tip_source == "app":
        tip_label = _t("news_tip_requested", lang).format(range=range_labels[time_range])
        hot_meta = ""
    else:
        tip_label = _t("news_tip_suggestions", lang)
        hot_meta = ""
    tip_card = f'''
        <div class="tip-card">
            <div class="tip-header">
                <span class="tip-label">{_html_escape(_t("hot_tags_title", lang))}</span>
                <span class="segment-pills-wrap">{segment_pills}</span>
            </div>
            <div class="tip-header">
                <span class="tip-label">{_html_escape(tip_label)}</span>
                <span class="tip-range-wrap">{range_links}{" · " + hot_meta if hot_meta else ""}</span>
            </div>
            <div class="tip-chips">{tip_chips}</div>
        </div>'''
    card_rows = _news_card_rows(lang, time_range, segment, force_hot, topic_filter)
    for row in card_rows:
        path = OUTPUT / row["file"]
        display_topic = row["display_topic"]
        query_topic = row["query_topic"]
        stem = row["stem"]
        try:
            raw_text = path.read_text(encoding="utf-8", errors="replace") if path.exists() else f"# No news found for: {query_topic}\n"
        except Exception:
            news_cards.append(
                f'<div class="card card-error"><h3>{_html_escape(display_topic)}</h3>'
                f"<p class=\"muted\">{_html_escape(_t('news_parse_error', lang))}</p></div>"
            )
            continue
        raw_head = raw_text.strip().split("\n", 1)[0] if raw_text.strip() else ""
        try:
            parsed = parse_raw_news(raw_text)
        except Exception:
            news_cards.append(
                f'<div class="card card-error"><h3>{_html_escape(display_topic)}</h3>'
                f"<p class=\"muted\">{_html_escape(_t('news_parse_error', lang))}</p></div>"
            )
            continue

        if not parsed:
            if raw_head.startswith("# Search error"):
                news_cards.append(
                    f'<div class="card card-error"><h3>{_html_escape(display_topic)}</h3>'
                    f'<p class="muted">{_t("news_search_failed", lang)}</p></div>'
                )
            else:
                if display_topic:
                    news_cards.append(
                        f'<div class="topic-label">{_t("news_topic_label", lang)}: {_html_escape(display_topic)}</div>'
                    )
                msg = (
                    _t("news_topic_no_hits", lang)
                    if raw_head.startswith("# No news found")
                    else _t("news_topic_unparsed", lang)
                )
                news_cards.append(
                    f'<div class="card" style="border:1px dashed var(--border);background:var(--cream-dark);">'
                    f'<p class="muted" style="margin:0 0 0.65rem 0;">{_html_escape(msg)}</p>'
                    f'<form method="post" action="/api/search-news" style="margin:0;">'
                    f'<input type="hidden" name="topic" value="{_html_escape(query_topic)}">'
                    f'<input type="hidden" name="range" value="{_html_escape(time_range)}">'
                    f'<button type="submit" class="v2t-submit-btn">{_html_escape(_t("news_rescan_topic", lang))}</button>'
                    f"</form></div>"
                )
            continue

        try:
            if display_topic:
                news_cards.append(f'<div class="topic-label">{_t("news_topic_label", lang)}: {_html_escape(display_topic)}</div>')
            run_type = f"raw_{stem}"
            parsed_life = add_lifecycle_to_items(run_type, parsed[:5])
            for item in parsed_life:
                snip = item.get("snippet", "") or ""
                c_score = evaluate_contagious_article(item.get("title", ""), snip, lang=lang)
                c_total = c_score.get("total", 0)
                c_top = " + ".join(c_score.get("top_signals", []))
                c_badge = (
                    f'<span class="berger {_score_badge_class(c_total)}" title="{_html_escape(c_top)}">{_html_escape(_t("contagious_score", lang))} {c_total}</span>'
                    if c_total > 0
                    else ""
                )
                angles = generate_angles(display_topic or "", item.get("title", ""), snip, count=2, lang=lang)
                angles_html = "".join(f'<li>{_html_escape(a)}</li>' for a in angles)
                life = item.get("lifecycle", "peaking")
                src = _get_source_for_item(item)
                snip_html = f'<p class="card-snippet">{_html_escape(_sanitize_snippet(snip))}</p>' if snip else ""
                src_label = _t("source", lang)
                src_html = f'<p class="card-source">{_html_escape(src_label)}: {_html_escape(src)}</p>' if src else ""
                life_label = _t(f"lifecycle_{life}", lang) if life in ("rising", "peaking", "fading") else life
                life_tt = _lifecycle_tooltip_html(lang, life)
                news_cards.append(f'''<div class="card">
                    <div class="card-header"><h3><a href="{_html_escape(item["url"])}" target="_blank" rel="noopener">{_html_escape(item["title"])}</a></h3><span class="badges">{c_badge}<span class="lifecycle-wrap"><span class="lifecycle lifecycle-{life}">{_html_escape(life_label)}</span><span class="lifecycle-tooltip">{life_tt}</span></span></span></div>
                    {snip_html}
                    {src_html}
                    <div class="angles-wrap"><span class="angles-label">{_html_escape(_t("content_angles", lang))}:</span><ul class="angles-list">{angles_html}</ul></div>
                </div>''')
        except Exception:
            news_cards.append(
                f'<div class="card card-error"><h3>{_html_escape(display_topic)}</h3>'
                f"<p class=\"muted\">{_html_escape(_t('news_parse_error', lang))}</p></div>"
            )

    if news_html_override:
        r = request.args.get("range", "1d")
        msg_key = "news_chinese_topic_in_en" if lang == "en" else "news_english_topic_in_zh"
        switch_url = f"/news?lang=zh&range={r}" if lang == "en" else f"/news?range={r}"
        news_html = f'<div class="card"><p class="muted">{_t(msg_key, lang)}</p><p class="browse-more"><a href="{switch_url}">{_t("news_show_all", lang)}</a></p></div>'
    else:
        if news_cards:
            news_html = "\n".join(news_cards)
        elif topic_filter:
            news_html = f'<div class="card"><p class="muted">{_t("news_searching", lang)}</p><p class="browse-more"><a href="{request.url}">↻ {_t("news_refresh_all", lang)}</a></p></div>'
        else:
            news_html = f'<div class="card"><p class="muted">{_t("news_no_results", lang)}</p></div>'

    refresh_label = _t("news_refresh_all", lang)
    refresh_interval = _t("news_refresh_interval", lang)
    refresh_meta = f'<div class="content-meta"><p class="sources-label"><a href="/news/refresh" class="refresh-link">{_html_escape(refresh_label)}</a> — {_html_escape(refresh_interval)}</p></div>'

    show_all_link = f'<p class="browse-more"><a href="/news?range={time_range}">{_html_escape(_t("news_show_all", lang))}</a></p>' if topic_filter else ""
    body = f"""
    <section>
        <div class="section-title">📰 {_html_escape(_t("news_title", lang))}</div>
        <p class="section-desc muted">{_html_escape(_t("news_desc", lang))}</p>
        {show_all_link}
        {search_form}
        {tip_card}
        {refresh_meta}
        <div class="cards">{news_html}</div>
    </section>
    """
    return _base(_t("news_title", lang), body, "/news", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


def _render_longform():
    """Long-form and podcast page. Focused deep-content feed."""
    lang = _get_lang()
    INITIAL_SHOW = 12
    cards = []
    force_refresh = request.args.get("refresh_longform") == "1"
    fetched_items, fetched_sources, from_cache, cache_age = _get_longform_feed(
        lang=lang, max_per_feed=6, force_refresh=force_refresh
    )
    deduped_items = []
    seen_titles = set()
    for item in fetched_items:
        title_key = (item.get("title") or "").lower().strip()
        if not title_key or title_key in seen_titles:
            continue
        seen_titles.add(title_key)
        deduped_items.append(item)

    theme_counts = {
        "all": 0,
        "transition": 0,
        "one_person_business": 0,
        "design_creativity_ai": 0,
    }
    for i, item in enumerate(deduped_items[:48], 1):
        snip = item.get("snippet", "") or ""
        title = item.get("title", "N/A")
        c_score = evaluate_contagious_article(title, snip, lang=lang)
        c_total = c_score.get("total", 0)
        c_top = " + ".join(c_score.get("top_signals", []))
        c_badge = (
            f'<span class="berger {_score_badge_class(c_total)}" title="{_html_escape(c_top)}">{_html_escape(_t("contagious_score", lang))} {c_total}</span>'
            if c_total > 0
            else ""
        )
        theme = classify_focus_theme(title, snip, lang)
        theme_counts["all"] += 1
        if theme in theme_counts:
            theme_counts[theme] += 1
        src = _get_source_for_item(item)
        src_type = _source_type_for_item(item)
        type_label = _t("daily_focus_podcast", lang) if src_type == "podcast" else _t("daily_focus_longform", lang)
        snip_html = f'<p class="card-snippet">{_html_escape(_sanitize_snippet(snip))}</p>' if snip else ""
        src_html = f'<p class="card-source">{_html_escape(_t("source", lang))}: {_html_escape(src)} · {_html_escape(type_label)}</p>' if src else ""
        hidden_class = " load-more-hidden" if i > INITIAL_SHOW else ""
        cards.append(
            f'''<div class="card longform-card{hidden_class}" data-focus-theme="{_html_escape(theme)}">
            <div class="card-header"><h3><a href="{_html_escape(item.get("url", ""))}" target="_blank" rel="noopener">{_html_escape(title)}</a></h3><span class="badges">{c_badge}</span></div>
            {snip_html}
            {src_html}
        </div>'''
        )

