Merge pull request #287 from bellingcat/fix/insta_tbot_empty

Only return success for instagram_tbot_extractor.py with content.

Author: erinhmclark

src/auto_archiver/modules/instagram_tbot_extractor/instagram_tbot_extractor.py

@@ -88,6 +88,9 @@ def download(self, item: Metadata) -> Metadata:
 
             if message:
                 result.set_content(message).set_title(message[:128])
+            elif result.is_empty():
+                logger.debug(f"No media found for link {url=} for {self.name}: {message}")
+                return False
             return result.success("insta-via-bot")
 
     def _send_url_to_bot(self, url: str):
@@ -104,13 +107,13 @@ def _process_messages(self, chat, since_id, tmp_dir, result):
         message = ""
         time.sleep(3)
         # media is added before text by the bot so it can be used as a stop-logic mechanism
-        while attempts < max(self.timeout - 3, 3) and (not message or not len(seen_media)):
+        while attempts < max(self.timeout - 3, 15) and (not message or not len(seen_media)):
             attempts += 1
             time.sleep(1)
             for post in self.client.iter_messages(chat, min_id=since_id):
                 since_id = max(since_id, post.id)
                 # Skip known filler message:
-                if post.message == "The bot receives information through https://hikerapi.com/p/hJqpppqi":
+                if "The bot receives information through https://hikerapi.com/" in post.message:
                     continue
                 if post.media and post.id not in seen_media:
                     filename_dest = os.path.join(tmp_dir, f"{chat.id}_{post.id}")

tests/extractors/test_instagram_tbot_extractor.py

@@ -68,6 +68,12 @@ def test_download_invalid(extractor, metadata_sample, mocker):
     assert extractor.download(metadata_sample) is False
 
 
+def test_fails_with_empty_response(extractor, metadata_sample, mocker):
+    mocker.patch.object(extractor, "_send_url_to_bot", return_value=(mocker.MagicMock(), 101))
+    mocker.patch.object(extractor, "_process_messages", return_value="")
+    assert extractor.download(metadata_sample) is False
+
+
 @pytest.mark.skip(reason="Requires authentication.")
 class TestInstagramTbotExtractorReal(TestExtractorBase):
     # To run these tests set the TELEGRAM_API_ID and TELEGRAM_API_HASH environment variables, and ensure the session file exists.