Merge branch 'master' into subtitles_rework
[youtube-dl] / youtube_dl / PostProcessor.py
index f2e2aa1fadf5e8c8c1962ef11851820802550ce0..fddf58606015b92cc21a9f89818c90852c365e83 100644 (file)
-#!/usr/bin/env python
-# -*- coding: utf-8 -*-
-
 import os
 import subprocess
 import sys
 import time
 
-from utils import *
+from .utils import *
 
 
 class PostProcessor(object):
-       """Post Processor class.
-
-       PostProcessor objects can be added to downloaders with their
-       add_post_processor() method. When the downloader has finished a
-       successful download, it will take its internal chain of PostProcessors
-       and start calling the run() method on each one of them, first with
-       an initial argument and then with the returned value of the previous
-       PostProcessor.
-
-       The chain will be stopped if one of them ever returns None or the end
-       of the chain is reached.
-
-       PostProcessor objects follow a "mutual registration" process similar
-       to InfoExtractor objects.
-       """
-
-       _downloader = None
-
-       def __init__(self, downloader=None):
-               self._downloader = downloader
-
-       def set_downloader(self, downloader):
-               """Sets the downloader for this PP."""
-               self._downloader = downloader
-
-       def run(self, information):
-               """Run the PostProcessor.
-
-               The "information" argument is a dictionary like the ones
-               composed by InfoExtractors. The only difference is that this
-               one has an extra field called "filepath" that points to the
-               downloaded file.
-
-               When this method returns None, the postprocessing chain is
-               stopped. However, this method may return an information
-               dictionary that will be passed to the next postprocessing
-               object in the chain. It can be the one it received after
-               changing some fields.
-
-               In addition, this method may raise a PostProcessingError
-               exception that will be taken into account by the downloader
-               it was called from.
-               """
-               return information # by default, do nothing
-
-class AudioConversionError(BaseException):
-       def __init__(self, message):
-               self.message = message
-
-class FFmpegExtractAudioPP(PostProcessor):
-       def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, keepvideo=False):
-               PostProcessor.__init__(self, downloader)
-               if preferredcodec is None:
-                       preferredcodec = 'best'
-               self._preferredcodec = preferredcodec
-               self._preferredquality = preferredquality
-               self._keepvideo = keepvideo
-               self._exes = self.detect_executables()
-
-       @staticmethod
-       def detect_executables():
-               def executable(exe):
-                       try:
-                               subprocess.check_output([exe, '-version'])
-                       except OSError:
-                               return False
-                       return exe
-               programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
-               return dict((program, executable(program)) for program in programs)
-
-       def get_audio_codec(self, path):
-               if not self._exes['ffprobe'] and not self._exes['avprobe']: return None
-               try:
-                       cmd = [self._exes['avprobe'] or self._exes['ffprobe'], '-show_streams', '--', encodeFilename(path)]
-                       handle = subprocess.Popen(cmd, stderr=file(os.path.devnull, 'w'), stdout=subprocess.PIPE)
-                       output = handle.communicate()[0]
-                       if handle.wait() != 0:
-                               return None
-               except (IOError, OSError):
-                       return None
-               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
-               return None
-
-       def run_ffmpeg(self, path, out_path, codec, more_opts):
-               if not self._exes['ffmpeg'] and not self._exes['avconv']:
-                       raise AudioConversionError('ffmpeg or avconv not found. Please install one.')   