    if cards:
        longform_html = "\n".join(cards)
    else:
        longform_html = (
            f'<div class="card"><p class="muted">{_html_escape(_t("longform_no_items", lang))}</p>'
            f'<p class="browse-more"><a href="/daily">{_html_escape(_t("longform_view_daily", lang))}</a></p></div>'
        )

    sources_label = ""
    if fetched_sources:
        sources_text = ", ".join(fetched_sources)
        sources_label = f'<p class="sources-label">{_html_escape(_t("longform_desc", lang))} {_html_escape(sources_text)}</p>'
    scope_label = f'<p class="sources-label">{_html_escape(_t("longform_scope", lang))}</p>'
    active_regions = set(region_tags_for_lang(lang))
    region_pills = "".join(
        f'<span class="segment-pill{" active" if tag in active_regions else ""}" style="cursor:default;">{_html_escape(tag)}</span>'
        for tag in REGION_TAGS
    )
    region_tags_html = f'''
    <div class="tip-card" style="margin-top:0.75rem;">
        <div class="tip-header">
            <span class="tip-label">Source regions</span>
        </div>
        <div class="tip-chips">{region_pills}</div>
    </div>'''
    theme_pills = "".join(
        [
            f'<button type="button" class="tip-chip tip-chip-active longform-theme-btn" data-theme="all">{_html_escape(_t("longform_theme_all", lang))} ({theme_counts["all"]})</button>',
            f'<button type="button" class="tip-chip longform-theme-btn" data-theme="transition">{_html_escape(_t("longform_theme_transition", lang))} ({theme_counts["transition"]})</button>',
            f'<button type="button" class="tip-chip longform-theme-btn" data-theme="one_person_business">{_html_escape(_t("longform_theme_one_person", lang))} ({theme_counts["one_person_business"]})</button>',
            f'<button type="button" class="tip-chip longform-theme-btn" data-theme="design_creativity_ai">{_html_escape(_t("longform_theme_design_ai", lang))} ({theme_counts["design_creativity_ai"]})</button>',
        ]
    )
    theme_filter_html = f'''
    <div class="tip-card" style="margin-top:0.75rem;">
        <div class="tip-header">
            <span class="tip-label">{_html_escape(_t("longform_theme_label", lang))}</span>
        </div>
        <div class="tip-chips">{theme_pills}</div>
    </div>'''
    hot_tags = _hot_tags_for_lang(lang, limit=10)
    hot_tags_links = "".join(
        f'<a class="tip-chip" href="/news?topic={quote(raw_topic_file_stem(t), safe="")}{"&lang=zh" if lang=="zh" else ""}">{_html_escape(t)}</a>'
        for t in hot_tags
    )
    hot_tags_card = f'''
    <div class="tip-card" style="margin-top:0.75rem;">
        <div class="tip-header">
            <span class="tip-label">{_html_escape(_t("hot_tags_title", lang))}</span>
            <span class="tip-range-wrap">{_html_escape(_t("hot_tags_desc", lang))}</span>
        </div>
        <div class="tip-chips">{hot_tags_links}</div>
    </div>'''
    load_more_btn = ""
    if cards and len(cards) > INITIAL_SHOW:
        load_more_btn = f'''
        <button type="button" class="btn-load-more" id="longformLoadMore">{_t("load_more", lang)}</button>
        <script>
        (function() {{
            var btn = document.getElementById("longformLoadMore");
            var hidden = document.querySelectorAll(".longform-card.load-more-hidden");
            if (!btn || hidden.length === 0) return;
            btn.addEventListener("click", function() {{
                var toShow = 12;
                for (var i = 0; i < toShow && hidden.length > 0; i++) {{
                    hidden[0].classList.remove("load-more-hidden");
                    hidden = document.querySelectorAll(".longform-card.load-more-hidden");
                }}
                if (hidden.length === 0) btn.style.display = "none";
            }});
        }})();
        </script>'''

    refresh_url = "/longform?refresh_longform=1" + ("&lang=zh" if lang == "zh" else "")
    cache_meta = ""
    if from_cache and cache_age >= 0:
        cache_meta = " · " + _t("longform_cache_age", lang).format(mins=max(1, cache_age // 60))
    body = f"""
    <section>
        <div class="section-title">📚 {_html_escape(_t("longform_title", lang))}</div>
        <div class="content-meta">
            <p class="sources-label"><a href="{refresh_url}" class="refresh-link">{_html_escape(_t("refresh_now", lang))}</a> — {_html_escape(_t("longform_refresh_desc", lang))}{_html_escape(cache_meta)}</p>
            {sources_label}
            {scope_label}
        </div>
        {region_tags_html}
        {theme_filter_html}
        {hot_tags_card}
        <div class="cards">{longform_html}</div>
        {load_more_btn}
    </section>
    <script>
    (function() {{
        var buttons = document.querySelectorAll(".longform-theme-btn");
        var cards = document.querySelectorAll(".longform-card");
        if (!buttons.length || !cards.length) return;
        function applyTheme(theme) {{
            cards.forEach(function(card) {{
                var t = card.getAttribute("data-focus-theme") || "all";
                card.style.display = (theme === "all" || t === theme) ? "" : "none";
            }});
            buttons.forEach(function(btn) {{
                btn.classList.toggle("tip-chip-active", btn.getAttribute("data-theme") === theme);
            }});
        }}
        buttons.forEach(function(btn) {{
            btn.addEventListener("click", function() {{
                applyTheme(btn.getAttribute("data-theme") || "all");
            }});
        }});
    }})();
    </script>
    """
    return _base(_t("longform_title", lang), body, "/longform", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


def _render_viral():
    """Viral Videos page only."""
    lang = _get_lang()
    items = _list_digests()
    video_files = [x for x in items if x["type"] == "videos"]
    video_cards = []
    for x in video_files:
        path = OUTPUT / x["file"]
        _, parsed = parse_file(path)
        query = x.get("query", "")
        run_type = f"videos_{query}" if query else "videos_unknown"
        parsed_life = add_lifecycle_to_items(run_type, parsed[:6])
        for item in parsed_life:
            berger_val = item.get("berger")
            berger_display = f"{berger_val}/100" if berger_val is not None else "—"
            berger_class = "high" if (berger_val or 0) >= 50 else "mid" if (berger_val or 0) >= 25 else "low"
            life = item.get("lifecycle", "peaking")
            life_label = _t(f"lifecycle_{life}", lang) if life in ("rising", "peaking", "fading") else life
            life_tt = _lifecycle_tooltip_html(lang, life)
            desc = item.get("desc", "") or ""
            url = item.get("url", "") or ""
            platform = "Bilibili" if "bilibili" in url else "YouTube"
            angles = generate_angles(query or "", item.get("title", ""), desc, count=2, lang=lang)
            angles_html = "".join(f'<li>{_html_escape(a)}</li>' for a in angles)
            video_cards.append(f'''<div class="card">
                <div class="card-header">
                    <h3><a href="{_html_escape(url)}" target="_blank" rel="noopener">{_html_escape(item["title"])}</a></h3>
                    <span class="badges"><span class="berger berger-{berger_class}" title="{_html_escape(_t("viral_berger_methodology", lang))}">{berger_display}</span><span class="lifecycle-wrap"><span class="lifecycle lifecycle-{life}">{_html_escape(life_label)}</span><span class="lifecycle-tooltip">{life_tt}</span></span></span>
                </div>
                <p class="card-source">{platform} · Views: {_html_escape(item["views"])} · Magic: {_html_escape(item["magic"]) or "—"}</p>
                <p class="card-snippet">{_html_escape(desc)}</p>
                <div class="angles-wrap"><span class="angles-label">{_html_escape(_t("content_angles", lang))}:</span><ul class="angles-list">{angles_html}</ul></div>
            </div>''')

    content_angles_label = _t("content_angles", lang)
    viral_suggestions_label = _t("viral_suggestions", lang)
    viral_loading_t = _t("viral_loading", lang)
    viral_empty_title_t = _t("viral_empty_title", lang)
    viral_empty_body_t = _t("viral_empty_body", lang)
    viral_error_t = _t("viral_error", lang)
    viral_timeout_t = _t("viral_timeout", lang)
    # When lang=zh, only show Bilibili content — never English YouTube
    if lang == "zh":
        video_cards = [c for c in video_cards if "bilibili" in c]
    video_html = "\n".join(video_cards) if video_cards else f'<div class="card empty-state-loading"><p class="muted"><span class="loader-spinner"></span> {_html_escape(viral_loading_t)}</p></div>'

    default_source = "china" if lang == "zh" else "global"
    segment = _get_segment()
    seg_viral = SEGMENT_VIRAL_TIPS.get(segment, {}).get(lang)
    viral_tips = (seg_viral or (DEFAULT_VIRAL_TIPS_ZH if lang == "zh" else DEFAULT_VIRAL_TIPS))[:10]
    hot_tags = _hot_tags_for_lang(lang, limit=8)
    merged_tips = []
    for t in viral_tips + hot_tags:
        if t not in merged_tips:
            merged_tips.append(t)
    merged_tips = merged_tips[:14]
    tip_chips = "".join(
        f'<button type="button" class="tip-chip viral-tip-chip" data-query="{_html_escape(t)}">{_html_escape(t)}</button>'
        for t in merged_tips
    )
    viral_segment_pills = "".join(
        f'<a href="/viral?segment={seg_id}" class="segment-pill{" active" if segment == seg_id else ""}">{AUDIENCE_SEGMENTS[seg_id]["name_zh"] if lang == "zh" else AUDIENCE_SEGMENTS[seg_id]["name_en"]}</a>'
        for seg_id in ("general", "tech", "entrepreneurs", "creator")
    )

