def parseSTRF(self, t, strh):
fccType = strh['fccType']
retval = {}
if fccType == 'auds':
(
retval['wFormatTag'],
retval['nChannels'],
retval['nSamplesPerSec'],
retval['nAvgBytesPerSec'],
retval['nBlockAlign'],
retval['nBitsPerSample'],
) = struct.unpack('<HHHHHH', t[0:12])
ai = mediainfo.AudioInfo()
ai.samplerate = retval['nSamplesPerSec']
ai.channels = retval['nChannels']
ai.samplebits = retval['nBitsPerSample']
ai.bitrate = retval['nAvgBytesPerSec'] * 8
# TODO: set code if possible
# http://www.stats.uwa.edu.au/Internal/Specs/DXALL/FileSpec/Languages
# ai.language = strh['wLanguage']
try:
ai.codec = fourcc.RIFFWAVE[retval['wFormatTag']]
except:
ai.codec = "Unknown"
self.audio.append(ai)
elif fccType == 'vids':
v = struct.unpack('<IIIHH', t[0:16])
(
retval['biSize'],
retval['biWidth'],
retval['biHeight'],
retval['biPlanes'],
retval['biBitCount'],
) = v
retval['fourcc'] = t[16:20]
v = struct.unpack('IIIII', t[20:40])
(
retval['biSizeImage'],
retval['biXPelsPerMeter'],
retval['biYPelsPerMeter'],
retval['biClrUsed'],
retval['biClrImportant'],
) = v
vi = mediainfo.VideoInfo()
try:
vi.codec = fourcc.RIFFCODEC[t[16:20]]
except:
vi.codec = "Unknown"
vi.width = retval['biWidth']
vi.height = retval['biHeight']
vi.bitrate = strh['dwRate']
vi.fps = round(float(strh['dwRate'] * 100) / strh['dwScale']) / 100
vi.length = strh['dwLength'] / vi.fps
self.video.append(vi)
return retval
def _read_header(self, object_id, s):
if object_id == 'PROP':
prop = struct.unpack('>9IHH', s)
_print(prop)
if object_id == 'MDPR':
mdpr = struct.unpack('>H7I', s[:30])
_print(mdpr)
self.length = mdpr[7] / 1000
(stream_name_size, ) = struct.unpack('>B', s[30:31])
stream_name = s[31:31 + stream_name_size]
pos = 31 + stream_name_size
(mime_type_size, ) = struct.unpack('>B', s[pos:pos + 1])
mime = s[pos + 1:pos + 1 + mime_type_size]
pos += mime_type_size + 1
(type_specific_len, ) = struct.unpack('>I', s[pos:pos + 4])
type_specific = s[pos + 4:pos + 4 + type_specific_len]
pos += 4 + type_specific_len
if mime[:5] == 'audio':
ai = mediainfo.AudioInfo()
ai.id = mdpr[0]
ai.bitrate = mdpr[2]
self.audio.append(ai)
elif mime[:5] == 'video':
vi = mediainfo.VideoInfo()
vi.id = mdpr[0]
vi.bitrate = mdpr[2]
self.video.append(vi)
else:
_print("Unknown: %s" % mime)
if object_id == 'CONT':
pos = 0
(title_len, ) = struct.unpack('>H', s[pos:pos + 2])
self.title = s[2:title_len + 2]
pos += title_len + 2
(author_len, ) = struct.unpack('>H', s[pos:pos + 2])
self.artist = s[pos + 2:pos + author_len + 2]
pos += author_len + 2
(copyright_len, ) = struct.unpack('>H', s[pos:pos + 2])
self.copyright = s[pos + 2:pos + copyright_len + 2]
pos += copyright_len + 2
(comment_len, ) = struct.unpack('>H', s[pos:pos + 2])
self.comment = s[pos + 2:pos + comment_len + 2]
def _readatom(self, file):
s = file.read(8)
if len(s) < 8:
return 0
atomsize,atomtype = struct.unpack('>I4s', s)
if not str(atomtype).decode('latin1').isalnum():
# stop at nonsense data
return 0
if mediainfo.DEBUG or ATOM_DEBUG:
print "%s [%X]" % (atomtype,atomsize)
if atomtype == 'udta':
# Userdata (Metadata)
pos = 0
tabl = {}
i18ntabl = {}
atomdata = file.read(atomsize-8)
while pos < atomsize-12:
(datasize,datatype) = struct.unpack('>I4s', atomdata[pos:pos+8])
if ord(datatype[0]) == 169:
