[ffmpeg] Download and merge in a single step if possible

README.md

@@ -130,6 +130,7 @@ ### Differences in default behavior
 * Youtube live chat (if available) is considered as a subtitle. Use `--sub-langs all,-live_chat` to download all subtitles except live chat. You can also use `--compat-options no-live-chat` to prevent live chat from downloading
 * Youtube channel URLs are automatically redirected to `/video`. Append a `/featured` to the URL to download only the videos in the home page. If the channel does not have a videos tab, we try to download the equivalent `UU` playlist instead. Also, `/live` URLs raise an error if there are no live videos instead of silently downloading the entire channel. You may use `--compat-options no-youtube-channel-redirect` to revert all these redirections
 * Unavailable videos are also listed for youtube playlists. Use `--compat-options no-youtube-unavailable-videos` to remove this
+* If `ffmpeg` is used as the downloader, the downloading and merging of formats happen in a single step when possible. Use `--compat-options no-direct-merge` to revert this
 
 For ease of use, a few more compat options are available:
 * `--compat-options all`: Use all compat options

yt_dlp/YoutubeDL.py

@@ -387,8 +387,9 @@ class YoutubeDL(object):
                        if True, otherwise use ffmpeg/avconv if False, otherwise
                        use downloader suggested by extractor if None.
     compat_opts:       Compatibility options. See "Differences in default behavior".
-                       Note that only format-sort, format-spec, no-live-chat, no-attach-info-json
+                       Note that only format-sort, format-spec, no-live-chat,
-                       playlist-index, list-formats, no-youtube-channel-redirect
+                       no-attach-info-json, playlist-index, list-formats,
+                       no-direct-merge, no-youtube-channel-redirect,
                        and no-youtube-unavailable-videos works when used via the API
 
     The following parameters are not used by YoutubeDL itself, they are used by
@@ -2294,7 +2295,8 @@ def dl(self, name, info, subtitle=False, test=False):
         if not test:
             for ph in self._progress_hooks:
                 fd.add_progress_hook(ph)
-            self.write_debug('Invoking downloader on %r' % info.get('url'))
+            urls = '", "'.join([f['url'] for f in info.get('requested_formats', [])] or [info['url']])
+            self.write_debug('Invoking downloader on "%s"' % urls)
         new_info = dict(info)
         if new_info.get('http_headers') is None:
             new_info['http_headers'] = self._calc_headers(new_info)
@@ -2533,17 +2535,6 @@ def existing_file(*filepaths):
 
                 success = True
                 if info_dict.get('requested_formats') is not None:
-                    downloaded = []
-                    merger = FFmpegMergerPP(self)
-                    if self.params.get('allow_unplayable_formats'):
-                        self.report_warning(
-                            'You have requested merging of multiple formats '
-                            'while also allowing unplayable formats to be downloaded. '
-                            'The formats won\'t be merged to prevent data corruption.')
-                    elif not merger.available:
-                        self.report_warning(
-                            'You have requested merging of multiple formats but ffmpeg is not installed. '
-                            'The formats won\'t be merged.')
 
