fix(db): normalize stale positive-ID media file paths

alembic/versions/20260524_013_fix_media_file_paths.py

@@ -0,0 +1,126 @@
+"""Fix stale positive-ID folder prefixes in media.file_path.
+
+Before v4.0.5, the backup stored media in directories named with raw (positive)
+entity IDs. After v4.0.5, directories use Telethon's marked IDs (negative for
+groups/channels). The v4.0.6 migration fixed chat_id columns but never updated
+file_path strings, leaving ~37% of media rows with paths pointing to the wrong
+folder name.
+
+This migration rewrites file_path values so the folder component matches the
+marked chat_id already stored in media.chat_id.
+
+Revision ID: 013
+Revises: 012
+Create Date: 2026-05-24
+"""
+
+import sqlalchemy as sa
+
+from alembic import op
+
+revision = "013"
+down_revision = "012"
+branch_labels = None
+depends_on = None
+
+_BATCH_SIZE = 5000
+
+
+def _derive_stale_folder(chat_id: int) -> str | None:
+    """Derive the old positive folder name from a marked chat_id.
+
+    Basic groups: chat_id = -X  →  old folder = "X"
+    Channels:     chat_id = -(1000000000000 + X)  →  old folder = "X"
+    Users:        chat_id > 0  →  no mismatch possible, return None
+    """
+    if chat_id >= 0:
+        return None
+    raw = -chat_id
+    if raw > 1000000000000:
+        return str(raw - 1000000000000)
+    return str(raw)
+
+
+def upgrade():
+    conn = op.get_bind()
+    dialect = conn.dialect.name
+
+    if dialect == "postgresql":
+        # Get all distinct negative chat_ids that have media
+        result = conn.execute(sa.text("SELECT DISTINCT chat_id FROM media WHERE chat_id < 0 AND file_path IS NOT NULL"))
+        chat_ids = [row[0] for row in result]
+
+        for chat_id in chat_ids:
+            stale_folder = _derive_stale_folder(chat_id)
+            if stale_folder is None:
+                continue
+            correct_folder = str(chat_id)
+
+            # Only update rows where file_path contains the stale folder
+            # Use pattern: .../<stale_folder>/... → .../<correct_folder>/...
+            stale_pattern = f"%/{stale_folder}/%"
+            conn.execute(
+                sa.text(
+                    "UPDATE media SET file_path = REPLACE(file_path, :old_seg, :new_seg) "
+                    "WHERE chat_id = :cid AND file_path LIKE :pattern"
+                ),
+                {
+                    "old_seg": f"/{stale_folder}/",
+                    "new_seg": f"/{correct_folder}/",
+                    "cid": chat_id,
+                    "pattern": stale_pattern,
+                },
+            )
+
+    elif dialect == "sqlite":
+        result = conn.execute(sa.text("SELECT DISTINCT chat_id FROM media WHERE chat_id < 0 AND file_path IS NOT NULL"))
+        chat_ids = [row[0] for row in result]
+
+        for chat_id in chat_ids:
+            stale_folder = _derive_stale_folder(chat_id)
+            if stale_folder is None:
+                continue
+            correct_folder = str(chat_id)
+
+            stale_pattern = f"%/{stale_folder}/%"
+            conn.execute(
+                sa.text(
+                    "UPDATE media SET file_path = REPLACE(file_path, :old_seg, :new_seg) "
+                    "WHERE chat_id = :cid AND file_path LIKE :pattern"
+                ),
+                {
+                    "old_seg": f"/{stale_folder}/",
+                    "new_seg": f"/{correct_folder}/",
+                    "cid": chat_id,
+                    "pattern": stale_pattern,
+                },
+            )
+
+
+def downgrade():
+    # Reversible: swap the folder components back
+    conn = op.get_bind()
+    dialect = conn.dialect.name
+
+    result = conn.execute(sa.text("SELECT DISTINCT chat_id FROM media WHERE chat_id < 0 AND file_path IS NOT NULL"))
+    chat_ids = [row[0] for row in result]
+
+    for chat_id in chat_ids:
+        stale_folder = _derive_stale_folder(chat_id)
+        if stale_folder is None:
+            continue
+        correct_folder = str(chat_id)
+        pattern = f"%/{correct_folder}/%"
+
+        conn.execute(
+            sa.text(
+                "UPDATE media SET file_path = REPLACE(file_path, :new_seg, :old_seg) "
+                "WHERE chat_id = :cid AND file_path LIKE :pattern"
+            ),
+            {
+                "new_seg": f"/{correct_folder}/",
+                "old_seg": f"/{stale_folder}/",
+                "cid": chat_id,
+                "pattern": pattern,
+            },
+        )