    search_form = f'''
        <form class="form" id="viralSearchForm">
            <input type="text" name="q" placeholder="{_html_escape(_t("viral_placeholder", lang))}" id="viralQuery">
            <button type="submit">{_html_escape(_t("viral_search_btn", lang))}</button>
        </form>'''
    source_select = f'''
        <label class="viral-source-label">{_html_escape(_t("viral_source_label", lang))}</label>
        <select name="source" id="viralSource" class="source-select viral-source-select">
            <option value="global"{' selected' if default_source == 'global' else ''}>{_html_escape(_t("viral_source_global", lang))}</option>
            <option value="china"{' selected' if default_source == 'china' else ''}>{_html_escape(_t("viral_source_china", lang))}</option>
        </select>'''
    tip_card = f'''
        <div class="tip-card">
            <div class="tip-header">
                <span class="tip-label">{_html_escape(_t("audience_segment_label", lang))}</span>
                <span class="segment-pills-wrap">{viral_segment_pills}</span>
            </div>
            <div class="tip-header">
                <span class="tip-label">{_html_escape(viral_suggestions_label)}</span>
                <span class="tip-range-wrap viral-source-wrap">{source_select}</span>
            </div>
            <div class="tip-chips">{tip_chips}</div>
        </div>'''
    berger_methodology = _t("viral_berger_methodology", lang)
    viral_empty_suggestions = merged_tips[:6]
    source_desc = f'<p class="sources-label">{_html_escape(_t("viral_what_we_serve", lang))} {_html_escape(_t("viral_china_note", lang))}</p>'

    ct_video = _t("content_type_video", lang)
    ct_post = _t("content_type_post", lang)
    ct_article = _t("content_type_article", lang)
    ct_note = _t("content_type_note", lang)
    body = f"""
    <section>
        <div class="section-title">▶️ {_html_escape(_t("viral_title", lang))} · {_html_escape(_t("viral_subtitle", lang))}</div>
        <p class="section-desc muted">{_html_escape(_t("viral_desc", lang))}</p>
        {search_form}
        {tip_card}
        {source_desc}
        <div id="viralLangHint" class="card" style="display:none; margin-bottom:1rem; border-left:4px solid var(--accent);"></div>
        <div class="cards" id="viralVideos">{video_html}</div>
    </section>
    <script>
    (function initViral() {{
        const form = document.getElementById('viralSearchForm');
        const input = document.getElementById('viralQuery');
        if (!form || !input) return;
        document.body.addEventListener('click', function(e) {{
            const chip = e.target.closest('.viral-tip-chip');
            if (chip) {{
                e.preventDefault();
                e.stopPropagation();
                const q = chip.getAttribute('data-query') || '';
                input.value = q;
                form.dispatchEvent(new Event('submit', {{ bubbles: true }}));
            }}
        }}, true);
        const params = new URLSearchParams(location.search);
        const q = params.get('q');
        const isZh = document.documentElement.lang === 'zh';
        if (q) {{
            input.value = q;
            setTimeout(function() {{ form.dispatchEvent(new Event('submit', {{ bubbles: true }})); }}, 100);
        }} else {{
            input.value = isZh ? '热门' : 'viral trending';
            setTimeout(function() {{ form.dispatchEvent(new Event('submit', {{ bubbles: true }})); }}, 100);
        }}
    }})();
    document.getElementById('viralSearchForm')?.addEventListener('submit', async (e) => {{
        e.preventDefault();
        const q = document.getElementById('viralQuery').value || 'viral trending';
        const src = document.getElementById('viralSource')?.value || 'global';
        const el = document.getElementById('viralVideos');
        const hintEl = document.getElementById('viralLangHint');
        if (hintEl) {{ hintEl.style.display = 'none'; hintEl.innerHTML = ''; }}
        el.innerHTML = '<div class="card empty-state-loading"><p class="muted"><span class="loader-spinner"></span> ' + {json.dumps(viral_loading_t)} + '</p></div>';
        const FETCH_TIMEOUT = 15000;
        const ctrl = new AbortController();
        const t = setTimeout(() => ctrl.abort(), FETCH_TIMEOUT);
        try {{
            const lang = document.documentElement.lang || 'en';
            const url = '/api/viral-videos?q=' + encodeURIComponent(q) + '&source=' + encodeURIComponent(src) + '&lang=' + encodeURIComponent(lang);
            const r = await fetch(url, {{ signal: ctrl.signal }});
            clearTimeout(t);
            const d = await r.json();
            if (d.error) {{
                const msg = (d.china_hint || (d.error || '').replace(/</g, '&lt;'));
                el.innerHTML = '<p class="muted">' + (d.china_hint ? msg : ({json.dumps(viral_error_t)} + ': ' + msg)) + '</p>';
                return;
            }}
            if (d.lang_hint && hintEl) {{
                hintEl.innerHTML = '<p class="muted">' + d.lang_hint + '</p>';
                hintEl.style.display = 'block';
            }}
            let html = '';
            (d.videos || []).forEach(v => {{
                const bergerNum = v.berger;
                const bergerDisplay = (bergerNum != null && bergerNum !== '') ? (bergerNum + '/100') : '—';
                const bc = (bergerNum != null && bergerNum >= 50) ? 'high' : (bergerNum != null && bergerNum >= 25) ? 'mid' : 'low';
                const bergerTitle = (v.interpretation || '').trim() || {json.dumps(berger_methodology)};
                const angles = (v.angles || []).map(a => '<li>' + a + '</li>').join('');
                const anglesLabel = {json.dumps(content_angles_label)};
                const platform = v.platform || 'YouTube';
                const ct = v.content_type || 'video';
                const ctLabel = ({{ video: {json.dumps(ct_video)}, post: {json.dumps(ct_post)}, article: {json.dumps(ct_article)}, note: {json.dumps(ct_note)} }})[ct] || ct;
                const url = (v.url || '').trim();
                const hasUrl = url && (url.startsWith('http://') || url.startsWith('https://'));
                const titleEsc = (v.title || '').replace(/</g, '&lt;').replace(/>/g, '&gt;');
                const titleEl = hasUrl ? '<a href="' + url.replace(/"/g, '&quot;') + '" target="_blank" rel="noopener">' + titleEsc + '</a>' : '<span>' + titleEsc + '</span>';
                html += '<div class="card"><div class="card-header"><h3>' + titleEl + '</h3><span class="badges"><span class="berger berger-' + bc + '" title="' + bergerTitle.replace(/"/g, '&quot;') + '">' + bergerDisplay + '</span></span></div><p class="card-source">' + platform + ' · ' + ctLabel + ' · Views: ' + (v.views || '') + ' · Magic: ' + (v.magic || '—') + '</p><p class="card-snippet">' + (v.desc || '').replace(/</g, '&lt;').replace(/>/g, '&gt;') + '</p>' + (angles ? '<div class="angles-wrap"><span class="angles-label">' + anglesLabel + ':</span><ul class="angles-list">' + angles + '</ul></div>' : '') + '</div>';
            }});
            if (!d.videos || d.videos.length === 0) {{
                const suggestions = {json.dumps(viral_empty_suggestions)};
                const emptyTitle = {json.dumps(viral_empty_title_t)};
                const emptyBody = {json.dumps(viral_empty_body_t)};
                const emptyHtml = '<div class="empty-state"><div class="empty-state-icon">🔍</div><div class="empty-state-title">' + emptyTitle + '</div><p class="empty-state-body">' + emptyBody + '</p><div class="empty-state-chips">' + suggestions.map(s => '<button type="button" class="tip-chip viral-tip-chip" data-query="' + String(s).replace(/"/g, '&quot;') + '">' + s + '</button>').join('') + '</div></div>';
                el.innerHTML = emptyHtml;
            }} else {{
                el.innerHTML = html;
            }}
        }} catch (err) {{
            const errMsg = err.name === 'AbortError' ? {json.dumps(viral_timeout_t)} : (err.message || '');
            const isChina = document.getElementById('viralSource')?.value === 'china';
            if (isChina) {{
                const suggestions = {json.dumps(viral_empty_suggestions)};
                const emptyTitle = {json.dumps(viral_empty_title_t)};
                const emptyBody = {json.dumps(viral_empty_body_t)};
                el.innerHTML = '<div class="empty-state"><div class="empty-state-icon">🔍</div><div class="empty-state-title">' + emptyTitle + '</div><p class="empty-state-body">' + emptyBody + '</p><div class="empty-state-chips">' + suggestions.map(s => '<button type="button" class="tip-chip viral-tip-chip" data-query="' + String(s).replace(/"/g, '&quot;') + '">' + s + '</button>').join('') + '</div></div>';
            }} else {{
                el.innerHTML = '<p class="muted">' + {json.dumps(viral_error_t)} + ': ' + (errMsg || '').replace(/</g, '&lt;') + '</p>';
            }}
        }}
    }});
    </script>
    """
    return _base(_t("viral_title", lang), body, "/viral", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/set-lang")
def set_lang():
    """Set language cookie and redirect. ?lang=zh|en&next=/daily"""
    lang = request.args.get("lang", "en")
    lang = "zh" if lang == "zh" else "en"
    next_url = request.args.get("next", "/")
    resp = make_response(redirect(next_url))
    resp.set_cookie("virallab-lang", lang, max_age=365 * 24 * 3600, path="/")
    return resp


