weekly-update.yml

name: Weekly Auto-Update

on:
  schedule:
    - cron: '0 8 * * 1'  # Every Monday at 8:00 AM UTC
  workflow_dispatch:

permissions:
  contents: write

jobs:
  update:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout
        uses: actions/checkout@v4

      - name: Fetch network data and generate briefing
        env:
          ANTHROPIC_API_KEY: ${{ secrets.ANTHROPIC_API_KEY }}
        run: |
          python3 << 'PYEOF'
          import json, urllib.request, urllib.error, os, re
          from datetime import datetime, timedelta, timezone

          # ── 1. Load Instagram posts (updated daily by fetch-instagram.yml) ──
          posts = []
          try:
              with open("assets/data/instagram.json") as f:
                  ig_data = json.load(f)
              posts = ig_data.get("posts", [])
              print(f"Loaded {len(posts)} posts from instagram.json")
          except Exception as e:
              print(f"Failed to load instagram.json: {e}")

          now_iso = datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")

          # ── 2. Fetch network node count ──
          NODES_URL = "https://cluster.1.pools.functionyard.fula.network/nodes"
          total_nodes = 900  # fallback
          try:
              req = urllib.request.Request(NODES_URL, headers={"User-Agent": "Mozilla/5.0"})
              with urllib.request.urlopen(req, timeout=15) as resp:
                  nodes_data = resp.read()
                  # The endpoint might return a JSON array or object
                  parsed = json.loads(nodes_data)
                  if isinstance(parsed, list):
                      total_nodes = len(parsed)
                  elif isinstance(parsed, dict) and "count" in parsed:
                      total_nodes = parsed["count"]
                  elif isinstance(parsed, dict):
                      total_nodes = len(parsed)
              print(f"Network nodes: {total_nodes}")
          except Exception as e:
              print(f"Node count fetch failed, using fallback: {e}")

          network_data = {"updated": now_iso, "total_nodes": total_nodes}
          with open("assets/data/network.json", "w") as f:
              json.dump(network_data, f, indent=2)
          print("Wrote network.json")

          # ── 3. Determine content seed for Claude briefing ──
          now_utc = datetime.now(timezone.utc)
          seven_days_ago = now_utc - timedelta(days=7)

          def parse_date(d):
              # Strip milliseconds if present (e.g. ".000" in "2026-02-23T22:13:00.000Z")
              d_clean = re.sub(r'\.\d+', '', d)
              for fmt in ["%Y-%m-%dT%H:%M:%SZ", "%Y-%m-%dT%H:%M:%S%z", "%a, %d %b %Y %H:%M:%S %z"]:
                  try:
                      dt = datetime.strptime(d_clean, fmt)
                      return dt.replace(tzinfo=timezone.utc) if dt.tzinfo is None else dt
                  except ValueError:
                      continue
              return None

          recent_posts = []
          for p in posts:
              dt = parse_date(p.get("timestamp", ""))
              if dt and dt >= seven_days_ago:
                  recent_posts.append(p)

          if recent_posts:
              seed_posts = recent_posts
              prompt_mode = "recent"
          else:
              seed_posts = posts[:3]
              prompt_mode = "quiet"

          print(f"Content seed: {len(seed_posts)} posts ({prompt_mode} mode)")

          # ── 4. Load previous pulse for continuity ──
          prev_summary = ""
          if os.path.exists("assets/data/pulse.json"):
              try:
                  with open("assets/data/pulse.json") as f:
                      prev_pulse = json.load(f)
                  prev_summary = prev_pulse.get("briefing", "")
              except Exception:
                  pass

          # ── 5. Call Claude API with web search ──
          api_key = os.environ.get("ANTHROPIC_API_KEY", "")
          if not api_key:
              print("No ANTHROPIC_API_KEY set, skipping briefing generation")
              pulse_data = {
                  "updated": now_iso,
                  "week_of": (now_utc - timedelta(days=now_utc.weekday())).strftime("%Y-%m-%d"),
                  "headline": "Weekly Update",
                  "subtitle": "Check back for this week's briefing.",
                  "briefing": "",
                  "sources": []
              }
          else:
              seed_text = "\n".join([f"- [{p.get('timestamp','')}] {p.get('caption','')}" for p in seed_posts])

              if prompt_mode == "recent":
                  user_context = f"""## This Week's Posts from @functionland
          {seed_text}

          ## Network Stats
          Active nodes: {total_nodes}

          ## Previous Briefing (for continuity, do not repeat)
          {prev_summary[:500]}"""
              else:
                  user_context = f"""## Recent Posts from @functionland (no new posts this week)
          {seed_text}

          ## Network Stats
          Active nodes: {total_nodes}

          ## Previous Briefing (for continuity, do not repeat)
          {prev_summary[:500]}

          Note: There were no new posts this week. Based on Functionland's recent activity and current industry developments, write a briefing on where the project stands in the broader DePIN landscape."""

              system_prompt = """You are a senior technology journalist writing a weekly briefing for Functionland's website. Your style is authoritative, clear, and concise — like a Wall Street Journal technology correspondent.