# i18n Metadata...
mypos = 8+pos
while mypos < datasize+pos:
# first 4 Bytes are i18n header
(tlen,lang) = struct.unpack('>HH', atomdata[mypos:mypos+4])
i18ntabl[lang] = i18ntabl.get(lang, {})
i18ntabl[lang][datatype[1:]] = atomdata[mypos+4:mypos+tlen+4]
mypos += tlen+4
elif datatype == 'WLOC':
# Drop Window Location
pass
else:
if ord(atomdata[pos+8:pos+datasize][0]) > 1:
tabl[datatype] = atomdata[pos+8:pos+datasize]
pos += datasize
if len(i18ntabl.keys()) > 0:
for k in i18ntabl.keys():
if QTLANGUAGES.has_key(k):
self.appendtable('QTUDTA', i18ntabl[k], QTLANGUAGES[k])
self.appendtable('QTUDTA', tabl, QTLANGUAGES[k])
else:
#print "NO i18"
self.appendtable('QTUDTA', tabl)
elif atomtype == 'trak':
atomdata = file.read(atomsize-8)
pos = 0
vi = None
ai = None
info = None
while pos < atomsize-8:
(datasize,datatype) = struct.unpack('>I4s', atomdata[pos:pos+8])
if datatype == 'tkhd':
tkhd = struct.unpack('>6I8x4H36xII', atomdata[pos+8:pos+datasize])
vi = mediainfo.VideoInfo()
vi.width = tkhd[10] >> 16
vi.height = tkhd[11] >> 16
vi.id = tkhd[3]
ai = mediainfo.AudioInfo()
ai.id = tkhd[3]
try:
# XXX Date number of Seconds is since January 1st 1904!!!
# XXX 2082844800 is the difference between Unix and Apple time
# XXX Fix me to work on Apple, too
self.date = int(tkhd[1]) - 2082844800
self.date = time.strftime('%y/%m/%d', time.gmtime(self.date))
except Exception, e:
print 'ex', e
elif datatype == 'mdia':
pos += 8
datasize -= 8
if ATOM_DEBUG:
print '--> mdia information'
while datasize:
mdia = struct.unpack('>I4s', atomdata[pos:pos+8])
if mdia[1] == 'mdhd':
mdhd = struct.unpack('>IIIIIhh', atomdata[pos+8:pos+8+24])
# duration / time scale
if vi:
vi.length = mdhd[4] / mdhd[3]
if ai:
ai.length = mdhd[4] / mdhd[3]
if mdhd[5] in QTLANGUAGES:
ai.language = QTLANGUAGES[mdhd[5]]
# mdhd[6] == quality
self.length = max(self.length, mdhd[4] / mdhd[3])
elif mdia[1] == 'minf':
# minf has only atoms inside
pos -= (mdia[0] - 8)
datasize += (mdia[0] - 8)
elif mdia[1] == 'stbl':
# stbl has only atoms inside
pos -= (mdia[0] - 8)
datasize += (mdia[0] - 8)
elif mdia[1] == 'hdlr':
hdlr = struct.unpack('>I4s4s', atomdata[pos+8:pos+8+12])
if hdlr[1] == 'mhlr':
if hdlr[2] == 'vide' and not vi in self.video:
self.video.append(vi)
info = vi
if hdlr[2] == 'soun' and not ai in self.audio:
self.audio.append(ai)
info = ai
elif mdia[1] == 'stsd':
stsd = struct.unpack('>2I', atomdata[pos+8:pos+8+8])
if stsd[1] > 0 and info:
codec = struct.unpack('>I4s', atomdata[pos+16:pos+16+8])
info.codec = codec[1]
if info.codec == 'jpeg':
# jpeg is no video, remove it from the list
self.video.remove(vi)
info = None
elif mdia[1] == 'dinf':
dref = struct.unpack('>I4s', atomdata[pos+8:pos+8+8])
if ATOM_DEBUG:
print ' --> %s, %s' % mdia
print ' --> %s, %s (reference)' % dref
elif ATOM_DEBUG:
if mdia[1].startswith('st'):
print ' --> %s, %s (sample)' % mdia
elif mdia[1] in ('vmhd', 'smhd'):
print ' --> %s, %s (media information header)' % mdia
else:
print ' --> %s, %s (unknown)' % mdia
pos += mdia[0]
datasize -= mdia[0]
elif datatype == 'udta' and ATOM_DEBUG:
print struct.unpack('>I4s', atomdata[:8])
elif ATOM_DEBUG:
if datatype == 'edts':
print "--> %s [%d] (edit list)" % (datatype, datasize)
else:
print "--> %s [%d] (unknown)" % (datatype, datasize)
pos += datasize
def ReadPESHeader(self, offset, buffer, id=0):
"""
Parse a PES header.
Since it starts with 0x00 0x00 0x01 like 'normal' mpegs, this
function will return (0, -1) when it is no PES header or
(packet length, timestamp position (maybe -1))
http://dvd.sourceforge.net/dvdinfo/pes-hdr.html
"""
if not buffer[0:3] == '\x00\x00\x01':
return 0, -1
packet_length = (ord(buffer[4]) << 8) + ord(buffer[5]) + 6
align = ord(buffer[6]) & 4
header_length = ord(buffer[8])
# PES ID (starting with 001)
if ord(buffer[3]) & 0xE0 == 0xC0:
id = id or ord(buffer[3]) & 0x1F
for a in self.audio:
if a.id == id:
break
else:
self.audio.append(mediainfo.AudioInfo())
self.audio[-1].id = id
self.audio[-1].keys.append('id')
elif ord(buffer[3]) & 0xF0 == 0xE0:
id = id or ord(buffer[3]) & 0xF
for v in self.video:
if v.id == id:
break
else:
self.video.append(mediainfo.VideoInfo())
self.video[-1].id = id
self.video[-1].keys.append('id')
# new mpeg starting
if buffer[header_length+9:header_length+13] == \
'\x00\x00\x01\xB3' and not self.sequence_header_offset:
# yes, remember offset for later use
self.sequence_header_offset = offset + header_length + 9
elif ord(buffer[3]) == 189 or ord(buffer[3]) == 191:
# private stream. we don't know, but maybe we can guess later
id = id or ord(buffer[3]) & 0xF
if align and buffer[header_length + 9:header_length +
11] == '\x0b\x77':
# AC3 stream
for a in self.audio:
if a.id == id:
break
else:
self.audio.append(mediainfo.AudioInfo())
self.audio[-1].id = id
self.audio[-1].codec = 'AC3'
self.audio[-1].keys.append('id')
else:
# unknown content
pass
ptsdts = ord(buffer[7]) >> 6
if ptsdts and ptsdts == ord(buffer[9]) >> 4:
if ord(buffer[9]) >> 4 != ptsdts:
print 'WARNING: bad PTS/DTS, please contact us'
return packet_length, -1
# timestamp = self.ReadPTS(buffer[9:14])
high = ((ord(buffer[9]) & 0xF) >> 1)
med = (ord(buffer[10]) << 7) + (ord(buffer[11]) >> 1)
low = (ord(buffer[12]) << 7) + (ord(buffer[13]) >> 1)
return packet_length, 9
return packet_length, -1
def ReadHeader(self, buffer, offset):
"""
Handle MPEG header in buffer on position offset
Return -1 on error, new offset or 0 if the new offset can't be scanned
"""
if buffer[offset:offset + 3] != '\x00\x00\x01':
return -1
id = ord(buffer[offset + 3])
if id == PADDING_PKT:
return offset + (ord(buffer[offset + 4]) << 8) + ord(
buffer[offset + 5]) + 6
if id == PACK_PKT:
if ord(buffer[offset + 4]) & 0xF0 == 0x20:
self.type = 'MPEG1 video'
self.get_time = self.ReadSCRMpeg1
return offset + 12
elif (ord(buffer[offset + 4]) & 0xC0) == 0x40:
