Fixed bug in file:// downloader handler with uris containing percent-escaped chars

scrapy/core/downloader/handlers/file.py

@@ -1,6 +1,8 @@
 """Download handler for file:// scheme"""
 from __future__ import with_statement
 
+from urllib import url2pathname
+
 from twisted.internet import defer
 from scrapy.core.downloader.responsetypes import responsetypes
 
@@ -10,7 +12,7 @@ def download_file(request, spider):
     return defer.maybeDeferred(_all_in_one_read_download_file, request, spider)
 
 def _all_in_one_read_download_file(request, spider):
-    filepath = request.url.split("file://")[1]
+    filepath = url2pathname(request.url.split("file://")[1])
     with open(filepath) as f:
         body = f.read()
     respcls = responsetypes.from_args(filename=filepath, body=body)

scrapy/tests/test_downloader_handlers.py

@@ -20,7 +20,7 @@ class FileTestCase(unittest.TestCase):
 
     def setUp(self):
         self.tmpname = self.mktemp()
-        fd = open(self.tmpname, 'w')
+        fd = open(self.tmpname + '<', 'w')
         fd.write('0123456789')
         fd.close()
 
@@ -30,7 +30,8 @@ class FileTestCase(unittest.TestCase):
             self.assertEquals(response.status, 200)
             self.assertEquals(response.body, '0123456789')
 
-        request = Request('file://%s' % self.tmpname)
+        request = Request('file://%s' % self.tmpname + '<')
+        assert request.url.upper().endswith('%3C')
         return download_file(request, BaseSpider()).addCallback(_test)
 
     def test_non_existent(self):