          Your task:
          1. Review the provided posts and network data from Functionland this week
          2. Use web search to research relevant context: DePIN industry trends, competing projects, regulatory developments, technology breakthroughs related to the topics in the posts
          3. Write a 150-250 word weekly briefing that:
             - Leads with Functionland's most newsworthy update from the posts
             - Provides industry context from your research (why this matters in the broader DePIN/Web3 landscape)
             - Includes specific facts, numbers, or comparisons from your research
             - Ends with a forward-looking sentence about what to watch next
             - If there are recent announcements or news or developments, focus on those instead of generic update
          4. Also generate a short headline (max 10 words) and a one-sentence subtitle

          Rules:
          - Never fabricate statistics or quotes
          - If you cannot verify something, don't include it
          - Do not use exclamation marks or hype language
          - Write for an informed but not expert audience
          - Keep the tone professional and credible
          - Every claim should be grounded in the post data or your web research
          - Every claim should be grounded in the tweet data or your web research
          - It should have positive tone and be encouraging for Functionland focused on positive news and not with a negative tone

          Respond with ONLY valid JSON in this exact format:
          {"headline": "...", "subtitle": "...", "briefing": "...", "sources": ["url1", "url2"]}"""

              request_body = json.dumps({
                  "model": "claude-sonnet-4-6",
                  "max_tokens": 1024,
                  "system": system_prompt,
                  "tools": [{"type": "web_search_20250305", "name": "web_search", "max_uses": 5}],
                  "messages": [{"role": "user", "content": user_context}]
              }).encode()

              headers = {
                  "Content-Type": "application/json",
                  "X-Api-Key": api_key,
                  "Anthropic-Version": "2023-06-01"
              }

              req = urllib.request.Request("https://api.anthropic.com/v1/messages", data=request_body, headers=headers, method="POST")
              print("Calling Claude API with web search...")

              try:
                  with urllib.request.urlopen(req, timeout=120) as resp:
                      result = json.loads(resp.read())

                  # Extract text from response content blocks
                  text_content = ""
                  for block in result.get("content", []):
                      if block.get("type") == "text":
                          text_content += block["text"]

                  # Parse JSON from response
                  json_match = re.search(r'\{[^{}]*"headline"[^{}]*\}', text_content, re.DOTALL)
                  if json_match:
                      briefing_data = json.loads(json_match.group())
                  else:
                      # Try parsing the entire text as JSON
                      briefing_data = json.loads(text_content.strip())

                  week_of = (now_utc - timedelta(days=now_utc.weekday())).strftime("%Y-%m-%d")
                  pulse_data = {
                      "updated": now_iso,
                      "week_of": week_of,
                      "headline": briefing_data.get("headline", "Weekly Update"),
                      "subtitle": briefing_data.get("subtitle", ""),
                      "briefing": briefing_data.get("briefing", ""),
                      "sources": briefing_data.get("sources", [])
                  }
                  print("Claude briefing generated successfully")
              except urllib.error.HTTPError as e:
                  error_body = e.read().decode() if e.fp else ""
                  print(f"Claude API call failed: {e} — {error_body}")
                  pulse_data = {
                      "updated": now_iso,
                      "week_of": (now_utc - timedelta(days=now_utc.weekday())).strftime("%Y-%m-%d"),
                      "headline": "Weekly Update",
                      "subtitle": "Check back for this week's briefing.",
                      "briefing": "",
                      "sources": []
                  }
              except Exception as e:
                  print(f"Claude API call failed: {e}")
                  pulse_data = {
                      "updated": now_iso,
                      "week_of": (now_utc - timedelta(days=now_utc.weekday())).strftime("%Y-%m-%d"),
                      "headline": "Weekly Update",
                      "subtitle": "Check back for this week's briefing.",
                      "briefing": "",
                      "sources": []
                  }

          # Archive previous pulse before overwriting
          try:
              with open("assets/data/pulse.json", "r") as f:
                  old_pulse = json.load(f)
              if old_pulse.get("briefing", "").strip():
                  archive_path = f"assets/data/history/pulse-before-{pulse_data['week_of']}.json"
                  os.makedirs("assets/data/history", exist_ok=True)
                  with open(archive_path, "w") as f:
                      json.dump(old_pulse, f, indent=2)
                  print(f"Archived previous pulse to {archive_path}")
          except FileNotFoundError:
              print("No existing pulse.json to archive")
          except Exception as e:
              print(f"Pulse archive failed (non-fatal): {e}")

          with open("assets/data/pulse.json", "w") as f:
              json.dump(pulse_data, f, indent=2)
          print("Wrote pulse.json")

          # ── 6. Update sitemap.xml homepage lastmod ──
          today = now_utc.strftime("%Y-%m-%d")
          try:
              with open("sitemap.xml", "r") as f:
                  sitemap = f.read()
              import re as re2
              sitemap = re2.sub(
                  r'(<loc>https://fx\.land/</loc>\s*<lastmod>)\d{4}-\d{2}-\d{2}(</lastmod>)',
                  rf'\g<1>{today}\2',
                  sitemap
              )
              with open("sitemap.xml", "w") as f:
                  f.write(sitemap)
              print(f"Updated sitemap.xml homepage lastmod to {today}")
          except Exception as e:
              print(f"Sitemap update failed: {e}")

          print("Done!")
          PYEOF

      - name: Commit and push changes
        run: |
          git config user.name "github-actions[bot]"
          git config user.email "github-actions[bot]@users.noreply.github.com"
          git add assets/data/network.json assets/data/pulse.json assets/data/history/ sitemap.xml
          if git diff --staged --quiet; then
            echo "No changes to commit"
          else
            git commit -m "chore: weekly auto-update ($(date -u +%Y-%m-%d))"
            git push
          fi