Merge pull request scrapy#1254 from scrapy/bugfix-1232

[MRG +1] Unquote request path before passing to FTPClient, it already escape paths

Author: kmike

scrapy/core/downloader/handlers/ftp.py

@@ -30,7 +30,7 @@
 
 import re
 from io import BytesIO
-from six.moves.urllib.parse import urlparse
+from six.moves.urllib.parse import urlparse, unquote
 
 from twisted.internet import reactor
 from twisted.protocols.ftp import FTPClient, CommandFailed
@@ -73,7 +73,7 @@ def download_request(self, request, spider):
                                     request.meta["ftp_password"],
                                     passive=request.meta.get("ftp_passive", 1))
         return creator.connectTCP(parsed_url.hostname, parsed_url.port or 21).addCallback(self.gotClient,
-                                request, parsed_url.path)
+                                request, unquote(parsed_url.path))
 
     def gotClient(self, client, request, filepath):
         self.client = client

tests/test_downloader_handlers.py

@@ -511,7 +511,9 @@ def setUp(self):
         os.mkdir(self.directory)
         userdir = os.path.join(self.directory, self.username)
         os.mkdir(userdir)
-        FilePath(userdir).child('file.txt').setContent("I have the power!")
+        fp = FilePath(userdir)
+        fp.child('file.txt').setContent("I have the power!")
+        fp.child('file with spaces.txt').setContent("Moooooooooo power!")
 
         # setup server
         realm = FTPRealm(anonymousRoot=self.directory, userHome=self.directory)
@@ -547,6 +549,19 @@ def _test(r):
             self.assertEqual(r.headers, {'Local Filename': [''], 'Size': ['17']})
         return self._add_test_callbacks(d, _test)
 
+    def test_ftp_download_path_with_spaces(self):
+        request = Request(
+            url="ftp://127.0.0.1:%s/file with spaces.txt" % self.portNum,
+            meta={"ftp_user": self.username, "ftp_password": self.password}
+        )
+        d = self.download_handler.download_request(request, None)
+
+        def _test(r):
+            self.assertEqual(r.status, 200)
+            self.assertEqual(r.body, 'Moooooooooo power!')
+            self.assertEqual(r.headers, {'Local Filename': [''], 'Size': ['18']})
+        return self._add_test_callbacks(d, _test)
+
     def test_ftp_download_notexist(self):
         request = Request(url="ftp://127.0.0.1:%s/notexist.txt" % self.portNum,
                 meta={"ftp_user": self.username, "ftp_password": self.password})