Subtitles extracted through OCR can be remuxed with the final cut movie (in addition to image based ones).

removeads.py

@@ -16,7 +16,7 @@ import shutil
 from tqdm import tqdm, trange
 from select import select
 from math import floor, ceil, log
-from shutil import copyfile, which
+from shutil import copyfile, which, move
 import hexdump
 from iso639 import Lang
 from iso639.exceptions import InvalidLanguageValue
@@ -107,9 +107,10 @@ def getSubTitlesTracks(ffprobe, mkvPath):
     
     return tracks
 
-def extractSRT(mkvextract, mkvPath, subtitles, langs):
+def extractSRT(mkvextract, fileName, subtitles, langs):
     logger = logging.getLogger(__name__)
-    params = [mkvextract, mkvPath, 'tracks']
+    
+    params = [mkvextract, fileName, 'tracks']
     
     res = []
 
@@ -133,8 +134,20 @@ def extractSRT(mkvextract, mkvPath, subtitles, langs):
                 count = count+1
 
     env = {**os.environ, 'LANG': 'C'}
-    with Popen(params, env=env) as extract:
-        extract.wait()
+    with Popen(params, stdout=PIPE, close_fds=False, env=env) as extract:
+        pb = tqdm(TextIOWrapper(extract.stdout, encoding="utf-8"), total=100, unit='%', desc='Extraction:')
+        for line in pb:
+            if line.startswith('Progress :'):
+                p = re.compile('^Progress : (?P<progress>[0-9]{1,3})%$')
+                m = p.match(line)
+                if m == None:
+                    logger.error('Impossible to parse progress')
+                pb.update(int(m['progress'])-pb.n)
+        pb.update(100-pb.n)
+        pb.refresh()
+        pb.close()
+    
+    extract.wait()
         
     if extract.returncode != 0:
         logger.error('Mkvextract returns an error code: %d' % extract.returncode)
@@ -982,12 +995,15 @@ def dumpPPM(pictures, prefix, temporaries):
         pos+=length
         picture+=1
 
+
+        
 def extractAllStreams(ffmpeg, ffprobe, inputFile, begin, end, streams, filesPrefix, nbFrames, width, height, temporaries, dumpMemFD=False):
     logger = logging.getLogger(__name__)
     encoderParams = [ 'ffmpeg', '-y', '-loglevel', 'quiet' ]
     inputParams = []
     codecsParams = []
     
+    
     if begin < end:
         videoID=0
         audioID=0
@@ -1001,6 +1017,8 @@ def extractAllStreams(ffmpeg, ffprobe, inputFile, begin, end, streams, filesPref
                 m = pattern.match(frameRate)
                 if m != None:
                     frameRate = float(m['numerator']) / float(m['denominator'])
+                # TODO: Framerate estimation seems broken ...
+                frameRate = 25.
                 sar = stream['sample_aspect_ratio']
                 dar = stream['display_aspect_ratio']
                 pixelFormat = stream['pix_fmt']
@@ -1217,6 +1235,52 @@ def extractSubTitleTrack(mkvmerge, inputFileName, index, lang):
         for lines in out:
             logger.info(out)
 
