deploy: auto-deploy 17:13:13

app/api/v1/reconcile.py

@@ -88,14 +88,32 @@ def _failed_result(pair_index: int, error: str) -> Dict[str, Any]:
 
 
 def _validate_extensions(src_ext: str, dst_ext: str, idx: int) -> None:
+    if not src_ext:
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Source file has no recognizable extension."})
+    if not dst_ext:
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Destination file has no recognizable extension."})
     if src_ext not in SUPPORTED_EXTENSIONS:
-        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Unsupported source format '{src_ext}'."})
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Unsupported source format '{src_ext}'. Supported: {', '.join(sorted(SUPPORTED_EXTENSIONS))}."})
     if dst_ext not in SUPPORTED_EXTENSIONS:
-        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Unsupported destination format '{dst_ext}'."})
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Unsupported destination format '{dst_ext}'. Supported: {', '.join(sorted(SUPPORTED_EXTENSIONS))}."})
     if src_ext != dst_ext:
         raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: File type mismatch - source '{src_ext}' != destination '{dst_ext}'."})
 
 
+def _validate_file_data(data: bytes, label: str, pair_index: int) -> None:
+    if not data:
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {pair_index}: {label} file is empty."})
+    if len(data) > _MAX_UPLOAD_BYTES:
+        raise HTTPException(status_code=413, detail={"success": False, "message": f"Pair {pair_index}: {label} file exceeds {_settings.max_upload_mb} MB limit."})
+
+
+def _validate_url(url: str, label: str, pair_index: int) -> None:
+    if not url or not url.strip():
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {pair_index}: {label} URL is empty."})
+    if not url.lower().startswith(("http://", "https://")):
+        raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {pair_index}: {label} URL must start with http:// or https://."})
+
+
 def _build_response(results: List[Dict[str, Any]], start_time: float) -> ReconciliationResponse:
     failed = sum(1 for r in results if r["status"] == "FAILED")
     return ReconciliationResponse(
@@ -240,10 +258,16 @@ async def reconcile_files(
     for idx, (src, dst, cm) in enumerate(files, 1):
         if src is None and dst is None:
             continue
-        if src is None or dst is None:
-            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Both source and destination files required."})
-        src_ext = extract_extension(src.filename or "")
-        dst_ext = extract_extension(dst.filename or "")
+        if src is None:
+            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Source file is missing."})
+        if dst is None:
+            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Destination file is missing."})
+        if not src.filename:
+            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Source file has no filename."})
+        if not dst.filename:
+            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Destination file has no filename."})
+        src_ext = extract_extension(src.filename)
+        dst_ext = extract_extension(dst.filename)
         _validate_extensions(src_ext, dst_ext, idx)
         pairs.append((src, dst, cm, idx, src_ext, dst_ext))
 
@@ -254,8 +278,8 @@ async def reconcile_files(
         async with semaphore_jobs:
             src_data = await src.read()
             dst_data = await dst.read()
-            if len(src_data) > _MAX_UPLOAD_BYTES or len(dst_data) > _MAX_UPLOAD_BYTES:
-                return _failed_result(idx, "File size exceeds maximum allowed limit")
+            _validate_file_data(src_data, f"Pair {idx} source", idx)
+            _validate_file_data(dst_data, f"Pair {idx} destination", idx)
             mapping = _parse_column_mapping_str(cm_str)
             return await _process_file_pair(src_data, dst_data, src_ext, dst_ext, mapping, idx)
 
@@ -280,7 +304,15 @@ async def reconcile_urls(
         raise HTTPException(status_code=400, detail={"success": False, "message": "At least 1 pair is required."})
 
     for idx, pair in enumerate(body.pairs, 1):
-        _validate_extensions(extract_extension(pair.source), extract_extension(pair.destination), idx)
+        if not pair.source or not pair.source.strip():
+            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Source URL is empty."})
+        if not pair.destination or not pair.destination.strip():
+            raise HTTPException(status_code=400, detail={"success": False, "message": f"Pair {idx}: Destination URL is empty."})
+        _validate_url(pair.source, "Source", idx)
+        _validate_url(pair.destination, "Destination", idx)
+        src_ext = extract_extension(pair.source)
+        dst_ext = extract_extension(pair.destination)
+        _validate_extensions(src_ext, dst_ext, idx)
 
     async with aiohttp.ClientSession() as session:
         async def process_url_pair(pair: ReconciliationPair, idx: int) -> Dict[str, Any]:
@@ -296,8 +328,8 @@ async def reconcile_urls(
                 except Exception as e:
                     return _failed_result(idx, str(e))
 
-                if len(src_data) > _MAX_UPLOAD_BYTES or len(dst_data) > _MAX_UPLOAD_BYTES:
-                    return _failed_result(idx, "File size exceeds maximum allowed limit")
+                _validate_file_data(src_data, f"Pair {idx} source", idx)
+                _validate_file_data(dst_data, f"Pair {idx} destination", idx)
 
                 mapping = _parse_column_mapping_model(pair.column_mapping)
                 return await _process_file_pair(src_data, dst_data, src_ext, dst_ext, mapping, idx)