routes

app/routes.py

@@ -54,11 +54,20 @@ logger = logging.getLogger(__name__)
 router = APIRouter()
 
 
+MAX_WEBHOOK_BODY_BYTES = (
+    2 * 1024 * 1024
+)
+
+
 RETRIEVAL_UNAVAILABLE_DETAIL = (
     "Vyhľadávací index alebo embeddingový "
     "model momentálne nie je dostupný."
 )
 
+INVALID_QUERY_DETAIL = (
+    "Neplatný vyhľadávací dotaz."
+)
+
 INTERNAL_ERROR_DETAIL = (
     "Nastala interná chyba servera."
 )
@@ -166,21 +175,16 @@ class HealthResponse(BaseModel):
     ready: bool
 
     database_exists: bool
-    database_path: str
 
     search_engine: str
     rag_enabled: bool
 
-    zpwiki_root: str
     zpwiki_exists: bool
 
     security_configured: bool
 
 
 class RagSource(BaseModel):
-    # RAG source sa môže v budúcnosti
-    # rozšíriť o ďalšie retrieval metadata
-    # bez rozbitia response modelu.
     model_config = ConfigDict(
         extra="allow",
     )
@@ -302,12 +306,17 @@ def raise_retrieval_error(
         error,
         ValueError,
     ):
+        logger.warning(
+            "Neplatný retrieval dotaz: %s",
+            error,
+        )
+
         raise HTTPException(
             status_code=(
                 status.HTTP_400_BAD_REQUEST
             ),
-            detail=str(
+            detail=(
-                error
+                INVALID_QUERY_DETAIL
             ),
         ) from error
 
@@ -420,6 +429,79 @@ def configured_repository() -> str:
         ) from error
 
 
+async def read_webhook_body(
+    request: Request,
+) -> bytes:
+    content_length = request.headers.get(
+        "content-length"
+    )
+
+    if content_length:
+        try:
+            declared_length = int(
+                content_length
+            )
+
+        except ValueError as error:
+            raise HTTPException(
+                status_code=400,
+                detail=(
+                    "Neplatná Content-Length "
+                    "hlavička"
+                ),
+            ) from error
+
+        if declared_length < 0:
+            raise HTTPException(
+                status_code=400,
+                detail=(
+                    "Neplatná Content-Length "
+                    "hlavička"
+                ),
+            )
+
+        if (
+            declared_length
+            > MAX_WEBHOOK_BODY_BYTES
+        ):
+            raise HTTPException(
+                status_code=413,
+                detail=(
+                    "Webhook payload je "
+                    "príliš veľký"
+                ),
+            )
+
+    chunks: list[bytes] = []
+
+    total_size = 0
+
+    async for chunk in request.stream():
+        total_size += len(
+            chunk
+        )
+
+        if (
+            total_size
+            > MAX_WEBHOOK_BODY_BYTES
+        ):
+            raise HTTPException(
+                status_code=413,
+                detail=(
+                    "Webhook payload je "
+                    "príliš veľký"
+                ),
+            )
+
+        chunks.append(
+            chunk
+        )
+
+    return b"".join(
+        chunks
+    )
+
+
 @router.get(
     "/health",
     include_in_schema=False,
@@ -436,6 +518,7 @@ def health() -> dict[str, Any]:
 
     try:
         validate_security_configuration()
+
         security_configured = True
 
     except RuntimeError:
@@ -453,16 +536,10 @@ def health() -> dict[str, Any]:
         "database_exists": (
             database_exists
         ),
-        "database_path": str(
-            DB_FILE
-        ),
         "search_engine": (
             "hybrid_fts5_embeddings"
         ),
         "rag_enabled": True,
-        "zpwiki_root": str(
-            ZPWIKI_ROOT
-        ),
         "zpwiki_exists": (
             zpwiki_exists
         ),
@@ -665,10 +742,14 @@ async def gitea_webhook(
     dict[str, Any]
     | JSONResponse
 ):
-    raw_body = await request.body()
-
     secret = webhook_secret()
 
+    raw_body = await read_webhook_body(
+        request
+    )
+
+    # Gitea podpis overujeme nad presným raw
+    # request body ešte pred JSON parsingom.
     if not verify_gitea_signature(
         raw_body,
         x_gitea_signature,