@app.route("/set-region")
def set_region():
    """Set region cookie and redirect. ?region=global|americas|europe|asia&next=/field"""
    r = request.args.get("region", "global")
    r = r if r in ("global", "americas", "europe", "asia") else "global"
    next_url = request.args.get("next", "/field")
    if "//" in next_url or not next_url.startswith("/"):
        next_url = "/field"
    resp = make_response(redirect(next_url))
    resp.set_cookie("virallab-region", r, max_age=365 * 24 * 3600, path="/")
    return resp


@app.route("/", methods=["GET", "HEAD", "POST"])
def index():
    return _render_home()


@app.route("/campaign")
def campaign():
    return redirect("/")


@app.route("/daily")
def daily():
    return _render_daily()


@app.route("/daily/refresh")
def daily_refresh():
    """Manual refresh: run daily news fetch in background, redirect immediately."""
    next_url = request.args.get("next", "/daily")
    if next_url.startswith("/") and "//" not in next_url:
        pass  # safe relative path
    else:
        next_url = "/daily"
    import threading
    def _bg_refresh():
        try:
            _run_refresh_daily()
        except Exception:
            pass
    threading.Thread(target=_bg_refresh, daemon=True).start()
    return redirect(next_url)


def _run_refresh_videos():
    """Refresh trending viral videos (videos_trending_viral.md). Used by scheduler and /api/refresh-videos."""
    script = Path(__file__).parent / "scripts" / "video_trending.py"
    env = _env_no_proxy()
    env["PYTHONPATH"] = str(Path(__file__).parent)
    try:
        subprocess.run(
            [sys.executable, str(script), "trending viral"],
            capture_output=True,
            env=env,
            cwd=Path(__file__).parent,
            timeout=45,
        )
    except Exception:
        pass


def _run_refresh_news():
    """Refresh platform hot list, then re-fetch news only for current hot-tag set (not every raw_*.md)."""
    try:
        script = Path(__file__).parent / "scripts" / "fetch_hot_trending.py"
        env = _env_no_proxy()
        env["PYTHONPATH"] = str(Path(__file__).parent)
        subprocess.run([sys.executable, str(script)], capture_output=True, env=env, cwd=Path(__file__).parent, timeout=30)
    except Exception:
        pass
    topics = _topics_for_news_refresh()
    if not topics:
        return
    script = Path(__file__).parent / "scripts" / "search_only.py"
    env = _env_for_topic_search_subprocess()
    timeout_sec = _topic_search_subprocess_timeout_sec()
    for topic in topics:
        q = (topic or "").strip()
        if not q:
            continue
        try:
            subprocess.run(
                [sys.executable, str(script), q],
                check=True,
                env=env,
                cwd=Path(__file__).parent,
                timeout=timeout_sec,
            )
        except Exception as e:
            print(f"[news/refresh] topic {q!r}: {e}")


def _run_scheduled_china_crawl():
    """Lightweight B站 + 知乎 search for 「热门」; writes china_crawler_*.json. Opt-in via CHINA_SCHEDULED_CRAWL_MINUTES."""
    root = Path(__file__).parent
    OUTPUT.mkdir(parents=True, exist_ok=True)
    env = {**_env_no_proxy(), "PYTHONPATH": str(root), "CHINA_CRAWLER_OUTPUT_DIR": str(OUTPUT)}
    for platform, kw in (("bilibili", "热门"), ("zhihu", "热门")):
        try:
            subprocess.run(
                [
                    sys.executable,
                    "-m",
                    "tools.china_crawler",
                    "run",
                    "--platform",
                    platform,
                    "--type",
                    "search",
                    "--keywords",
                    kw,
                    "--max",
                    "12",
                ],
                capture_output=True,
                text=True,
                env=env,
                cwd=root,
                timeout=180,
            )
        except Exception as e:
            print(f"[scheduler] china crawl {platform}: {e}")


def _create_and_start_virallab_scheduler():
    """Build APScheduler jobs for daily digest, topic news, viral videos, longform, optional China crawl."""
    from apscheduler.schedulers.background import BackgroundScheduler

    def _job_daily():
        try:
            if not _run_refresh_daily():
                print("[scheduler] daily refresh failed")
        except Exception as e:
            print(f"[scheduler] daily refresh exception: {e}")

    def _job_news():
        try:
            _run_refresh_news()
        except Exception:
            pass

    def _job_videos():
        try:
            _run_refresh_videos()
        except Exception:
            pass

    def _job_longform():
        try:
            _kick_longform_refresh("en", max_per_feed=6)
            _kick_longform_refresh("zh", max_per_feed=6)
        except Exception:
            pass

    def _job_china():
        try:
            _run_scheduled_china_crawl()
        except Exception:
            pass

    sched = BackgroundScheduler()
    job_opts = {"coalesce": True, "max_instances": 1, "misfire_grace_time": 7200}
    try:
        daily_min = int(os.environ.get("DAILY_REFRESH_INTERVAL_MINUTES", "30"))
    except ValueError:
        daily_min = 30
    daily_min = max(10, min(daily_min, 180))
    try:
        loop_min = int(os.environ.get("NEWS_VIDEOS_LONGFORM_REFRESH_MINUTES", "60"))
    except ValueError:
        loop_min = 60
    loop_min = max(15, min(loop_min, 180))
    sched.add_job(_job_daily, "interval", minutes=daily_min, id="daily_news", **job_opts)
    sched.add_job(_job_news, "interval", minutes=loop_min, id="news_by_topic", **job_opts)
    sched.add_job(_job_videos, "interval", minutes=loop_min, id="viral_videos", **job_opts)
    sched.add_job(_job_longform, "interval", minutes=loop_min, id="longform_refresh", **job_opts)
    try:
        china_min = int(os.environ.get("CHINA_SCHEDULED_CRAWL_MINUTES", "0"))
    except ValueError:
        china_min = 0
    if china_min > 0:
        china_min = max(60, min(china_min, 10080))
        sched.add_job(_job_china, "interval", minutes=china_min, id="china_crawler", **job_opts)
    sched.start()
    extra = f" · China crawl: every {china_min} min" if china_min > 0 else ""
    print(
        f"ViralLab scheduler: daily every {daily_min} min · news/videos/longform every {loop_min} min{extra} "
        f"(DAILY_REFRESH_INTERVAL_MINUTES / NEWS_VIDEOS_LONGFORM_REFRESH_MINUTES / CHINA_SCHEDULED_CRAWL_MINUTES)"
    )
    if os.environ.get("LONGFORM_SKIP_STARTUP_REFRESH", "").strip().lower() not in ("1", "true", "yes"):

        def _startup_longform():
            try:
                _kick_longform_refresh("en", max_per_feed=6)
                _kick_longform_refresh("zh", max_per_feed=6)
                print("Startup: longform en/zh RSS refresh queued (background)")
            except Exception as e:
                print(f"Startup longform kick failed: {e}")

        threading.Thread(target=_startup_longform, daemon=True).start()
    if (
        not (OUTPUT / "daily_news.md").exists()
        or not (OUTPUT / "videos_trending_viral.md").exists()
        or not _longform_cache_path("en").exists()
        or not _longform_cache_path("zh").exists()
    ):

        def _bootstrap():
            try:
                _run_refresh_daily()
                _run_refresh_news()
                _run_refresh_videos()
                _kick_longform_refresh("en", max_per_feed=6)
                _kick_longform_refresh("zh", max_per_feed=6)
                print("Bootstrap: daily, topic news, videos, longform populated")
            except Exception as e:
                print(f"Bootstrap refresh failed: {e}")

        threading.Thread(target=_bootstrap, daemon=True).start()
    return sched


def ensure_virallab_background_scheduler() -> None:
    """Start APScheduler once per process (python server.py, gunicorn, or first HTTP request as fallback)."""
    global _virallab_apscheduler
    if os.environ.get("VIRALLAB_DISABLE_SCHEDULER", "").strip().lower() in ("1", "true", "yes"):
        return
    if os.environ.get("WERKZEUG_RUN_MAIN") == "false":
        return
    with _virallab_scheduler_lock:
        if _virallab_apscheduler is not None:
            return
        _virallab_apscheduler = _create_and_start_virallab_scheduler()


@app.before_request
def _virallab_scheduler_before_request():
    ensure_virallab_background_scheduler()


@app.route("/news/refresh")
def news_refresh():
    """Manual refresh: re-search all topics in background, redirect immediately."""
    import threading
    def _bg_refresh():
        try:
            _run_refresh_news()
        except Exception:
            pass
    threading.Thread(target=_bg_refresh, daemon=True).start()
    return redirect("/news")


@app.route("/field")
def field():
    return _render_field()


@app.route("/news")
def news():
    return _render_news()


@app.route("/longform")
def longform():
    return _render_longform()


