Prepare to upload

[youtubedl] / youtube_dl / postprocessor / ffmpeg.py

diff --git a/youtube_dl/postprocessor/ffmpeg.py b/youtube_dl/postprocessor/ffmpeg.py
index 96ddb3b36f2b9219af7edfe209b09ca3f34ae546..70416c25ee1a3fb333c63ff7334f87250122fb07 100644 (file)
--- a/youtube_dl/postprocessor/ffmpeg.py
+++ b/youtube_dl/postprocessor/ffmpeg.py
@@ -4,13 +4,11 @@ import io
 import os
 import subprocess
 import time
 import os
 import subprocess
 import time

+import re

 
 
 from .common import AudioConversionError, PostProcessor
 
 
 
 from .common import AudioConversionError, PostProcessor
 

-from ..compat import (
-    compat_subprocess_get_DEVNULL,
-)

 from ..utils import (
     encodeArgument,
     encodeFilename,
 from ..utils import (
     encodeArgument,
     encodeFilename,


@@ -22,19 +20,30 @@ from ..utils import (
     subtitles_filename,
     dfxp2srt,
     ISO639Utils,
     subtitles_filename,
     dfxp2srt,
     ISO639Utils,

+    replace_extension,

 )
 
 
 EXT_TO_OUT_FORMATS = {
 )
 
 
 EXT_TO_OUT_FORMATS = {

-    "aac": "adts",
-    "m4a": "ipod",
-    "mka": "matroska",
-    "mkv": "matroska",
-    "mpg": "mpeg",
-    "ogv": "ogg",
-    "ts": "mpegts",
-    "wma": "asf",
-    "wmv": "asf",
+    'aac': 'adts',
+    'flac': 'flac',
+    'm4a': 'ipod',
+    'mka': 'matroska',
+    'mkv': 'matroska',
+    'mpg': 'mpeg',
+    'ogv': 'ogg',
+    'ts': 'mpegts',
+    'wma': 'asf',
+    'wmv': 'asf',
+}
+ACODECS = {
+    'mp3': 'libmp3lame',
+    'aac': 'aac',
+    'flac': 'flac',
+    'm4a': 'aac',
+    'opus': 'libopus',
+    'vorbis': 'libvorbis',
+    'wav': None,

 }
 
 
 }
 
 


@@ -65,7 +74,21 @@ class FFmpegPostProcessor(PostProcessor):
 
     def _determine_executables(self):
         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
 
     def _determine_executables(self):
         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']

-        prefer_ffmpeg = False
+        prefer_ffmpeg = True
+
+        def get_ffmpeg_version(path):
+            ver = get_exe_version(path, args=['-version'])
+            if ver:
+                regexs = [
+                    r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$',  # Ubuntu, see [1]
+                    r'n([0-9.]+)$',  # Arch Linux
+                    # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
+                ]
+                for regex in regexs:
+                    mobj = re.match(regex, ver)
+                    if mobj:
+                        ver = mobj.group(1)
+            return ver

 
         self.basename = None
         self.probe_basename = None
 
         self.basename = None
         self.probe_basename = None


@@ -73,7 +96,7 @@ class FFmpegPostProcessor(PostProcessor):
         self._paths = None
         self._versions = None
         if self._downloader:
         self._paths = None
         self._versions = None
         if self._downloader:

-            prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', False)
+            prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', True)

             location = self._downloader.params.get('ffmpeg_location')
             if location is not None:
                 if not os.path.exists(location):
             location = self._downloader.params.get('ffmpeg_location')
             if location is not None:
                 if not os.path.exists(location):


@@ -98,26 +121,25 @@ class FFmpegPostProcessor(PostProcessor):
                 self._paths = dict(
                     (p, os.path.join(location, p)) for p in programs)
                 self._versions = dict(
                 self._paths = dict(
                     (p, os.path.join(location, p)) for p in programs)
                 self._versions = dict(

-                    (p, get_exe_version(self._paths[p], args=['-version']))
-                    for p in programs)
+                    (p, get_ffmpeg_version(self._paths[p])) for p in programs)

         if self._versions is None:
             self._versions = dict(
         if self._versions is None:
             self._versions = dict(

-                (p, get_exe_version(p, args=['-version'])) for p in programs)
+                (p, get_ffmpeg_version(p)) for p in programs)

             self._paths = dict((p, p) for p in programs)
 
             self._paths = dict((p, p) for p in programs)
 

-        if prefer_ffmpeg:
-            prefs = ('ffmpeg', 'avconv')
-        else:
+        if prefer_ffmpeg is False:

             prefs = ('avconv', 'ffmpeg')
             prefs = ('avconv', 'ffmpeg')

+        else:
+            prefs = ('ffmpeg', 'avconv')

         for p in prefs:
             if self._versions[p]:
                 self.basename = p
                 break
 
         for p in prefs:
             if self._versions[p]:
                 self.basename = p
                 break
 

-        if prefer_ffmpeg:
-            prefs = ('ffprobe', 'avprobe')
-        else:
+        if prefer_ffmpeg is False:

             prefs = ('avprobe', 'ffprobe')
             prefs = ('avprobe', 'ffprobe')

