Allow merging formats (closes #1612)
[youtube-dl] / youtube_dl / YoutubeDL.py
index fd26f00b721c05f8889ad07ad9e119451ab96251..8731279c114923a3ab91173ee2f75d0734a8244a 100644 (file)
@@ -51,9 +51,11 @@ from .utils import (
     write_json_file,
     write_string,
     YoutubeDLHandler,
+    prepend_extension,
 )
 from .extractor import get_info_extractor, gen_extractors
-from .FileDownloader import FileDownloader
+from .downloader import get_suitable_downloader
+from .PostProcessor import FFmpegMergerPP
 from .version import __version__
 
 
@@ -148,6 +150,7 @@ class YoutubeDL(object):
     socket_timeout:    Time to wait for unresponsive hosts, in seconds
     bidi_workaround:   Work around buggy terminals without bidirectional text
                        support, using fridibi
+    debug_printtraffic:Print out sent and received HTTP traffic
 
     The following parameters are not used by YoutubeDL itself, they are used by
     the FileDownloader:
@@ -164,6 +167,8 @@ class YoutubeDL(object):
 
     def __init__(self, params=None):
         """Create a FileDownloader object with the given options."""
+        if params is None:
+            params = {}
         self._ies = []
         self._ies_instances = {}
         self._pps = []
@@ -172,7 +177,7 @@ class YoutubeDL(object):
         self._num_downloads = 0
         self._screen_file = [sys.stdout, sys.stderr][params.get('logtostderr', False)]
         self._err_file = sys.stderr
-        self.params = {} if params is None else params
+        self.params = params
 
         if params.get('bidi_workaround', False):
             try:
@@ -183,12 +188,18 @@ class YoutubeDL(object):
                     width_args = []
                 else:
                     width_args = ['-w', str(width)]
-                self._fribidi = subprocess.Popen(
-                    ['fribidi', '-c', 'UTF-8'] + width_args,
+                sp_kwargs = dict(
                     stdin=subprocess.PIPE,
                     stdout=slave,
                     stderr=self._err_file)
-                self._fribidi_channel = os.fdopen(master, 'rb')
+                try:
+                    self._output_process = subprocess.Popen(
+                        ['bidiv'] + width_args, **sp_kwargs
+                    )
+                except OSError:
+                    self._output_process = subprocess.Popen(
+                        ['fribidi', '-c', 'UTF-8'] + width_args, **sp_kwargs)
+                self._output_channel = os.fdopen(master, 'rb')
             except OSError as ose:
                 if ose.errno == 2:
                     self.report_warning(u'Could not find fribidi executable, ignoring --bidi-workaround . Make sure that  fribidi  is an executable file in one of the directories in your $PATH.')
@@ -205,8 +216,6 @@ class YoutubeDL(object):
                 u'Set the LC_ALL environment variable to fix this.')
             self.params['restrictfilenames'] = True
 
-        self.fd = FileDownloader(self, self.params)
-
         if '%(stitle)s' in self.params.get('outtmpl', ''):
             self.report_warning(u'%(stitle)s is deprecated. Use the %(title)s and the --restrict-filenames flag(which also secures %(uploader)s et al) instead.')
 
@@ -242,15 +251,20 @@ class YoutubeDL(object):
         self._pps.append(pp)
         pp.set_downloader(self)
 
+    def add_progress_hook(self, ph):
+        """Add the progress hook (currently only for the file downloader)"""
+        self._progress_hooks.append(ph)
+
     def _bidi_workaround(self, message):
-        if not hasattr(self, '_fribidi_channel'):
+        if not hasattr(self, '_output_channel'):
             return message
 
+        assert hasattr(self, '_output_process')
         assert type(message) == type(u'')
         line_count = message.count(u'\n') + 1
-        self._fribidi.stdin.write((message + u'\n').encode('utf-8'))
-        self._fribidi.stdin.flush()
-        res = u''.join(self._fribidi_channel.readline().decode('utf-8')
+        self._output_process.stdin.write((message + u'\n').encode('utf-8'))
+        self._output_process.stdin.flush()
+        res = u''.join(self._output_channel.readline().decode('utf-8')
                        for _ in range(line_count))
         return res[:-len(u'\n')]
 
