Use rfc6266 to handle Content-Disposition in downloads

pkg/PKGBUILD.qutebrowser-git

@@ -9,7 +9,7 @@ arch=(any)
 url="http://www.qutebrowser.org/"
 license=('GPL')
 depends=('python>=3.3' 'python-setuptools' 'python-pyqt5>=5.2' 'qt5-base>=5.2'
-         'qt5-webkit>=5.2' 'libxkbcommon-x11')
+         'qt5-webkit>=5.2' 'libxkbcommon-x11' 'python-rfc6266')
 makedepends=('python' 'python-setuptools')
 optdepends=('python-colorlog: colored logging output'
             'python-ipdb: better debugging')

qutebrowser/browser/downloads.py

@@ -24,6 +24,7 @@ import os.path
 from functools import partial
 from collections import deque
 
+import rfc6266
 from PyQt5.QtCore import pyqtSlot, pyqtSignal, QObject, QCoreApplication
 from PyQt5.QtNetwork import QNetworkRequest, QNetworkReply
 
@@ -32,12 +33,16 @@ import qutebrowser.utils.message as message
 import qutebrowser.utils.url as urlutils
 import qutebrowser.commands.utils as cmdutils
 from qutebrowser.utils.log import downloads as logger
+from qutebrowser.utils.log import fix_rfc2622
 from qutebrowser.utils.usertypes import PromptMode, Question, Timer
 from qutebrowser.utils.misc import (interpolate_color, format_seconds,
-                                    format_size, get_http_header)
+                                    format_size)
 from qutebrowser.commands.exceptions import CommandError
 
 
+fix_rfc2622()
+
+
 class DownloadItem(QObject):
 
     """A single download currently running.
@@ -328,7 +333,11 @@ class DownloadManager(QObject):
         """
         # First check if the Content-Disposition header has a filename
         # attribute.
-        filename = get_http_header(reply, 'Content-Disposition', 'filename')
+        if reply.hasRawHeader('Content-Disposition'):
+            # We use the unsafe variant of the filename as we sanitize it via
+            # os.path.basename later.
+            filename = rfc6266.parse_headers(
+                bytes(reply.rawHeader('Content-Disposition'))).filename_unsafe
         # Then try to get filename from url
         if not filename:
             filename = reply.url().path()

qutebrowser/browser/webpage.py

@@ -163,6 +163,7 @@ class BrowserPage(QWebPage):
             start_download: Emitted with the QNetworkReply associated with the
                             passed request.
         """
+        from qutebrowser.utils.debug import set_trace; set_trace()
         reply = self.networkAccessManager().get(request)
         self.start_download.emit(reply)
 

qutebrowser/utils/earlyinit.py

@@ -218,3 +218,37 @@ def check_pkg_resources():
         msgbox.exec_()
         app.quit()
         sys.exit(1)
+
+
+def check_rfc6266():
+    """Check if rfc6266 is installed."""
+    from PyQt5.QtWidgets import QApplication, QMessageBox
+    try:
+        import rfc6266  # pylint: disable=unused-variable
+    except ImportError:
+        app = QApplication(sys.argv)
+        msgbox = QMessageBox(QMessageBox.Critical, "qutebrowser: Fatal error!",
+                             textwrap.dedent("""
+            Fatal error: rfc6266 is required to run qutebrowser but could
+            not be imported! Maybe it's not installed?
+
+            On Debian/Ubuntu:
+                No package available, try:
+                pip install rfc6266
+
+            On Archlinux:
+                pacman -S python-rfc6266
+
+            On Windows:
+                pip install rfc6266
+
+            For other distributions:
+                Check your package manager for similiarly named packages.
+            """).strip())
+        if '--debug' in sys.argv:
+            print(file=sys.stderr)
+            traceback.print_exc()
+        msgbox.resize(msgbox.sizeHint())
+        msgbox.exec_()
+        app.quit()
+        sys.exit(1)

qutebrowser/utils/log.py

@@ -72,6 +72,7 @@ downloads = getLogger('downloads')
 js = getLogger('js')
 qt = getLogger('qt')
 style = getLogger('style')
+rfc6266 = getLogger('rfc6266')
 
 
 ram_handler = None
@@ -86,18 +87,37 @@ def init_log(args):
         raise ValueError("Invalid log level: {}".format(args.loglevel))
 
     console, ram = _init_handlers(numeric_level, args.color, args.loglines)
-    if args.logfilter is not None and numeric_level <= logging.DEBUG:
-        console.addFilter(LogFilter(args.logfilter.split(',')))
     root = getLogger()
     if console is not None:
+        if args.logfilter is not None and numeric_level <= logging.DEBUG:
+            console.addFilter(LogFilter(args.logfilter.split(',')))
+            console.addFilter(LeplFilter())
         root.addHandler(console)
     if ram is not None:
         root.addHandler(ram)
+        console.addFilter(LeplFilter())
     root.setLevel(logging.NOTSET)
     logging.captureWarnings(True)
     qInstallMessageHandler(qt_message_handler)
 
 
+def fix_rfc2622():
+    """Fix the rfc6266 logger.
+
+    In rfc2622 <= v0.04, a NullHandler class is added as handler, instead of an
+    object, which causes an exception later.
+
+    This was fixed in [1], but since v0.05 is not out yet, we work around the
+    issue by deleting the wrong handler.
+
+    This should be executed after init_log is done and rfc6266 is imported, but
+    before using it.
+
+    [1]: https://github.com/g2p/rfc6266/commit/cad58963ed13f5e1068fcc9e4326123b6b2bdcf8
+    """
+    rfc6266.removeHandler(logging.NullHandler)
+
+
 def _init_handlers(level, color, ram_capacity):
     """Init log handlers.
 
@@ -255,6 +275,22 @@ class LogFilter(logging.Filter):
         return False
 
 
+class LeplFilter(logging.Filter):
+
+    """Filter to filter debug log records by the lepl library."""
+
+    def filter(self, record):
+        """Determine if the specified record is to be logged."""
+        if (record.levelno == logging.INFO and
+                record.name == 'lepl.lexer.rewriters.AddLexer'):
+            # Special useless info message triggered by rfc6266
+            return False
+        if record.levelno > logging.DEBUG:
+            # More important than DEBUG, so we won't filter at all
+            return True
+        return not record.name.startswith('lepl.')
+
+
 class RAMHandler(logging.Handler):
 
     """Logging handler which keeps the messages in a deque in RAM.

qutebrowser/utils/misc.py

@@ -28,8 +28,6 @@ import os
 import re
 import sys
 import shlex
-import email
-import email.policy
 import os.path
 import operator
 import urllib.request
@@ -397,33 +395,6 @@ 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.

scripts/setupcommon.py

@@ -92,6 +92,7 @@ setupdata = {
                     "QtWebKit."),
     'long_description': read_file('README'),
     'url': 'http://www.qutebrowser.org/',
+    'requires': 'rfc6266',
     'author': _get_constant('author'),
     'author_email': _get_constant('email'),
     'license': _get_constant('license'),