self.type = 'MPEG2 video'
self.get_time = self.ReadSCRMpeg2
return offset + (ord(buffer[offset + 13]) & 0x07) + 14
else:
# WTF? Very strange
return -1
if 0xC0 <= id <= 0xDF:
# code for audio stream
for a in self.audio:
if a.id == id:
break
else:
self.audio.append(mediainfo.AudioInfo())
self.audio[-1].id = id
self.audio[-1].keys.append('id')
return 0
if 0xE0 <= id <= 0xEF:
# code for video stream
for v in self.video:
if v.id == id:
break
else:
self.video.append(mediainfo.VideoInfo())
self.video[-1].id = id
self.video[-1].keys.append('id')
return 0
if id == SEQ_HEAD:
# sequence header, remember that position for later use
self.sequence_header_offset = offset
return 0
if id in (PRIVATE_STREAM1, PRIVATE_STREAM2):
# private stream. we don't know, but maybe we can guess later
add = ord(buffer[offset + 8])
# if (ord(buffer[offset+6]) & 4) or 1:
# id = ord(buffer[offset+10+add])
if buffer[offset + 11 + add:offset + 15 +
add].find('\x0b\x77') != -1:
# AC3 stream
for a in self.audio:
if a.id == id:
break
else:
self.audio.append(mediainfo.AudioInfo())
self.audio[-1].id = id
self.audio[-1].codec = 'AC3'
self.audio[-1].keys.append('id')
return 0
if id == SYS_PKT:
return 0
if id == EXT_START:
return 0
return 0
def _getnextheader(self, s):
r = struct.unpack('<16sQ', s[:24])
(guidstr, objsize) = r
guid = self._parseguid(guidstr)
if guid == GUIDS['ASF_File_Properties_Object']:
_print("File Properties Object")
val = struct.unpack('<16s6Q4I', s[24:24 + 80])
(fileid, size, date, packetcount, duration, \
senddur, preroll, flags, minpack, maxpack, maxbr) = \
val
self.length = duration / 10000000
elif guid == GUIDS['ASF_Stream_Properties_Object']:
_print("Stream Properties Object [%d]" % objsize)
streamtype = self._parseguid(s[24:40])
errortype = self._parseguid(s[40:56])
offset, typelen, errorlen, flags = struct.unpack(
'>QIIH4x', s[56:78])
strno = flags & 63
encrypted = flags >> 15
if streamtype == GUIDS['ASF_Video_Media']:
vi = mediainfo.VideoInfo()
#vi.width, vi.height, formatsize = struct.unpack('<IIxH', s[78:89])
vi.width, vi.height, depth, codec, = struct.unpack(
'<4xII2xH4s', s[89:89 + 20])
vi.codec = fourcc.RIFFCODEC[codec]
vi.id = strno
self.video.append(vi)
elif streamtype == GUIDS['ASF_Audio_Media']:
ai = mediainfo.AudioInfo()
twocc, ai.channels, ai.samplerate, bitrate, block, ai.samplebits, = struct.unpack(
'<HHIIHH', s[78:78 + 16])
ai.bitrate = 8 * bitrate # XXX Is this right?
