by-qa 0.4.4.dev1__tar.gz → 0.4.4.dev3__tar.gz

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: by-qa
-Version: 0.4.4.dev1
+Version: 0.4.4.dev3
 Summary: Open-source knowledge-base ingestion and retrieval service.
 Author: AI Assistant
 License: MIT

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "by-qa"
-version = "0.4.4.dev1"
+version = "0.4.4.dev3"
 description = "Open-source knowledge-base ingestion and retrieval service."
 readme = "README.md"
 requires-python = ">=3.12"

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/src/by_qa/knowledge_base/api/routes.py

@@ -42,6 +42,7 @@ from by_qa.knowledge_base.api.schemas import (
     DocumentUpdateRequest,
     FileBuildStatusRequest,
     FileToMarkdownIndexRequest,
+    KnowledgeBaseFileCountRequest,
     KnowledgeItemDownloadRequest,
     KnowledgeItemGlobRequest,
     KnowledgeItemListDirRequest,
@@ -98,15 +99,17 @@ def _documented_error_response(
     result_msg: str,
     result_object: dict[str, Any] | None = None,
     status_code: int = 422,
+    preserve_http_status: bool = False,
 ) -> JSONResponse:
-    """Return the documented error envelope with HTTP status normalized to 200."""
-    logger.info(
-        "knowledge_base error response normalized to HTTP 200: business_status_code=%s, result_msg=%s",
-        status_code,
-        result_msg,
-    )
+    """Return the error envelope, normalizing legacy endpoints to HTTP 200."""
+    if not preserve_http_status:
+        logger.info(
+            "knowledge_base error response normalized to HTTP 200: business_status_code=%s, result_msg=%s",
+            status_code,
+            result_msg,
+        )
     return JSONResponse(
-        status_code=200,
+        status_code=status_code if preserve_http_status else 200,
         content=jsonable_encoder(
             {
                 "resultCode": "-1",
@@ -501,6 +504,61 @@ def register_routes(
             )
         return _documented_success_response(result_object={})
 
+    @app.post(
+        "/api/v1/knowledgeBases/fileCount",
+        openapi_extra={
+            "requestBody": {
+                "required": True,
+                "content": {
+                    "application/json": {
+                        "schema": KnowledgeBaseFileCountRequest.model_json_schema(
+                            by_alias=True
+                        )
+                    }
+                },
+            }
+        },
+    )
+    async def knowledge_base_file_count(http_request: Request):
+        # Parse locally so malformed bodies also honor this endpoint's HTTP 422
+        # contract rather than the application's legacy HTTP 200 error handler.
+        try:
+            request = KnowledgeBaseFileCountRequest.model_validate(
+                await http_request.json()
+            )
+        except ValueError:
+            return _documented_error_response(
+                result_msg="request validation failed",
+                status_code=422,
+                preserve_http_status=True,
+            )
+
+        logger.info(
+            "knowledge_base_file_count request received: kb_code=%s", request.kb_code
+        )
+        try:
+            service = await get_knowledge_base_service()
+            result = await service.file_count(request)
+        except KnowledgeBaseConfigurationError as exc:
+            return _documented_error_response(
+                result_msg=str(exc), status_code=503, preserve_http_status=True
+            )
+        except KnowledgeBaseValidationError as exc:
+            return _documented_error_response(
+                result_msg=str(exc), status_code=422, preserve_http_status=True
+            )
+        except Exception as exc:
+            logger.exception(
+                "knowledge_base_file_count unexpected error: kb_code=%s",
+                request.kb_code,
+            )
+            return _documented_error_response(
+                result_msg=str(exc) or "internal error",
+                status_code=500,
+                preserve_http_status=True,
+            )
+        return _documented_success_response(result_object=result)
+
     @app.post("/api/v1/directories/create")
     async def create_directory(
         background_tasks: BackgroundTasks,
@@ -694,6 +752,7 @@ def register_routes(
         file_description: str | None = Body(None, alias="fileDescription"),
         file_content: UploadFile | None = File(None, alias="fileContent"),
         process_front_matter: bool = Form(True, alias="processFrontMatter"),
+        extract_zip: bool = Form(True, alias="extractZip"),
         skip_if_duplicate: bool = Form(False, alias="skipIfDuplicate"),
         metadata: str | None = Form(None, alias="metadata"),
     ):
@@ -709,6 +768,7 @@ def register_routes(
                         file_content.content_type if file_content is not None else None
                     ),
                     "processFrontMatter": process_front_matter,
+                    "extractZip": extract_zip,
                     "skipIfDuplicate": skip_if_duplicate,
                     "metadata": _parse_optional_metadata_form(metadata),
                 }
@@ -724,16 +784,17 @@ def register_routes(
                 status_code=422,
             )
         logger.info(
-            "upload_file request received: kb_code=%s, file_path=%s, has_description=%s, process_front_matter=%s",
+            "upload_file request received: kb_code=%s, file_path=%s, has_description=%s, process_front_matter=%s, extract_zip=%s",
             request.kb_code,
             request.file_path,
             request.file_description is not None,
             request.process_front_matter,
+            request.extract_zip,
         )
         filename = (file_content.filename or "") if file_content is not None else ""
         try:
             service = await get_knowledge_item_ingestion_service()
-            if filename.lower().endswith(".zip"):
+            if request.extract_zip and filename.lower().endswith(".zip"):
                 if not zipfile.is_zipfile(io.BytesIO(payload or b"")):
                     return _documented_error_response(
                         result_msg="invalid zip file",

