Parse Content-Disposition header properly

qutebrowser/browser/downloads.py

@@ -30,7 +30,7 @@ import qutebrowser.utils.message as message
 from qutebrowser.utils.log import downloads as logger
 from qutebrowser.utils.usertypes import PromptMode, Question, Timer
 from qutebrowser.utils.misc import (interpolate_color, format_seconds,
-                                    format_size)
+                                    format_size, get_http_header)
 
 
 class DownloadItem(QObject):
@@ -321,18 +321,10 @@ class DownloadManager(QObject):
         Args:
             reply: The QNetworkReply to get a filename for.
         """
-        filename = None
         # First check if the Content-Disposition header has a filename
         # attribute.
-        if reply.hasRawHeader('Content-Disposition'):
+        from qutebrowser.utils.debug import set_trace; set_trace()
-            header = reply.rawHeader('Content-Disposition')
+        filename = get_http_header(reply, 'Content-Disposition', 'filename')
-            data = header.split(':', maxsplit=1)[1].strip()
-            for pair in data.split(';'):
-                if '=' in pair:
-                    key, value = pair.split('=')
-                    if key == 'filename':
-                        filename = value.strip('"')
-                        break
         # Then try to get filename from url
         if not filename:
             filename = reply.url().path()

qutebrowser/utils/misc.py

@@ -26,6 +26,8 @@ import os
 import re
 import sys
 import shlex
+import email
+import email.policy
 import os.path
 import operator
 import urllib.request
@@ -393,6 +395,33 @@ def check_print_compat():
     return not (os.name == 'nt' and qt_version_check('5.3.0', operator.lt))
 
 
+def get_http_header(reply, headername, param=None):
+    """Get a parameter from a HTTP header.
+
+    Note we use the email value to get a HTTP header, because they're both MIME
+    headers and email supports that.
+
+    Args:
+        reply: The QNetworkReply to get the header from.
+        headername: The name of the header.
+        param: The name of the param to get, or None to get the whole contents.
+
+    Return:
+        The data as a string, or None if the data wasn't found.
+
+    FIXME add tests
+    """
+    if not reply.hasRawHeader(headername):
+        return None
+    header = (headername.encode('ascii') + b': ' +
+              bytes(reply.rawHeader(headername)))
+    msg = email.message_from_bytes(header, policy=email.policy.HTTP)
+    if param is not None:
+        return msg.get_param(param, header=headername)
+    else:
+        return msg.get(headername, None)
+
+
 class EventLoop(QEventLoop):
 
     """A thin wrapper around QEventLoop.