ai.codec = fourcc.RIFFWAVE[twocc]
ai.id = strno
self.audio.append(ai)
pass
elif guid == GUIDS['ASF_Header_Extension_Object']:
_print("ASF_Header_Extension_Object %d" % objsize)
size = struct.unpack('<I', s[42:46])[0]
data = s[46:46 + size]
while len(data):
_print("Sub:")
h = self._getnextheader(data)
data = data[h[1]:]
elif guid == GUIDS['ASF_Codec_List_Object']:
_print("List Object")
pass
elif guid == GUIDS['ASF_Error_Correction_Object']:
_print("Error Correction")
pass
elif guid == GUIDS['ASF_Content_Description_Object']:
_print("Content Description Object")
val = struct.unpack('<5H', s[24:24 + 10])
pos = 34
strings = []
for i in val:
strings.append(s[pos:pos + i].replace('\0',
'').lstrip().rstrip())
pos += i
(self.title, self.artist, self.copyright, self.caption,
rating) = tuple(strings)
elif guid == GUIDS['ASF_Extended_Content_Description_Object']:
(count, ) = struct.unpack('<H', s[24:26])
pos = 26
descriptor = {}
for i in range(0, count):
# Read additional content descriptors
d = self._parsekv(s[pos:])
pos += d[0]
descriptor[d[1]] = d[2]
self.appendtable('ASFDESCRIPTOR', descriptor)
elif guid == GUIDS['ASF_Metadata_Object']:
(count, ) = struct.unpack('<H', s[24:26])
pos = 26
descriptor = {}
for i in range(0, count):
# Read additional content descriptors
d = self._parsekv2(s[pos:])
pos += d[0]
descriptor[d[1]] = d[2]
# TODO: Find the stream in self.audio and self.video and
# append it there instead of here
self.appendtable('ASFMETADATA%d' % d[3], descriptor)
elif guid == GUIDS['ASF_Language_List_Object']:
count = struct.unpack('<H', s[24:26])[0]
pos = 26
lang = []
for i in range(0, count):
idlen = struct.unpack('<B', s[pos:pos + 1])[0]
idstring = s[pos + 1:pos + 1 + idlen]
_print("Language: %d/%d: %s" % (i + 1, count, idstring))
lang.append(idstring)
pos += 1 + idlen
if len(lang) == 1:
self.language = lang[0]
else:
self.language = tuple(lang)
# TODO: Find the stream in self.audio and self.video and
# set it there instead of here
elif guid == GUIDS['ASF_Stream_Bitrate_Properties_Object']:
(count, ) = struct.unpack('<H', s[24:26])
pos = 26
for i in range(0, count):
strno, avbitrate = struct.unpack('<HI', s[pos:pos + 6])
strno &= 63
_print("Stream %d Bitrate: %d" % (strno, avbitrate))
# TODO: Find the stream in self.audio and self.video and
# set it there instead of here
else:
# Just print the type:
bfail = 1
for h in GUIDS.keys():
if GUIDS[h] == guid:
_print("Unparsed %s [%d]" % (h, objsize))
bfail = 0
if bfail:
_print("unknown: %s [%d]" % (self._printguid(guid), objsize))
return r
def _parseHeader(self, header, granule):
headerlen = len(header)
flags = ord(header[0])
if headerlen >= 30 and header[1:7] == 'vorbis':
#print("Vorbis Audio Header")
ai = mediainfo.AudioInfo()
ai.version, ai.channels, ai.samplerate, bitrate_max, ai.bitrate, \
bitrate_min, blocksize, framing = \
struct.unpack('<IBIiiiBB',header[7:7+23])
ai.codec = 'Vorbis'
#ai.granule = granule
#ai.length = granule / ai.samplerate
self.audio.append(ai)
self.all_streams.append(ai)
elif headerlen >= 7 and header[1:7] == 'theora':
#print "Theora Header"
# Theora Header
# XXX Finish Me
vi = mediainfo.VideoInfo()
vi.codec = 'theora'
self.video.append(vi)
self.all_streams.append(vi)
elif headerlen >= 142 and header[
1:36] == 'Direct Show Samples embedded in Ogg':
#print 'Direct Show Samples embedded in Ogg'
# Old Directshow format
# XXX Finish Me
vi = mediainfo.VideoInfo()