-               if codec is None:
-                       acodec_opts = []
-               else:
-                       acodec_opts = ['-acodec', codec]
-               cmd = ([self._exes['avconv'] or self._exes['ffmpeg'], '-y', '-i', encodeFilename(path), '-vn']
-                          + acodec_opts + more_opts +
-                          ['--', encodeFilename(out_path)])
-               p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-               stdout,stderr = p.communicate()
-               if p.returncode != 0:
-                       msg = stderr.strip().split('\n')[-1]
-                       raise AudioConversionError(msg)
-
-       def run(self, information):
-               path = information['filepath']
-
-               filecodec = self.get_audio_codec(path)
-               if filecodec is None:
-                       self._downloader.to_stderr(u'WARNING: unable to obtain file audio codec with ffprobe')
-                       return None
-
-               more_opts = []
-               if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
-                       if self._preferredcodec == 'm4a' and filecodec == 'aac':
-                               # Lossless, but in another container
-                               acodec = 'copy'
-                               extension = self._preferredcodec
-                               more_opts = [self._exes['avconv'] and '-bsf:a' or '-absf', 'aac_adtstoasc']
-                       elif filecodec in ['aac', 'mp3', 'vorbis']:
-                               # Lossless if possible
-                               acodec = 'copy'
-                               extension = filecodec
-                               if filecodec == 'aac':
-                                       more_opts = ['-f', 'adts']
-                               if filecodec == 'vorbis':
-                                       extension = 'ogg'
-                       else:
-                               # MP3 otherwise.
-                               acodec = 'libmp3lame'
-                               extension = 'mp3'
-                               more_opts = []
-                               if self._preferredquality is not None:
-                                       if int(self._preferredquality) < 10:
-                                               more_opts += [self._exes['avconv'] and '-q:a' or '-aq', self._preferredquality]
-                                       else:
-                                               more_opts += [self._exes['avconv'] and '-b:a' or '-ab', self._preferredquality]
-               else:
-                       # We convert the audio (lossy)
-                       acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
-                       extension = self._preferredcodec
-                       more_opts = []
-                       if self._preferredquality is not None:
-                               if int(self._preferredquality) < 10:
-                                       more_opts += [self._exes['avconv'] and '-q:a' or '-aq', self._preferredquality]
-                               else:
-                                       more_opts += [self._exes['avconv'] and '-b:a' or '-ab', self._preferredquality]
-                       if self._preferredcodec == 'aac':
-                               more_opts += ['-f', 'adts']
-                       if self._preferredcodec == 'm4a':
-                               more_opts += [self._exes['avconv'] and '-bsf:a' or '-absf', 'aac_adtstoasc']
-                       if self._preferredcodec == 'vorbis':
-                               extension = 'ogg'
-                       if self._preferredcodec == 'wav':
-                               extension = 'wav'
-                               more_opts += ['-f', 'wav']
-
-               prefix, sep, ext = path.rpartition(u'.') # not os.path.splitext, since the latter does not work on unicode in all setups
-               new_path = prefix + sep + extension
-               self._downloader.to_screen(u'[' + (self._exes['avconv'] and 'avconv' or 'ffmpeg') + '] Destination: ' + new_path)
-               try:
-                       self.run_ffmpeg(path, new_path, acodec, more_opts)
-               except:
-                       etype,e,tb = sys.exc_info()
-                       if isinstance(e, AudioConversionError):
-                               self._downloader.to_stderr(u'ERROR: audio conversion failed: ' + e.message)
-                       else:
-                               self._downloader.to_stderr(u'ERROR: error running ' + (self._exes['avconv'] and 'avconv' or 'ffmpeg'))
-                       return None
-
-               # Try to update the date time for extracted audio file.
-               if information.get('filetime') is not None:
-                       try:
-                               os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
-                       except:
-                               self._downloader.to_stderr(u'WARNING: Cannot update utime of audio file')
-
-               if not self._keepvideo:
-                       try:
-                               os.remove(encodeFilename(path))
-                       except (IOError, OSError):
-                               self._downloader.to_stderr(u'WARNING: Unable to remove downloaded video file')
-                               return None
-
-               information['filepath'] = new_path
-               return information
+    """Post Processor class.
+
+    PostProcessor objects can be added to downloaders with their
+    add_post_processor() method. When the downloader has finished a
+    successful download, it will take its internal chain of PostProcessors
+    and start calling the run() method on each one of them, first with
+    an initial argument and then with the returned value of the previous
+    PostProcessor.
+
+    The chain will be stopped if one of them ever returns None or the end
+    of the chain is reached.
