Feat/add title metadata

youtube_transcript_api/_transcripts.py

@@ -3,7 +3,7 @@
 from itertools import chain
 
 from html import unescape
-from typing import List, Dict, Iterator, Iterable, Pattern, Optional
+from typing import List, Dict, Iterator, Iterable, Pattern, Optional, Tuple
 
 from defusedxml import ElementTree
 
@@ -55,6 +55,7 @@ class FetchedTranscript:
 
     snippets: List[FetchedTranscriptSnippet]
     video_id: str
+    title: str
     language: str
     language_code: str
     is_generated: bool
@@ -105,6 +106,7 @@ def __init__(
         self,
         http_client: Session,
         video_id: str,
+        title: str,
         url: str,
         language: str,
         language_code: str,
@@ -117,6 +119,7 @@ def __init__(
         """
         self._http_client = http_client
         self.video_id = video_id
+        self.title = title
         self._url = url
         self.language = language
         self.language_code = language_code
@@ -141,6 +144,7 @@ def fetch(self, preserve_formatting: bool = False) -> FetchedTranscript:
         return FetchedTranscript(
             snippets=snippets,
             video_id=self.video_id,
+            title=self.title,
             language=self.language,
             language_code=self.language_code,
             is_generated=self.is_generated,
@@ -167,6 +171,7 @@ def translate(self, language_code: str) -> "Transcript":
         return Transcript(
             self._http_client,
             self.video_id,
+            self.title,
             "{url}&tlang={language_code}".format(
                 url=self._url, language_code=language_code
             ),
@@ -205,14 +210,15 @@ def __init__(
 
     @staticmethod
     def build(
-        http_client: Session, video_id: str, captions_json: Dict
+        http_client: Session, video_id: str, captions_json: Dict, title: str
     ) -> "TranscriptList":
         """
         Factory method for TranscriptList.
 
         :param http_client: http client which is used to make the transcript retrieving http calls
         :param video_id: the id of the video this TranscriptList is for
         :param captions_json: the JSON parsed from the YouTube pages static HTML
+        :param title: the title of the video
         :return: the created TranscriptList
         """
         translation_languages = [
@@ -235,6 +241,7 @@ def build(
             transcript_dict[caption["languageCode"]] = Transcript(
                 http_client,
                 video_id,
+                title,
                 caption["baseUrl"].replace("&fmt=srv3", ""),
                 caption["name"]["runs"][0]["text"],
                 caption["languageCode"],
@@ -350,26 +357,25 @@ def __init__(self, http_client: Session, proxy_config: Optional[ProxyConfig]):
         self._proxy_config = proxy_config
 
     def fetch(self, video_id: str) -> TranscriptList:
-        return TranscriptList.build(
-            self._http_client,
-            video_id,
-            self._fetch_captions_json(video_id),
-        )
+        captions_json, title = self._fetch_video_data(video_id)
+        return TranscriptList.build(self._http_client, video_id, captions_json, title)
 
-    def _fetch_captions_json(self, video_id: str, try_number: int = 0) -> Dict:
+    def _fetch_video_data(self, video_id: str, try_number: int = 0) -> Tuple[Dict, str]:
         try:
             html = self._fetch_video_html(video_id)
             api_key = self._extract_innertube_api_key(html, video_id)
             innertube_data = self._fetch_innertube_data(video_id, api_key)
-            return self._extract_captions_json(innertube_data, video_id)
+            captions_json = self._extract_captions_json(innertube_data, video_id)
+            title = self._extract_video_title(innertube_data)
+            return captions_json, title
         except RequestBlocked as exception:
             retries = (
                 0
                 if self._proxy_config is None
                 else self._proxy_config.retries_when_blocked
             )
             if try_number + 1 < retries:
-                return self._fetch_captions_json(video_id, try_number=try_number + 1)
+                return self._fetch_video_data(video_id, try_number=try_number + 1)
             raise exception.with_proxy_config(self._proxy_config)
 
     def _extract_innertube_api_key(self, html: str, video_id: str) -> str:
@@ -392,6 +398,9 @@ def _extract_captions_json(self, innertube_data: Dict, video_id: str) -> Dict:
 
         return captions_json
 
+    def _extract_video_title(self, innertube_data: Dict) -> str:
+        return innertube_data.get("videoDetails", {}).get("title")
+
     def _assert_playability(self, playability_status_data: Dict, video_id: str) -> None:
         playability_status = playability_status_data.get("status")
         if (

youtube_transcript_api/test/test_api.py

@@ -72,6 +72,7 @@ def setUp(self):
             language_code="en",
             is_generated=False,
             video_id="GJLlxj_dtq8",
+            title="Surface Go Review - It’s Awesome",
         )
         self.ref_transcript_raw = self.ref_transcript.to_raw_data()
 
@@ -91,7 +92,6 @@ def setUp(self):
 
     def test_fetch(self):
         transcript = YouTubeTranscriptApi().fetch("GJLlxj_dtq8")
-
         self.assertEqual(
             transcript,
             self.ref_transcript,
@@ -114,6 +114,7 @@ def test_fetch__with_altered_user_agent(self):
             responses.POST,
             "https://www.youtube.com/youtubei/v1/player",
             body=load_asset("youtube_altered_user_agent.innertube.json.static"),
+            content_type="application/json",
         )
 
         transcript = YouTubeTranscriptApi().fetch("GJLlxj_dtq8")

youtube_transcript_api/test/test_cli.py

@@ -41,6 +41,7 @@ def setUp(self):
                 language_code="en",
                 is_generated=True,
                 video_id="GJLlxj_dtq8",
+                title="Video Title",
             )
         )
         self.transcript_mock.translate = MagicMock(return_value=self.transcript_mock)

youtube_transcript_api/test/test_formatters.py

@@ -31,6 +31,7 @@ def setUp(self):
             language_code="en",
             is_generated=True,
             video_id="12345",
+            title="Video Title",
         )
         self.transcripts = [self.transcript, self.transcript]
         self.transcript_raw = self.transcript.to_raw_data()