ATProto._convert: bug fix for when fetching a link preview gets no data

for 2363a48, fixes https://console.cloud.google.com/errors/detail/CNrq2-2F78_HWg;time=P1D;locations=global?project=bridgy-federated
snarfed · Jan 16, 2025 · 19ffd30 · 19ffd30
1 parent 8796b9e
commit 19ffd30
Show file tree

Hide file tree

Showing 2 changed files with 31 additions and 9 deletions.
diff --git a/atproto.py b/atproto.py
@@ -906,7 +906,7 @@ def fetch_blob(url, blob_field, name, check_size=True, check_type=True):
                             logger.warning(e)
                             continue
 
-                        if link:
+                        if link and link.as1:
                             if img := util.get_url(link.as1, 'image'):
                                 props = appview.defs['app.bsky.embed.external#external']['properties']
                                 fetch_blob(img, props, name='thumb',

diff --git a/tests/test_atproto.py b/tests/test_atproto.py
@@ -1575,15 +1575,15 @@ def test_send_note_existing_repo(self, mock_create_task):
   <meta property="og:title" content="Titull" />
   <meta property="og:description" content="Descrypshun" />
 </head>
-</html>""", url='http://orig.co/inal'),
+</html>""", url='http://orig.co/post'),
         requests_response('blob contents', content_type='image/png'),
     ])
     def test_send_note_first_link_preview_embed(self, _, __):
         user = self.make_user_and_repo()
 
         obj = Object(id='fake:post', source_protocol='fake', our_as1={
             **NOTE_AS,
-            'content': 'My <a href="http://orig.co/inal">original</a> post',
+            'content': 'My <a href="http://orig.co/post">original</a> post',
         })
         self.assertTrue(ATProto.send(obj, 'https://bsky.brid.gy'))
 
@@ -1593,14 +1593,14 @@ def test_send_note_first_link_preview_embed(self, _, __):
         last_tid = arroba.util.int_to_tid(arroba.util._tid_ts_last)
         self.assert_equals({
             **NOTE_BSKY,
-            'bridgyOriginalText': 'My <a href="http://orig.co/inal">original</a> post',
+            'bridgyOriginalText': 'My <a href="http://orig.co/post">original</a> post',
             'embed': {
                 '$type': 'app.bsky.embed.external',
                 'external': {
                     '$type': 'app.bsky.embed.external#external',
                     'description': 'Descrypshun',
                     'title': 'Titull',
-                    'uri': 'http://orig.co/inal',
+                    'uri': 'http://orig.co/post',
                     'thumb': {
                         '$type': 'blob',
                         'mimeType': 'image/png',
@@ -1611,10 +1611,32 @@ def test_send_note_first_link_preview_embed(self, _, __):
             },
         }, repo.get_record('app.bsky.feed.post', last_tid), ignore=['facets'])
 
+    @patch.object(tasks_client, 'create_task', return_value=Task(name='my task'))
+    def test_send_note_link_preview_existing_object_no_data(self, _):
+        user = self.make_user_and_repo()
+
+        # no data!
+        Object(id='http://orig.co/post', source_protocol='web').put()
+
+        obj = Object(id='fake:post', source_protocol='fake', our_as1={
+            **NOTE_AS,
+            'content': 'My <a href="http://orig.co/post">original</a> post',
+        })
+        self.assertTrue(ATProto.send(obj, 'https://bsky.brid.gy'))
+
+        # check repo, record
+        did = user.key.get().get_copy(ATProto)
+        repo = self.storage.load_repo(did)
+        last_tid = arroba.util.int_to_tid(arroba.util._tid_ts_last)
+        self.assert_equals({
+            **NOTE_BSKY,
+            'bridgyOriginalText': 'My <a href="http://orig.co/post">original</a> post',
+        }, repo.get_record('app.bsky.feed.post', last_tid), ignore=['facets'])
+
     @patch.object(tasks_client, 'create_task', return_value=Task(name='my task'))
     @patch('requests.get', side_effect=[
         requests_response(f'<html><head><title>A poast</title></head></html>',
-                          url='http://orig.co/inal'),
+                          url='http://orig.co/post'),
     ])
     def test_send_note_link_preview_non_web_url(self, mock_get, mock_create_task):
         user = self.make_user_and_repo()
@@ -1664,7 +1686,7 @@ def test_send_note_link_preview_blocklisted_domain(self, mock_get, __):
   <meta property="og:title" content="Titull" />
   <meta property="og:description" content="Descrypshun" />
 </head>
-</html>""", url='http://orig.co/inal'),
+</html>""", url='http://orig.co/post'),
         requests_response('blob contents', content_type='image/png'),
     ])
     @patch.dict(
@@ -1676,7 +1698,7 @@ def test_send_note_truncated_original_post_embed_overrides_first_link_preview(
 
         obj = Object(id='fake:post', source_protocol='fake', our_as1={
             **NOTE_AS,
-            'content': 'My <a href="http://orig.co/inal">original</a> poaaast',
+            'content': 'My <a href="http://orig.co/post">original</a> poaaast',
         })
         self.assertTrue(ATProto.send(obj, 'https://bsky.brid.gy'))
 
@@ -1687,7 +1709,7 @@ def test_send_note_truncated_original_post_embed_overrides_first_link_preview(
         self.assert_equals({
             **NOTE_BSKY,
             'text': 'My original […]',
-            'bridgyOriginalText': 'My <a href="http://orig.co/inal">original</a> poaaast',
+            'bridgyOriginalText': 'My <a href="http://orig.co/post">original</a> poaaast',
             'embed': {
                 '$type': 'app.bsky.embed.external',
                 'external': {