feat: filter site bodies before media upload

tests/test_source_adapters.py

@@ -35,6 +35,13 @@ class FollowedPageResponse(FakeResponse):
         return data
 
 
+class EmptyBodyResponse(FakeResponse):
+    async def json(self, **_kwargs):
+        data = await super().json(**_kwargs)
+        data["items"][0]["text"] = ""
+        return data
+
+
 class FakeSession:
     def post(self, *_args, **_kwargs):
         return FakeResponse()
@@ -45,6 +52,11 @@ class FollowedPageSession(FakeSession):
         return FollowedPageResponse()
 
 
+class EmptyBodySession(FakeSession):
+    def post(self, *_args, **_kwargs):
+        return EmptyBodyResponse()
+
+
 class SourceAdapterTests(unittest.IsolatedAsyncioTestCase):
     async def test_worker_response_is_normalized(self) -> None:
         client = SiteParserClient(FakeSession(), "http://worker", "token", "captcha", 30)
@@ -66,6 +78,17 @@ class SourceAdapterTests(unittest.IsolatedAsyncioTestCase):
             "runtime_state_json": "{}",
         })
         self.assertEqual(items[0].text, "Title\nAuthor")
+        self.assertEqual(items[0].body_text, "Title\nAuthor")
+
+    async def test_title_is_not_counted_as_site_body(self) -> None:
+        client = SiteParserClient(EmptyBodySession(), "http://worker", "token", "captcha", 30)
+        items, _ = await client.fetch({
+            "url": "https://example.test/rss.xml",
+            "settings_json": '{"format":"rss"}',
+            "runtime_state_json": "{}",
+        })
+        self.assertEqual(items[0].text, "Title")
+        self.assertEqual(items[0].body_text, "")
 
     def test_site_config_is_required(self) -> None:
         with self.assertRaisesRegex(ValueError, "нужен конфиг"):
@@ -75,6 +98,10 @@ class SourceAdapterTests(unittest.IsolatedAsyncioTestCase):
         with self.assertRaisesRegex(ValueError, "content_selector"):
             validate_source_config("site", {"format": "rss", "follow_links": True})
 
+    def test_site_min_text_length_is_validated(self) -> None:
+        with self.assertRaisesRegex(ValueError, "min_text_length"):
+            validate_source_config("site", {"format": "rss", "min_text_length": "many"})
+
 
 if __name__ == "__main__":
     unittest.main()

tests/test_storage_uploader.py

@@ -1,7 +1,7 @@
 import asyncio
 from pathlib import Path
 
-from vk_parser_app.workers.vk_storage_uploader import TelegramStorageUploader, video_provider
+from vk_parser_app.workers.media_uploader import MediaUploader, video_provider
 
 
 class FakeContent:
@@ -35,7 +35,7 @@ def test_video_provider() -> None:
 
 async def test_http_video_download() -> None:
     path = "/tmp/vkparser_tg_media_http_test.mp4"
-    result = await TelegramStorageUploader().download_http_video(FakeSession(), "https://example.test/a.mp4", path)
+    result = await MediaUploader().download_http_video(FakeSession(), "https://example.test/a.mp4", path)
     assert result["size_bytes"] == 5
     assert Path(path).read_bytes() == b"video"
     Path(path).unlink()