From 7dbbfedd3b765f9c0ac95424c8864f2ea4feee5f Mon Sep 17 00:00:00 2001
From: Florian Bruhin <git@the-compiler.org>
Date: Sat, 3 Jan 2015 17:50:59 +0100
Subject: [PATCH] Replace unencodable chars in download filenames.

Fixes #427.
---
 qutebrowser/browser/downloads.py     | 13 ++++++++++++-
 qutebrowser/test/utils/test_utils.py | 20 ++++++++++++++++++++
 qutebrowser/utils/utils.py           |  8 ++++++++
 3 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/qutebrowser/browser/downloads.py b/qutebrowser/browser/downloads.py
index d7b5f7c35..f9aaa022e 100644
--- a/qutebrowser/browser/downloads.py
+++ b/qutebrowser/browser/downloads.py
@@ -21,6 +21,7 @@
 
 import io
 import os
+import sys
 import os.path
 import shutil
 import functools
@@ -385,6 +386,10 @@ class DownloadItem(QObject):
                              "existing: {}, fileobj {}".format(
                                  filename, self._filename, self.fileobj))
         filename = os.path.expanduser(filename)
+        # Remove chars which can't be encoded in the filename encoding.
+        # See https://github.com/The-Compiler/qutebrowser/issues/427
+        encoding = sys.getfilesystemencoding()
+        filename = utils.force_encoding(filename, encoding)
         if os.path.isabs(filename) and os.path.isdir(filename):
             # We got an absolute directory from the user, so we save it under
             # the default filename in that directory.
@@ -639,7 +644,10 @@ class DownloadManager(QAbstractListModel):
             return self.fetch_request(request, filename, fileobj, page,
                                       auto_remove)
         q = self._prepare_question()
-        q.default = urlutils.filename_from_url(request.url())
+        filename = urlutils.filename_from_url(request.url())
+        encoding = sys.getfilesystemencoding()
+        filename = utils.force_encoding(filename, encoding)
+        q.default = filename
         message_bridge = objreg.get('message-bridge', scope='window',
                                     window=self._win_id)
         q.answered.connect(
@@ -719,6 +727,9 @@ class DownloadManager(QAbstractListModel):
             download.autoclose = False
         else:
             q = self._prepare_question()
+            encoding = sys.getfilesystemencoding()
+            suggested_filename = utils.force_encoding(suggested_filename,
+                                                      encoding)
             q.default = suggested_filename
             q.answered.connect(download.set_filename)
             q.cancelled.connect(download.cancel)
diff --git a/qutebrowser/test/utils/test_utils.py b/qutebrowser/test/utils/test_utils.py
index 027e0d516..a0db50b51 100644
--- a/qutebrowser/test/utils/test_utils.py
+++ b/qutebrowser/test/utils/test_utils.py
@@ -383,5 +383,25 @@ class RaisesTests(unittest.TestCase):
             utils.raises(ValueError, self.do_raise)
 
 
+class ForceEncodingTests(unittest.TestCase):
+
+    """Test force_encoding."""
+
+    def test_fitting_ascii(self):
+        """Test with a text fitting into ascii."""
+        text = 'hello world'
+        self.assertEqual(utils.force_encoding(text, 'ascii'), text)
+
+    def test_fitting_utf8(self):
+        """Test with a text fitting into utf-8."""
+        text = 'hellö wörld'
+        self.assertEqual(utils.force_encoding(text, 'utf-8'), text)
+
+    def test_not_fitting_ascii(self):
+        """Test with a text not fitting into ascii."""
+        text = 'hellö wörld'
+        self.assertEqual(utils.force_encoding(text, 'ascii'), 'hell? w?rld')
+
+
 if __name__ == '__main__':
     unittest.main()
diff --git a/qutebrowser/utils/utils.py b/qutebrowser/utils/utils.py
index 567c84e9c..3d264a335 100644
--- a/qutebrowser/utils/utils.py
+++ b/qutebrowser/utils/utils.py
@@ -542,3 +542,11 @@ def raises(exc, func, *args):
         return True
     else:
         return False
+
+
+def force_encoding(text, encoding):
+    """Make sure a given text is encodable with the given encoding.
+
+    This replaces all chars not encodable with question marks.
+    """
+    return text.encode(encoding, errors='replace').decode(encoding)