fix: ревью Opus — баги #1-2 + Protocol расширен + BatchGroupingResult

deploy/repository.py

@@ -63,6 +63,30 @@ class Repository(Protocol):
         """Текущая спецификация контракта."""
         ...
 
+    def get_document(self, doc_id: str) -> dict | None:
+        """Получить документ по id."""
+        ...
+
+    def list_by_batch(self, batch_id: str) -> list[dict]:
+        """Все документы батча с полями классификации."""
+        ...
+
+    def count_by_status(self, batch_id: str) -> dict[str, int]:
+        """Количество документов по classify_status."""
+        ...
+
+    def reset_classify_status(self, batch_id: str) -> None:
+        """Сбросить classify_status на 'pending'."""
+        ...
+
+    def set_classify_processing(self, doc_id: str) -> None:
+        """Пометить документ как обрабатываемый."""
+        ...
+
+    def delete_document(self, doc_id: str) -> None:
+        """Удалить документ."""
+        ...
+
 
 # ── Продакшен: обёртка над db/*.py ──────────────────────────────
 
@@ -76,8 +100,7 @@ class PgRepository:
 
     def set_document_parsed(self, doc_id, elements):
         from db import documents
-        import json as _json
-        documents.set_parsed(doc_id, _json.dumps(elements, ensure_ascii=False))
+        documents.set_parsed(doc_id, elements)  # documents.set_parsed уже делает json.dumps
 
     def set_document_error(self, doc_id, error):
         from db import documents
@@ -118,6 +141,30 @@ class PgRepository:
         from db import spec_current
         return spec_current.list_by_contract(contract_id)
 
+    def get_document(self, doc_id):
+        from db import documents
+        return documents.get(doc_id)
+
+    def list_by_batch(self, batch_id):
+        from db import documents
+        return documents.list_by_batch(batch_id)
+
+    def count_by_status(self, batch_id):
+        from db import documents
+        return documents.count_by_status(batch_id)
+
+    def reset_classify_status(self, batch_id):
+        from db import documents
+        documents.reset_classify_status(batch_id)
+
+    def set_classify_processing(self, doc_id):
+        from db import documents
+        documents.set_classify_processing(doc_id)
+
+    def delete_document(self, doc_id):
+        from db import documents
+        documents.delete(doc_id)
+
 
 # ── Тестовый: in-memory заглушка ────────────────────────────────
 
@@ -191,3 +238,29 @@ class MemRepository:
 
     def get_spec_current(self, contract_id):
         return self.spec_current.get(contract_id, [])
+
+    def get_document(self, doc_id):
+        return self.documents.get(doc_id)
+
+    def list_by_batch(self, batch_id):
+        return [d for d in self.documents.values() if d.get("batch_id") == batch_id]
+
+    def count_by_status(self, batch_id):
+        counts = {}
+        for d in self.documents.values():
+            if d.get("batch_id") == batch_id:
+                s = d.get("classify_status", "unknown")
+                counts[s] = counts.get(s, 0) + 1
+        return counts
+
+    def reset_classify_status(self, batch_id):
+        for d in self.documents.values():
+            if d.get("batch_id") == batch_id:
+                d["classify_status"] = "pending"
+
+    def set_classify_processing(self, doc_id):
+        if doc_id in self.documents:
+            self.documents[doc_id]["classify_status"] = "processing"
+
+    def delete_document(self, doc_id):
+        self.documents.pop(doc_id, None)