A comment to remember how to extract SPS and PPS from a file with ffmpeg.

removeads.py

@@ -26,11 +26,15 @@ from shutil import copyfile, which
 
 def checkRequiredTools():
     logger = logging.getLogger(__name__)
-    tools = ['ffmpeg', 'ffprobe', 'mkvmerge']
+    required = ['ffmpeg', 'ffprobe', 'mkvmerge']
-    for tool in tools:
+    optional = ['mkvextract', 'vobusbocr']
+    for tool in required:
         if which(tool) == None:
             logger.error('Required tool: %s is missing.' % tool)
             exit(-1)
+    for tool in optional:
+        if which(tool) == None:
+            logger.info('Optional tool: %s is missing.' % tool)
 
 
 @unique
@@ -49,6 +53,11 @@ class SupportedFormat(IntEnum):
         else:
             return 'Unsupported format'
 
+# Extract SPS/PPS 
+# https://gitlab.com/mbunkus/mkvtoolnix/-/issues/2390
+# ffmpeg -i <InputFile (before concatenation)> -c:v copy -an -sn -bsf:v trace_headers -t 0.01 -report -loglevel 0 -f null -
+
+
 def getFormat(inputFile):
     logger = logging.getLogger(__name__)
         
@@ -352,10 +361,11 @@ def extractPictures(inputFile, begin, nbFrames, width=640, height=480):
     # "P6\nWIDTH HEIGHT\n255\n"
     headerLen=2+1+ceil(log(width, 10))+1+ceil(log(height, 10))+1+3+1
     logger.debug('Header length: %d' % headerLen)
-    length = (width*height*3+headerLen)*nbFrames
+    imageLength = width*height*3+headerLen
+    length = imageLength*nbFrames
     logger.debug("Estimated length: %d" % length)
     
-    pg = trange(length)
+    pg = trange(nbFrames)
     images = bytes()
     with Popen(['ffmpeg', '-loglevel', 'quiet' ,'-y', '-ss', '%s'%begin, '-i', '/proc/self/fd/%d' % infd, '-s', '%dx%d'%(width, height), '-vframes', '%d'%nbFrames, '-c:v', 'ppm', '-f', 'image2pipe', '/proc/self/fd/%d' % fdw ], stdout=PIPE, close_fds=False) as ffmpeg:
         while ffmpeg.poll() == None:
@@ -363,11 +373,11 @@ def extractPictures(inputFile, begin, nbFrames, width=640, height=480):
             # fds, _, _ = select([fdr, ffmpeg.stdout], [], [], .1)
             fds, _, _ = select([fdr], [], [], .1)
             if fdr in fds:
-                buf = os.read(fdr, 1000000)
+                buf = os.read(fdr, imageLength)
                 # print("Read %d bytes of image. ffmpeg finished: %s" % (len(buf), ffmpeg.poll()))
                 if len(buf) == 0:
                     break
-                pg.update(len(buf))
+                pg.update(len(buf)/imageLength)
                 images=images+buf
             if ffmpeg.stdout in fds:
                 for line in TextIOWrapper(ffmpeg.stdout, encoding="utf-8"):
@@ -379,11 +389,11 @@ def extractPictures(inputFile, begin, nbFrames, width=640, height=480):
     while True:
         fd, _, _ = select([fdr], [], [], .1)
         if fd != []:
-            buf = os.read(fdr, 1000000)
+            buf = os.read(fdr, imageLength)
             # print("Read %d bytes of image" % len(buf))
             if len(buf) == 0:
                 break
-            pg.update(len(buf))
+            pg.update(len(buf)/imageLength)
             images=images+buf
         else:
             # Nothing more to read
@@ -661,6 +671,28 @@ def mergeMKVs(inputs, outputName):
     
     return out
 
+def findSubtitlesTracks(filename):
+    # ffprobe -loglevel quiet -select_streams s -show_entries stream=index:stream_tags=language -of json corgi.ts
+    logger = logging.getLogger(__name__)
+    
+    with Popen(['ffprobe', '-i', filename, '-select_streams', 's', '-show_entries', 'stream=index:stream_tags=language', '-of', 'json'], stdout=PIPE, close_fds=False) as ffprobe:
+        out, _ = ffprobe.communicate()
+        out = json.load(BytesIO(out))
+        if 'streams' in out:
+            return out['streams']
+        else:
+            logger.error('Impossible to retrieve format of file')
+    pass
+
+def extractSubTitleTrack(inputFileName, index, lang):
+    #  mkvextract video.mkv tracks position:nom [position:nom]
+    logger = logging.getLogger(__name__)
+    
+    with Popen(['mkvextract', inputFileName, 'tracks', '%d:%s' % (index,lang)], stdout=PIPE, close_fds=False) as mkvextract:
+        out, _ = mkvextract.communicate()
+        for lines in out:
+            logger.info(out)
+
     
     
 def main():
@@ -673,6 +705,7 @@ def main():
     parser.add_argument("-p", "--part", dest='parts', nargs='+', required=False, action='append', metavar="hh:mm:ss[.mmm]-hh:mm:ss[.mmm]", help="Extract this exact part of the original file.")
     parser.add_argument("-k", "--keep", action='store_true', help="Do not cleanup temporary files after processing.")
     parser.add_argument("--dump-pictures", action='store_true', dest='dump', help="For debug purpose, dump pictures of headers (and trailers) before (after) each part. They are kept in memory only otherwise.")
+    parser.add_argument("-s","--srt", action='store_true', dest='srt', help="Dump subtitles ")
   
     args = parser.parse_args()
     logger.debug("Arguments: %s" % args)
@@ -857,6 +890,38 @@ def main():
     else:
         logger.info("Nothing else to do.")
   
+    if args.srt:
+        logger.info("Find subtitles tracks and language.")
+        subtitles = findSubtitlesTracks(args.outputFile)
+        sts = {}
+        for subtitle in subtitles:
+            index = subtitle['index']
+            if 'tags' in subtitle:
+                if 'language' in subtitle['tags']:
+                    lang = subtitle['tags']['language']
+                    if lang in sts:
+                        sts[lang].append(index)
+                    else:
+                        sts[lang] = [index]
+                else:
+                    logger.error("Dropping subtitle: %s because it is missing language indication")
+            else:
+                logger.error("Dropping subtitle: %s because it is missing language indication")
+     
+    for lang in sts:
+        indexes = sts[lang]
+        if len(indexes) == 0:
+            # Nothing to do. This should not happen.
+            continue
+        if len(indexes) == 1:
+            index = indexes[0]
+            filename = 'essai-%s.srt' % lang
+        elif len(indexes) > 1:
+            nbsrt = 1
+            for index in indexes:
+                filename = 'essai-%s-%d.srt' % (lang, nbsrt)
+                nbsrt+=1
+  
     if not args.keep:
         logger.info("Cleaning temporary files")
         for f in temporaries: