Added new option '--sub-format' to choose the format of the subtitles to downloade (defaut=srt)
This commit is contained in:
parent
553d097442
commit
9e62bc4439
|
@ -29,6 +29,7 @@
|
||||||
"simulate": false,
|
"simulate": false,
|
||||||
"skip_download": false,
|
"skip_download": false,
|
||||||
"subtitleslang": null,
|
"subtitleslang": null,
|
||||||
|
"subtitlesformat": "srt",
|
||||||
"test": true,
|
"test": true,
|
||||||
"updatetime": true,
|
"updatetime": true,
|
||||||
"usenetrc": false,
|
"usenetrc": false,
|
||||||
|
|
|
@ -42,7 +42,7 @@ class TestYoutubeSubtitles(unittest.TestCase):
|
||||||
DL = FakeDownloader()
|
DL = FakeDownloader()
|
||||||
DL.params['allsubtitles'] = False
|
DL.params['allsubtitles'] = False
|
||||||
DL.params['writesubtitles'] = False
|
DL.params['writesubtitles'] = False
|
||||||
|
DL.params['subtitlesformat'] = 'srt'
|
||||||
def test_youtube_no_subtitles(self):
|
def test_youtube_no_subtitles(self):
|
||||||
DL = FakeDownloader()
|
DL = FakeDownloader()
|
||||||
DL.params['writesubtitles'] = False
|
DL.params['writesubtitles'] = False
|
||||||
|
@ -80,6 +80,14 @@ class TestYoutubeSubtitles(unittest.TestCase):
|
||||||
info_dict = IE.extract('QRS8MkLhQmM')
|
info_dict = IE.extract('QRS8MkLhQmM')
|
||||||
subtitles = info_dict[0]['subtitles']
|
subtitles = info_dict[0]['subtitles']
|
||||||
self.assertEqual(len(subtitles), 12)
|
self.assertEqual(len(subtitles), 12)
|
||||||
|
def test_youtube_subtitles_format(self):
|
||||||
|
DL = FakeDownloader()
|
||||||
|
DL.params['writesubtitles'] = True
|
||||||
|
DL.params['subtitlesformat'] = 'sbv'
|
||||||
|
IE = YoutubeIE(DL)
|
||||||
|
info_dict = IE.extract('QRS8MkLhQmM')
|
||||||
|
sub = info_dict[0]['subtitles'][0]
|
||||||
|
self.assertEqual(md5(sub[2]), '13aeaa0c245a8bed9a451cb643e3ad8b')
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
unittest.main()
|
unittest.main()
|
||||||
|
|
|
@ -78,9 +78,10 @@ class FileDownloader(object):
|
||||||
updatetime: Use the Last-modified header to set output file timestamps.
|
updatetime: Use the Last-modified header to set output file timestamps.
|
||||||
writedescription: Write the video description to a .description file
|
writedescription: Write the video description to a .description file
|
||||||
writeinfojson: Write the video description to a .info.json file
|
writeinfojson: Write the video description to a .info.json file
|
||||||
writesubtitles: Write the video subtitles to a file (default=srt)
|
writesubtitles: Write the video subtitles to a file
|
||||||
onlysubtitles: Downloads only the subtitles of the video
|
onlysubtitles: Downloads only the subtitles of the video
|
||||||
allsubtitles: Downloads all the subtitles of the video
|
allsubtitles: Downloads all the subtitles of the video
|
||||||
|
subtitlesformat: Subtitle format [sbv/srt] (default=srt)
|
||||||
subtitleslang: Language of the subtitles to download
|
subtitleslang: Language of the subtitles to download
|
||||||
test: Download only first bytes to test the downloader.
|
test: Download only first bytes to test the downloader.
|
||||||
keepvideo: Keep the video file after post-processing
|
keepvideo: Keep the video file after post-processing
|
||||||
|
@ -445,8 +446,9 @@ class FileDownloader(object):
|
||||||
# that way it will silently go on when used with unsupporting IE
|
# that way it will silently go on when used with unsupporting IE
|
||||||
subtitle = info_dict['subtitles'][0]
|
subtitle = info_dict['subtitles'][0]
|
||||||
(sub_error, sub_lang, sub) = subtitle
|
(sub_error, sub_lang, sub) = subtitle
|
||||||
|
sub_format = self.params.get('subtitlesformat')
|
||||||
try:
|
try:
|
||||||
sub_filename = filename.rsplit('.', 1)[0] + u'.' + sub_lang + u'.srt'
|
sub_filename = filename.rsplit('.', 1)[0] + u'.' + sub_lang + u'.' + sub_format
|
||||||
self.report_writesubtitles(sub_filename)
|
self.report_writesubtitles(sub_filename)
|
||||||
with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8') as subfile:
|
with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8') as subfile:
|
||||||
subfile.write(sub)
|
subfile.write(sub)
|
||||||
|
@ -458,10 +460,11 @@ class FileDownloader(object):
|
||||||
|
|
||||||
if self.params.get('allsubtitles', False) and 'subtitles' in info_dict and info_dict['subtitles']:
|
if self.params.get('allsubtitles', False) and 'subtitles' in info_dict and info_dict['subtitles']:
|
||||||
subtitles = info_dict['subtitles']
|
subtitles = info_dict['subtitles']
|
||||||
|
sub_format = self.params.get('subtitlesformat')
|
||||||
for subtitle in subtitles:
|
for subtitle in subtitles:
|
||||||
(sub_error, sub_lang, sub) = subtitle
|
(sub_error, sub_lang, sub) = subtitle
|
||||||
try:
|
try:
|
||||||
sub_filename = filename.rsplit('.', 1)[0] + u'.' + sub_lang + u'.srt'
|
sub_filename = filename.rsplit('.', 1)[0] + u'.' + sub_lang + u'.' + sub_format
|
||||||
self.report_writesubtitles(sub_filename)
|
self.report_writesubtitles(sub_filename)
|
||||||
with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8') as subfile:
|
with io.open(encodeFilename(sub_filename), 'w', encoding='utf-8') as subfile:
|
||||||
subfile.write(sub)
|
subfile.write(sub)
|
||||||
|
|
|
@ -244,7 +244,7 @@ class YoutubeIE(InfoExtractor):
|
||||||
return (u'WARNING: video has no closed captions', None)
|
return (u'WARNING: video has no closed captions', None)
|
||||||
return sub_lang_list
|
return sub_lang_list
|
||||||
|
|
||||||
def _request_subtitle(self, sub_lang, sub_name, video_id, format = 'srt'):
|
def _request_subtitle(self, sub_lang, sub_name, video_id, format):
|
||||||
self.report_video_subtitles_request(video_id, sub_lang)
|
self.report_video_subtitles_request(video_id, sub_lang)
|
||||||
params = compat_urllib_parse.urlencode({
|
params = compat_urllib_parse.urlencode({
|
||||||
'lang': sub_lang,
|
'lang': sub_lang,
|
||||||
|
@ -264,7 +264,7 @@ class YoutubeIE(InfoExtractor):
|
||||||
def _extract_subtitle(self, video_id):
|
def _extract_subtitle(self, video_id):
|
||||||
self.report_video_subtitles_download(video_id)
|
self.report_video_subtitles_download(video_id)
|
||||||
sub_lang_list = self._get_available_subtitles(video_id)
|
sub_lang_list = self._get_available_subtitles(video_id)
|
||||||
|
sub_format = self._downloader.params.get('subtitlesformat')
|
||||||
if self._downloader.params.get('subtitleslang', False):
|
if self._downloader.params.get('subtitleslang', False):
|
||||||
sub_lang = self._downloader.params.get('subtitleslang')
|
sub_lang = self._downloader.params.get('subtitleslang')
|
||||||
elif 'en' in sub_lang_list:
|
