Refactored MoveFilesPP to respect non-video files

yt_dlp/YoutubeDL.py

@@ -3236,13 +3236,15 @@ class YoutubeDL:
         sub_files = self._write_subtitles(info_dict, temp_filename)
         if sub_files is None:
             return
-        files_to_move.update(dict(sub_files))
+
+        files_to_move['requested_subtitles'] = 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))
+
+        files_to_move['thumbnails'] = thumb_files
 
         infofn = self.prepare_filename(info_dict, 'infojson')
         _infojson_written = self._write_info_json('video', info_dict, infofn)
@@ -3452,7 +3454,6 @@ 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('unable to download video data: %s' % error_to_compat_str(err))
                 return
@@ -3644,8 +3645,6 @@ 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):
@@ -3675,10 +3674,7 @@ 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
@@ -4294,10 +4290,17 @@ 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({
+                    'current_filepath': existing_sub,
+                    'final_filepath': sub_filename_final,
+                    'lang': sub_lang,
+                    'ext': sub_info['ext']
+                })
+
                 continue
 
             self.to_screen(f'[info] Writing video subtitles to: {sub_filename}')
@@ -4308,7 +4311,13 @@ 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({
+                        'current_filepath': sub_filename,
+                        'final_filepath': sub_filename_final,
+                        'lang': sub_lang,
+                        'ext': sub_info['ext']
+                    })
+
                     continue
                 except OSError:
                     self.report_error(f'Cannot write video subtitles file {sub_filename}')
@@ -4319,7 +4328,13 @@ 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({
+                    'current_filepath': sub_filename,
+                    'final_filepath': sub_filename_final,
+                    'lang': sub_lang,
+                    'ext': sub_info['ext']
+                })
+
             except (DownloadError, ExtractorError, IOError, 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'
@@ -4360,7 +4375,12 @@ class YoutubeDL:
                 self.to_screen('[info] %s is already present' % (
                     thumb_display_id if multiple else f'{label} thumbnail').capitalize())
                 t['filepath'] = existing_thumb
-                ret.append((existing_thumb, thumb_filename_final))
+                ret.append({
+                    'current_filepath': existing_thumb,
+                    'final_filepath': thumb_filename_final,
+                    'id': t['id']
+                })
+
             else:
                 self.to_screen(f'[info] Downloading {thumb_display_id} ...')
                 try:
@@ -4368,7 +4388,13 @@ 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({
+                        'current_filepath': thumb_filename,
+                        'final_filepath': thumb_filename_final,
+                        'id': t['id']
+                    })
+
                     t['filepath'] = thumb_filename
                 except network_exceptions as err:
                     if isinstance(err, HTTPError) and err.status == 404:
@@ -4378,4 +4404,5 @@ class YoutubeDL:
                     thumbnails.pop(idx)
             if ret and not write_all:
                 break
+
         return ret

yt_dlp/postprocessor/ffmpeg.py

@@ -1016,8 +1016,10 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                     'filepath': new_file,
                 }
 
-            info['__files_to_move'][new_file] = replace_extension(
-                info['__files_to_move'][sub['filepath']], new_ext)
+            for sub_info in info['__files_to_move']['requested_subtitles']:
+                if sub_info['lang'] == lang and sub_info['ext'] == sub['ext']:
+                    sub_info['current_filepath'] = replace_extension(sub_info['current_filepath'], new_ext)
+                    sub_info['final_filepath'] = replace_extension(sub_info['final_filepath'], new_ext)
 
         return sub_filenames, info
 
@@ -1083,16 +1085,20 @@ 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('Correcting thumbnail "%s" extension to webp' % thumbnail_filename)
                 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
+
+                for thumb_info in info['__files_to_move']['thumbnails']:
+                    if thumb_info['id'] == thumbnail['id']:
+                        thumb_info['current_filepath'] = replace_extension(thumbnail_filename, 'webp')
+                        thumb_info['final_filepath'] = replace_extension(thumb_info['final_filepath'], 'webp')
 
