Update app.py

app.py

@@ -72,7 +72,13 @@ except Exception:
     hf_hub_download = None  # type: ignore
 
 
-APP_VERSION = "2026-02-11_app_g_signals"
+APP_VERSION = "2026-02-11_app_h_dataset_defaults"
+
+# Defaults for your separated dataset + space setup
+DEFAULT_DATASET_REPO_ID = "cjc0013/EpsteinWithAnomScore"
+DEFAULT_DATASET_FILENAME = "corpus.sqlite"
+DEFAULT_SIGNALS_DATASET_REPO_ID = "cjc0013/EpsteinWithAnomScore"
+DEFAULT_SIGNALS_FILENAME = "public_method_sanitized_topN.jsonl"
 
 
 # -----------------------------
@@ -101,6 +107,8 @@ def _parse_dataset_ref(repo_like: str) -> Tuple[str, Optional[str]]:
     Accept either:
       - "user/repo"
       - "https://huggingface.co/datasets/user/repo/blob/main/file.ext"
+      - "https://huggingface.co/datasets/user/repo/resolve/main/file.ext"
+      - "https://huggingface.co/spaces/user/repo"
 
     Returns: (repo_id, inferred_filename_or_None)
     """
@@ -113,17 +121,21 @@ def _parse_dataset_ref(repo_like: str) -> Tuple[str, Optional[str]]:
         p = (u.path or "").strip("/")
         parts = p.split("/")
 
-        if len(parts) >= 3 and parts[0] == "datasets":
+        # Accept both datasets and spaces URL shapes; return owner/repo
+        if len(parts) >= 3 and parts[0] in ("datasets", "spaces"):
             repo_id = f"{parts[1]}/{parts[2]}"
             inferred_file: Optional[str] = None
 
-            if "blob" in parts:
-                try:
-                    i = parts.index("blob")
-                    if i + 2 < len(parts):
-                        inferred_file = "/".join(parts[i + 2 :])
-                except Exception:
-                    inferred_file = None
+            # Try extracting filename from blob/resolve URL forms
+            for marker in ("blob", "resolve"):
+                if marker in parts:
+                    try:
+                        i = parts.index(marker)
+                        if i + 2 < len(parts):
+                            inferred_file = "/".join(parts[i + 2 :])
+                            break
+                    except Exception:
+                        pass
 
             return repo_id, inferred_file
 
@@ -200,10 +212,10 @@ def ensure_db_file() -> Path:
             print(f"[db] using local file: {p}")
             return p
 
-    ds_repo_raw = _env("DATASET_REPO_ID", "")
+    ds_repo_raw = _env("DATASET_REPO_ID", DEFAULT_DATASET_REPO_ID)
     ds_repo, inferred_file = _parse_dataset_ref(ds_repo_raw)
 
-    ds_file_raw = _env("DATASET_FILENAME", "corpus.sqlite")
+    ds_file_raw = _env("DATASET_FILENAME", DEFAULT_DATASET_FILENAME)
     ds_file = _clean_env_value(ds_file_raw)
 
     if inferred_file and (not os.environ.get("DATASET_FILENAME") or not ds_file):
@@ -214,7 +226,7 @@ def ensure_db_file() -> Path:
 
     local_dir = Path(_env("DB_LOCAL_DIR", "./data")).expanduser().resolve()
     local_dir.mkdir(parents=True, exist_ok=True)
-    target = (local_dir / (ds_file if ds_file else "corpus.sqlite")).resolve()
+    target = (local_dir / (ds_file if ds_file else DEFAULT_DATASET_FILENAME)).resolve()
 
     print(f"[db] DATASET_REPO_ID={ds_repo!r}")
     print(f"[db] DATASET_FILENAME={ds_file!r}")
@@ -225,7 +237,7 @@ def ensure_db_file() -> Path:
             raise RuntimeError("DATASET_REPO_ID is set, but huggingface_hub is not installed. Add it to requirements.txt.")
 
         if not ds_file:
-            ds_file = "corpus.sqlite"
+            ds_file = DEFAULT_DATASET_FILENAME
 
         cached_path = hf_hub_download(
             repo_id=ds_repo,
@@ -303,11 +315,11 @@ def ensure_signals_file() -> Optional[Path]:
             return p
 
     # optional dataset download
-    ds_repo_raw = _env("METHOD_SIGNALS_DATASET_REPO_ID", "")
+    ds_repo_raw = _env("METHOD_SIGNALS_DATASET_REPO_ID", DEFAULT_SIGNALS_DATASET_REPO_ID)
     ds_repo, inferred_file = _parse_dataset_ref(ds_repo_raw)
     ds_repo = _clean_env_value(ds_repo)
 
-    ds_file_raw = _env("METHOD_SIGNALS_FILENAME", "public_method_sanitized_topN.jsonl")
+    ds_file_raw = _env("METHOD_SIGNALS_FILENAME", DEFAULT_SIGNALS_FILENAME)
     ds_file = _clean_env_value(ds_file_raw)
     if inferred_file and (not os.environ.get("METHOD_SIGNALS_FILENAME") or not ds_file):
         ds_file = inferred_file
@@ -322,7 +334,7 @@ def ensure_signals_file() -> Optional[Path]:
 
     local_dir = Path(_env("METHOD_SIGNALS_LOCAL_DIR", "./data")).expanduser().resolve()
     local_dir.mkdir(parents=True, exist_ok=True)
-    target = (local_dir / (ds_file if ds_file else "public_method_sanitized_topN.jsonl")).resolve()
+    target = (local_dir / (ds_file if ds_file else DEFAULT_SIGNALS_FILENAME)).resolve()
 
     print(f"[signals] METHOD_SIGNALS_DATASET_REPO_ID={ds_repo!r}")
     print(f"[signals] METHOD_SIGNALS_FILENAME={ds_file!r}")
@@ -1301,7 +1313,8 @@ def ui_signals_pick(choice: str, state: Dict[str, Dict[str, Any]]):
             order_index=_safe_int(r.get("order_index"), -1),
         )
 
-        details = _signal_details_text(r, SIGNALS_PATH)
+        src_path = SIGNALS_PATH if SIGNALS_PATH is not None else ensure_signals_file()
+        details = _signal_details_text(r, src_path)
         return uid, details, f"✅ Picked {cid}", ""
     except Exception as e:
         return "", "", f"⚠️ {type(e).__name__}: {e}", _fmt_debug(e)
@@ -1329,6 +1342,12 @@ def build_ui() -> gr.Blocks:
 - **Search:** type words -> Search -> pick result -> Open
 - **Clusters:** Load clusters -> pick one -> Load chunks -> pick chunk -> Open
 - **Signals (optional):** Load signal cards -> pick card -> Open linked chunk
+
+**Default dataset repo (override with env vars):**
+- <code>DATASET_REPO_ID={DEFAULT_DATASET_REPO_ID}</code>
+- <code>DATASET_FILENAME={DEFAULT_DATASET_FILENAME}</code>
+- <code>METHOD_SIGNALS_DATASET_REPO_ID={DEFAULT_SIGNALS_DATASET_REPO_ID}</code>
+- <code>METHOD_SIGNALS_FILENAME={DEFAULT_SIGNALS_FILENAME}</code>
 """
         )