ID

08c1bad

date

2024-02-16 07:18:53+00:00

author

Alex Chan <alex@alexwlchan.net>

parent

20ff716

message

Download the subtitles, if available

changed files

2 files, 18 additions, 2 deletions

• web/save_youtube_videos.py
• web/test_save_youtube_videos.py

Changed files

web/save_youtube_videos.py (4983) → web/save_youtube_videos.py (5294)

diff --git a/web/save_youtube_videos.py b/web/save_youtube_videos.py
index 5ae3104..fd228ae 100755
--- a/web/save_youtube_videos.py
+++ b/web/save_youtube_videos.py
@@ -88,7 +88,7 @@ def log_result(format_template):
 
 def classify_file_type(
     video_id: str, filename: str
-) -> Literal["video", "info", "thumbnail"] | None:
+) -> Literal["video", "info", "thumbnail", "subtitles"] | None:
     """
     Given an already-downloaded file, work out what sort of file it is.
     """
@@ -120,6 +120,17 @@ def classify_file_type(
     if filename.endswith((f"-{video_id}.info.json", f" [{video_id}].info.json")):
         return "info"
 
+    if filename.endswith(
+        (
+            f" [{video_id}].en.vtt",
+            f" [{video_id}].en-US.vtt",
+            f" [{video_id}].en-GB.vtt",
+            f" [{video_id}].en-CA.vtt",
+            f" [{video_id}].live_chat.json",
+        )
+    ):
+        return "subtitles"
+
     raise ValueError(f"Unrecognised filename: {filename}")
 
 
@@ -152,7 +163,7 @@ def download_video(*, video_id, download_root):
     # Construct the command.  The expensive bit is redownloading the
     # video file, so don't do that if it's already downloaded.
     video_url = f"https://youtube.com/watch?v={video_id}"
-    cmd = [video_url]
+    cmd = [video_url, "--write-sub"]
 
     if has_video:
         cmd.append("--skip-download")

web/test_save_youtube_videos.py (874) → web/test_save_youtube_videos.py (1030)

diff --git a/web/test_save_youtube_videos.py b/web/test_save_youtube_videos.py
index c182a8a..b8157c9 100644
--- a/web/test_save_youtube_videos.py
+++ b/web/test_save_youtube_videos.py
@@ -23,6 +23,11 @@ from save_youtube_videos import classify_file_type
             "The World's Most Remote Buildings-wGS53t8ZbO8.f251.webm.part",
             None,
         ),
+        (
+            "IjCylxs8hZU",
+            "Soviet Flying Aircraft Carriers Were Ingenious [IjCylxs8hZU].en.vtt",
+            "subtitles",
+        ),
     ],
 )
 def test_classify_file_type(video_id, filename, file_type):