+def remuxSRTSubtitles(mkvmerge, inputFile, outputFileName, subtitles):
+    logger = logging.getLogger(__name__)
+    
+    try:
+        out = open(outputFileName, 'w')
+    except IOError:
+        logger.error('Impossible to create file: %s' % outputFileName)
+        return None
+    
+    outfd = out.fileno()
+    infd = inputFile.fileno()
+    lseek(infd, 0, SEEK_SET)
+    set_inheritable(infd, True)
+    set_inheritable(outfd, True)
+    
+    mkvmergeParams = [mkvmerge, '/proc/self/fd/%d' % infd]
+    for fd, lang in subtitles:
+        lseek(fd, 0, SEEK_SET)
+        set_inheritable(fd, True)
+        mkvmergeParams.extend(['--language', '0:%s' % lang, '/proc/self/fd/%d' % fd])
+    
+    mkvmergeParams.extend(['-o', '/proc/self/fd/%d' % outfd])
+    
+    warnings = []
+    env = {**os.environ, 'LANG': 'C'}
+    logger.info('Remux subtitles: %s' % mkvmergeParams)
+    with Popen(mkvmergeParams, stdout=PIPE, close_fds=False, env=env) as mkvmerge:
+        pb = tqdm(TextIOWrapper(mkvmerge.stdout, encoding="utf-8"), total=100, unit='%', desc='Remux subtitles:')
+        for line in pb:
+            if line.startswith('Progress :'):
+                p = re.compile('^Progress : (?P<progress>[0-9]{1,3})%$')
+                m = p.match(line)
+                if m == None:
+                    logger.error('Impossible to parse progress')
+                pb.n = int(m['progress'])
+                pb.update()
+            elif line.startswith('Warning'):
+                warnings.append(line)
+                
+        status = mkvmerge.wait()
+        if status == 1:
+            logger.warning('Extraction returns warning')
+            for w in warnings:
+                logger.warning(w)
+        elif status == 2:
+            logger.error('Extraction returns errors')
     
     
 def main():
@@ -1431,30 +1495,35 @@ def main():
     # We need to check the end also
     checks.append(pos)
     
+    finalCutName = '%s-cut.mkv' % basename
+    
     nbMKVParts = len(mkvparts)
     if nbMKVParts > 1:
         logger.info('Merging: %s' % mkvparts)
-        mergeMKVs(inputs=mkvparts, outputName=args.outputFile)
+        mergeMKVs(inputs=mkvparts, outputName=finalCutName)
     elif nbMKVParts == 1:
-        copyfile('part-1.mkv', args.outputFile)
+        copyfile('part-1.mkv', finalCutName)
     else:
         logger.info("Nothing else to do.")
         
+    try:
+        finalCut = open(finalCutName, mode='r')
+    except IOError:
+        logger.error("Impossible to open %s to finalize processing." % finalCutName)
+        exit(-1)
+  
     if args.srt:
         if not allOptionalTools:
             logger.warning("Missing tools for extracting subtitles.")
+            move(finalCutName, args.outputFile)
         else:
-            try:
-                final = open(args.outputFile, mode='r')
-            except IOError:
-                logger.error("Impossible to open %s to finalize processing." % args.outputFile)
-                exit(-1)
-            
-            duration = getMovieDuration(paths['ffprobe'], final)
+            # Final cut is not any more the final step.
+            temporaries.append(finalCut)
+            duration = getMovieDuration(paths['ffprobe'], finalCut)
             supportedLangs = getTesseractSupportedLang(paths['tesseract'])
             logger.info('Supported lang: %s' % supportedLangs)
             logger.info('Find subtitles tracks and language.')
-            subtitles = findSubtitlesTracks(paths['ffprobe'], final)
+            subtitles = findSubtitlesTracks(paths['ffprobe'], finalCut)
             logger.info(subtitles)
             sts = {}
             for subtitle in subtitles:
@@ -1472,7 +1541,7 @@ def main():
                     logger.error("Dropping subtitle: %s because it is missing language indication")
             
             logger.debug(sts)
-            listOfSubtitles = extractSRT(paths['mkvextract'], args.outputFile, sts, supportedLangs)
+            listOfSubtitles = extractSRT(paths['mkvextract'], finalCutName, sts, supportedLangs)
             logger.info(listOfSubtitles)
             for idxName, subName, _, _ in listOfSubtitles:
                 try:
@@ -1492,6 +1561,10 @@ def main():
             ocr = doOCR(paths['vobsubocr'], listOfSubtitles, duration, temporaries, args.dump)
             logger.info(ocr)
     
+            # Remux SRT subtitles
+            remuxSRTSubtitles(paths['mkvmerge'], finalCut, args.outputFile, ocr)
+    else:
+        move(finalCutName, args.outputFile)
   
     if not args.keep:
         logger.info("Cleaning temporary files")