Merge 7bac30e475 into f2a4983df7

test/test_YoutubeDL.py

@@ -929,32 +929,33 @@ class TestYoutubeDL(unittest.TestCase):
         }), r'^30fps$')
 
     def test_postprocessors(self):
-        filename = 'post-processor-testfile.mp4'
-        audiofile = filename + '.mp3'
+        filename = 'post-processor-testfile'
+        video_file = filename + '.mp4'
+        audio_file = filename + '.mp3'
 
         class SimplePP(PostProcessor):
             def run(self, info):
-                with open(audiofile, 'w') as f:
+                with open(audio_file, 'w') as f:
                     f.write('EXAMPLE')
                 return [info['filepath']], info
 
         def run_pp(params, pp):
-            with open(filename, 'w') as f:
+            with open(video_file, 'w') as f:
                 f.write('EXAMPLE')
             ydl = YoutubeDL(params)
             ydl.add_post_processor(pp())
-            ydl.post_process(filename, {'filepath': filename})
+            ydl.post_process(video_file, {'filepath': video_file})
 
-        run_pp({'keepvideo': True}, SimplePP)
-        self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
-        self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
-        os.unlink(filename)
-        os.unlink(audiofile)
+        run_pp({'keepvideo': True, 'outtmpl': filename}, SimplePP)
+        self.assertTrue(os.path.exists(video_file), f'{video_file} doesn\'t exist')
+        self.assertTrue(os.path.exists(audio_file), f'{audio_file} doesn\'t exist')
+        os.unlink(video_file)
+        os.unlink(audio_file)
 
-        run_pp({'keepvideo': False}, SimplePP)
-        self.assertFalse(os.path.exists(filename), f'{filename} exists')
-        self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
-        os.unlink(audiofile)
+        run_pp({'keepvideo': False, 'outtmpl': filename}, SimplePP)
+        self.assertFalse(os.path.exists(video_file), f'{video_file} exists')
+        self.assertTrue(os.path.exists(audio_file), f'{audio_file} doesn\'t exist')
+        os.unlink(audio_file)
 
         class ModifierPP(PostProcessor):
             def run(self, info):
@@ -962,9 +963,9 @@ class TestYoutubeDL(unittest.TestCase):
                     f.write('MODIFIED')
                 return [], info
 
-        run_pp({'keepvideo': False}, ModifierPP)
-        self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
-        os.unlink(filename)
+        run_pp({'keepvideo': False, 'outtmpl': filename}, ModifierPP)
+        self.assertTrue(os.path.exists(video_file), f'{video_file} doesn\'t exist')
+        os.unlink(video_file)
 
     def test_match_filter(self):
         first = {

yt_dlp/YoutubeDL.py

@@ -3239,7 +3239,6 @@ class YoutubeDL:
         # info_dict['_filename'] needs to be set for backward compatibility
         info_dict['_filename'] = full_filename = self.prepare_filename(info_dict, warn=True)
         temp_filename = self.prepare_filename(info_dict, 'temp')
-        files_to_move = {}
 
         # Forced printings
         self.__forced_printings(info_dict, full_filename, incomplete=('format' not in info_dict))
@@ -3267,13 +3266,11 @@ class YoutubeDL:
         sub_files = self._write_subtitles(info_dict, temp_filename)
         if sub_files is None:
             return
-        files_to_move.update(dict(sub_files))
 
         thumb_files = self._write_thumbnails(
             'video', info_dict, temp_filename, self.prepare_filename(info_dict, 'thumbnail'))
         if thumb_files is None:
             return
-        files_to_move.update(dict(thumb_files))
 
         infofn = self.prepare_filename(info_dict, 'infojson')
         _infojson_written = self._write_info_json('video', info_dict, infofn)
@@ -3347,13 +3344,12 @@ class YoutubeDL:
                for link_type, should_write in write_links.items()):
             return
 