+
+    PostProcessor objects follow a "mutual registration" process similar
+    to InfoExtractor objects.
+    """
+
+    _downloader = None
+
+    def __init__(self, downloader=None):
+        self._downloader = downloader
+
+    def set_downloader(self, downloader):
+        """Sets the downloader for this PP."""
+        self._downloader = downloader
+
+    def run(self, information):
+        """Run the PostProcessor.
+
+        The "information" argument is a dictionary like the ones
+        composed by InfoExtractors. The only difference is that this
+        one has an extra field called "filepath" that points to the
+        downloaded file.
+
+        This method returns a tuple, the first element of which describes
+        whether the original file should be kept (i.e. not deleted - None for
+        no preference), and the second of which is the updated information.
+
+        In addition, this method may raise a PostProcessingError
+        exception if post processing fails.
+        """
+        return None, information # by default, keep file and do nothing
+
+class FFmpegPostProcessorError(PostProcessingError):
+    pass
+
+class AudioConversionError(PostProcessingError):
+    pass
+
+class FFmpegPostProcessor(PostProcessor):
+    def __init__(self,downloader=None):
+        PostProcessor.__init__(self, downloader)
+        self._exes = self.detect_executables()
+
+    @staticmethod
+    def detect_executables():
+        def executable(exe):
+            try:
+                subprocess.Popen([exe, '-version'], stdout=subprocess.PIPE, stderr=subprocess.PIPE).communicate()
+            except OSError:
+                return False
+            return exe
+        programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
+        return dict((program, executable(program)) for program in programs)
+
+    def run_ffmpeg(self, path, out_path, opts):
+        if not self._exes['ffmpeg'] and not self._exes['avconv']:
+            raise FFmpegPostProcessorError(u'ffmpeg or avconv not found. Please install one.')
+        cmd = ([self._exes['avconv'] or self._exes['ffmpeg'], '-y', '-i', encodeFilename(path)]
+               + opts +
+               [encodeFilename(self._ffmpeg_filename_argument(out_path))])
+        p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+        stdout,stderr = p.communicate()
+        if p.returncode != 0:
+            stderr = stderr.decode('utf-8', 'replace')
+            msg = stderr.strip().split('\n')[-1]
+            raise FFmpegPostProcessorError(msg)
+
+    def _ffmpeg_filename_argument(self, fn):
+        # ffmpeg broke --, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details
+        if fn.startswith(u'-'):
+            return u'./' + fn
+        return fn
+
+class FFmpegExtractAudioPP(FFmpegPostProcessor):
+    def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
+        FFmpegPostProcessor.__init__(self, downloader)
+        if preferredcodec is None:
+            preferredcodec = 'best'
+        self._preferredcodec = preferredcodec
+        self._preferredquality = preferredquality
+        self._nopostoverwrites = nopostoverwrites
+
+    def get_audio_codec(self, path):
+        if not self._exes['ffprobe'] and not self._exes['avprobe']:
+            raise PostProcessingError(u'ffprobe or avprobe not found. Please install one.')
+        try:
+            cmd = [self._exes['avprobe'] or self._exes['ffprobe'], '-show_streams', encodeFilename(self._ffmpeg_filename_argument(path))]
+            handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE)
+            output = handle.communicate()[0]
+            if handle.wait() != 0:
+                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
+        return None
+
+    def run_ffmpeg(self, path, out_path, codec, more_opts):
+        if not self._exes['ffmpeg'] and not self._exes['avconv']:
+            raise AudioConversionError('ffmpeg or avconv not found. Please install one.')