# Default audience keywords; overridden by config/china_audience_keywords.json if present
_CHINA_AUDIENCE_KEYWORDS_DEFAULT = {
    "laid_off": {
        "name_zh": "大厂被裁员",
        "name_en": "Laid-off (big tech)",
        "keywords": ["一人公司", "超级个体", "裁员", "职场", "工作流", "副业", "再就业", "转型"],
    },
    "overseas_student": {
        "name_zh": "海外留学生",
        "name_en": "Overseas students",
        "keywords": ["留学", "求职", "一人公司", "远程工作", "海外", "实习", "职场"],
    },
    "small_boss": {
        "name_zh": "小老板",
        "name_en": "Small business owners",
        "keywords": ["一人公司", "超级个体", "创业", "副业", "小成本", "变现", "职场"],
    },
}

CONFIG_DIR = Path(__file__).parent / "config"
CHINA_AUDIENCE_KEYWORDS_FILE = CONFIG_DIR / "china_audience_keywords.json"


def get_china_audience_keywords():
    """Load audience keywords from config file, or return built-in default."""
    if CHINA_AUDIENCE_KEYWORDS_FILE.exists():
        try:
            data = json.loads(CHINA_AUDIENCE_KEYWORDS_FILE.read_text(encoding="utf-8"))
            if isinstance(data, dict) and data:
                return data
        except Exception:
            pass
    return _CHINA_AUDIENCE_KEYWORDS_DEFAULT


def _render_china_content():
    """China content page: Free/compliant first, hot topics, then audience keywords + optional fetch."""
    from urllib.parse import quote
    from src.china_sources import get_china_search_url