+        else:
+            prefs = ('ffprobe', 'avprobe')

         for p in prefs:
             if self._versions[p]:
                 self.probe_basename = p
         for p in prefs:
             if self._versions[p]:
                 self.probe_basename = p


@@ -140,27 +162,45 @@ class FFmpegPostProcessor(PostProcessor):
         return self._paths[self.probe_basename]
 
     def get_audio_codec(self, path):
         return self._paths[self.probe_basename]
 
     def get_audio_codec(self, path):

-        if not self.probe_available:
-            raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
+        if not self.probe_available and not self.available:
+            raise PostProcessingError('ffprobe/avprobe and ffmpeg/avconv not found. Please install one.')

         try:
         try:

-            cmd = [
-                encodeFilename(self.probe_executable, True),
-                encodeArgument('-show_streams'),
-                encodeFilename(self._ffmpeg_filename_argument(path), True)]
+            if self.probe_available:
+                cmd = [
+                    encodeFilename(self.probe_executable, True),
+                    encodeArgument('-show_streams')]
+            else:
+                cmd = [
+                    encodeFilename(self.executable, True),
+                    encodeArgument('-i')]
+            cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))

             if self._downloader.params.get('verbose', False):
             if self._downloader.params.get('verbose', False):

-                self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
-            handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
-            output = handle.communicate()[0]
-            if handle.wait() != 0:
+                self._downloader.to_screen(
+                    '[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
+            handle = subprocess.Popen(
+                cmd, stderr=subprocess.PIPE,
+                stdout=subprocess.PIPE, stdin=subprocess.PIPE)
+            stdout_data, stderr_data = handle.communicate()
+            expected_ret = 0 if self.probe_available else 1
+            if handle.wait() != expected_ret:

                 return None
         except (IOError, OSError):
             return None
                 return None
         except (IOError, OSError):
             return None

-        audio_codec = None
-        for line in output.decode('ascii', 'ignore').split('\n'):
-            if line.startswith('codec_name='):
-                audio_codec = line.split('=')[1].strip()
-            elif line.strip() == 'codec_type=audio' and audio_codec is not None:
-                return audio_codec
+        output = (stdout_data if self.probe_available else stderr_data).decode('ascii', 'ignore')
+        if self.probe_available:
+            audio_codec = None
+            for line in output.split('\n'):
+                if line.startswith('codec_name='):
+                    audio_codec = line.split('=')[1].strip()
+                elif line.strip() == 'codec_type=audio' and audio_codec is not None:
+                    return audio_codec
+        else:
+            # Stream #FILE_INDEX:STREAM_INDEX[STREAM_ID](LANGUAGE): CODEC_TYPE: CODEC_NAME
+            mobj = re.search(
+                r'Stream\s*#\d+:\d+(?:\[0x[0-9a-f]+\])?(?:\([a-z]{3}\))?:\s*Audio:\s*([0-9a-z]+)',
+                output)
+            if mobj:
+                return mobj.group(1)

         return None
 
     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
         return None
 
     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):


@@ -177,10 +217,13 @@ class FFmpegPostProcessor(PostProcessor):
                 encodeArgument('-i'),
                 encodeFilename(self._ffmpeg_filename_argument(path), True)
             ])
                 encodeArgument('-i'),
                 encodeFilename(self._ffmpeg_filename_argument(path), True)
             ])

-        cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
-               files_cmd +
-               [encodeArgument(o) for o in opts] +
-               [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
+        cmd = [encodeFilename(self.executable, True), encodeArgument('-y')]
+        # avconv does not have repeat option
+        if self.basename == 'ffmpeg':
+            cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
+        cmd += (files_cmd
+                + [encodeArgument(o) for o in opts]
+                + [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])

 
         if self._downloader.params.get('verbose', False):
             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 
         if self._downloader.params.get('verbose', False):
             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))


@@ -237,7 +280,7 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor):
                 acodec = 'copy'
                 extension = 'm4a'
                 more_opts = ['-bsf:a', 'aac_adtstoasc']
                 acodec = 'copy'
                 extension = 'm4a'
                 more_opts = ['-bsf:a', 'aac_adtstoasc']

-            elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
+            elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:

                 # Lossless if possible
                 acodec = 'copy'
                 extension = filecodec
                 # Lossless if possible
                 acodec = 'copy'
                 extension = filecodec


@@ -256,8 +299,8 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor):
                     else:
                         more_opts += ['-b:a', self._preferredquality + 'k']
         else:
                     else:
                         more_opts += ['-b:a', self._preferredquality + 'k']
         else:

-            # We convert the audio (lossy)
-            acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
+            # We convert the audio (lossy if codec is lossy)
+            acodec = ACODECS[self._preferredcodec]

             extension = self._preferredcodec
             more_opts = []
             if self._preferredquality is not None:
             extension = self._preferredcodec
             more_opts = []
             if self._preferredquality is not None:


@@ -283,8 +326,8 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor):
         information['ext'] = extension
 
         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
         information['ext'] = extension
 
         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.

