[postprocessor/embedthumbnail] Fix issues (closes #25717)
[youtube-dl] / youtube_dl / postprocessor / embedthumbnail.py
index b6507db272a81f1e0ce3880be74c6015ec006f93..5a33595886853e95c52ffbfa465193260d8f8893 100644 (file)
@@ -1,19 +1,19 @@
-# -*- coding: utf-8 -*-
+# coding: utf-8
 from __future__ import unicode_literals
 
 
 import os
 import subprocess
 
-from .common import PostProcessor
-from ..compat import (
-    compat_urlretrieve,
-)
+from .ffmpeg import FFmpegPostProcessor
+
 from ..utils import (
     check_executable,
+    encodeArgument,
     encodeFilename,
     PostProcessingError,
     prepend_extension,
+    replace_extension,
     shell_quote
 )
 
@@ -22,44 +22,82 @@ class EmbedThumbnailPPError(PostProcessingError):
     pass
 
 
-class EmbedThumbnailPP(PostProcessor):
+class EmbedThumbnailPP(FFmpegPostProcessor):
+    def __init__(self, downloader=None, already_have_thumbnail=False):
+        super(EmbedThumbnailPP, self).__init__(downloader)
+        self._already_have_thumbnail = already_have_thumbnail
+
     def run(self, info):
         filename = info['filepath']
         temp_filename = prepend_extension(filename, 'temp')
-        temp_thumbnail = prepend_extension(filename, 'thumb')
-
-        if not info.get('thumbnail'):
-            raise EmbedThumbnailPPError('Thumbnail was not found. Nothing to do.')
 
-        compat_urlretrieve(info['thumbnail'], temp_thumbnail)
+        if not info.get('thumbnails'):
+            self._downloader.to_screen('[embedthumbnail] There aren\'t any thumbnails to embed')
+            return [], info
+
+        thumbnail_filename = info['thumbnails'][-1]['filename']
+
+        if not os.path.exists(encodeFilename(thumbnail_filename)):
+            self._downloader.report_warning(
+                'Skipping embedding the thumbnail because the file is missing.')
+            return [], info
+
+        def is_webp(path):
+            with open(encodeFilename(path), 'rb') as f:
+                b = f.read(12)
+            return b[0:4] == b'RIFF' and b[8:] == b'WEBP'
+
+        # Correct extension for WebP file with wrong extension (see #25687, #25717)
+        _, thumbnail_ext = os.path.splitext(thumbnail_filename)
+        if thumbnail_ext:
+            thumbnail_ext = thumbnail_ext[1:].lower()
+            if thumbnail_ext != 'webp' and is_webp(thumbnail_filename):
+                self._downloader.to_screen(
+                    '[ffmpeg] Correcting extension to webp and escaping path for thumbnail "%s"' % thumbnail_filename)
+                thumbnail_webp_filename = replace_extension(thumbnail_filename, 'webp')
+                os.rename(encodeFilename(thumbnail_filename), encodeFilename(thumbnail_webp_filename))
+                thumbnail_filename = thumbnail_webp_filename
+                thumbnail_ext = 'webp'
+
+        # Convert unsupported thumbnail formats to JPEG (see #25687, #25717)
+        if thumbnail_ext not in ['jpg', 'png']:
+            # NB: % is supposed to be escaped with %% but this does not work
+            # for input files so working around with standard substitution
+            escaped_thumbnail_filename = thumbnail_filename.replace('%', '#')
+            os.rename(encodeFilename(thumbnail_filename), encodeFilename(escaped_thumbnail_filename))
+            escaped_thumbnail_jpg_filename = replace_extension(escaped_thumbnail_filename, 'jpg')
+            self._downloader.to_screen('[ffmpeg] Converting thumbnail "%s" to JPEG' % escaped_thumbnail_filename)
+            self.run_ffmpeg(escaped_thumbnail_filename, escaped_thumbnail_jpg_filename, ['-bsf:v', 'mjpeg2jpeg'])
+            os.remove(encodeFilename(escaped_thumbnail_filename))
+            thumbnail_jpg_filename = replace_extension(thumbnail_filename, 'jpg')
+            # Rename back to unescaped for further processing
+            os.rename(encodeFilename(escaped_thumbnail_jpg_filename), encodeFilename(thumbnail_jpg_filename))
+            thumbnail_filename = thumbnail_jpg_filename
 
         if info['ext'] == 'mp3':
-            if not check_executable('ffmpeg', ['-version']):
-                raise AtomicParsleyPPError('FFmpeg was not found. Please install.')
-
-            cmd = ['ffmpeg', '-i', filename, '-i', temp_thumbnail, '-c', 'copy', '-map', '0', '-map', '1', '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (Front)"', temp_filename]
+            options = [
+                '-c', 'copy', '-map', '0', '-map', '1',
+                '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (Front)"']
 
             self._downloader.to_screen('[ffmpeg] Adding thumbnail to "%s"' % filename)
 
-            if self._downloader.params.get('verbose', False):
-                self._downloader.to_screen('[debug] FFmpeg command line: %s' % shell_quote(cmd))
-
-            p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-            stdout, stderr = p.communicate()
-
-            if p.returncode != 0:
-                msg = stderr.decode('utf-8', 'replace').strip()
-                raise EmbedThumbnailPPError(msg)
+            self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
 
-            os.remove(encodeFilename(temp_thumbnail))
+            if not self._already_have_thumbnail:
+                os.remove(encodeFilename(thumbnail_filename))
             os.remove(encodeFilename(filename))
             os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 
-        elif info['ext'] == 'm4a':
+        elif info['ext'] in ['m4a', 'mp4']:
             if not check_executable('AtomicParsley', ['-v']):
                 raise EmbedThumbnailPPError('AtomicParsley was not found. Please install.')
 
-            cmd = ['AtomicParsley', filename, '--artwork', temp_thumbnail, '-o', temp_filename]
+            cmd = [encodeFilename('AtomicParsley', True),
+                   encodeFilename(filename, True),
+                   encodeArgument('--artwork'),
+                   encodeFilename(thumbnail_filename, True),
+                   encodeArgument('-o'),
+                   encodeFilename(temp_filename, True)]
 
             self._downloader.to_screen('[atomicparsley] Adding thumbnail to "%s"' % filename)
 
@@ -73,7 +111,8 @@ class EmbedThumbnailPP(PostProcessor):
                 msg = stderr.decode('utf-8', 'replace').strip()
                 raise EmbedThumbnailPPError(msg)
 
-            os.remove(encodeFilename(temp_thumbnail))
+            if not self._already_have_thumbnail:
+                os.remove(encodeFilename(thumbnail_filename))
             # for formats that don't support thumbnails (like 3gp) AtomicParsley
             # won't create to the temporary file
             if b'No changes' in stdout:
@@ -82,6 +121,6 @@ class EmbedThumbnailPP(PostProcessor):
                 os.remove(encodeFilename(filename))
                 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
         else:
-            raise EmbedThumbnailPPError('Only mp3 and m4a are supported for thumbnail embedding for now.')
+            raise EmbedThumbnailPPError('Only mp3 and m4a/mp4 are supported for thumbnail embedding for now.')
 
         return [], info