-        new_info, files_to_move = self.pre_process(info_dict, 'before_dl', files_to_move)
+        new_info, _ = self.pre_process(info_dict, 'before_dl')
         replace_info_dict(new_info)
 
         if self.params.get('skip_download'):
             info_dict['filepath'] = temp_filename
             info_dict['__finaldir'] = os.path.dirname(os.path.abspath(encodeFilename(full_filename)))
-            info_dict['__files_to_move'] = files_to_move
             replace_info_dict(self.run_pp(MoveFilesAfterDownloadPP(self, False), info_dict))
             info_dict['__write_download_archive'] = self.params.get('force_write_download_archive')
         else:
@@ -3467,9 +3463,6 @@ class YoutubeDL:
                         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_video_file(full_filename, temp_filename)
@@ -3483,7 +3476,6 @@ class YoutubeDL:
 
                 dl_filename = dl_filename or temp_filename
                 info_dict['__finaldir'] = os.path.dirname(os.path.abspath(encodeFilename(full_filename)))
-
             except network_exceptions as err:
                 self.report_error(f'unable to download video data: {err}')
                 return
@@ -3554,7 +3546,7 @@ class YoutubeDL:
 
                 fixup()
                 try:
-                    replace_info_dict(self.post_process(dl_filename, info_dict, files_to_move))
+                    replace_info_dict(self.post_process(dl_filename, info_dict))
                 except PostProcessingError as err:
                     self.report_error(f'Postprocessing: {err}')
                     return
@@ -3677,8 +3669,6 @@ class YoutubeDL:
                 os.remove(filename)
             except OSError:
                 self.report_warning(f'Unable to delete file {filename}')
-            if filename in info.get('__files_to_move', []):  # NB: Delete even if None
-                del info['__files_to_move'][filename]
 
     @staticmethod
     def post_extract(info_dict):
@@ -3695,8 +3685,7 @@ class YoutubeDL:
 
     def run_pp(self, pp, infodict):
         files_to_delete = []
-        if '__files_to_move' not in infodict:
-            infodict['__files_to_move'] = {}
+
         try:
             files_to_delete, infodict = pp.run(infodict)
         except PostProcessingError as e:
@@ -3708,10 +3697,7 @@ class YoutubeDL:
 
         if not files_to_delete:
             return infodict
-        if self.params.get('keepvideo', False):
-            for f in files_to_delete:
-                infodict['__files_to_move'].setdefault(f, '')
-        else:
+        if not self.params.get('keepvideo', False):
             self._delete_downloaded_files(
                 *files_to_delete, info=infodict, msg='Deleting original file %s (pass -k to keep)')
         return infodict
@@ -3724,23 +3710,27 @@ class YoutubeDL:
         return info
 
     def pre_process(self, ie_info, key='pre_process', files_to_move=None):
+        if files_to_move is not None:
+            self.report_warning('[pre_process] "files_to_move" is deprecated and may be removed in a future version')
+
         info = dict(ie_info)
-        info['__files_to_move'] = files_to_move or {}
         try:
             info = self.run_all_pps(key, info)
         except PostProcessingError as err:
             msg = f'Preprocessing: {err}'
             info.setdefault('__pending_error', msg)
             self.report_error(msg, is_error=False)
-        return info, info.pop('__files_to_move', None)
+        return info, files_to_move
 
     def post_process(self, filename, info, files_to_move=None):
         """Run all the postprocessors on the given file."""
+        if files_to_move is not None:
+            self.report_warning('[post_process] "files_to_move" is deprecated and may be removed in a future version')
+
         info['filepath'] = filename
-        info['__files_to_move'] = files_to_move or {}
         info = self.run_all_pps('post_process', info, additional_pps=info.get('__postprocessors'))
         info = self.run_pp(MoveFilesAfterDownloadPP(self), info)
-        del info['__files_to_move']
+        info.pop('__multiple_thumbnails', None)
         return self.run_all_pps('after_move', info)
 
     def _make_archive_id(self, info_dict):
@@ -4325,10 +4315,11 @@ class YoutubeDL:
             sub_filename = subtitles_filename(filename, sub_lang, sub_format, info_dict.get('ext'))
             sub_filename_final = subtitles_filename(sub_filename_base, sub_lang, sub_format, info_dict.get('ext'))
             existing_sub = self.existing_file((sub_filename_final, sub_filename))
+
             if existing_sub:
                 self.to_screen(f'[info] Video subtitle {sub_lang}.{sub_format} is already present')
                 sub_info['filepath'] = existing_sub
-                ret.append((existing_sub, sub_filename_final))
+                ret.append(existing_sub)
                 continue
 
             self.to_screen(f'[info] Writing video subtitles to: {sub_filename}')
@@ -4339,7 +4330,7 @@ class YoutubeDL:
                     with open(sub_filename, 'w', encoding='utf-8', newline='') as subfile:
                         subfile.write(sub_info['data'])
                     sub_info['filepath'] = sub_filename
-                    ret.append((sub_filename, sub_filename_final))
+                    ret.append(sub_filename)
                     continue
                 except OSError:
                     self.report_error(f'Cannot write video subtitles file {sub_filename}')
@@ -4350,7 +4341,7 @@ class YoutubeDL:
                 sub_copy.setdefault('http_headers', info_dict.get('http_headers'))
                 self.dl(sub_filename, sub_copy, subtitle=True)
                 sub_info['filepath'] = sub_filename
-                ret.append((sub_filename, sub_filename_final))
+                ret.append(sub_filename)
             except (DownloadError, ExtractorError, OSError, ValueError, *network_exceptions) as err:
                 msg = f'Unable to download video subtitles for {sub_lang!r}: {err}'
                 if self.params.get('ignoreerrors') is not True:  # False or 'only_download'
@@ -4370,6 +4361,7 @@ class YoutubeDL:
                 self.to_screen(f'[info] There are no {label} thumbnails to download')
                 return ret
         multiple = write_all and len(thumbnails) > 1
+        info_dict['__multiple_thumbnails'] = multiple
 
         if thumb_filename_base is None:
             thumb_filename_base = filename
@@ -4391,7 +4383,7 @@ class YoutubeDL:
                 self.to_screen('[info] {} is already present'.format((
                     thumb_display_id if multiple else f'{label} thumbnail').capitalize()))
                 t['filepath'] = existing_thumb
-                ret.append((existing_thumb, thumb_filename_final))
+                ret.append(existing_thumb)
             else:
                 self.to_screen(f'[info] Downloading {thumb_display_id} ...')
                 try:
@@ -4399,7 +4391,7 @@ class YoutubeDL:
                     self.to_screen(f'[info] Writing {thumb_display_id} to: {thumb_filename}')
                     with open(encodeFilename(thumb_filename), 'wb') as thumbf:
                         shutil.copyfileobj(uf, thumbf)
-                    ret.append((thumb_filename, thumb_filename_final))
+                    ret.append(thumb_filename)
                     t['filepath'] = thumb_filename
                 except network_exceptions as err:
                     if isinstance(err, HTTPError) and err.status == 404:
@@ -4409,4 +4401,5 @@ class YoutubeDL:
                     thumbnails.pop(idx)
             if ret and not write_all:
                 break
+
         return ret

yt_dlp/postprocessor/embedthumbnail.py

@@ -231,4 +231,8 @@ class EmbedThumbnailPP(FFmpegPostProcessor):
             thumbnail_filename if converted or not self._already_have_thumbnail else None,
             original_thumbnail if converted and not self._already_have_thumbnail else None,
             info=info)
+
+        if not self._already_have_thumbnail:
+            info['thumbnails'][idx].pop('filepath', None)
+
         return [], info

yt_dlp/postprocessor/ffmpeg.py

@@ -663,6 +663,10 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
         os.replace(temp_filename, filename)
 
