diff options
| author | Yen Chi Hsuan <yan12125@gmail.com> | 2015-12-19 18:21:42 +0800 |
|---|---|---|
| committer | Yen Chi Hsuan <yan12125@gmail.com> | 2015-12-19 18:21:42 +0800 |
| commit | d631d5f9f27f93767226192e4288990413fa9dbd (patch) | |
| tree | 055755c210c91b43d2491087764560c2f9c0ab39 | |
| parent | 4f29fa99069760dc47ef9ca5dbf607a567d2982f (diff) | |
| download | youtube-dl-d631d5f9f27f93767226192e4288990413fa9dbd.tar.xz | |
[utils] Fix TTML conversion
Tolerate invalid timestamps (closes #7909)
| -rw-r--r-- | test/test_utils.py | 7 | ||||
| -rw-r--r-- | youtube_dl/utils.py | 11 |
2 files changed, 13 insertions, 5 deletions
diff --git a/test/test_utils.py b/test/test_utils.py index 8fc74e591..86045e680 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -661,8 +661,8 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4') {'like_count': 190, 'dislike_count': 10})) def test_parse_dfxp_time_expr(self): - self.assertEqual(parse_dfxp_time_expr(None), 0.0) - self.assertEqual(parse_dfxp_time_expr(''), 0.0) + self.assertEqual(parse_dfxp_time_expr(None), None) + self.assertEqual(parse_dfxp_time_expr(''), None) self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1) self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1) self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0) @@ -676,6 +676,9 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4') <p begin="0" end="1">The following line contains Chinese characters and special symbols</p> <p begin="1" end="2">第二行<br/>♪♪</p> <p begin="2" dur="1"><span>Third<br/>Line</span></p> + <p begin="3" end="-1">Lines with invalid timestamps are ignored</p> + <p begin="-1" end="-1">Ignore, two</p> + <p begin="3" dur="-1">Ignored, three</p> </div> </body> </tt>''' diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 91917fc96..ee20c3d9b 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -1976,7 +1976,7 @@ def match_filter_func(filter_str): def parse_dfxp_time_expr(time_expr): if not time_expr: - return 0.0 + return mobj = re.match(r'^(?P<time_offset>\d+(?:\.\d+)?)s?$', time_expr) if mobj: @@ -2020,10 +2020,15 @@ def dfxp2srt(dfxp_data): raise ValueError('Invalid dfxp/TTML subtitle') for para, index in zip(paras, itertools.count(1)): - begin_time = parse_dfxp_time_expr(para.attrib['begin']) + begin_time = parse_dfxp_time_expr(para.attrib.get('begin')) end_time = parse_dfxp_time_expr(para.attrib.get('end')) + dur = parse_dfxp_time_expr(para.attrib.get('dur')) + if begin_time is None: + continue if not end_time: - end_time = begin_time + parse_dfxp_time_expr(para.attrib['dur']) + if not dur: + continue + end_time = begin_time + dur out.append('%d\n%s --> %s\n%s\n\n' % ( index, srt_subtitles_timecode(begin_time), |