-        if (new_path == path or
-                (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
+        if (new_path == path
+                or (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):

             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
             return [], information
 
             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
             return [], information
 


@@ -367,14 +410,16 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
             # Don't copy the existing subtitles, we may be running the
             # postprocessor a second time
             '-map', '-0:s',
             # Don't copy the existing subtitles, we may be running the
             # postprocessor a second time
             '-map', '-0:s',

+            # Don't copy Apple TV chapters track, bin_data (see #19042, #19024,
+            # https://trac.ffmpeg.org/ticket/6016)
+            '-map', '-0:d',

         ]
         if information['ext'] == 'mp4':
             opts += ['-c:s', 'mov_text']
         for (i, lang) in enumerate(sub_langs):
             opts.extend(['-map', '%d:0' % (i + 1)])
         ]
         if information['ext'] == 'mp4':
             opts += ['-c:s', 'mov_text']
         for (i, lang) in enumerate(sub_langs):
             opts.extend(['-map', '%d:0' % (i + 1)])

-            lang_code = ISO639Utils.short2long(lang)
-            if lang_code is not None:
-                opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
+            lang_code = ISO639Utils.short2long(lang) or lang
+            opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])

 
         temp_filename = prepend_extension(filename, 'temp')
         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 
         temp_filename = prepend_extension(filename, 'temp')
         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)


@@ -419,17 +464,40 @@ class FFmpegMetadataPP(FFmpegPostProcessor):
 
         filename = info['filepath']
         temp_filename = prepend_extension(filename, 'temp')
 
         filename = info['filepath']
         temp_filename = prepend_extension(filename, 'temp')

+        in_filenames = [filename]
+        options = []

 
         if info['ext'] == 'm4a':
 
         if info['ext'] == 'm4a':

-            options = ['-vn', '-acodec', 'copy']
+            options.extend(['-vn', '-acodec', 'copy'])

         else:
         else:

-            options = ['-c', 'copy']
+            options.extend(['-c', 'copy'])

 
         for (name, value) in metadata.items():
             options.extend(['-metadata', '%s=%s' % (name, value)])
 
 
         for (name, value) in metadata.items():
             options.extend(['-metadata', '%s=%s' % (name, value)])
 

+        chapters = info.get('chapters', [])
+        if chapters:
+            metadata_filename = replace_extension(filename, 'meta')
+            with io.open(metadata_filename, 'wt', encoding='utf-8') as f:
+                def ffmpeg_escape(text):
+                    return re.sub(r'(=|;|#|\\|\n)', r'\\\1', text)
+
+                metadata_file_content = ';FFMETADATA1\n'
+                for chapter in chapters:
+                    metadata_file_content += '[CHAPTER]\nTIMEBASE=1/1000\n'
+                    metadata_file_content += 'START=%d\n' % (chapter['start_time'] * 1000)
+                    metadata_file_content += 'END=%d\n' % (chapter['end_time'] * 1000)
+                    chapter_title = chapter.get('title')
+                    if chapter_title:
+                        metadata_file_content += 'title=%s\n' % ffmpeg_escape(chapter_title)
+                f.write(metadata_file_content)
+                in_filenames.append(metadata_filename)
+                options.extend(['-map_metadata', '1'])
+

         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)

-        self.run_ffmpeg(filename, temp_filename, options)
+        self.run_ffmpeg_multiple_files(in_filenames, temp_filename, options)
+        if chapters:
+            os.remove(metadata_filename)

         os.remove(encodeFilename(filename))
         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
         return [], info
         os.remove(encodeFilename(filename))
         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
         return [], info


@@ -507,7 +575,7 @@ class FFmpegFixupM3u8PP(FFmpegPostProcessor):
             temp_filename = prepend_extension(filename, 'temp')
 
             options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
             temp_filename = prepend_extension(filename, 'temp')
 
             options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']

-            self._downloader.to_screen('[ffmpeg] Fixing malformated aac bitstream in "%s"' % filename)
+            self._downloader.to_screen('[ffmpeg] Fixing malformed AAC bitstream in "%s"' % filename)

             self.run_ffmpeg(filename, temp_filename, options)
 
             os.remove(encodeFilename(filename))
             self.run_ffmpeg(filename, temp_filename, options)
 
             os.remove(encodeFilename(filename))


@@ -542,7 +610,7 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
             sub_filenames.append(old_file)
             new_file = subtitles_filename(filename, lang, new_ext)
 
             sub_filenames.append(old_file)
             new_file = subtitles_filename(filename, lang, new_ext)
 

-            if ext == 'dfxp' or ext == 'ttml' or ext == 'tt':
+            if ext in ('dfxp', 'ttml', 'tt'):

                 self._downloader.report_warning(
                     'You have requested to convert dfxp (TTML) subtitles into another format, '
                     'which results in style information loss')
                 self._downloader.report_warning(
                     'You have requested to convert dfxp (TTML) subtitles into another format, '
                     'which results in style information loss')


@@ -550,7 +618,7 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                 dfxp_file = old_file
                 srt_file = subtitles_filename(filename, lang, 'srt')
 
                 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:
                     srt_data = dfxp2srt(f.read())
 
                 with io.open(srt_file, 'wt', encoding='utf-8') as f: