v1.0.162: auto-classify backend — classify/grouping services, 7 new columns, /classify-batch, /api/groups, /apply-groups

deploy/db/documents.py

@@ -2,12 +2,12 @@
 from .connection import query, execute, execute_returning
 
 
-def insert(filename, mime_type, original_bytes, status="uploaded"):
+def insert(filename, mime_type, original_bytes, status="uploaded", batch_id=None):
     """Insert document, return row dict."""
     return execute_returning(
-        """INSERT INTO documents (filename, mime_type, original_bytes, status)
-           VALUES (%s, %s, %s, %s) RETURNING *""",
-        (filename, mime_type, original_bytes, status),
+        """INSERT INTO documents (filename, mime_type, original_bytes, status, batch_id)
+           VALUES (%s, %s, %s, %s, %s) RETURNING *""",
+        (filename, mime_type, original_bytes, status, batch_id),
     )
 
 
@@ -34,3 +34,47 @@ def set_error(doc_id, error_message):
 
 def delete(doc_id):
     return execute("DELETE FROM documents WHERE id = %s", (doc_id,))
+
+
+def set_classification(doc_id, doc_type, own_number, parent_number, doc_date, counterparty):
+    """Store LLM classification results."""
+    return execute(
+        """UPDATE documents SET doc_type=%s, own_number=%s, parent_number=%s,
+           doc_date=%s, counterparty=%s, classify_status='classified'
+           WHERE id=%s""",
+        (doc_type, own_number, parent_number, doc_date, counterparty, doc_id),
+    )
+
+
+def set_classify_failed(doc_id, error):
+    return execute(
+        "UPDATE documents SET classify_status='failed', error_message=%s WHERE id=%s",
+        (error, doc_id),
+    )
+
+
+def list_pending(batch_id):
+    """Documents waiting for classification."""
+    return query(
+        "SELECT id, filename, elements_json FROM documents WHERE batch_id=%s AND classify_status='pending' AND status='parsed'",
+        (batch_id,),
+    )
+
+
+def list_by_batch(batch_id):
+    """All documents in a batch with classification fields."""
+    return query(
+        """SELECT id, filename, status, doc_type, own_number, parent_number,
+           doc_date, counterparty, classify_status, error_message
+           FROM documents WHERE batch_id=%s ORDER BY created_at""",
+        (batch_id,),
+    )
+
+
+def count_by_status(batch_id):
+    """Count documents by classify_status."""
+    rows = query(
+        "SELECT classify_status, count(*) as cnt FROM documents WHERE batch_id=%s GROUP BY classify_status",
+        (batch_id,),
+    )
+    return {r["classify_status"]: r["cnt"] for r in rows}

deploy/db/prompts.py

@@ -35,3 +35,18 @@ def seed_defaults():
             true, 'Авто-создан из llm_prompt.py _build_diff'
         )
     """)
+    _ensure_classify_prompt()
+
+
+def _ensure_classify_prompt():
+    """Ensure classify prompt exists (idempotent)."""
+    existing = query("SELECT id FROM prompts WHERE role = 'classify' AND is_active = true LIMIT 1")
+    if existing:
+        return
+    execute("""
+        INSERT INTO prompts (role, name, body, is_active, notes) VALUES (
+            'classify', 'default-v1',
+            'Ты — классификатор договорных документов. Ниже фрагмент текста документа.\n\nОпредели:\n1. doc_type: "contract" (договор), "supplement" (допсоглашение/допник), "specification" (спецификация/приложение), "other"\n2. own_number: номер ЭТОГО документа (например "МЭС-123-2024" или "ДС №2")\n3. parent_number: номер родительского договора (для допников/спецификаций — "к Договору №..."). Для самого договора — null.\n4. doc_date: дата документа в формате YYYY-MM-DD\n5. counterparty: название контрагента (из поля "Заказчик", "Арендатор" или аналогичного)\n6. confidence: "ok" если уверен, "low" если сомневаешься\n\nВерни СТРОГО JSON без пояснений:\n{\n  "doc_type": "...",\n  "own_number": "..." или null,\n  "parent_number": "..." или null,\n  "doc_date": "YYYY-MM-DD" или null,\n  "counterparty": "..." или null,\n  "confidence": "ok" или "low"\n}\n\nДОКУМЕНТ:\n---\n{header_text}\n---',
+            true, 'Авто-создан: classify prompt'
+        )
+    """)