fix(deploy): support legacy data layout for runtime assets

Author: ieduer

.github/workflows/deploy.yml

@@ -17,8 +17,8 @@ jobs:
             cd /root/cross-subject-knowledge
             git pull origin main
             mkdir -p data/index state/logs state/cache state/tmp state/batch
-            if [ ! -f data/textbook_chunks.index ]; then
-              echo "ERROR: missing required runtime asset data/textbook_chunks.index"
+            if [ ! -f data/textbook_chunks.index ] && [ ! -f data/index/textbook_chunks.index ]; then
+              echo "ERROR: missing required runtime asset textbook_chunks.index (checked data/ and data/index/)"
               exit 1
             fi
             docker build -t textbook-knowledge .

backend/main.py

@@ -36,9 +36,19 @@
 for d in ("logs", "cache", "tmp", "batch"):
     (STATE_ROOT / d).mkdir(parents=True, exist_ok=True)
 
-DB_PATH = DATA_ROOT / "index/textbook_mineru_fts.db"
+def _resolve_data_asset(filename: str) -> Path:
+    primary = DATA_ROOT / "index" / filename
+    legacy = DATA_ROOT / filename
+    if primary.exists():
+        return primary
+    if legacy.exists():
+        return legacy
+    return primary
+
+
+DB_PATH = _resolve_data_asset("textbook_mineru_fts.db")
 FRONTEND = Path(__file__).parent.parent / "frontend"
-FAISS_INDEX_PATH = DATA_ROOT / "index/textbook_chunks.index"
+FAISS_INDEX_PATH = _resolve_data_asset("textbook_chunks.index")
 
 app = FastAPI(title="跨学科教材知识平台", version="0.1.0")
 app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_methods=["*"], allow_headers=["*"])

backend/preflight.py

@@ -12,6 +12,16 @@
 from pathlib import Path
 
 
+def _resolve_data_asset(data_root: Path, filename: str) -> Path:
+    primary = data_root / "index" / filename
+    legacy = data_root / filename
+    if primary.exists():
+        return primary
+    if legacy.exists():
+        return legacy
+    return primary
+
+
 def main() -> int:
     project_root = Path(os.getenv("PROJECT_ROOT", Path(__file__).resolve().parents[1])).expanduser().resolve()
     default_data_root = project_root / "data"
@@ -26,8 +36,8 @@ def main() -> int:
         d.mkdir(parents=True, exist_ok=True)
 
     required_files = [
-        data_root / "index" / "textbook_mineru_fts.db",
-        data_root / "index" / "textbook_chunks.index",
+        _resolve_data_asset(data_root, "textbook_mineru_fts.db"),
+        _resolve_data_asset(data_root, "textbook_chunks.index"),
     ]
 
     missing = [str(p) for p in required_files if not p.exists()]

backend/sync_db.py

@@ -8,7 +8,9 @@
 if not (_DEFAULT_DATA_ROOT / "index").exists() and (_ALT_DATA_ROOT / "index").exists():
     _DEFAULT_DATA_ROOT = _ALT_DATA_ROOT
 DATA_ROOT = Path(os.getenv("DATA_ROOT", _DEFAULT_DATA_ROOT)).expanduser().resolve()
-DB_PATH = DATA_ROOT / "index/textbook_mineru_fts.db"
+DB_PATH_PRIMARY = DATA_ROOT / "index" / "textbook_mineru_fts.db"
+DB_PATH_LEGACY = DATA_ROOT / "textbook_mineru_fts.db"
+DB_PATH = DB_PATH_PRIMARY if DB_PATH_PRIMARY.exists() or not DB_PATH_LEGACY.exists() else DB_PATH_LEGACY
 
 def download_db():
     print(f"Checking for DB updates from {DB_URL}...")