[ffmpeg] Add dfxp (TTML) subtitles support (#3432, #5146)

9 years ago · bf6427d2fb
parent 672f1bd849
commit bf6427d2fb
3 changed files with 116 additions and 0 deletions
--- a/test/test_utils.py
+++ b/test/test_utils.py
@ -58,6 +58,8 @@ from youtube_dl.utils import (
    xpath_text,
    render_table,
    match_str,
+    parse_dfxp_time_expr,
+    dfxp2srt,
 )


@ -581,6 +583,42 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
            'like_count > 100 & dislike_count <? 50 & description',
            {'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('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)
+        self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
+
+    def test_dfxp2srt(self):
+        dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
+            <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
+            <body>
+                <div xml:lang="en">
+                    <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" end="3"><span>Third<br/>Line</span></p>
+                </div>
+            </body>
+            </tt>'''
+        srt_data = '''1
+00:00:00,000 --> 00:00:01,000
+The following line contains Chinese characters and special symbols
+
+2
+00:00:01,000 --> 00:00:02,000
+第二行
+♪♪
+
+3
+00:00:02,000 --> 00:00:03,000
+Third
+Line
+
+'''
+        self.assertEqual(dfxp2srt(dfxp_data), srt_data)
+

 if __name__ == '__main__':
    unittest.main()
--- a/youtube_dl/postprocessor/ffmpeg.py
+++ b/youtube_dl/postprocessor/ffmpeg.py
@ -20,6 +20,7 @@ from ..utils import (
    prepend_extension,
    shell_quote,
    subtitles_filename,
+    dfxp2srt,
 )


@ -651,6 +652,30 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                    'format' % new_ext)
                continue
            new_file = subtitles_filename(filename, lang, new_ext)
+
+            if ext == 'dfxp' or ext == 'ttml':
+                self._downloader.report_warning(
+                    'You have requested to convert dfxp (TTML) subtitles into another format, '
+                    'which results in style information loss')
+
+                dfxp_file = subtitles_filename(filename, lang, ext)
+                srt_file = subtitles_filename(filename, lang, 'srt')
+
+                with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
+                    srt_data = dfxp2srt(f.read())
+
+                with io.open(srt_file, 'wt', encoding='utf-8') as f:
+                    f.write(srt_data)
+
+                ext = 'srt'
+                subs[lang] = {
+                    'ext': 'srt',
+                    'data': srt_data
+                }
+
+                if new_ext == 'srt':
+                    continue
+
            self.run_ffmpeg(
                subtitles_filename(filename, lang, ext),
                new_file, ['-f', new_format])
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@ -1800,6 +1800,59 @@ def match_filter_func(filter_str):
    return _match_func


+def parse_dfxp_time_expr(time_expr):
+    if not time_expr:
+        return 0.0
+
+    mobj = re.match(r'^(?P<time_offset>\d+(?:\.\d+)?)s?$', time_expr)
+    if mobj:
+        return float(mobj.group('time_offset'))
+
+    mobj = re.match(r'^(\d+):(\d\d):(\d\d(?:\.\d+)?)$', time_expr)
+    if mobj:
+        return 3600 * int(mobj.group(1)) + 60 * int(mobj.group(2)) + float(mobj.group(3))
+
+
+def format_srt_time(seconds):
+    (mins, secs) = divmod(seconds, 60)
+    (hours, mins) = divmod(mins, 60)
+    millisecs = (secs - int(secs)) * 1000
+    secs = int(secs)
+    return '%02d:%02d:%02d,%03d' % (hours, mins, secs, millisecs)
+
+
+def dfxp2srt(dfxp_data):
+    _x = functools.partial(xpath_with_ns, ns_map={'ttml': 'http://www.w3.org/ns/ttml'})
+
+    def parse_node(node):
+        str_or_empty = functools.partial(str_or_none, default='')
+
+        out = str_or_empty(node.text)
+
+        for child in node:
+            if child.tag == _x('ttml:br'):
+                out += '\n' + str_or_empty(child.tail)
+            elif child.tag == _x('ttml:span'):
+                out += str_or_empty(parse_node(child))
+            else:
+                out += str_or_empty(xml.etree.ElementTree.tostring(child))
+
+        return out
+
+    dfxp = xml.etree.ElementTree.fromstring(dfxp_data.encode('utf-8'))
+    out = []
+    paras = dfxp.findall(_x('.//ttml:p'))
+
+    for para, index in zip(paras, itertools.count(1)):
+        out.append('%d\n%s --> %s\n%s\n\n' % (
+            index,
+            format_srt_time(parse_dfxp_time_expr(para.attrib.get('begin'))),
+            format_srt_time(parse_dfxp_time_expr(para.attrib.get('end'))),
+            parse_node(para)))
+
+    return ''.join(out)
+
+
 class PerRequestProxyHandler(compat_urllib_request.ProxyHandler):
    def __init__(self, proxies=None):
        # Set default handlers