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/src/by_qa/knowledge_base/api/schemas.py

@@ -54,6 +54,17 @@ class DeleteKnowledgeBaseRequest(BaseModel):
     )
 
 
+class KnowledgeBaseFileCountRequest(BaseModel):
+    """Request body for counting all live files in a knowledge base."""
+
+    model_config = ConfigDict(populate_by_name=True)
+
+    kb_code: str = Field(
+        min_length=1,
+        validation_alias=AliasChoices("knCode", "kb_code"),
+    )
+
+
 class UpdateKnowledgeBaseRequest(BaseModel):
     """Request body for updating knowledge base business fields."""
 
@@ -329,6 +340,10 @@ class KnowledgeItemUploadRequest(BaseModel):
         default=True,
         validation_alias=AliasChoices("processFrontMatter", "process_front_matter"),
     )
+    extract_zip: bool = Field(
+        default=True,
+        validation_alias=AliasChoices("extractZip", "extract_zip"),
+    )
     metadata: dict[str, Any] | None = None
     skip_if_duplicate: bool = Field(
         default=False,

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/src/by_qa/knowledge_base/repositories/knowledge_fs_entry_repository.py

@@ -1261,6 +1261,21 @@ class KnowledgeFsEntryRepository:
         )
         return [dict(row) for row in await self._fetchall(cursor)]
 
+    async def count_file_entries_by_knowledge_base_id(
+        self, cursor, *, knowledge_base_id: int
+    ) -> int:
+        """Count all live files, regardless of directory depth or build state."""
+        await cursor.execute(
+            """SELECT COUNT(*) AS file_count
+            FROM knowledge_fs_entry
+            WHERE knowledge_base_id = %(knowledge_base_id)s
+              AND is_deleted = FALSE
+              AND entry_type = 'FILE'""",
+            {"knowledge_base_id": knowledge_base_id},
+        )
+        row = await cursor.fetchone()
+        return int(row["file_count"])
+
     async def list_file_entries_by_knowledge_base_id(
         self, cursor, *, knowledge_base_id: int
     ) -> list[dict[str, Any]]:

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/src/by_qa/knowledge_base/services/knowledge_base_service.py

@@ -19,6 +19,7 @@ from by_qa.knowledge_base.api.schemas import (
     DeleteDirectoryRequest,
     DeleteKnowledgeBaseRequest,
     FileBuildStatusRequest,
+    KnowledgeBaseFileCountRequest,
     KnowledgeItemDownloadRequest,
     KnowledgeItemGlobRequest,
     KnowledgeItemListDirItem,
@@ -150,7 +151,8 @@ class KnowledgeBaseService:
         completed_build_batch_ids: list[str] = []
         terminated_semantic_tasks: list[dict[str, Any]] = []
         completed_semantic_batches: dict[
-            str, tuple[dict[str, Any], dict[str, int]]
+            str,
+            tuple[dict[str, Any], dict[str, int]],
         ] = {}
         try:
             cursor = connection.cursor()
@@ -989,6 +991,27 @@ class KnowledgeBaseService:
             return target_path
         return target_path.rstrip("/") + current_path[len(source_path.rstrip("/")) :]
 
+    async def file_count(
+        self, request: KnowledgeBaseFileCountRequest
+    ) -> dict[str, int]:
+        """Return the number of non-deleted files in an active knowledge base."""
+        connection = await self.connection_factory()
+        try:
+            cursor = connection.cursor()
+            kb_row = await self.knowledge_base_repository.get_by_code(
+                cursor, request.kb_code
+            )
+            if not kb_row:
+                raise KnowledgeBaseValidationError(
+                    f"knowledge base not found: {request.kb_code}"
+                )
+            count = await self.knowledge_fs_entry_repository.count_file_entries_by_knowledge_base_id(
+                cursor, knowledge_base_id=self._row_id(kb_row)
+            )
+            return {"fileCount": count}
+        finally:
+            await connection.close()
+
     async def list_dir(
         self, request: KnowledgeItemListDirRequest
     ) -> KnowledgeItemListDirResponse:

{by_qa-0.4.4.dev1 → by_qa-0.4.4.dev3}/uv.lock

@@ -1,5 +1,5 @@
 version = 1
-revision = 3
+revision = 5
 requires-python = ">=3.12"
 resolution-markers = [
     "python_full_version >= '3.14' and sys_platform == 'win32'",
@@ -313,7 +313,7 @@ wheels = [
 
 [[package]]
 name = "by-qa"
-version = "0.4.4.dev1"
+version = "0.4.4.dev3"
 source = { editable = "." }
 dependencies = [
     { name = "httpx" },