Spaces:
Runtime error
Runtime error
Remove unused and check subtitle_url.
Browse files- fetchYoutubeSubtitle.py +6 -32
fetchYoutubeSubtitle.py
CHANGED
@@ -24,12 +24,14 @@ async def fetchSubtitle(url: str, lang: Optional[str] = 'en', vttType="vtt") ->
|
|
24 |
if info_dict.get("subtitles"):
|
25 |
# get first available subtitle
|
26 |
subtitle_url = getVttUrlFromSubtitles(info_dict.get("subtitles"), lang, vttType)
|
27 |
-
|
28 |
-
|
|
|
29 |
if info_dict.get("automatic_captions"):
|
30 |
subtitle_url = getVttUrlFromSubtitles(info_dict.get("automatic_captions"), lang, vttType)
|
31 |
-
|
32 |
-
|
|
|
33 |
return None
|
34 |
|
35 |
async def fetchSubtitleUrls(url: str) -> json:
|
@@ -48,31 +50,3 @@ async def fetchSubtitleUrls(url: str) -> json:
|
|
48 |
if info_dict.get("automatic_captions"):
|
49 |
return info_dict.get("automatic_captions")
|
50 |
return None
|
51 |
-
|
52 |
-
def get_subtitle(url, lang='en'):
|
53 |
-
if lang is None:
|
54 |
-
lang = 'en'
|
55 |
-
# Download subtitles if available
|
56 |
-
ydl_opts = {
|
57 |
-
'writesubtitles': True,
|
58 |
-
'outtmpl': '%(id)s.%(ext)s',
|
59 |
-
'subtitleslangs': [lang],
|
60 |
-
'skip_download': True,
|
61 |
-
}
|
62 |
-
try:
|
63 |
-
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
|
64 |
-
info_dict = ydl.extract_info(url, download=True)
|
65 |
-
video_id = info_dict.get("id", None)
|
66 |
-
if video_id is None:
|
67 |
-
return None
|
68 |
-
|
69 |
-
subtitle_file = f"{video_id}.{lang}.vtt"
|
70 |
-
with open(subtitle_file, 'r') as f:
|
71 |
-
subtitle_content = f.read()
|
72 |
-
subtitle_content = re.sub(r"<[^>]+>", "", subtitle_content)
|
73 |
-
return subtitle_content
|
74 |
-
except error:
|
75 |
-
print(error)
|
76 |
-
return None
|
77 |
-
|
78 |
-
return None
|
|
|
24 |
if info_dict.get("subtitles"):
|
25 |
# get first available subtitle
|
26 |
subtitle_url = getVttUrlFromSubtitles(info_dict.get("subtitles"), lang, vttType)
|
27 |
+
if subtitle_url:
|
28 |
+
with ydl.urlopen(subtitle_url) as subtitle:
|
29 |
+
return subtitle.read().decode()
|
30 |
if info_dict.get("automatic_captions"):
|
31 |
subtitle_url = getVttUrlFromSubtitles(info_dict.get("automatic_captions"), lang, vttType)
|
32 |
+
if subtitle_url:
|
33 |
+
with ydl.urlopen(subtitle_url) as subtitle:
|
34 |
+
return subtitle.read().decode()
|
35 |
return None
|
36 |
|
37 |
async def fetchSubtitleUrls(url: str) -> json:
|
|
|
50 |
if info_dict.get("automatic_captions"):
|
51 |
return info_dict.get("automatic_captions")
|
52 |
return None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|