                     def compatible_formats(formats):
                         # TODO: some formats actually allow this (mkv, webm, ogg, mp4), but not all of them.
@@ -2591,27 +2582,57 @@ def correct_ext(filename):
                     temp_filename = correct_ext(temp_filename)
                     dl_filename = existing_file(full_filename, temp_filename)
                     info_dict['__real_download'] = False
-                    if dl_filename is None:
+
-                        for f in requested_formats:
+                    _protocols = set(determine_protocol(f) for f in requested_formats)
-                            new_info = dict(info_dict)
+                    if len(_protocols) == 1:
-                            new_info.update(f)
+                        info_dict['protocol'] = _protocols.pop()
-                            fname = prepend_extension(
+                    directly_mergable = (
-                                self.prepare_filename(new_info, 'temp'),
+                        'no-direct-merge' not in self.params.get('compat_opts', [])
-                                'f%s' % f['format_id'], new_info['ext'])
+                        and info_dict.get('protocol') is not None  # All requested formats have same protocol
-                            if not self._ensure_dir_exists(fname):
+                        and not self.params.get('allow_unplayable_formats')
-                                return
+                        and get_suitable_downloader(info_dict, self.params).__name__ == 'FFmpegFD')
-                            downloaded.append(fname)
+                    if directly_mergable:
-                            partial_success, real_download = self.dl(fname, new_info)
+                        info_dict['url'] = requested_formats[0]['url']
-                            info_dict['__real_download'] = info_dict['__real_download'] or real_download
+                        # Treat it as a single download
-                            success = success and partial_success
+                        dl_filename = existing_file(full_filename, temp_filename)
-                        if merger.available and not self.params.get('allow_unplayable_formats'):
+                        if dl_filename is None:
-                            info_dict['__postprocessors'].append(merger)
+                            success, real_download = self.dl(temp_filename, info_dict)
-                            info_dict['__files_to_merge'] = downloaded
+                            info_dict['__real_download'] = real_download
-                            # Even if there were no downloads, it is being merged only now
+                    else:
-                            info_dict['__real_download'] = True
+                        downloaded = []
-                        else:
+                        merger = FFmpegMergerPP(self)
-                            for file in downloaded:
+                        if self.params.get('allow_unplayable_formats'):
-                                files_to_move[file] = None
+                            self.report_warning(
+                                'You have requested merging of multiple formats '
+                                'while also allowing unplayable formats to be downloaded. '
+                                'The formats won\'t be merged to prevent data corruption.')
+                        elif not merger.available:
+                            self.report_warning(
+                                'You have requested merging of multiple formats but ffmpeg is not installed. '
+                                'The formats won\'t be merged.')
+
+                        if dl_filename is None:
+                            for f in requested_formats:
+                                new_info = dict(info_dict)
+                                del new_info['requested_formats']
+                                new_info.update(f)
+                                fname = prepend_extension(
+                                    self.prepare_filename(new_info, 'temp'),
+                                    'f%s' % f['format_id'], new_info['ext'])
+                                if not self._ensure_dir_exists(fname):
+                                    return
+                                downloaded.append(fname)
+                                partial_success, real_download = self.dl(fname, new_info)
+                                info_dict['__real_download'] = info_dict['__real_download'] or real_download
+                                success = success and partial_success
+                            if merger.available and not self.params.get('allow_unplayable_formats'):
+                                info_dict['__postprocessors'].append(merger)
+                                info_dict['__files_to_merge'] = downloaded
+                                # Even if there were no downloads, it is being merged only now
+                                info_dict['__real_download'] = True
+                            else:
+                                for file in downloaded:
+                                    files_to_move[file] = None
                 else:
                     # Just a single file
                     dl_filename = existing_file(full_filename, temp_filename)

yt_dlp/__init__.py

@@ -264,8 +264,8 @@ def parse_compat_opts():
         return parsed_compat_opts
 
     all_compat_opts = [
-        'filename', 'format-sort', 'abort-on-error', 'format-spec', 'multistreams',
+        'filename', 'format-sort', 'abort-on-error', 'format-spec', 'no-playlist-metafiles',
-        'no-playlist-metafiles', 'no-live-chat', 'playlist-index', 'list-formats',
+        'multistreams', 'no-live-chat', 'playlist-index', 'list-formats', 'no-direct-merge',
         'no-youtube-channel-redirect', 'no-youtube-unavailable-videos', 'no-attach-info-json',
     ]
     compat_opts = parse_compat_opts()

yt_dlp/downloader/external.py

@@ -346,7 +346,7 @@ def available(cls, path=None):
         return FFmpegPostProcessor().available
 
     def _call_downloader(self, tmpfilename, info_dict):
-        url = info_dict['url']
+        urls = [f['url'] for f in info_dict.get('requested_formats', [])] or [info_dict['url']]
         ffpp = FFmpegPostProcessor(downloader=self)
         if not ffpp.available:
             self.report_error('m3u8 download detected but ffmpeg could not be found. Please install')
@@ -378,7 +378,7 @@ def _call_downloader(self, tmpfilename, info_dict):
         # if end_time:
         #     args += ['-t', compat_str(end_time - start_time)]
 
-        if info_dict.get('http_headers') is not None and re.match(r'^https?://', url):
+        if info_dict.get('http_headers') is not None and re.match(r'^https?://', urls[0]):
             # Trailing \r\n after each HTTP header is important to prevent warning from ffmpeg/avconv:
             # [http @ 00000000003d2fa0] No trailing CRLF found in HTTP header.
             headers = handle_youtubedl_headers(info_dict['http_headers'])
@@ -436,7 +436,15 @@ def _call_downloader(self, tmpfilename, info_dict):
             elif isinstance(conn, compat_str):
                 args += ['-rtmp_conn', conn]
 
-        args += ['-i', url, '-c', 'copy']
+        for url in urls:
+            args += ['-i', url]
+        args += ['-c', 'copy']
+        if info_dict.get('requested_formats'):
+            for (i, fmt) in enumerate(info_dict['requested_formats']):
+                if fmt.get('acodec') != 'none':
+                    args.extend(['-map', '%d:a:0' % i])
+                if fmt.get('vcodec') != 'none':
+                    args.extend(['-map', '%d:v:0' % i])
 
         if self.params.get('test', False):
             args += ['-fs', compat_str(self._TEST_FILE_SIZE)]