Improvement of how we determine the initial timestamp.

README.md

@@ -1,2 +1,9 @@
-# panasonic
+# Panasonic SMART TV download recordings
 
+This project allows to download recordings from a Panasonic smart TV (model ...).
+This TV model is DLNA compatible. To download a recordings it is possible to request
+the TV to stream a MPEG-TS file through a HTTP get request.
+
+# Usage
+
+# Progress bar

panasonic.py

@@ -150,12 +150,15 @@ def dumpFilms(films):
         logger.info('%d - %s enregistré le %s sur %s. Durée: %s. %s' % (fid, film['title'], film['date'], film['channel'], film['duration'], film['url']))
         fid = fid + 1
 
-def getInitialTSFromBuffer(content):
+def getInitialTSFromBuffer(content, delta=0.1):
     logger = logging.getLogger(__name__)
 
-    with Popen(['ffprobe','/dev/stdin','-loglevel', 'quiet', '-read_intervals', '0%+0.000001', '-show_entries', 'frame=frame_no,best_effort_timestamp_time', '-of','json'], stdout=PIPE, stdin=PIPE) as ffprobe:
-        out, _ = ffprobe.communicate(input=content)
+    logger.debug('Determining initial timestamp from content')
+    length = len(content)
+    with Popen(['ffprobe','/dev/stdin','-loglevel', 'quiet', '-read_intervals', ('0%%+%f' % delta), '-show_entries', 'frame=frame_no,best_effort_timestamp_time', '-of','json'], stdout=PIPE, stdin=PIPE) as ffprobe:
+        out, _ = ffprobe.communicate(input=content[0:min(length,200000)])
         frames = json.load(BytesIO(out))
+        logger.debug('Frames: %s' % frames)
         if 'frames' in frames:
             frames = frames['frames']
             if len(frames) > 0:
@@ -173,10 +176,12 @@ def getInitialTSFromBuffer(content):
 def getLastTSFromBuffer(content):
     logger = logging.getLogger(__name__)
     
+    logger.debug('Determining last timestamp from content.')
     length = len(content)
     with Popen(['ffprobe', '/dev/stdin', '-loglevel', 'quiet', '-show_entries', 'frame=frame_no,best_effort_timestamp_time', '-of','json'], stdout=PIPE, stdin=PIPE) as ffprobe:
-        out, _ = ffprobe.communicate(input=content[max(0,length-100000):])
+        out, _ = ffprobe.communicate(input=content[max(0,length-200000):])
         frames = json.load(BytesIO(out))
+        logger.debug('Frames: %s' % frames)
         if 'frames' in frames:
             frames = frames['frames']
             if len(frames) > 0:
@@ -220,7 +225,7 @@ def getLastTSFromFD(fd):
         else:
             return None
 
-def downloadMovie(url, outputFileName, duration):
+def downloadMovie(url, outputFileName):
     logger = logging.getLogger(__name__)
     
     try:
@@ -230,10 +235,11 @@ def downloadMovie(url, outputFileName, duration):
         exit(-1)
         
     outputFD = output.fileno()
+    session = requests.Session()
     
     sample = 2.
     headers = {'TimeSeekRange.dlna.org': 'npt=0.000-%.03f' % sample}
-    r = requests.get(url, headers=headers)
+    r = session.get(url, headers=headers)
     if r.status_code != 200:
         logger.error('Impossible to download first %f seconds: %d' % (sample,r.status_code))
         exit(-1)
@@ -242,7 +248,7 @@ def downloadMovie(url, outputFileName, duration):
         logger.error('TimeSeekRange.dlna.org is not in header: %s' % r.headers)
         exit(-1)
     
-    p = re.compile('^.*/(?P<totalsecs>[0-9]+)\.(?P<totalms>[0-9]+)$')
+    p = re.compile(r'^.*/(?P<totalsecs>[0-9]+)\.(?P<totalms>[0-9]+)$')
     m = p.match(r.headers['TimeSeekRange.dlna.org'])
     if m == None:
         logger.error('Impossible to parse timestamps' % r.headers['TimeSeekRange.dlna.org'])
@@ -250,9 +256,20 @@ def downloadMovie(url, outputFileName, duration):
   
     total = int(m.group('totalsecs'))+(int(m.group('totalms')))/1000
     total = timedelta(seconds=total)
+    logger.debug('Header: %s' % r.headers)
 
-    initialTS = getInitialTSFromBuffer(r.content)
+    initialTS = None
+    nbRetries = 0
+    maxRetries = 10
+    while initialTS == None and nbRetries < maxRetries: 
+        initialTS = getInitialTSFromBuffer(r.content, delta=0.1*nbRetries)
+        nbRetries = nbRetries+1
+    logger.debug('Initial timestamp: %s' % initialTS)
+    if initialTS == None:
+        logger.error('Impossible to retrieve initial timestamp')
+        exit(-1)
     lastTS = getLastTSFromBuffer(r.content)
+    logger.debug('Last timestamp: %s' % lastTS)
     delta = lastTS-initialTS    
     ratio = delta/total
     estimatedLength = ceil(len(r.content)/ratio)
@@ -262,17 +279,19 @@ def downloadMovie(url, outputFileName, duration):
     pb = tqdm(total=estimatedLength, unit='bytes', desc='Download', unit_scale=True, unit_divisor=1024)
 
     nbiters = 0
-    with requests.get(url, stream=True) as r:
+    
+    with session.get(url, stream=True) as r:
         chunks = r.iter_content(chunk_size = 100000)
         for chunk in chunks:
             output.write(chunk)
             if nbiters == 1000:
                 lastTS = getLastTSFromFD(outputFD)
-                delta = lastTS-initialTS    
-                ratio = delta/total
-                length = fstat(outputFD).st_size
-                estimatedLength = ceil(length/ratio)
-                pb.total = estimatedLength
+                if lastTS != None:
+                    delta = lastTS-initialTS    
+                    ratio = delta/total
+                    length = fstat(outputFD).st_size
+                    estimatedLength = ceil(length/ratio)
+                    pb.total = estimatedLength
                 lseek(outputFD, 0, SEEK_END)
                 nbiters = 0
             pb.update(len(chunk))
@@ -350,7 +369,7 @@ def main():
         t = timedelta(seconds=duration)
         
         logger.info('Downloading movie "%s" whose duration is %s' % (title, t))
-        downloadMovie(url=films[mid]['url'], outputFileName=args.output, duration=duration)
+        downloadMovie(url=films[mid]['url'], outputFileName=args.output)
 
 
 if __name__ == "__main__":