Upload summaries.py

local_rag/summaries.py

@@ -29,24 +29,14 @@ def _year_from_filename(file_name: str) -> str | None:
 
 def _download_cloud_summary(file_name: str) -> Path | None:
     """Download one summary from the HF Dataset and cache it locally."""
-    print(f"[summary] requested file = {file_name!r}", flush=True)
-
     if not HF_SUMMARY_ON_DEMAND:
-        print("[summary] on-demand download disabled", flush=True)
         return None
 
     year = _year_from_filename(file_name)
-    print(f"[summary] detected year = {year!r}", flush=True)
-
     if not year:
-        print("[summary] could not detect year from filename; download skipped", flush=True)
         return None
 
     repo_file = f"{HF_SUMMARY_PREFIX}/{year}/{Path(file_name).name}"
-    print(f"[summary] dataset repo = {HF_DATASET_REPO}", flush=True)
-    print(f"[summary] repo file = {repo_file!r}", flush=True)
-    print(f"[summary] cache dir = {HF_SUMMARY_CACHE}", flush=True)
-
     try:
         downloaded = hf_hub_download(
             repo_id=HF_DATASET_REPO,
@@ -54,20 +44,74 @@ def _download_cloud_summary(file_name: str) -> Path | None:
             filename=repo_file,
             local_dir=str(HF_SUMMARY_CACHE),
         )
-    except Exception as exc:
-        print(
-            f"[summary] download ERROR: {type(exc).__name__}: {exc}",
-            flush=True,
-        )
+    except Exception:
         return None
 
     path = Path(downloaded)
-    print(f"[summary] downloaded path = {path}", flush=True)
-    print(f"[summary] downloaded exists = {path.is_file()}", flush=True)
-
     return path if path.is_file() else None
 
 
+def summary_search_dirs() -> list[Path]:
+    """Directories to search for summary .txt files (first match wins)."""
+    dirs: list[Path] = []
+    for raw in (os.getenv("PAPER_SUMMARIES_DIRS") or "").split(":"):
+        raw = raw.strip()
+        if raw:
+            dirs.append(Path(raw).expanduser().resolve())
+
+    override = os.getenv("PAPER_SUMMARIES_DIR")
+    if override:
+        dirs.append(Path(override).expanduser().resolve())
+
+    project_root = Path(__file__).resolve().parents[1]
+    dirs.append(project_root / "paper_summaries")
+
+    unique: list[Path] = []
+    seen: set[str] = set()
+    for path in dirs:
+        key = str(path)
+        if key in seen:
+            continue
+        seen.add(key)
+        unique.append(path)
+    return unique
+
+
+def summaries_dir() -> Path:
+    for path in summary_search_dirs():
+        if path.exists():
+            return path
+    return Path(__file__).resolve().parents[1] / "paper_summaries"
+
+
+def _manifest_path(config: LocalRagConfig | None = None) -> Path:
+    cfg = config or LocalRagConfig.from_env()
+    return cfg.manifest_path
+
+
+def load_manifest(config: LocalRagConfig | None = None) -> dict:
+    path = _manifest_path(config)
+    if not path.exists():
+        raise FileNotFoundError(f"Manifest not found: {path}. Rebuild local_rag_data/manifest.json first.")
+    return json.loads(path.read_text(encoding="utf-8"))
+
+
+def title_from_filename(file_name: str) -> str:
+    stem = Path(file_name).stem
+    match = _FILENAME_TITLE.match(stem)
+    if match:
+        return match.group(1).strip()
+    return stem
+
+
+def extract_title(file_name: str, summary: str | None = None) -> str:
+    if summary:
+        match = TITLE_PATTERN.search(summary)
+        if match:
+            return match.group(1).strip()
+    return title_from_filename(file_name)
+
+
 def summary_path(file_name: str) -> Path | None:
     # Preserve existing local behavior first.
     for root in summary_search_dirs():