Group search results by owner; auto-resume stuck pending documents

main.py

@@ -38,6 +38,24 @@ async def lifespan(app: FastAPI):
         cleanup_expired_sessions()
     except Exception:
         pass
+    # Resume any documents that were left in 'pending' state from a previous run
+    # (e.g. server restart killed the background extraction task).
+    try:
+        import asyncio
+        from database.connection import get_db
+        from services.extractor import get_default_provider
+        from routers.upload import _extract_and_save
+
+        default_provider = get_default_provider()
+        with get_db() as conn:
+            stuck = conn.execute(
+                "SELECT id, image_path, provider FROM documents WHERE status='pending'"
+            ).fetchall()
+        for row in stuck:
+            provider = row["provider"] or default_provider
+            asyncio.create_task(_extract_and_save(row["id"], row["image_path"], provider))
+    except Exception:
+        pass
     yield
 
 

routers/documents.py

@@ -233,6 +233,27 @@ async def retry_all_errors():
     return JSONResponse({"ok": True, "retried": len(rows)})
 
 
+@router.post("/documents/resume-pending")
+async def resume_pending():
+    """Re-fire background extraction for any docs stuck in 'pending'.
+    Useful when a previous server restart killed in-flight extraction tasks."""
+    import asyncio
+    from services.extractor import get_default_provider
+    from routers.upload import _extract_and_save
+
+    default_provider = get_default_provider()
+    with get_db() as conn:
+        rows = conn.execute(
+            "SELECT id, image_path, provider FROM documents WHERE status='pending'"
+        ).fetchall()
+
+    for row in rows:
+        provider = row["provider"] or default_provider
+        asyncio.create_task(_extract_and_save(row["id"], row["image_path"], provider))
+
+    return JSONResponse({"ok": True, "resumed": len(rows)})
+
+
 @router.post("/documents/scan-duplicates")
 async def scan_duplicates():
     """

routers/search.py

@@ -31,12 +31,9 @@ async def search(
     if property_number or district or block:
         properties = search_properties(property_number, district, block)
 
-    # If exactly one person found, preload their full details
+    # If exactly one person found, preload their full details (across all scopes)
     if len(persons) == 1 and not properties:
-        selected_person = get_person_with_properties(
-            persons[0]["id"],
-            persons[0].get("search_scope"),
-        )
+        selected_person = get_person_with_properties(persons[0]["id"], None)
 
     return render_template(
         templates,

services/search_service.py

@@ -20,7 +20,7 @@ def _normalize_scope(text: str | None) -> str | None:
 
 
 def search_persons(query: str) -> list[dict]:
-    """Search persons by name, keeping separate result rows per search scope."""
+    """Search persons by name. One row per person, aggregating all their search scopes."""
     norm = normalize_arabic(query.strip())
     pattern = f"%{norm}%"
     raw_pattern = f"%{query.strip()}%"
@@ -30,9 +30,9 @@ def search_persons(query: str) -> list[dict]:
         rows = conn.execute(
             f"""
             SELECT p.*,
-                   {scope_expr} AS search_scope,
                    COUNT(DISTINCT pr.id) AS property_count,
-                   COUNT(DISTINCT d.id) AS document_count
+                   COUNT(DISTINCT d.id) AS document_count,
+                   GROUP_CONCAT(DISTINCT {scope_expr}) AS search_scopes_raw
             FROM persons p
             LEFT JOIN documents d ON d.person_id = p.id
             LEFT JOIN properties pr ON pr.document_id = d.id
@@ -41,14 +41,13 @@ def search_persons(query: str) -> list[dict]:
                OR p.father_name LIKE ?
                OR p.first_name LIKE ?
                OR p.family_name LIKE ?
-            GROUP BY p.id, {scope_expr}
+            GROUP BY p.id
             ORDER BY
                 CASE WHEN p.first_name_norm = ? THEN 0
                      WHEN p.family_name_norm = ? THEN 0
                      ELSE 1 END,
                 p.first_name,
-                p.family_name,
-                search_scope
+                p.family_name
             """,
             (pattern, pattern, raw_pattern, raw_pattern, raw_pattern, norm, norm),
         ).fetchall()
@@ -56,7 +55,17 @@ def search_persons(query: str) -> list[dict]:
     results = []
     for row in rows:
         person = dict(row)
-        person["search_scope"] = _normalize_scope(person.get("search_scope"))
+        raw = person.pop("search_scopes_raw", None) or ""
+        scopes = [s.strip() for s in raw.split(",") if s and s.strip()]
+        # Deduplicate while preserving order
+        seen = set()
+        unique_scopes = []
+        for s in scopes:
+            if s not in seen:
+                seen.add(s)
+                unique_scopes.append(s)
+        person["search_scopes"] = unique_scopes
+        person["search_scope"] = "، ".join(unique_scopes) if unique_scopes else None
         results.append(person)
     return results
 

templates/documents.html

@@ -58,6 +58,12 @@
 </div>
 {% endif %}
 
+{% if stats.processing %}
+<div class="alert-banner">
+  <button class="btn btn-sm btn-primary" onclick="resumePending(this)">استئناف معالجة الوثائق المعلّقة ({{ stats.processing }})</button>
+</div>
+{% endif %}
+
 {% if stats.pending_review %}
 <div class="alert-banner">
   <a href="/review/next">ابدأ مراجعة {{ stats.pending_review }} وثيقة →</a>
@@ -182,6 +188,22 @@ async function retryAllErrors(btn) {
   }
 }
 
+async function resumePending(btn) {
+  btn.disabled = true;
+  const original = btn.textContent;
+  btn.textContent = 'جارٍ الاستئناف...';
+  const res = await fetch('/documents/resume-pending', { method: 'POST' });
+  if (res.ok) {
+    const data = await res.json();
+    alert(`تم استئناف ${data.resumed} وثيقة. قد تستغرق المعالجة بضع لحظات.`);
+    location.reload();
+  } else {
+    alert('فشل الاستئناف.');
+    btn.disabled = false;
+    btn.textContent = original;
+  }
+}
+
 async function scanDuplicates(btn) {
   btn.disabled = true;
   const original = btn.textContent;

templates/search.html

@@ -144,7 +144,7 @@
   {% else %}
     <div class="persons-grid">
       {% for p in persons %}
-      <a href="/persons/{{ p.id }}{% if p.search_scope %}?search_scope={{ p.search_scope|urlencode }}{% endif %}" class="person-card">
+      <a href="/persons/{{ p.id }}" class="person-card">
         <div class="person-name">{{ p.first_name }} {{ p.father_name or '' }} {{ p.family_name or '' }}</div>
         <div class="person-meta">
           {% if p.family_origin %}{{ p.family_origin }} · {% endif %}