@@ -533,7 +547,7 @@ class YoutubeDL(object):
             def make_result(embedded_info):
                 new_result = ie_result.copy()
                 for f in ('_type', 'url', 'ext', 'player_url', 'formats',
-                          'entries', 'urlhandle', 'ie_key', 'duration',
+                          'entries', 'ie_key', 'duration',
                           'subtitles', 'annotations', 'format',
                           'thumbnail', 'thumbnails'):
                     if f in new_result:
@@ -667,17 +681,15 @@ class YoutubeDL(object):
             formats = list(takewhile_inclusive(
                 lambda f: f['format_id'] != format_limit, formats
             ))
-        if self.params.get('prefer_free_formats'):
-            def _free_formats_key(f):
-                try:
-                    ext_ord = [u'flv', u'mp4', u'webm'].index(f['ext'])
-                except ValueError:
-                    ext_ord = -1
-                # We only compare the extension if they have the same height and width
-                return (f.get('height'), f.get('width'), ext_ord)
-            formats = sorted(formats, key=_free_formats_key)
-
-        info_dict['formats'] = formats
+
+        # TODO Central sorting goes here
+
+        if formats[0] is not info_dict: 
+            # only set the 'formats' fields if the original info_dict list them
+            # otherwise we end up with a circular reference, the first (and unique)
+            # element in the 'formats' field in info_dict is info_dict itself, 
+            # wich can't be exported to json
+            info_dict['formats'] = formats
         if self.params.get('listformats', None):
             self.list_formats(info_dict)
             return
@@ -694,7 +706,17 @@ class YoutubeDL(object):
             # the first that is available, starting from left
             req_formats = req_format.split('/')
             for rf in req_formats:
-                selected_format = self.select_format(rf, formats)
+                if re.match(r'.+?\+.+?', rf) is not None:
+                    # Two formats have been requested like '137+139'
+                    format_1, format_2 = rf.split('+')
+                    formats_info = (self.select_format(format_1, formats),
+                        self.select_format(format_2, formats))
+                    if all(formats_info):
+                        selected_format = {'requested_formats': formats_info}
+                    else:
+                        selected_format = None
+                else:
+                    selected_format = self.select_format(rf, formats)
                 if selected_format is not None:
                     formats_to_download = [selected_format]
                     break
@@ -840,8 +862,7 @@ class YoutubeDL(object):
             else:
                 self.to_screen(u'[info] Writing video description metadata as JSON to: ' + infofn)
                 try:
-                    json_info_dict = dict((k, v) for k, v in info_dict.items() if not k in ['urlhandle'])
-                    write_json_file(json_info_dict, encodeFilename(infofn))
+                    write_json_file(info_dict, encodeFilename(infofn))
                 except (OSError, IOError):
                     self.report_error(u'Cannot write metadata to JSON file ' + infofn)
                     return
@@ -871,7 +892,27 @@ class YoutubeDL(object):
                 success = True
             else:
                 try:
-                    success = self.fd._do_download(filename, info_dict)
+                    def dl(name, info):
+                        fd = get_suitable_downloader(info)(self, self.params)
+                        for ph in self._progress_hooks:
+                            fd.add_progress_hook(ph)
+                        return fd.download(name, info)
+                    if info_dict.get('requested_formats') is not None:
+                        downloaded = []
+                        success = True
+                        for f in info_dict['requested_formats']:
+                            new_info = dict(info_dict)
+                            new_info.update(f)
+                            fname = self.prepare_filename(new_info)
+                            fname = prepend_extension(fname, 'f%s' % f['format_id'])
+                            downloaded.append(fname)
+                            partial_success = dl(fname, new_info)
+                            success = success and partial_success
+                        info_dict['__postprocessors'] = [FFmpegMergerPP(self)]
+                        info_dict['__files_to_merge'] = downloaded
+                    else:
+                        # Just a single file
+                        success = dl(filename, info_dict)
                 except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err:
                     self.report_error(u'unable to download video data: %s' % str(err))
                     return
@@ -928,7 +969,11 @@ class YoutubeDL(object):
         info = dict(ie_info)
         info['filepath'] = filename
         keep_video = None
-        for pp in self._pps:
+        pps_chain = []
+        if ie_info.get('__postprocessors') is not None:
+            pps_chain.extend(ie_info['__postprocessors'])
+        pps_chain.extend(self._pps)
+        for pp in pps_chain:
             try:
                 keep_video_wish, new_info = pp.run(info)
                 if keep_video_wish is not None:
@@ -989,13 +1034,15 @@ class YoutubeDL(object):
     def format_resolution(format, default='unknown'):
         if format.get('vcodec') == 'none':
             return 'audio only'
-        if format.get('_resolution') is not None:
-            return format['_resolution']
+        if format.get('resolution') is not None:
+            return format['resolution']
         if format.get('height') is not None:
             if format.get('width') is not None:
                 res = u'%sx%s' % (format['width'], format['height'])
             else:
                 res = u'%sp' % format['height']
+        elif format.get('width') is not None:
+            res = u'?x%d' % format['width']
         else:
             res = default
         return res
@@ -1003,15 +1050,19 @@ class YoutubeDL(object):
     def list_formats(self, info_dict):
         def format_note(fdict):
             res = u''
+            if fdict.get('ext') in ['f4f', 'f4m']:
+                res += u'(unsupported) '
             if fdict.get('format_note') is not None:
                 res += fdict['format_note'] + u' '
+            if fdict.get('tbr') is not None:
+                res += u'%4dk ' % fdict['tbr']
             if (fdict.get('vcodec') is not None and
                     fdict.get('vcodec') != 'none'):
-                res += u'%-5s' % fdict['vcodec']
-            elif fdict.get('vbr') is not None:
-                res += u'video'
+                res += u'%-5s@' % fdict['vcodec']
+            elif fdict.get('vbr') is not None and fdict.get('abr') is not None:
+                res += u'video@'
             if fdict.get('vbr') is not None:
-                res += u'@%4dk' % fdict['vbr']
+                res += u'%4dk' % fdict['vbr']
             if fdict.get('acodec') is not None:
                 if res:
                     res += u', '
@@ -1046,7 +1097,7 @@ class YoutubeDL(object):
 