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "telegram-archive"
-version = "7.10.13"
+version = "7.10.14"
 description = "Automated Telegram backup with Docker. Performs incremental backups of messages and media on a configurable schedule."
 readme = "README.md"
 requires-python = ">=3.14"

scripts/entrypoint.sh

@@ -92,6 +92,17 @@ if has_tables and not has_alembic:
             CONSTRAINT alembic_version_pkc PRIMARY KEY (version_num)
         );
     \"\"\")
+    # Check artifact from migration 013: file_path values use negative chat_id folders
+    # If any media row for a negative chat_id has a correctly-negative folder, 013 has run
+    cur.execute(\"\"\"
+        SELECT EXISTS (
+            SELECT 1 FROM media
+            WHERE chat_id < 0 AND file_path LIKE '%/' || CAST(chat_id AS TEXT) || '/%'
+            LIMIT 1
+        );
+    \"\"\")
+    has_013_paths = cur.fetchone()[0]
+
     # Check artifact from migration 012: idx_media_chat_type index
     cur.execute(\"\"\"
         SELECT EXISTS (
@@ -198,7 +209,9 @@ if has_tables and not has_alembic:
     has_push_subs = cur.fetchone()[0]
 
     # Determine which version to stamp based on existing schema
-    if has_012_index:
+    if has_013_paths:
+        stamp_version = '013'
+    elif has_012_index:
         stamp_version = '012'
     elif has_011_content_hash:
         stamp_version = '011'
@@ -286,6 +299,10 @@ if has_tables and not has_alembic:
         )
     ''')
 
+    # Check artifact from migration 013: file_path values use negative chat_id folders
+    cur.execute(\"SELECT EXISTS(SELECT 1 FROM media WHERE chat_id < 0 AND file_path LIKE '%/' || CAST(chat_id AS TEXT) || '/%' LIMIT 1)\")
+    has_013_paths = cur.fetchone()[0]
+
     # Check artifact from migration 012: idx_media_chat_type index
     cur.execute(\"SELECT name FROM sqlite_master WHERE type='index' AND name='idx_media_chat_type'\")
     has_012_index = cur.fetchone() is not None
@@ -336,7 +353,9 @@ if has_tables and not has_alembic:
     has_push_subs = cur.fetchone() is not None
 
     # Determine which version to stamp based on existing schema
-    if has_012_index:
+    if has_013_paths:
+        stamp_version = '013'
+    elif has_012_index:
         stamp_version = '012'
     elif has_011_content_hash:
         stamp_version = '011'

src/__init__.py

@@ -2,4 +2,4 @@
 Telegram Backup Automation - Main Package
 """
 
-__version__ = "7.10.13"
+__version__ = "7.10.14"

src/web/main.py

@@ -820,7 +820,11 @@
         logger.warning("Blocked restricted media request for non-chat folder: %s", parts[0])
         raise HTTPException(status_code=403, detail="Access denied")
     if media_chat_id not in user_chat_ids:
-        raise HTTPException(status_code=403, detail="Access denied")
+        # Legacy fallback: positive folder may correspond to negative marked ID
+        if media_chat_id > 0 and (-media_chat_id in user_chat_ids or -(1000000000000 + media_chat_id) in user_chat_ids):
+            pass
+        else:
+            raise HTTPException(status_code=403, detail="Access denied")
 
 
 def _strip_original_media_paths(messages: list[dict]) -> None:
@@ -918,7 +922,25 @@
     try:
         resolved = candidate.resolve(strict=True)
     except OSError, ValueError:
-        raise HTTPException(status_code=404, detail="File not found")
+        # Legacy fallback: pre-v4.0.5 paths used positive IDs, disk uses negative marked IDs.
+        # Try alternate folder names: X→-X (basic group), X→-100X (channel/supergroup)
+        parts = path.split("/", 1)
+        resolved = None
+        if len(parts) == 2:
+            folder, rest = parts
+            alt_folders = []
+            if not folder.startswith("-"):
+                alt_folders = [f"-{folder}", f"-100{folder}"]
+            else:
+                alt_folders = [folder[1:]]
+            for alt in alt_folders:
+                try:
+                    resolved = (_media_root / alt / rest).resolve(strict=True)
+                    break
+                except OSError, ValueError:
+                    continue
+        if resolved is None:
+            raise HTTPException(status_code=404, detail="File not found")
     if not resolved.is_relative_to(_media_root):
         raise HTTPException(status_code=403, detail="Access denied")
 

src/web/thumbnails.py

@@ -124,7 +124,22 @@
         return dest
 
     if not source.exists():
-        return None
+        # Legacy fallback: pre-v4.0.5 paths used positive IDs, disk uses negative marked IDs.
+        # Try alternate folder names: X→-X (basic group), X→-100X (channel/supergroup)
+        alt_folders = []
+        if not folder.startswith("-"):
+            alt_folders = [f"-{folder}", f"-100{folder}"]
+        else:
+            alt_folders = [folder[1:]]
+        found = False
+        for alt in alt_folders:
+            alt_source = (media_root / alt / filename).resolve()
+            if alt_source.is_relative_to(media_root_resolved) and alt_source.exists():
+                source = alt_source
+                found = True
+                break
+        if not found:
+            return None
 
     async with _generation_semaphore:
         loop = asyncio.get_running_loop()