Merge pull request #63 from RyouMon/fix-file-download-exception

Fix FileException when downloading media. (caused by OffsiteMiddleware)
RyouMon · Dec 2, 2024 · ecd5f2d · ecd5f2d
2 parents 6e828df + bc5ed0f
commit ecd5f2d
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 7 deletions.
diff --git a/src/favorites_crawler/pipelines.py b/src/favorites_crawler/pipelines.py
@@ -23,7 +23,10 @@ class BasePipeline(FilesPipeline):
     def get_media_requests(self, item, info):
         item_dict = ItemAdapter(item).asdict()
         referer = item_dict.get('referer')
-        return (Request(url, headers={'referer': referer}) for url in item_dict.get(self.files_urls_field, ()))
+        return (
+            Request(url, headers={'referer': referer}, dont_filter=True)
+            for url in item_dict.get(self.files_urls_field, ())
+        )
 
     def file_path(self, request, response=None, info=None, *, item=None):
         return item.get_filepath(request.url, info.spider)

diff --git a/tests/test_pipelines.py b/tests/test_pipelines.py
@@ -11,10 +11,10 @@ class TestBasePipeline:
     def test_should_set_referer_when_get_media_requests(self, mock_request):
         mock_item = {'referer': sentinel.referer, 'file_urls': [getattr(sentinel, f'url-{i}') for i in range(1, 10)]}
 
-        list(BasePipeline('mock_path').get_media_requests(mock_item, None))
+        list(BasePipeline('mock_path', crawler=None).get_media_requests(mock_item, None))
 
         calls = [
-            call(getattr(sentinel, f'url-{i}'), headers={'referer': sentinel.referer})
+            call(getattr(sentinel, f'url-{i}'), headers={'referer': sentinel.referer}, dont_filter=True)
             for i in range(1, 10)
         ]
         mock_request.assert_has_calls(calls, any_order=True)
@@ -24,7 +24,7 @@ def test_file_path_should_call_item_get_filepath(self):
         mock_info = MagicMock()
         mock_item = MagicMock()
 
-        BasePipeline('mock_path').file_path(mock_request, None, mock_info, item=mock_item)
+        BasePipeline('mock_path', crawler=None).file_path(mock_request, None, mock_info, item=mock_item)
 
         mock_item.get_filepath.assert_called_once_with(mock_request.url, mock_info.spider)
 
@@ -36,11 +36,11 @@ def test_process_item_should_drop_item_when_cbz_file_already_exist(self, tmp_pat
         mock_item.get_folder_name.return_value = 'abc'
 
         with pytest.raises(DropItem):
-            ComicPipeline(str(tmp_path)).process_item(mock_item, None)
+            ComicPipeline(str(tmp_path), crawler=None).process_item(mock_item, None)
 
     @patch('favorites_crawler.pipelines.create_comic_archive')
     def test_should_create_comic_archive_when_close_spider(self, mock_create_comic_archive, tmp_path):
-        pipeline = ComicPipeline('mock_path')
+        pipeline = ComicPipeline('mock_path', crawler=None)
         pipeline.files_path = tmp_path
         (tmp_path / 'comic').mkdir()
         pipeline.comic_comments = {'comic': b'comment'}
@@ -51,7 +51,7 @@ def test_should_create_comic_archive_when_close_spider(self, mock_create_comic_a
 
     @patch('favorites_crawler.pipelines.create_comic_archive')
     def test_should_not_create_comic_archive_when_comic_comments_is_empty(self, mock_create_comic_archive, tmp_path):
-        pipeline = ComicPipeline('mock_path')
+        pipeline = ComicPipeline('mock_path', crawler=None)
         pipeline.comic_comments = {}
 
         pipeline.close_spider(None)