     @staticmethod
     def _options(target_ext):
@@ -1130,8 +1136,11 @@ 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)
+
+            for thumb_info in info['__files_to_move']['thumbnails']:
+                if thumb_info['id'] == thumbnail_dict['id']:
+                    thumb_info['current_filepath'] = replace_extension(thumb_info['current_filepath'], target_ext)
+                    thumb_info['final_filepath'] = replace_extension(thumb_info['final_filepath'], target_ext)
 
         if not has_thumbnail:
             self.to_screen('There aren\'t any thumbnails to convert')

yt_dlp/postprocessor/movefilesafterdownload.py

@@ -4,13 +4,11 @@ from .common import PostProcessor
 from ..compat import shutil
 from ..utils import (
     PostProcessingError,
-    decodeFilename,
-    encodeFilename,
     make_dir,
 )
 
-
 class MoveFilesAfterDownloadPP(PostProcessor):
+    FILETYPE_KEYS = ['media', 'thumbnails', 'requested_subtitles']
 
     def __init__(self, downloader=None, downloaded=True):
         PostProcessor.__init__(self, downloader)
@@ -20,34 +18,77 @@ class MoveFilesAfterDownloadPP(PostProcessor):
     def pp_key(cls):
         return 'MoveFiles'
 
+    def expand_relative_paths(self, files_to_move, finaldir):
+        for filetype in self.FILETYPE_KEYS:
+            if filetype not in files_to_move:
+                continue
+            for file_attrs in files_to_move[filetype]:
+                if not os.path.isabs(file_attrs['final_filepath']):
+                    file_attrs['final_filepath'] = os.path.join(finaldir, file_attrs['final_filepath'])
+                if not os.path.isabs(file_attrs['current_filepath']):
+                    file_attrs['current_filepath'] = os.path.abspath(file_attrs['current_filepath'])
+
+        return files_to_move
+
+    def write_filepath_into_info(self, info, filetype, file_attrs):
+        if filetype == 'media':
+            info['filepath'] = file_attrs['final_filepath']
+            return
+
+        elif filetype == 'thumbnails':
+            for filetype_dict in info[filetype]:
+                if filetype_dict['id'] == file_attrs['id']:
+                    filetype_dict['filepath'] = file_attrs['final_filepath']
+
+        elif filetype == 'requested_subtitles':
+            lang = file_attrs['lang']
+            if lang in info[filetype]:
+                info[filetype][lang]['filepath'] = file_attrs['final_filepath']
+
     def run(self, info):
-        dl_path, dl_name = os.path.split(encodeFilename(info['filepath']))
+        dl_path, dl_name = os.path.split(info['filepath'])
         finaldir = info.get('__finaldir', dl_path)
-        finalpath = os.path.join(finaldir, dl_name)
+        info['__files_to_move']['media'] = []
+
         if self._downloaded:
-            info['__files_to_move'][info['filepath']] = decodeFilename(finalpath)
+            info['__files_to_move']['media'] = [{ 'current_filepath': info['filepath'], 'final_filepath': dl_name }]
+
+        files_to_move = self.expand_relative_paths(info['__files_to_move'], finaldir)
 
-        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 filetype in self.FILETYPE_KEYS:
+            if filetype not in files_to_move:
                 continue
-            if not os.path.exists(encodeFilename(oldfile)):
-                self.report_warning('File "%s" cannot be found' % oldfile)
+
+            for file_attrs in files_to_move[filetype]:
+                current_filepath = file_attrs['current_filepath']
+                final_filepath = file_attrs['final_filepath']
+
+                if not current_filepath or not final_filepath:
+                    continue
+
+                if current_filepath == final_filepath:
+                    # This ensures the infojson contains the full filepath even
+                    # when --no-overwrites is used
+                    self.write_filepath_into_info(info, filetype, file_attrs)
                     continue
-            if os.path.exists(encodeFilename(newfile)):
+
+                if not os.path.exists(current_filepath):
+                    self.report_warning('File "%s" cannot be found' % current_filepath)
+                    continue
+
+                if os.path.exists(final_filepath):
                     if self.get_param('overwrites', True):
-                    self.report_warning('Replacing existing file "%s"' % newfile)
-                    os.remove(encodeFilename(newfile))
+                        self.report_warning('Replacing existing file "%s"' % final_filepath)
+                        os.remove(final_filepath)
                     else:
                         self.report_warning(
                             'Cannot move file "%s" out of temporary directory since "%s" already exists. '
-                        % (oldfile, newfile))
+                            % (current_filepath, final_filepath))
                         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
+                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
+                self.write_filepath_into_info(info, filetype, file_attrs)
+
         return [], info