         header_line = line({
             'format_id': u'format code', 'ext': u'extension',
-            '_resolution': u'resolution', 'format_note': u'note'}, idlen=idlen)
+            'resolution': u'resolution', 'format_note': u'note'}, idlen=idlen)
         self.to_screen(u'[info] Available formats for %s:\n%s\n%s' %
                        (info_dict['id'], header_line, u"\n".join(formats_s)))
 
@@ -1109,10 +1160,13 @@ class YoutubeDL(object):
             if 'http' in proxies and 'https' not in proxies:
                 proxies['https'] = proxies['http']
         proxy_handler = compat_urllib_request.ProxyHandler(proxies)
+
+        debuglevel = 1 if self.params.get('debug_printtraffic') else 0
         https_handler = make_HTTPS_handler(
-            self.params.get('nocheckcertificate', False))
+            self.params.get('nocheckcertificate', False), debuglevel=debuglevel)
+        ydlh = YoutubeDLHandler(debuglevel=debuglevel)
         opener = compat_urllib_request.build_opener(
-            https_handler, proxy_handler, cookie_processor, YoutubeDLHandler())
+            https_handler, proxy_handler, cookie_processor, ydlh)
         # Delete the default user-agent header, which would otherwise apply in
         # cases where our custom HTTP handler doesn't come into play
         # (See https://github.com/rg3/youtube-dl/issues/1309 for details)