Merge pull request #288 from kxepal/update-multipart

Update multipart

aiohttp/multipart.py

@@ -53,8 +53,7 @@ def is_quoted(string):
         return string[0] == string[-1] == '"'
 
     def is_rfc5987(string):
-        # this isn't very correct
-        return "''" in string
+        return is_token(string) and string.count("'") == 2
 
     def is_extended_param(string):
         return string.endswith('*')
@@ -103,20 +102,12 @@ def unescape(text, *, chars=''.join(map(re.escape, CHAR))):
                 continue
 
         elif is_extended_param(key):
-            if is_quoted(value):
-                warnings.warn(BadContentDispositionParam(item))
-                continue
-            elif is_rfc5987(value):
+            if is_rfc5987(value):
                 encoding, _, value = value.split("'", 2)
                 encoding = encoding or 'utf-8'
-            elif "'":
-                warnings.warn(BadContentDispositionParam(item))
-                continue
-            elif not is_token(value):
+            else:
                 warnings.warn(BadContentDispositionParam(item))
                 continue
-            else:
-                encoding = 'utf-8'
 
             try:
                 value = unquote(value, encoding, 'strict')
@@ -569,17 +560,16 @@ def __init__(self, obj, headers=None, *, chunk_size=8192):
         }
 
     def _fill_headers_with_defaults(self):
-        """Updates part headers by """
-        if CONTENT_LENGTH not in self.headers:
-            content_length = self._guess_content_length(self.obj)
-            if content_length is not None:
-                self.headers[CONTENT_LENGTH] = str(content_length)
-
         if CONTENT_TYPE not in self.headers:
             content_type = self._guess_content_type(self.obj)
             if content_type is not None:
                 self.headers[CONTENT_TYPE] = content_type
 
+        if CONTENT_LENGTH not in self.headers:
+            content_length = self._guess_content_length(self.obj)
+            if content_length is not None:
+                self.headers[CONTENT_LENGTH] = str(content_length)
+
         if CONTENT_DISPOSITION not in self.headers:
             filename = self._guess_filename(self.obj)
             if filename is not None:
@@ -588,12 +578,20 @@ def _fill_headers_with_defaults(self):
     def _guess_content_length(self, obj):
         if isinstance(obj, bytes):
             return len(obj)
+        elif isinstance(obj, str):
+            *_, params = parse_mimetype(self.headers.get(CONTENT_TYPE))
+            charset = params.get('charset', 'us-ascii')
+            return len(obj.encode(charset))
+        elif isinstance(obj, io.StringIO):
+            *_, params = parse_mimetype(self.headers.get(CONTENT_TYPE))
+            charset = params.get('charset', 'us-ascii')
+            return len(obj.getvalue().encode(charset)) - obj.tell()
+        elif isinstance(obj, io.BytesIO):
+            return len(obj.getvalue()) - obj.tell()
         elif isinstance(obj, io.IOBase):
             try:
                 return os.fstat(obj.fileno()).st_size - obj.tell()
             except (AttributeError, OSError):
-                if isinstance(obj, io.BytesIO):
-                    return len(obj.getvalue()) - obj.tell()
                 return None
         else:
             return None
@@ -602,7 +600,7 @@ def _guess_content_type(self, obj, default='application/octet-stream'):
         if hasattr(obj, 'name'):
             name = getattr(obj, 'name')
             return mimetypes.guess_type(name)[0]
-        elif isinstance(obj, str):
+        elif isinstance(obj, (str, io.StringIO)):
             return 'text/plain; charset=utf-8'
         else:
             return default
@@ -635,21 +633,20 @@ def serialize(self):
                 for item in self.headers.items()
             )
         yield b'\r\n\r\n'
+        yield from self._maybe_encode_stream(self._serialize_obj())
+        yield b'\r\n'
 
+    def _serialize_obj(self):
         obj = self.obj
         mtype, stype, *_ = parse_mimetype(self.headers.get(CONTENT_TYPE))
         serializer = self._serialize_map.get((mtype, stype))
         if serializer is not None:
-            stream = serializer(obj)
-        else:
-            for key in self._serialize_map:
-                if not isinstance(key, tuple) and isinstance(obj, key):
-                    stream = self._serialize_map[key](obj)
-                    break
-            else:
-                stream = self._serialize_default(obj)
-        yield from self._maybe_encode_stream(stream)
-        yield b'\r\n'
+            return serializer(obj)
+
+        for key in self._serialize_map:
+            if not isinstance(key, tuple) and isinstance(obj, key):
+                return self._serialize_map[key](obj)
+        return self._serialize_default(obj)
 
     def _serialize_bytes(self, obj):
         yield obj

docs/api.rst

@@ -1,8 +1,10 @@
+.. _aiohttp-api:
+
 Helpers API
 ===========
 
 All public names from submodules ``client``, ``connector``,
-``errors``, ``parsers``, ``protocol``, ``server``, ``utils``,
+``errors``, ``multipart``, ``parsers``, ``protocol``, ``server``, ``utils``,
 ``websocket`` and ``wsgi`` are exported into ``aiohttp``
 namespace.
 
@@ -38,6 +40,14 @@ aiohttp.helpers module
     :undoc-members:
     :show-inheritance:
 
+aiohttp.multipart module
+------------------------
+
+.. automodule:: aiohttp.multipart
+    :members:
+    :undoc-members:
+    :show-inheritance:
+
 aiohttp.parsers module
 ----------------------
 

docs/client.rst

@@ -251,6 +251,7 @@ If you pass file object as data parameter, aiohttp will stream it to server
 automatically. Check :class:`aiohttp.stream.StreamReader` for supported format
 information.
 
+.. seealso:: :ref:`aiohttp-multipart`
 
 Streaming uploads
 -----------------

docs/index.rst

@@ -129,6 +129,7 @@ Contents:
    web_reference
    server
    multidict
+   multipart
    api
    gunicorn
    contributing