v1.0.161: cgi.FieldStorage for UTF-8 filenames

deploy/services/upload.py

@@ -1,6 +1,5 @@
 """Upload service — accept file, store to DB, parse."""
-import uuid, base64, json
-from email.parser import BytesParser
+import uuid, base64, json, io, cgi
 from db import documents, contracts, supplements
 from .parse import parse_file
 
@@ -9,23 +8,28 @@ def handle_upload(rfile, content_type, content_length):
     """Parse multipart upload, store in DB, return result dict."""
     body = rfile.read(content_length)
 
-    msg = BytesParser().parsebytes(
-        b"Content-Type: " + content_type.encode() + b"\r\n\r\n" + body
-    )
+    environ = {
+        "REQUEST_METHOD": "POST",
+        "CONTENT_TYPE": content_type,
+        "CONTENT_LENGTH": str(content_length),
+    }
+    fs = cgi.FieldStorage(fp=io.BytesIO(body), environ=environ, keep_blank_values=True)
+
     filename = None
     file_data = None
     contract_id = ""
 
-    if msg.is_multipart():
-        for part in msg.get_payload():
-            fn = part.get_filename()
-            if fn:
-                filename = fn
-                file_data = part.get_payload(decode=True)
-            else:
-                disp = part.get("Content-Disposition", "")
-                if 'name="contract_id"' in disp:
-                    contract_id = part.get_payload(decode=True).decode("utf-8", errors="ignore").strip()
+    if "files" in fs:
+        item = fs["files"]
+        if isinstance(item, list):
+            item = item[0]
+        filename = item.filename
+        file_data = item.file.read() if hasattr(item, "file") else item.value
+        if isinstance(file_data, str):
+            file_data = file_data.encode("utf-8")
+
+    if "contract_id" in fs:
+        contract_id = fs.getfirst("contract_id", "")
 
     if not filename or not file_data:
         return {"ok": False, "error": "no file in request"}
@@ -33,13 +37,11 @@ def handle_upload(rfile, content_type, content_length):
     mime = _mime_for(filename)
     b64 = base64.b64encode(file_data).decode()
 
-    # Удалить старый файл с тем же именем в этом контракте
     if contract_id:
         supplements.delete_by_document(contract_id, filename)
 
     doc = documents.insert(filename, mime, b64)
 
-    # Создать контракт если нет
     if not contract_id:
         from datetime import datetime
         now = datetime.now()
@@ -51,7 +53,6 @@ def handle_upload(rfile, content_type, content_length):
 
     supplements.insert(contract_id, doc["id"], supp_type)
 
-    # Парсинг на VM
     parsed = parse_file(filename, file_data)
     if parsed and parsed.get("status") == "parsed":
         documents.set_parsed(doc["id"], parsed["elements"])