    lang = _get_lang()
    title = _t("china_content_page_title", lang)
    audience_keywords = get_china_audience_keywords()
    hot_topics: list[str] = []
    try:
        # Cache-only read: avoid blocking page render on slow upstream hot-list APIs.
        from src.hot_trending import CACHE_FILE, rerank_topics_by_segment
        if CACHE_FILE.exists():
            data = json.loads(CACHE_FILE.read_text(encoding="utf-8"))
            hot_topics = data.get("topics_zh", []) or []
        hot_topics = rerank_topics_by_segment(hot_topics or [], "zh", "creator")[:12]
    except Exception:
        pass
    audience_buttons = []
    for key, info in audience_keywords.items():
        name = info["name_zh"] if lang == "zh" else info["name_en"]
        audience_buttons.append(
            f'<button type="button" class="tip-chip china-audience-btn" data-audience="{key}">{_html_escape(name)}</button>'
        )
    audience_html = "".join(audience_buttons)
    fetch_desc = _t("viral_china_fetch_desc", lang)
    free_query = "热门"
    free_links = [
        ("/viral?source=china", _t("china_content_free_bilibili", lang)),
        (get_china_search_url("xiaohongshu", free_query), "小红书"),
        (get_china_search_url("douyin", free_query), "抖音"),
        ("https://www.zhihu.com/search?type=content&q=" + quote(free_query), "知乎"),
        (get_china_search_url("shipinhao", free_query), "视频号"),
        (get_china_search_url("xiaoyuzhou", free_query), "小宇宙"),
        (get_china_search_url("tiktok", free_query), "TikTok"),
    ]
    free_links_html = " ".join(
        f'<a href="{_html_escape(url)}" target="_blank" rel="noopener" class="segment-pill active" style="margin-right:0.35rem;">{_html_escape(label)}</a>'
        for url, label in free_links
    )
    hot_topics_html = ""
    if hot_topics:
        hot_topics_html = (
            '<div class="card china-panel">'
            '<div class="tip-header"><span class="tip-label">' + _html_escape(_t("china_content_hot_topics_label", lang)) + "</span></div>"
            '<div class="tip-chips" style="flex-wrap:wrap; gap:0.35rem;">'
            + "".join(
                '<button type="button" class="tip-chip china-hot-topic" data-keyword="' + _html_escape(t).replace('"', "&quot;") + '">' + _html_escape(t) + "</button>"
                for t in hot_topics
            )
            + "</div></div>"
        )
    seo_tags = _hot_tags_for_lang("zh", limit=12) or (hot_topics[:12] if hot_topics else [])
    seo_tags_html = ""
    if seo_tags:
        seo_tags_html = (
            '<div class="card china-panel">'
            '<div class="tip-header"><span class="tip-label">' + _html_escape(_t("hot_tags_title", lang)) + "</span></div>"
            '<p class="section-desc muted" style="margin:0 0 0.55rem 0;">' + _html_escape(_t("hot_tags_desc", lang)) + "</p>"
            '<div class="tip-chips" style="flex-wrap:wrap; gap:0.35rem;">'
            + "".join(
                '<button type="button" class="tip-chip china-hot-topic" data-keyword="' + _html_escape(t).replace('"', "&quot;") + '">' + _html_escape(t) + "</button>"
                for t in seo_tags
            )
            + "</div></div>"
        )
    body = f"""
    <section class="china-content-wrap">
        <div class="section-title">📥 {_html_escape(title)}</div>
        <p class="section-desc muted china-title-desc">{_html_escape(fetch_desc)}</p>
        <div class="china-layout-grid">
            <div class="china-stack">
                <div class="card china-panel" style="border-left: 4px solid var(--accent);">
                    <div class="tip-header"><span class="tip-label">{_html_escape(_t("china_content_free_compliant_title", lang))}</span></div>
                    <p class="section-desc muted" style="margin-bottom:0.45rem;">{_html_escape(_t("china_content_free_compliant_desc", lang))}</p>
                    <p class="section-desc muted" style="font-size:0.875rem; margin-bottom:0;">{_html_escape(_t("china_content_free_search_links", lang))}: {free_links_html}</p>
                </div>
                {hot_topics_html}
                {seo_tags_html}
                <div class="card china-panel">
                    <div class="tip-header"><span class="tip-label">{_html_escape(_t("china_content_audience_label", lang))}</span></div>
                    <div class="tip-chips" style="margin-bottom:0.7rem;">{audience_html}</div>
                    <div class="tip-header"><span class="tip-label">{_html_escape(_t("china_content_keyword_input_placeholder", lang))}</span></div>
                    <div class="tip-header" style="flex-wrap:wrap; gap:0.5rem; align-items:center;">
                        <input type="text" id="chinaKeywordInput" placeholder="e.g. open claw" style="max-width:16rem; padding:0.35rem 0.5rem;">
                        <button type="button" id="chinaAddTagBtn" class="btn">{_html_escape(_t("china_content_add_tag", lang))}</button>
                    </div>
                    <div class="tip-header" style="margin-top:0.45rem;"><span class="tip-label">{_html_escape(_t("china_content_tags_label", lang))}</span></div>
                    <div id="chinaActiveTags" class="tip-chips" style="min-height:2rem; margin-bottom:0.7rem;"></div>
                    <div class="tip-header"><span class="tip-label">{_html_escape(_t("viral_china_platforms", lang))}</span></div>
                    <div class="china-platform-row" style="margin-bottom:0.55rem;">
                        <label><input type="checkbox" name="cp" value="bilibili" class="china-cp" checked> B站</label>
                        <label><input type="checkbox" name="cp" value="xhs" class="china-cp" checked> 小红书</label>
                        <label><input type="checkbox" name="cp" value="douyin" class="china-cp" checked> 抖音</label>
                        <label><input type="checkbox" name="cp" value="zhihu" class="china-cp" checked> 知乎</label>
                        <label><input type="checkbox" name="cp" value="shipinhao" class="china-cp"> 视频号</label>
                    </div>
                    <div class="china-actions">
                        <button type="button" id="chinaFetchBtn" class="btn">{_html_escape(_t("viral_china_fetch_btn", lang))}</button>
                        <button type="button" id="chinaTestDouyinBtn" class="btn">{_html_escape(_t("douyin_session_test_btn", lang))}</button>
                        <p class="china-help-text">建议先从 B站 + 知乎开始，速度更快；小红书/抖音未登录时可能为空。</p>
                    </div>
                    <p id="chinaDouyinSessionResult" class="section-desc muted" style="margin:0.35rem 0 0.2rem 0;"></p>
                    <div class="tip-header" style="margin-top:0.4rem; flex-wrap:wrap; gap:0.5rem; align-items:center;">
                        <input type="text" id="chinaRewriteApiUrl" placeholder="模型改写 API URL（可选，例如 http://127.0.0.1:11434/api/generate）" style="max-width:28rem; width:100%; padding:0.35rem 0.5rem;">
                    </div>
                    <p class="section-desc muted" style="margin-top:0.35rem; margin-bottom:0;">不填也可先导出逐字稿 Markdown，再手动粘贴到任意 AI 改写。</p>
                    <div id="chinaResult" class="muted china-result-box" style="margin-top:0.65rem; display:none;"></div>
                </div>
            </div>
            <div class="china-stack">
                <div class="card china-panel">
                    <div class="tip-header"><span class="tip-label">{_html_escape(_t("china_content_login_title", lang))}</span></div>
                    <p class="section-desc muted" style="margin-bottom:0.45rem;">{_html_escape(_t("china_content_login_desc", lang))}</p>
                    <pre class="muted" style="background:var(--bg); padding:0.5rem 0.75rem; border-radius:6px; font-size:0.85rem; overflow-x:auto; white-space:pre-wrap; word-break:break-all; max-width:100%; box-sizing:border-box;">{_html_escape(_t("china_content_login_cmd", lang))}</pre>
                    <p class="section-desc muted" style="margin-top:0.35rem; font-size:0.875rem;">{_html_escape(_t("china_content_login_cmd_note", lang))}</p>
                    <p class="section-desc muted" style="margin-top:0.75rem; font-size:0.8rem; color:var(--muted); margin-bottom:0;">{_html_escape(_t("china_content_login_risk", lang))}</p>
                </div>
                <div class="card china-panel">
                    <p class="section-desc muted" style="margin:0 0 0.55rem 0;">{_html_escape(_t("china_content_after_fetch", lang))} <a href="/viral?source=china">Viral →</a></p>
                    <p class="section-desc muted" style="margin:0 0 0.55rem 0;">{_html_escape(_t("china_content_crawl_hint", lang))}</p>
                </div>
            </div>
            <div class="china-wide">
                <div class="section-title" style="margin-top:0.3rem;">{_html_escape(_t("china_content_fetched_title", lang))}</div>
                <div id="chinaContentList" class="cards" style="margin-top:0.45rem;">
                    <p class="muted">{_html_escape(_t("china_content_no_items_yet", lang))}</p>
                </div>
            </div>
        </div>
    </section>
    <script>
    (function() {{
        const durationTpl = {json.dumps(_t("china_crawl_duration", lang))};
        const audienceData = {json.dumps({k: v.get("keywords", []) for k, v in audience_keywords.items()})};
        let activeTags = [];
        function renderActiveTags() {{
            const el = document.getElementById('chinaActiveTags');
            if (!el) return;
            el.innerHTML = activeTags.map(function(kw, i) {{
                const esc = (kw || '').replace(/</g, '&lt;').replace(/>/g, '&gt;').replace(/"/g, '&quot;');
                return '<span class="segment-pill active" style="display:inline-flex;align-items:center;gap:0.25rem;"><span>' + esc + '</span><button type="button" class="china-tag-remove" data-i="' + i + '" aria-label="Remove">×</button></span>';
            }}).join(' ') || '<span class="muted">' + {json.dumps(_t("china_content_no_tags_yet", lang))} + '</span>';
            el.querySelectorAll('.china-tag-remove').forEach(function(btn) {{
                btn.addEventListener('click', function() {{
                    const i = parseInt(this.getAttribute('data-i'), 10);
                    activeTags = activeTags.filter(function(_, j) {{ return j !== i; }});
                    renderActiveTags();
                }});
            }});
        }}
        document.querySelectorAll('.china-audience-btn').forEach(function(btn) {{
            btn.addEventListener('click', function() {{
                const key = this.getAttribute('data-audience');
                const kws = audienceData[key] || [];
                kws.forEach(function(k) {{ if (k && activeTags.indexOf(k) === -1) activeTags.push(k); }});
                activeTags = activeTags.slice(0, 20);
                renderActiveTags();
            }});
        }});
        document.querySelectorAll('.china-hot-topic').forEach(function(btn) {{
            btn.addEventListener('click', function() {{
                const kw = (this.getAttribute('data-keyword') || '').trim();
                if (kw && activeTags.indexOf(kw) === -1) {{ activeTags.push(kw); activeTags = activeTags.slice(0, 20); renderActiveTags(); }}
            }});
        }});
        const inputEl = document.getElementById('chinaKeywordInput');
        document.getElementById('chinaAddTagBtn').addEventListener('click', function() {{
            const kw = (inputEl && inputEl.value) ? inputEl.value.trim() : '';
            if (kw && activeTags.indexOf(kw) === -1) {{ activeTags.push(kw); activeTags = activeTags.slice(0, 20); if (inputEl) inputEl.value = ''; renderActiveTags(); }}
        }});
        if (inputEl) inputEl.addEventListener('keydown', function(e) {{ if (e.key === 'Enter') {{ e.preventDefault(); document.getElementById('chinaAddTagBtn').click(); }} }});
        const platformNames = {{ bilibili: 'B站', xhs: '小红书', xiaohongshu: '小红书', douyin: '抖音', zhihu: '知乎', shipinhao: '视频号' }};
        const typeNames = {{ video: '视频', post: '帖子', article: '文章', note: '笔记' }};
        const evalLabels = {{ lang: {json.dumps(lang)}, evalBtn: {json.dumps(_t("china_content_eval_btn", lang))}, score: {json.dumps(_t("china_content_eval_score", lang))}, rewrite: {json.dumps(_t("china_content_rewrite_directions", lang))}, minto: {json.dumps(_t("china_content_minto_title", lang))}, loading: {json.dumps(_t("china_content_eval_loading", lang))}, error: {json.dumps(_t("china_content_eval_error", lang))} }};
        function buildChinaItemMarkdown(item) {{
            const title = (item.title || '').trim() || 'Untitled';
            const platform = (platformNames[item.platform] || item.platform || '-');
            const typeLabel = (typeNames[item.content_type] || item.content_type || '-');
            const views = (item.views || '-');
            const url = (item.url || '-');
            const desc = (item.description || item.snippet || '').trim();
            const transcript = (title + '\\n\\n' + desc).trim();
            const prompts = (evalLabels.lang === 'zh')
                ? [
                    '- "基于逐字稿写 3 个短视频开场钩子（15字内）"',
                    '- "按小红书风格改写成种草笔记，保留核心信息"',
                    '- "按抖音口播节奏改写成 30 秒脚本：开场-冲突-结论"',
                    '- "用金字塔原理重写：结论先行 + 3 个要点 + 对应论据"',
                ].join('\\n')
                : [
                    '- "Rewrite into 3 short-form video hooks (under 12 words each)"',
                    '- "Rewrite for Xiaohongshu note style while preserving key facts"',
                    '- "Rewrite as a 30-second Douyin voiceover script: hook-conflict-payoff"',
                    '- "Restructure with Minto Pyramid: conclusion first, then 3 key points with evidence"',
                ].join('\\n');
            return [
                '# China Content Transcript Export',
                '',
                '## Source',
                '',
                '- Title: ' + title,
                '- Platform: ' + platform,
                '- Type: ' + typeLabel,
                '- Views: ' + views,
                '- URL: ' + url,
                '',
                '---',
                '',
                '## Transcript',
                '',
                transcript,
                '',
                '---',
                '',
                '## AI Remix Prompts',
                '',
                prompts,
                '',
            ].join('\\n');
        }}
        function safeFileName(name) {{
            const n = (name || 'china-content').replace(/\\s+/g, '-').replace(/[^\\w\\-]/g, '').replace(/\\-+/g, '-').replace(/^\\-+|\\-+$/g, '');
            return (n || 'china-content').slice(0, 40);
        }}
        function downloadMarkdown(filename, markdown) {{
            const blob = new Blob([markdown], {{ type: 'text/markdown;charset=utf-8' }});
            const a = document.createElement('a');
            a.href = URL.createObjectURL(blob);
            a.download = filename;
            document.body.appendChild(a);
            a.click();
            a.remove();
            setTimeout(function() {{ URL.revokeObjectURL(a.href); }}, 500);
        }}
        function renderItems(items) {{
            const el = document.getElementById('chinaContentList');
            if (!el) return;
            if (!items || items.length === 0) {{
                el.innerHTML = '<p class="muted">' + {json.dumps(_t("china_content_no_items_yet", lang))} + '</p>';
                return;
            }}
            let html = '';
            items.forEach(function(v) {{
                const platform = platformNames[v.platform] || v.platform || '';
                const typeLabel = typeNames[v.content_type] || v.content_type || '';
                const title = (v.title || '').replace(/</g, '&lt;').replace(/>/g, '&gt;');
                const url = (v.url || '').replace(/"/g, '&quot;');
                const desc = (v.description || '').substring(0, 120).replace(/</g, '&lt;').replace(/>/g, '&gt;');
                const dataTitle = (v.title || '').replace(/"/g, '&quot;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
                const dataDesc = (v.description || '').replace(/"/g, '&quot;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
                const dataUrl = (v.url || '').replace(/"/g, '&quot;');
                const dataPlatform = (v.platform || '').replace(/"/g, '&quot;');
                const dataType = (v.content_type || '').replace(/"/g, '&quot;');
                const dataViews = (v.views || '').replace(/"/g, '&quot;');
                html += '<div class="card china-item-card" data-title="' + dataTitle + '" data-desc="' + dataDesc + '" data-url="' + dataUrl + '" data-platform="' + dataPlatform + '" data-content-type="' + dataType + '" data-views="' + dataViews + '"><div class="card-header"><h3><a href="' + url + '" target="_blank" rel="noopener">' + title + '</a></h3></div><p class="card-source">' + platform + ' · ' + typeLabel + (v.views ? ' · ' + v.views : '') + '</p><p class="card-snippet">' + desc + '</p><div class="tip-header" style="gap:0.4rem; flex-wrap:wrap; margin-top:0.5rem;"><button type="button" class="btn china-eval-btn">' + evalLabels.evalBtn + '</button><button type="button" class="btn china-export-md-btn">导出逐字稿.md</button><button type="button" class="btn china-ai-rewrite-btn">模型改写</button></div><div class="china-eval-result" style="display:none; margin-top:0.75rem; padding:0.75rem; background:var(--bg); border-radius:6px; font-size:0.875rem;"></div></div>';
            }});
            el.innerHTML = html;
            el.querySelectorAll('.china-export-md-btn').forEach(function(btn) {{
                btn.addEventListener('click', async function() {{
                    const card = this.closest('.china-item-card');
                    const resultDiv = card && card.querySelector('.china-eval-result');
                    if (!card) return;
                    const item = {{
                        title: card.getAttribute('data-title') || '',
                        description: card.getAttribute('data-desc') || '',
