[utils] Use bytes-like objects in dfxp2srt

This fixes handling of non-UTF8 TTML subtitles Closes #14191
7 years ago · 3869028ffb
parent 68d43a61b5
commit 3869028ffb
4 changed files with 41 additions and 11 deletions
--- a/6
+++ b/6
@ -1,3 +1,9 @@
+version <unreleased>
+
+Core
+* [utils] Fix handling raw TTML subtitles (#14191)
+
+
 version 2017.09.15

 Core
--- a/test/test_utils.py
+++ b/test/test_utils.py
@ -1064,7 +1064,7 @@ ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
                    <p begin="3" dur="-1">Ignored, three</p>
                </div>
            </body>
-            </tt>'''
+            </tt>'''.encode('utf-8')
        srt_data = '''1
 00:00:00,000 --> 00:00:01,000
 The following line contains Chinese characters and special symbols
@ -1089,7 +1089,7 @@ Line
                    <p begin="0" end="1">The first line</p>
                </div>
            </body>
-            </tt>'''
+            </tt>'''.encode('utf-8')
        srt_data = '''1
 00:00:00,000 --> 00:00:01,000
 The first line
@ -1115,7 +1115,7 @@ The first line
      <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p>
    </div>
  </body>
-</tt>'''
+</tt>'''.encode('utf-8')
        srt_data = '''1
 00:00:02,080 --> 00:00:05,839
 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
@ -1138,6 +1138,26 @@ part 3</font></u>
 '''
        self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)

+        dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
+            <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">Line 1</p>
+                    <p begin="1" end="2">第二行</p>
+                </div>
+            </body>
+            </tt>'''.encode('utf-16')
+        srt_data = '''1
+00:00:00,000 --> 00:00:01,000
+Line 1
+
+2
+00:00:01,000 --> 00:00:02,000
+第二行
+
+'''
+        self.assertEqual(dfxp2srt(dfxp_data_non_utf8), 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'), [])
--- a/youtube_dl/postprocessor/ffmpeg.py
+++ b/youtube_dl/postprocessor/ffmpeg.py
@ -585,7 +585,7 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                dfxp_file = old_file
                srt_file = subtitles_filename(filename, lang, 'srt')

-                with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
+                with open(dfxp_file, 'rb') as f:
                    srt_data = dfxp2srt(f.read())

                with io.open(srt_file, 'wt', encoding='utf-8') as f:
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@ -2572,14 +2572,18 @@ def srt_subtitles_timecode(seconds):


 def dfxp2srt(dfxp_data):
+    '''
+    @param dfxp_data A bytes-like object containing DFXP data
+    @returns A unicode object containing converted SRT data
+    '''
    LEGACY_NAMESPACES = (
-        ('http://www.w3.org/ns/ttml', [
-            'http://www.w3.org/2004/11/ttaf1',
-            'http://www.w3.org/2006/04/ttaf1',
-            'http://www.w3.org/2006/10/ttaf1',
+        (b'http://www.w3.org/ns/ttml', [
+            b'http://www.w3.org/2004/11/ttaf1',
+            b'http://www.w3.org/2006/04/ttaf1',
+            b'http://www.w3.org/2006/10/ttaf1',
        ]),
-        ('http://www.w3.org/ns/ttml#styling', [
-            'http://www.w3.org/ns/ttml#style',
+        (b'http://www.w3.org/ns/ttml#styling', [
+            b'http://www.w3.org/ns/ttml#style',
        ]),
    )

@ -2674,7 +2678,7 @@ def dfxp2srt(dfxp_data):
        for ns in v:
            dfxp_data = dfxp_data.replace(ns, k)

-    dfxp = compat_etree_fromstring(dfxp_data.encode('utf-8'))
+    dfxp = compat_etree_fromstring(dfxp_data)
    out = []
    paras = dfxp.findall(_x('.//ttml:p')) or dfxp.findall('.//p')