changed admin screen

src/app.py

@@ -13,6 +13,7 @@ HuggingFace Space secrets required:
 """
 import os
 import sys
+import json
 import subprocess
 from pathlib import Path
 
@@ -109,11 +110,12 @@ from src.ui.screens_preference import screen_pair_intro
 def _screen_admin(cfg: dict) -> None:
     """
     Coverage dashboard — visit ?admin=1 to see this.
-    Shows how many items are covered, reserved, and outstanding per category.
+    Always scans the HF repo directly — ignores local completions cache
+    so the count reflects real accepted submissions only.
     """
     from src.data import (
-        _get_accepted_counts, _load_pool, _pool_path,
-        _load_reservations, _expire_reservations, _data_dir,
+        _load_pool, _pool_path, _data_dir,
+        _load_reservations, _expire_reservations,
     )
 
     st.markdown("## 📊 Study Coverage Dashboard")
@@ -124,32 +126,62 @@ def _screen_admin(cfg: dict) -> None:
     )
 
     if st.button("🔄 Refresh", type="primary"):
-        # Invalidate completion caches so we re-scan HF
-        data_dir = _data_dir(cfg)
-        for f in data_dir.glob("completion_cache*"):
-            f.unlink()
         st.rerun()
 
+    hf_token    = cfg.get("hf_token", "")
+    output_repo = cfg.get("output_dataset_repo", "")
+
     for cat_cfg in cfg["categories"]:
         cat  = cat_cfg["name"]
         pool = _load_pool(str(_pool_path(cat, cfg)))
+        total = len(pool)
 
-        counts       = _get_accepted_counts(cat, cfg)
+        # ── Scan HF directly (no cache) ──────────────────────────────────────
+        hf_counts = {str(i): 0 for i in range(total)}
+        n_json    = 0
+        if hf_token and output_repo:
+            try:
+                from huggingface_hub import HfApi
+                api        = HfApi(token=hf_token)
+                files      = list(api.list_repo_files(repo_id=output_repo, repo_type="dataset"))
+                json_files = [f for f in files if f.startswith("json/") and f.endswith(".json")]
+                n_json     = len(json_files)
+                for filepath in json_files:
+                    try:
+                        content = api.hf_hub_download(
+                            repo_id=output_repo,
+                            filename=filepath,
+                            repo_type="dataset",
+                            token=hf_token,
+                        )
+                        with open(content) as f:
+                            submission = json.load(f)
+                        for item in submission.get("items", []):
+                            if item.get("category") != cat:
+                                continue
+                            idx = item.get("_pool_index")
+                            if idx is not None:
+                                hf_counts[str(idx)] = hf_counts.get(str(idx), 0) + 1
+                    except Exception as e:
+                        st.warning(f"Could not parse {filepath}: {e}")
+            except Exception as e:
+                st.error(f"Could not scan HF repo: {e}")
+
+        # ── Reservations (active in-progress users) ───────────────────────────
         reservations = _load_reservations(cfg)
         _expire_reservations(reservations)
-
-        total = len(pool)
-        covered = sum(1 for v in counts.values() if v >= 1)
-        uncovered = total - covered
-
-        # Items currently reserved but not yet completed
         reserved_uncovered = sum(
             1 for k, v in reservations.items()
-            if counts.get(k, 0) == 0
+            if hf_counts.get(k, 0) == 0
         )
+
+        covered         = sum(1 for v in hf_counts.values() if v >= 1)
+        uncovered       = total - covered
         truly_uncovered = uncovered - reserved_uncovered
 
         st.markdown(f"### {cat.capitalize()}")
+        st.caption(f"{n_json} submission file(s) in HF repo")
+
         col1, col2, col3, col4 = st.columns(4)
         col1.metric("Total items",     total)
         col2.metric("Covered ✅",      covered)