                        url: card.getAttribute('data-url') || '',
                        platform: card.getAttribute('data-platform') || '',
                        content_type: card.getAttribute('data-content-type') || '',
                        views: card.getAttribute('data-views') || '',
                    }};
                    this.disabled = true;
                    if (resultDiv) {{
                        resultDiv.style.display = 'block';
                        resultDiv.textContent = '正在生成 Markdown…';
                    }}
                    try {{
                        const r = await fetch('/api/china-content-export-md', {{
                            method: 'POST',
                            headers: {{ 'Content-Type': 'application/json' }},
                            body: JSON.stringify({{ item: item, lang: evalLabels.lang || 'zh' }}),
                        }});
                        const d = await r.json().catch(function() {{ return {{ ok: false, error: 'Server error' }}; }});
                        if (!d.ok || !d.markdown) {{
                            if (resultDiv) resultDiv.textContent = (d.error || '导出失败');
                            return;
                        }}
                        const filename = d.filename || ('transcript_' + safeFileName(item.title || 'china-content') + '.md');
                        downloadMarkdown(filename, d.markdown);
                        if (resultDiv) resultDiv.textContent = '导出成功：' + filename;
                    }} catch (e) {{
                        if (resultDiv) resultDiv.textContent = '导出失败: ' + (e.message || '');
                    }} finally {{
                        this.disabled = false;
                    }}
                }});
            }});
            el.querySelectorAll('.china-ai-rewrite-btn').forEach(function(btn) {{
                btn.addEventListener('click', async function() {{
                    const card = this.closest('.china-item-card');
                    const resultDiv = card && card.querySelector('.china-eval-result');
                    if (!card || !resultDiv) return;
                    const rewriteInput = document.getElementById('chinaRewriteApiUrl');
                    const rewriteApiUrl = (rewriteInput && rewriteInput.value ? rewriteInput.value : '').trim();
                    if (!rewriteApiUrl) {{
                        resultDiv.style.display = 'block';
                        resultDiv.textContent = '请先填写模型改写 API URL（支持本地端口）。';
                        return;
                    }}
                    const item = {{
                        title: card.getAttribute('data-title') || '',
                        description: card.getAttribute('data-desc') || '',
                        url: card.getAttribute('data-url') || '',
                        platform: card.getAttribute('data-platform') || '',
                        content_type: card.getAttribute('data-content-type') || '',
                        views: card.getAttribute('data-views') || '',
                    }};
                    const markdown = buildChinaItemMarkdown(item);
                    this.disabled = true;
                    resultDiv.style.display = 'block';
                    resultDiv.textContent = '模型改写中，请稍候…';
                    try {{
                        const r = await fetch('/api/china-content-rewrite', {{
                            method: 'POST',
                            headers: {{ 'Content-Type': 'application/json' }},
                            body: JSON.stringify({{
                                text: markdown,
                                lang: evalLabels.lang || 'zh',
                                rewrite_api_url: rewriteApiUrl,
                            }}),
                        }});
                        const d = await r.json().catch(function() {{ return {{ ok: false, error: 'Server error' }}; }});
                        if (!d.ok) {{
                            resultDiv.textContent = d.error || '模型改写失败';
                            return;
                        }}
                        const out = (d.rewritten_text || '').replace(/</g, '&lt;').replace(/>/g, '&gt;');
                        resultDiv.innerHTML = '<strong>模型改写结果：</strong><pre style="white-space:pre-wrap; margin-top:0.45rem;">' + out + '</pre>';
                    }} catch (e) {{
                        resultDiv.textContent = '模型改写失败: ' + (e.message || '');
                    }} finally {{
                        this.disabled = false;
                    }}
                }});
            }});
            el.querySelectorAll('.china-eval-btn').forEach(function(btn) {{
                btn.addEventListener('click', async function() {{
                    const card = this.closest('.china-item-card');
                    const resultDiv = card && card.querySelector('.china-eval-result');
                    if (!card || !resultDiv) return;
                    const text = (card.getAttribute('data-title') || '') + '\\n' + (card.getAttribute('data-desc') || '');
                    if (!text.trim()) {{ resultDiv.textContent = evalLabels.error; resultDiv.style.display = 'block'; return; }}
                    this.disabled = true;
                    resultDiv.textContent = evalLabels.loading;
                    resultDiv.style.display = 'block';
                    try {{
                        const r = await fetch('/api/china-content-evaluate', {{ method: 'POST', headers: {{ 'Content-Type': 'application/json' }}, body: JSON.stringify({{ text: text, lang: evalLabels.lang || 'zh' }}) }});
                        const d = await r.json().catch(function() {{ return {{ ok: false }}; }});
                        if (!d.ok) {{ resultDiv.innerHTML = '<span class="muted">' + (d.error || evalLabels.error) + '</span>'; return; }}
                        let out = '<strong>' + evalLabels.score + ':</strong> ' + (d.score || 0) + '/100';
                        if (d.stepps && d.stepps.length) out += ' <span class="muted">(' + d.stepps.map(function(s) {{ return s.name + ' ' + s.score; }}).join(', ') + ')</span>';
                        if (d.minto && (d.minto.conclusion || (d.minto.key_points && d.minto.key_points.length))) {{
                            out += '<br><strong>' + evalLabels.minto + ':</strong> ';
                            if (d.minto.conclusion) out += d.minto.conclusion.substring(0, 80) + (d.minto.conclusion.length > 80 ? '…' : '');
                            if (d.minto.key_points && d.minto.key_points.length) out += ' · ' + d.minto.key_points.length + ' key points';
                        }}
                        if (d.rewrite_directions && d.rewrite_directions.length) {{
                            out += '<br><strong>' + evalLabels.rewrite + ':</strong><ul style="margin:0.35rem 0 0 1rem;">';
                            d.rewrite_directions.forEach(function(line) {{ out += '<li>' + (line || '').replace(/</g, '&lt;').replace(/>/g, '&gt;') + '</li>'; }});
                            out += '</ul>';
                        }}
                        resultDiv.innerHTML = out;
                    }} catch (e) {{
                        resultDiv.textContent = evalLabels.error + ': ' + (e.message || '');
                    }} finally {{
                        this.disabled = false;
                    }}
                }});
            }});
        }}
        const btn = document.getElementById('chinaFetchBtn');
        const testDouyinBtn = document.getElementById('chinaTestDouyinBtn');
        const resultEl = document.getElementById('chinaResult');
        const sessionEl = document.getElementById('chinaDouyinSessionResult');
        if (!btn || !resultEl) return;
        if (testDouyinBtn) {{
            testDouyinBtn.addEventListener('click', async function() {{
                const old = testDouyinBtn.textContent;
                testDouyinBtn.disabled = true;
                testDouyinBtn.textContent = {json.dumps(_t("douyin_session_testing", lang))};
                if (sessionEl) sessionEl.textContent = '';
                try {{
                    const r = await fetch('/api/test-douyin-session', {{
                        method: 'POST',
                        headers: {{ 'Content-Type': 'application/json' }},
                        body: JSON.stringify({{}})
                    }});
                    const d = await r.json().catch(function() {{ return {{ ok:false, message:'session check failed' }}; }});
                    if (sessionEl) {{
                        const extra = (d.reasons && d.reasons.length) ? (' · ' + d.reasons.join(' | ')) : '';
                        sessionEl.textContent = (d.message || '') + extra;
                    }}
                }} catch (e) {{
                    if (sessionEl) sessionEl.textContent = (e && e.message) ? e.message : 'session check failed';
                }} finally {{
                    testDouyinBtn.disabled = false;
                    testDouyinBtn.textContent = old;
                }}
            }});
        }}
        btn.addEventListener('click', async function() {{
            const platforms = Array.from(document.querySelectorAll('.china-cp:checked')).map(c => c.value);
            if (!platforms.length) {{ alert('Select at least one platform'); return; }}
            const keywords = activeTags.length > 0 ? activeTags.slice(0, 10) : ['热门'];
            btn.disabled = true;
            const oldBtnText = btn.textContent;
            btn.textContent = '拉取中…';
            resultEl.style.display = 'none';
            resultEl.innerHTML = '';
            try {{
                resultEl.style.display = 'block';
                resultEl.textContent = '正在拉取，请稍候… 平台: ' + platforms.join(', ') + ' · 关键词: ' + keywords.slice(0,3).join(', ');
                const controller = new AbortController();
                const timeoutId = setTimeout(function() {{ controller.abort(); }}, 65000);
                const r = await fetch('/api/china-crawl', {{
                    method: 'POST',
                    headers: {{ 'Content-Type': 'application/json' }},
                    body: JSON.stringify({{ platforms: platforms, keywords: keywords }}),
                    signal: controller.signal
                }});
                clearTimeout(timeoutId);
                const d = await r.json().catch(() => ({{ ok: false, error: 'Server error' }}));
                resultEl.style.display = 'block';
                if (!d.ok) {{
                    resultEl.innerHTML = '<span style="color:var(--cta);">' + (d.error || 'Crawl failed').replace(/</g, '&lt;') + '</span>';
                }} else {{
                    const names = {{ bilibili: 'B站', xhs: '小红书', douyin: '抖音', zhihu: '知乎', shipinhao: '视频号' }};
                    let msg = (d.message || '') + (d.fetched && Object.keys(d.fetched).length ? ' ' + Object.entries(d.fetched).map(([k,v]) => names[k] + ' ' + v).join(', ') : '');
                    if (d.duration_sec != null) msg += ' · ' + durationTpl.replace('{{sec}}', d.duration_sec);
                    if (d.errors && d.errors.length) msg += '<br><span class="china-result-box" style="display:block; margin-top:0.5rem;">' + d.errors.map(e => (e || '').replace(/</g, '&lt;').replace(/>/g, '&gt;')).join('<br>') + '</span>';
                    if (d.platform_status) {{
                        const statusLabel = {{ ok: 'ok', partial: 'partial', blocked_login: 'need login', blocked_risk: 'risk control', network_error: 'network/error' }};
                        const rows = Object.keys(d.platform_status).map(function(pk) {{
                            const s = d.platform_status[pk] || {{}};
                            const n = names[pk] || pk;
                            const label = statusLabel[s.status] || (s.status || 'unknown');
                            return '<span class="segment-pill" style="margin-right:0.3rem;">' + n + ': ' + label + '</span>';
                        }});
                        if (rows.length) msg += '<br><div style="margin-top:0.45rem;">' + rows.join('') + '</div>';
                    }}
                    if (d.fallback_links && d.fallback_links.length) {{
                        const fb = d.fallback_links.slice(0, 5).map(function(x) {{
                            const n = names[x.platform] || x.platform;
                            const u = (x.url || '').replace(/"/g, '&quot;');
                            const k = (x.keyword || '').replace(/</g, '&lt;').replace(/>/g, '&gt;');
                            return '<a class="segment-pill active" target="_blank" rel="noopener" href="' + u + '" style="margin-right:0.35rem;">' + n + ' 搜索 ' + k + '</a>';
                        }});
                        if (fb.length) msg += '<br><div style="margin-top:0.45rem;">' + fb.join('') + '</div>';
                    }}
                    resultEl.innerHTML = msg;
                    renderItems(d.items || []);
                }}
            }} catch (err) {{
                if (err && err.name === 'AbortError') {{
                    resultEl.textContent = '请求超时：平台响应较慢。请先勾选 B站 / 知乎，或减少关键词后重试。';
                }} else {{
                    resultEl.textContent = err.message || 'Request failed';
                }}
            }} finally {{
                btn.disabled = false;
                btn.textContent = oldBtnText;
            }}
        }});
        renderActiveTags();
    }})();
    </script>
    """
    return _base(title, body, "/china-content", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/docs/bilibili-vpn")
def docs_bilibili_vpn():
    """Help page: fix B站 loading when in China with VPN."""
    lang = _get_lang()
    title_zh = "解决 B 站加载问题"
    title_en = "Fix Bilibili loading"
    title = title_zh if lang == "zh" else title_en
    body_zh = """
    <section>
        <div class="section-title">中国境内 + VPN 时让 B 站可用</div>
        <p class="section-desc muted">你在国内、本机开 VPN 时，B 站请求会从 VPN 出口出去，被识别为海外导致无法访问。任选一种方式即可。</p>
        <div class="card" style="max-width:640px;">
            <div class="tip-header"><span class="tip-label">方式一：分流 / 直连（推荐）</span></div>
            <p class="section-desc">在 VPN 或代理客户端里，把 <strong>bilibili.com</strong>、<strong>api.bilibili.com</strong> 设为<strong>直连</strong>（不走代理）。</p>
            <ul class="section-desc" style="margin-left:1.25rem;">
                <li><strong>Clash / ClashX</strong>：规则里加 <code>DOMAIN-SUFFIX,bilibili.com,DIRECT</code></li>
                <li>其他客户端：在「分流」「直连名单」里加上 <code>bilibili.com</code></li>
            </ul>
        </div>
        <div class="card" style="max-width:640px; margin-top:1rem;">
            <div class="tip-header"><span class="tip-label">方式二：单独给 B 站设代理</span></div>
            <p class="section-desc">若无法分流，在项目根目录建 <code>.env</code>，写入：</p>
            <pre class="muted" style="background:var(--bg); padding:0.75rem; border-radius:6px; font-size:0.85rem; overflow-x:auto;">BILIBILI_PROXY=http://127.0.0.1:18080</pre>
            <p class="section-desc muted" style="margin-top:0.35rem;">与 Cursor 一致：本机 gost 固定口（Quick Queue 上游由 qq-http-bridge 自动对齐）。若你用的是别的 HTTP 前端，把端口改成你的监听口。</p>
            <p class="section-desc muted" style="margin-top:0.5rem;">保存后重启 <code>python server.py</code>，再在「热门视频」选中国来源重试。</p>
        </div>
        <p class="section-desc muted" style="margin-top:1rem;"><a href="/viral">← 返回热门视频</a></p>
    </section>
    """
    body_en = """
    <section>
        <div class="section-title">Fix Bilibili when in China with VPN</div>
        <p class="section-desc muted">When you use VPN, Bilibili requests go through it and get blocked. Use either method below.</p>
        <div class="card" style="max-width:640px;">
            <div class="tip-header"><span class="tip-label">Option 1: Split tunnel (recommended)</span></div>
            <p class="section-desc">In your VPN/proxy app, set <strong>bilibili.com</strong> to <strong>DIRECT</strong> (bypass proxy).</p>
            <ul class="section-desc" style="margin-left:1.25rem;">
                <li><strong>Clash</strong>: Add rule <code>DOMAIN-SUFFIX,bilibili.com,DIRECT</code></li>
            </ul>
        </div>
        <div class="card" style="max-width:640px; margin-top:1rem;">
            <div class="tip-header"><span class="tip-label">Option 2: Bilibili-only proxy</span></div>
            <p class="section-desc">Create <code>.env</code> in project root with:</p>
            <pre class="muted" style="background:var(--bg); padding:0.75rem; border-radius:6px; font-size:0.85rem;">BILIBILI_PROXY=http://127.0.0.1:18080</pre>
            <p class="section-desc muted" style="margin-top:0.35rem;">Same stable local HTTP front as Cursor (gost); Quick Queue upstream is updated by qq-http-bridge. Use your listen port if different.</p>
            <p class="section-desc muted" style="margin-top:0.5rem;">Restart the server, then try China source again on the Viral page.</p>
        </div>
        <p class="section-desc muted" style="margin-top:1rem;"><a href="/viral">← Back to Viral</a></p>
    </section>
    """
    body = body_zh if lang == "zh" else body_en
    return _base(title, body, "/viral", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/china-content")
@app.route("/china-content/")
def china_content():
    if request.path.rstrip("/") != request.path and request.path.endswith("/"):
        return redirect("/china-content", code=302)
    try:
        return _render_china_content()
    except Exception as e:
        lang = _get_lang()
        err_msg = str(e)[:200] if str(e) else "Unknown error"
        body = f'<section><p class="section-desc muted">{_html_escape(_t("china_content_load_error", lang))}</p><p class="muted">Error: {_html_escape(err_msg)}</p><p><a href="/">' + (_html_escape("返回首页") if lang == "zh" else "Home") + '</a> · <a href="/china-content">' + (_html_escape("重试") if lang == "zh" else "Retry") + "</a></p></section>"
        return _base(_t("china_content_page_title", lang), body, "/china-content", lang=lang), 200, {"Content-Type": "text/html; charset=utf-8"}


