1
0
mirror of https://github.com/moparisthebest/SickRage synced 2024-11-17 23:05:11 -05:00
SickRage/lib/enzyme/real.py

119 lines
4.4 KiB
Python
Raw Normal View History

# -*- coding: utf-8 -*-
# enzyme - Video metadata parser
# Copyright 2011-2012 Antoine Bertin <diaoulael@gmail.com>
# Copyright 2003-2006 Thomas Schueppel <stain@acm.org>
# Copyright 2003-2006 Dirk Meyer <dischi@freevo.org>
#
# This file is part of enzyme.
#
# enzyme is free software; you can redistribute it and/or modify it under
# the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 3 of the License, or
# (at your option) any later version.
#
# enzyme is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with enzyme. If not, see <http://www.gnu.org/licenses/>.
__all__ = ['Parser']
import struct
import logging
from exceptions import ParseError
import core
# http://www.pcisys.net/~melanson/codecs/rmff.htm
# http://www.pcisys.net/~melanson/codecs/
# get logging object
log = logging.getLogger(__name__)
class RealVideo(core.AVContainer):
def __init__(self, file):
core.AVContainer.__init__(self)
self.mime = 'video/real'
self.type = 'Real Video'
h = file.read(10)
try:
(object_id, object_size, object_version) = struct.unpack('>4sIH', h)
except struct.error:
# EOF.
raise ParseError()
if not object_id == '.RMF':
raise ParseError()
file_version, num_headers = struct.unpack('>II', file.read(8))
log.debug(u'size: %d, ver: %d, headers: %d' % \
(object_size, file_version, num_headers))
for _ in range(0, num_headers):
try:
oi = struct.unpack('>4sIH', file.read(10))
except (struct.error, IOError):
# Header data we expected wasn't there. File may be
# only partially complete.
break
if object_id == 'DATA' and oi[0] != 'INDX':
log.debug(u'INDX chunk expected after DATA but not found -- file corrupt')
break
(object_id, object_size, object_version) = oi
if object_id == 'DATA':
# Seek over the data chunk rather than reading it in.
file.seek(object_size - 10, 1)
else:
self._read_header(object_id, file.read(object_size - 10))
log.debug(u'%r [%d]' % (object_id, object_size - 10))
# Read all the following headers
def _read_header(self, object_id, s):
if object_id == 'PROP':
prop = struct.unpack('>9IHH', s)
log.debug(u'PROP: %r' % prop)
if object_id == 'MDPR':
mdpr = struct.unpack('>H7I', s[:30])
log.debug(u'MDPR: %r' % mdpr)
self.length = mdpr[7] / 1000.0
(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 = core.AudioStream()
ai.id = mdpr[0]
ai.bitrate = mdpr[2]
self.audio.append(ai)
elif mime[:5] == 'video':
vi = core.VideoStream()
vi.id = mdpr[0]
vi.bitrate = mdpr[2]
self.video.append(vi)
else:
log.debug(u'Unknown: %r' % 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]
Parser = RealVideo