vi.codec = 'dshow'
self.video.append(vi)
self.all_streams.append(vi)
elif flags & PACKET_TYPE_BITS == PACKET_TYPE_HEADER and headerlen >= struct.calcsize(
STREAM_HEADER_VIDEO) + 1:
#print "New Directshow Format"
# New Directshow Format
htype = header[1:9]
if htype[:5] == 'video':
streamheader = struct.unpack(
STREAM_HEADER_VIDEO,
header[9:struct.calcsize(STREAM_HEADER_VIDEO) + 9])
vi = mediainfo.VideoInfo()
(type, ssize, timeunit, samplerate, vi.length, buffersize, \
vi.bitrate, vi.width, vi.height) = streamheader
vi.width /= 65536
vi.height /= 65536
# XXX length, bitrate are very wrong
try:
vi.codec = fourcc.RIFFCODEC[type]
except:
vi.codec = 'Unknown (%s)' % type
vi.fps = 10000000 / timeunit
self.video.append(vi)
self.all_streams.append(vi)
elif htype[:5] == 'audio':
streamheader = struct.unpack(
STREAM_HEADER_AUDIO,
header[9:struct.calcsize(STREAM_HEADER_AUDIO) + 9])
ai = mediainfo.AudioInfo()
(type, ssize, timeunit, ai.samplerate, ai.length, buffersize,
ai.bitrate, ai.channels, bloc, ai.bitrate) = streamheader
self.samplerate = ai.samplerate
_print("Samplerate %d" % self.samplerate)
self.audio.append(ai)
self.all_streams.append(ai)
elif htype[:4] == 'text':
subtitle = mediainfo.MediaInfo()
subtitle.keys.append('language')
subtitle.type = 'subtitle'
subtitle.length = 0
self.all_streams.append(subtitle)
else:
_print("Unknown Header")
def process_one_track(self, track):
# Process all the items at the track level
tabelem = self.process_one_level(track)
# We have the dict of track eleme, now build the MMPYTHON information
type = tabelem[MATROSKA_TRACK_TYPE_ID]
mytype = type.get_value()
_print("Track type found with UID %d" % mytype)
if (mytype == MATROSKA_VIDEO_TRACK):
_print("VIDEO TRACK found !!")
#VIDEOCORE = ['length', 'encoder', 'bitrate', 'samplerate', 'codec', 'samplebits',
# 'width', 'height', 'fps', 'aspect']
vi = mediainfo.VideoInfo()
try:
elem = tabelem[MATROSKA_CODEC_ID]
vi.codec = elem.get_data()
except:
vi.codec = 'Unknown'
try:
elem = tabelem[MATROSKA_FRAME_DURATION_ID]
vi.fps = 1 / (pow(10, -9) * (elem.get_value()))
except:
vi.fps = 0
try:
vinfo = tabelem[MATROSKA_VIDEO_SETTINGS_ID]
vidtab = self.process_one_level(vinfo)
vi.width = vidtab[MATROSKA_VID_WIDTH_ID].get_value()
vi.height = vidtab[MATROSKA_VID_HEIGHT_ID].get_value()
except:
_print("No other info about video track !!!")
self.video.append(vi)
elif (mytype == MATROSKA_AUDIO_TRACK):
_print("AUDIO TRACK found !!")
#AUDIOCORE = ['channels', 'samplerate', 'length', 'encoder', 'codec', 'samplebits',
# 'bitrate', 'language']
ai = mediainfo.AudioInfo()
try:
elem = tabelem[MATROSKA_TRACK_LANGUAGE_ID]
ai.language = elem.get_data()
ai['language'] = elem.get_data()
except:
ai.language = 'en'
ai['language'] = 'en'
try:
elem = tabelem[MATROSKA_CODEC_ID]
ai.codec = elem.get_data()
except:
ai.codec = "Unknown"
try:
ainfo = tabelem[MATROSKA_AUDIO_SETTINGS_ID]
audtab = self.process_one_level(vinfo)
ai.samplerate = unpack(
'!f', audtab[MATROSKA_AUDIO_SAMPLERATE_ID].get_value())[0]
ai.channels = audtab[MATROSKA_AUDIO_CHANNELS_ID].get_value()
except:
_print("No other info about audio track !!!")
self.audio.append(ai)
elif (mytype == MATROSKA_SUBTITLES_TRACK):
try:
elem = tabelem[MATROSKA_TRACK_LANGUAGE_ID]
language = elem.get_data()
_print("Subtitle language found : %s" % elem.get_data())
except:
language = "en" # By default
self.subtitles.append(language)