elif 'en' in sub_lang_list:
|
||||||
|
@ -274,15 +274,16 @@ class YoutubeIE(InfoExtractor):
|
||||||
if not sub_lang in sub_lang_list:
|
if not sub_lang in sub_lang_list:
|
||||||
return (u'WARNING: no closed captions found in the specified language "%s"' % sub_lang, None)
|
return (u'WARNING: no closed captions found in the specified language "%s"' % sub_lang, None)
|
||||||
|
|
||||||
subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id)
|
subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format)
|
||||||
return [subtitle]
|
return [subtitle]
|
||||||
|
|
||||||
def _extract_all_subtitles(self, video_id):
|
def _extract_all_subtitles(self, video_id):
|
||||||
self.report_video_subtitles_download(video_id)
|
self.report_video_subtitles_download(video_id)
|
||||||
sub_lang_list = self._get_available_subtitles(video_id)
|
sub_lang_list = self._get_available_subtitles(video_id)
|
||||||
|
sub_format = self._downloader.params.get('subtitlesformat')
|
||||||
subtitles = []
|
subtitles = []
|
||||||
for sub_lang in sub_lang_list:
|
for sub_lang in sub_lang_list:
|
||||||
subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id)
|
subtitle = self._request_subtitle(sub_lang, sub_lang_list[sub_lang].encode('utf-8'), video_id, sub_format)
|
||||||
subtitles.append(subtitle)
|
subtitles.append(subtitle)
|
||||||
return subtitles
|
return subtitles
|
||||||
|
|
||||||
|
@ -505,7 +506,7 @@ class YoutubeIE(InfoExtractor):
|
||||||
else:
|
else:
|
||||||
video_description = ''
|
video_description = ''
|
||||||
|
|
||||||
# closed captions
|
# subtitles
|
||||||
video_subtitles = None
|
video_subtitles = None
|
||||||
|
|
||||||
if self._downloader.params.get('writesubtitles', False):
|
if self._downloader.params.get('writesubtitles', False):
|
||||||
|
|
|
@ -182,6 +182,9 @@ def parseOpts():
|
||||||
video_format.add_option('--all-subs',
|
video_format.add_option('--all-subs',
|
||||||
action='store_true', dest='allsubtitles',
|
action='store_true', dest='allsubtitles',
|
||||||
help='downloads all the available subtitles of the video (currently youtube only)', default=False)
|
help='downloads all the available subtitles of the video (currently youtube only)', default=False)
|
||||||
|
video_format.add_option('--sub-format',
|
||||||
|
action='store', dest='subtitlesformat', metavar='LANG',
|
||||||
|
help='subtitle format [srt/sbv] (default=srt) (currently youtube only)', default='srt')
|
||||||
video_format.add_option('--sub-lang',
|
video_format.add_option('--sub-lang',
|
||||||
action='store', dest='subtitleslang', metavar='LANG',
|
action='store', dest='subtitleslang', metavar='LANG',
|
||||||
help='language of the subtitles to download (optional) use IETF language tags like \'en\'')
|
help='language of the subtitles to download (optional) use IETF language tags like \'en\'')
|
||||||
|
@ -458,6 +461,7 @@ def _real_main():
|
||||||
'writesubtitles': opts.writesubtitles,
|
'writesubtitles': opts.writesubtitles,
|
||||||
'onlysubtitles': opts.onlysubtitles,
|
'onlysubtitles': opts.onlysubtitles,
|
||||||
'allsubtitles': opts.allsubtitles,
|
'allsubtitles': opts.allsubtitles,
|
||||||
|
'subtitlesformat': opts.subtitlesformat,
|
||||||
'subtitleslang': opts.subtitleslang,
|
'subtitleslang': opts.subtitleslang,
|
||||||
'matchtitle': decodeOption(opts.matchtitle),
|
'matchtitle': decodeOption(opts.matchtitle),
|
||||||
'rejecttitle': decodeOption(opts.rejecttitle),
|
'rejecttitle': decodeOption(opts.rejecttitle),
|
||||||
|
|
Loading…
Reference in New Issue