+        if codec is None:
+            acodec_opts = []
+        else:
+            acodec_opts = ['-acodec', codec]
+        opts = ['-vn'] + acodec_opts + more_opts
+        try:
+            FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
+        except FFmpegPostProcessorError as err:
+            raise AudioConversionError(err.message)
+
+    def run(self, information):
+        path = information['filepath']
+
+        filecodec = self.get_audio_codec(path)
+        if filecodec is None:
+            raise PostProcessingError(u'WARNING: unable to obtain file audio codec with ffprobe')
+
+        more_opts = []
+        if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
+            if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
+                # Lossless, but in another container
+                acodec = 'copy'
+                extension = 'm4a'
+                more_opts = [self._exes['avconv'] and '-bsf:a' or '-absf', 'aac_adtstoasc']
+            elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
+                # Lossless if possible
+                acodec = 'copy'
+                extension = filecodec
+                if filecodec == 'aac':
+                    more_opts = ['-f', 'adts']
+                if filecodec == 'vorbis':
+                    extension = 'ogg'
+            else:
+                # MP3 otherwise.
+                acodec = 'libmp3lame'
+                extension = 'mp3'
+                more_opts = []
+                if self._preferredquality is not None:
+                    if int(self._preferredquality) < 10:
+                        more_opts += [self._exes['avconv'] and '-q:a' or '-aq', self._preferredquality]
+                    else:
+                        more_opts += [self._exes['avconv'] and '-b:a' or '-ab', self._preferredquality + 'k']
+        else:
+            # We convert the audio (lossy)
+            acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
+            extension = self._preferredcodec
+            more_opts = []
+            if self._preferredquality is not None:
+                if int(self._preferredquality) < 10:
+                    more_opts += [self._exes['avconv'] and '-q:a' or '-aq', self._preferredquality]
+                else:
+                    more_opts += [self._exes['avconv'] and '-b:a' or '-ab', self._preferredquality + 'k']
+            if self._preferredcodec == 'aac':
+                more_opts += ['-f', 'adts']
+            if self._preferredcodec == 'm4a':
+                more_opts += [self._exes['avconv'] and '-bsf:a' or '-absf', 'aac_adtstoasc']
+            if self._preferredcodec == 'vorbis':
+                extension = 'ogg'
+            if self._preferredcodec == 'wav':
+                extension = 'wav'
+                more_opts += ['-f', 'wav']
+
+        prefix, sep, ext = path.rpartition(u'.') # not os.path.splitext, since the latter does not work on unicode in all setups
+        new_path = prefix + sep + extension
+
+        # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
+        if new_path == path:
+            self._nopostoverwrites = True
+
+        try:
+            if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
+                self._downloader.to_screen(u'[youtube] Post-process file %s exists, skipping' % new_path)
+            else:
+                self._downloader.to_screen(u'[' + (self._exes['avconv'] and 'avconv' or 'ffmpeg') + '] Destination: ' + new_path)
+                self.run_ffmpeg(path, new_path, acodec, more_opts)
+        except:
+            etype,e,tb = sys.exc_info()
+            if isinstance(e, AudioConversionError):
+                msg = u'audio conversion failed: ' + e.message
+            else:
+                msg = u'error running ' + (self._exes['avconv'] and 'avconv' or 'ffmpeg')
+            raise PostProcessingError(msg)
+
+        # Try to update the date time for extracted audio file.
+        if information.get('filetime') is not None:
+            try:
+                os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
+            except:
+                self._downloader.report_warning(u'Cannot update utime of audio file')
+
+        information['filepath'] = new_path
+        return self._nopostoverwrites,information
+
+class FFmpegVideoConvertor(FFmpegPostProcessor):
+    def __init__(self, downloader=None,preferedformat=None):
+        super(FFmpegVideoConvertor, self).__init__(downloader)
+        self._preferedformat=preferedformat
+
+    def run(self, information):
+        path = information['filepath']
+        prefix, sep, ext = path.rpartition(u'.')
+        outpath = prefix + sep + self._preferedformat
+        if information['ext'] == self._preferedformat:
+            self._downloader.to_screen(u'[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
+            return True,information
+        self._downloader.to_screen(u'['+'ffmpeg'+'] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) +outpath)
+        self.run_ffmpeg(path, outpath, [])
+        information['filepath'] = outpath
+        information['format'] = self._preferedformat
+        information['ext'] = self._preferedformat
+        return False,information