+        if not self._already_have_subtitle:
+            for _, subtitle in subtitles.items():
+                subtitle.pop('filepath', None)
+
         files_to_delete = [] if self._already_have_subtitle else sub_filenames
         return files_to_delete, info
 
@@ -699,6 +703,7 @@ class FFmpegMetadataPP(FFmpegPostProcessor):
                 infojson_filename = info.get('infojson_filename')
                 options.extend(self._get_infojson_opts(info, infojson_filename))
                 if not infojson_filename:
+                    info.pop('infojson_filename', None)
                     files_to_delete.append(info.get('infojson_filename'))
             elif self._add_infojson is True:
                 self.to_screen('The info-json can only be attached to mkv/mka files')
@@ -1016,9 +1021,6 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                     'filepath': new_file,
                 }
 
-            info['__files_to_move'][new_file] = replace_extension(
-                info['__files_to_move'][sub['filepath']], new_ext)
-
         return sub_filenames, info
 
 
@@ -1083,16 +1085,15 @@ class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
         return imghdr.what(path) == 'webp'
 
     def fixup_webp(self, info, idx=-1):
-        thumbnail_filename = info['thumbnails'][idx]['filepath']
+        thumbnail = info['thumbnails'][idx]
+        thumbnail_filename = thumbnail['filepath']
         _, thumbnail_ext = os.path.splitext(thumbnail_filename)
         if thumbnail_ext:
             if thumbnail_ext.lower() != '.webp' and imghdr.what(thumbnail_filename) == 'webp':
                 self.to_screen(f'Correcting thumbnail "{thumbnail_filename}" extension to webp')
                 webp_filename = replace_extension(thumbnail_filename, 'webp')
                 os.replace(thumbnail_filename, webp_filename)
-                info['thumbnails'][idx]['filepath'] = webp_filename
-                info['__files_to_move'][webp_filename] = replace_extension(
-                    info['__files_to_move'].pop(thumbnail_filename), 'webp')
+                thumbnail['filepath'] = webp_filename
 
     @staticmethod
     def _options(target_ext):
@@ -1130,8 +1131,6 @@ class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
                 continue
             thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, target_ext)
             files_to_delete.append(original_thumbnail)
-            info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
-                info['__files_to_move'][original_thumbnail], target_ext)
 
         if not has_thumbnail:
             self.to_screen('There aren\'t any thumbnails to convert')

yt_dlp/postprocessor/movefilesafterdownload.py

@@ -1,16 +1,22 @@
 import os
+from pathlib import Path
 
 from .common import PostProcessor
 from ..compat import shutil
 from ..utils import (
     PostProcessingError,
-    decodeFilename,
-    encodeFilename,
     make_dir,
+    replace_extension,
 )
 
 
 class MoveFilesAfterDownloadPP(PostProcessor):
+    # Map of the keys that contain moveable files and the 'type' of the file
+    # for generating the output filename
+    CHILD_KEYS = {
+        'thumbnails': 'thumbnail',
+        'requested_subtitles': 'subtitle',
+    }
 
     def __init__(self, downloader=None, downloaded=True):
         PostProcessor.__init__(self, downloader)
@@ -20,33 +26,77 @@ class MoveFilesAfterDownloadPP(PostProcessor):
     def pp_key(cls):
         return 'MoveFiles'
 
