[utils] Add support for DCSubtitle

test/test_utils.py

@@ -79,6 +79,7 @@ from youtube_dl.utils import (
     match_str,
     parse_dfxp_time_expr,
     dfxp2srt,
+    dc2srt,
     cli_option,
     cli_valueless_option,
     cli_bool_option,
@@ -970,6 +971,38 @@ The first line
 '''
         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
 
+    def test_dc2srt(self):
+        dc_data = '''<?xml version="1.0" encoding="UTF-8"?>
+            <DCSubtitle Version="1.0">
+                <SubtitleID>id</SubtitleID>
+                <MovieTitle>title</MovieTitle>
+                <ReelNumber>1</ReelNumber>
+                <Language>English</Language>
+                <Font Italic="no">
+                    <Subtitle SpotNumber="1" TimeIn="00:00:05:000" TimeOut="00:00:08:357" FadeUpTime="20" FadeDownTime="20">
+                        <Text Direction="horizontal" HAlign="center" HPosition="0.0" VAlign="bottom" VPosition="14.0">^_^</Text>
+                        <Text Direction="horizontal" HAlign="center" HPosition="0.0" VAlign="bottom" VPosition="6.0">second line</Text>
+                    </Subtitle>
+                    <Subtitle SpotNumber="2" TimeIn="00:00:08:357" TimeOut="00:00:09:000" FadeUpTime="20" FadeDownTime="20">
+                        <Text Direction="horizontal" HAlign="center" HPosition="0.0" VAlign="bottom" VPosition="6.0">single line</Text>
+                    </Subtitle>
+                </Font>
+            </DCSubtitle>'''
+        srt_data = '''1
+00:00:05,000 --> 00:00:08,356
+^_^
+second line
+
+
+2
+00:00:08,356 --> 00:00:09,000
+single line
+
+
+'''
+
+        self.assertEqual(dc2srt(dc_data), srt_data)
+
     def test_cli_option(self):
         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])

youtube_dl/postprocessor/ffmpeg.py

@@ -21,6 +21,7 @@ from ..utils import (
     shell_quote,
     subtitles_filename,
     dfxp2srt,
+    dc2srt,
     ISO639Utils,
 )
 
@@ -568,6 +569,30 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                     continue
                 else:
                     sub_filenames.append(srt_file)
+            # TODO: Distinguish between different xml-formats
+            elif ext == 'xml':
+                self._downloader.report_warning(
+                    'You have requested to convert DC (XML) subtitles into another format, '
+                    'which results in style information loss')
+
+                dc_file = old_file
+                srt_file = subtitles_filename(filename, lang, 'srt')
+
+                with io.open(dc_file, 'rt', encoding='utf-8') as f:
+                    srt_data = dc2srt(f.read())
+
+                with io.open(srt_file, 'wt', encoding='utf-8') as f:
+                    f.write(srt_data)
+                old_file = srt_file
+
+                subs[lang] = {
+                    'ext': 'srt',
+                    'data': srt_data,
+                }
+
+                if new_ext == 'srt':
+                    continue
+                sub_filenames.append(srt_file)
 
             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 

youtube_dl/utils.py

@@ -2432,6 +2432,32 @@ def dfxp2srt(dfxp_data):
 
     return ''.join(out)
 
+# See https://web.archive.org/web/20140924175755/http://www.dlp.com/downloads/pdf_dlp_cinema_CineCanvas_Rev_C.pdf
+def dc2srt(xml_data):
+    xml = compat_etree_fromstring(xml_data.encode('utf-8'))
+    out = []
+    subtitles = xml.find('Font').findall('Subtitle')
+
+    if not subtitles:
+        raise ValueError('Invalid DC/XML subtitle')
+
+    for subtitle, index in zip(subtitles, itertools.count(1)):
+        begin_time = parse_dfxp_time_expr(subtitle.attrib.get('TimeIn'))
+        end_time = parse_dfxp_time_expr(subtitle.attrib.get('TimeOut'))
+        if not begin_time or not end_time:
+            continue
+
+        text = ''
+        for line in subtitle.findall('Text'):
+            text += line.text + '\n'
+
+        out.append('%d\n%s --> %s\n%s\n\n' % (
+            index,
+            srt_subtitles_timecode(begin_time),
+            srt_subtitles_timecode(end_time),
+            text))
+
+    return ''.join(out)
 
 def cli_option(params, command_option, param):
     param = params.get(param)