@app.route("/viral")
def viral():
    return _render_viral()


if __name__ == "__main__":
    import os
    import socket

    OUTPUT.mkdir(exist_ok=True)
    host = "0.0.0.0"  # Bind all interfaces so 127.0.0.1:5001 works locally
    debug = not os.environ.get("PORT")  # Auto-reload when running locally
    port = int(os.environ.get("PORT", 5001))
    if not os.environ.get("PORT"):
        # Keep a stable port across Werkzeug reloader parent/child processes.
        fixed = os.environ.get("VIRALLAB_DEV_PORT")
        if fixed:
            port = int(fixed)
        elif os.environ.get("WERKZEUG_RUN_MAIN") == "true":
            port = 5001
        else:
            chosen = 5001
            for candidate in range(5001, 5010):
                try:
                    s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
                    s.bind(("", candidate))
                    s.close()
                    chosen = candidate
                    break
                except OSError:
                    continue
            os.environ["VIRALLAB_DEV_PORT"] = str(chosen)
            port = chosen
            if port != 5001:
                print(f"Port 5001 in use, using http://127.0.0.1:{port}")

    # Auto-refresh: same as gunicorn — start scheduler in this process (reloader child or no-reloader).
    if os.environ.get("WERKZEUG_RUN_MAIN") == "true" or not debug:
        ensure_virallab_background_scheduler()

    dev_url = f"http://127.0.0.1:{port}"
    print(f"ViralLab at {dev_url}" + (" (auto-restart on save)" if debug else ""))

    if debug:
        # Werkzeug reloader: restarts the Python process on every .py save so
        # code changes take effect immediately. The page JS polls and auto-refreshes.
        app.run(host="127.0.0.1", port=port, debug=True, use_reloader=True)
    else:
        app.run(host=host, port=port, debug=False, use_reloader=False)


# gunicorn/uvicorn load ``server:app`` without executing ``if __name__ == "__main__"`` — start jobs at import.
if __name__ != "__main__":
    _ss = os.environ.get("SERVER_SOFTWARE", "")
    if _ss.startswith("gunicorn") or _ss.startswith("uvicorn"):
        ensure_virallab_background_scheduler()