+    def move_file_and_write_to_info(self, info_dict, relevant_dict=None, output_file_type=None):
+        relevant_dict = relevant_dict or info_dict
+        if 'filepath' not in relevant_dict:
+            return
+
+        output_file_type = output_file_type or ''
+        current_filepath, final_filepath = self.determine_filepath(info_dict, relevant_dict, output_file_type)
+        move_result = self.move_file(info_dict, current_filepath, final_filepath)
+
+        if move_result:
+            relevant_dict['filepath'] = move_result
+        else:
+            del relevant_dict['filepath']
+
+    def determine_filepath(self, info_dict, relevant_dict, output_file_type):
+        current_filepath = relevant_dict['filepath']
+        prepared_filepath = self._downloader.prepare_filename(info_dict, output_file_type)
+
+        if (output_file_type == 'thumbnail' and info_dict['__multiple_thumbnails']) or output_file_type == 'subtitle':
+            desired_extension = ''.join(Path(current_filepath).suffixes[-2:])
+        else:
+            desired_extension = Path(current_filepath).suffix
+
+        return current_filepath, replace_extension(prepared_filepath, desired_extension[1:])
+
+    def move_file(self, info_dict, current_filepath, final_filepath):
+        if not current_filepath or not final_filepath:
+            return
+
+        dl_parent_folder = os.path.split(info_dict['filepath'])[0]
+        finaldir = info_dict.get('__finaldir', os.path.abspath(dl_parent_folder))
+
+        if not os.path.isabs(current_filepath):
+            current_filepath = os.path.join(finaldir, current_filepath)
+
+        if not os.path.isabs(final_filepath):
+            final_filepath = os.path.join(finaldir, final_filepath)
+
+        if current_filepath == final_filepath:
+            return final_filepath
+
+        if not os.path.exists(current_filepath):
+            self.report_warning(f'File "{current_filepath}" cannot be found')
+            return
+
+        if os.path.exists(final_filepath):
+            if self.get_param('overwrites', True):
+                self.report_warning(f'Replacing existing file "{final_filepath}"')
+                os.remove(final_filepath)
+            else:
+                self.report_warning(f'Cannot move file "{current_filepath}" out of temporary directory since "{final_filepath}" already exists. ')
+                return
+
+        make_dir(final_filepath, PostProcessingError)
+        self.to_screen(f'Moving file "{current_filepath}" to "{final_filepath}"')
+        shutil.move(current_filepath, final_filepath)  # os.rename cannot move between volumes
+
+        return final_filepath
+
     def run(self, info):
-        dl_path, dl_name = os.path.split(encodeFilename(info['filepath']))
-        finaldir = info.get('__finaldir', dl_path)
-        finalpath = os.path.join(finaldir, dl_name)
-        if self._downloaded:
-            info['__files_to_move'][info['filepath']] = decodeFilename(finalpath)
+        # This represents the main media file (using the 'filepath' key)
+        self.move_file_and_write_to_info(info)
 
-        make_newfilename = lambda old: decodeFilename(os.path.join(finaldir, os.path.basename(encodeFilename(old))))
-        for oldfile, newfile in info['__files_to_move'].items():
-            if not newfile:
-                newfile = make_newfilename(oldfile)
-            if os.path.abspath(encodeFilename(oldfile)) == os.path.abspath(encodeFilename(newfile)):
+        for key, output_file_type in self.CHILD_KEYS.items():
+            if key not in info:
                 continue
-            if not os.path.exists(encodeFilename(oldfile)):
-                self.report_warning(f'File "{oldfile}" cannot be found')
-                continue
-            if os.path.exists(encodeFilename(newfile)):
-                if self.get_param('overwrites', True):
-                    self.report_warning(f'Replacing existing file "{newfile}"')
-                    os.remove(encodeFilename(newfile))
-                else:
-                    self.report_warning(
-                        f'Cannot move file "{oldfile}" out of temporary directory since "{newfile}" already exists. ')
-                    continue
-            make_dir(newfile, PostProcessingError)
-            self.to_screen(f'Moving file "{oldfile}" to "{newfile}"')
-            shutil.move(oldfile, newfile)  # os.rename cannot move between volumes
 
-        info['filepath'] = finalpath
+            if isinstance(info[key], (dict, list)):
+                iterable = info[key].values() if isinstance(info[key], dict) else info[key]
+
+                for file_dict in iterable:
+                    self.move_file_and_write_to_info(info, file_dict, output_file_type)
+
         return [], info