mirror of
https://github.com/SickGear/SickGear.git
synced 2024-12-05 02:43:37 +00:00
0d9fbc1ad7
This version of SickBeard uses both TVDB and TVRage to search and gather it's series data from allowing you to now have access to and download shows that you couldn't before because of being locked into only what TheTVDB had to offer. Also this edition is based off the code we used in our XEM editon so it does come with scene numbering support as well as all the other features our XEM edition has to offer. Please before using this with your existing database (sickbeard.db) please make a backup copy of it and delete any other database files such as cache.db and failed.db if present, we HIGHLY recommend starting out with no database files at all to make this a fresh start but the choice is at your own risk! Enjoy!
131 lines
4.3 KiB
Python
131 lines
4.3 KiB
Python
"""
|
|
Parsers for the different streams and fragments found in an OLE2 file.
|
|
|
|
Documents:
|
|
- goffice source code
|
|
|
|
Author: Robert Xiao, Victor Stinner
|
|
Creation: 2006-04-23
|
|
"""
|
|
|
|
from lib.hachoir_parser import HachoirParser
|
|
from lib.hachoir_core.field import FieldSet, RootSeekableFieldSet, RawBytes
|
|
from lib.hachoir_core.endian import LITTLE_ENDIAN
|
|
from lib.hachoir_core.stream import StringInputStream
|
|
from lib.hachoir_parser.misc.msoffice_summary import SummaryFieldSet, CompObj
|
|
from lib.hachoir_parser.misc.word_doc import WordDocumentFieldSet
|
|
|
|
PROPERTY_NAME = {
|
|
u"\5DocumentSummaryInformation": "doc_summary",
|
|
u"\5SummaryInformation": "summary",
|
|
u"WordDocument": "word_doc",
|
|
}
|
|
|
|
class OfficeRootEntry(HachoirParser, RootSeekableFieldSet):
|
|
PARSER_TAGS = {
|
|
"description": "Microsoft Office document subfragments",
|
|
}
|
|
endian = LITTLE_ENDIAN
|
|
|
|
def __init__(self, stream, **args):
|
|
RootSeekableFieldSet.__init__(self, None, "root", stream, None, stream.askSize(self))
|
|
HachoirParser.__init__(self, stream, **args)
|
|
|
|
def validate(self):
|
|
return True
|
|
|
|
def createFields(self):
|
|
for index, property in enumerate(self.ole2.properties):
|
|
if index == 0:
|
|
continue
|
|
try:
|
|
name = PROPERTY_NAME[property["name"].value]
|
|
except LookupError:
|
|
name = property.name+"content"
|
|
for field in self.parseProperty(index, property, name):
|
|
yield field
|
|
|
|
def parseProperty(self, property_index, property, name_prefix):
|
|
ole2 = self.ole2
|
|
if not property["size"].value:
|
|
return
|
|
if property["size"].value >= ole2["header/threshold"].value:
|
|
return
|
|
name = "%s[]" % name_prefix
|
|
first = None
|
|
previous = None
|
|
size = 0
|
|
start = property["start"].value
|
|
chain = ole2.getChain(start, True)
|
|
blocksize = ole2.ss_size
|
|
desc_format = "Small blocks %s..%s (%s)"
|
|
while True:
|
|
try:
|
|
block = chain.next()
|
|
contiguous = False
|
|
if not first:
|
|
first = block
|
|
contiguous = True
|
|
if previous and block == (previous+1):
|
|
contiguous = True
|
|
if contiguous:
|
|
previous = block
|
|
size += blocksize
|
|
continue
|
|
except StopIteration:
|
|
block = None
|
|
self.seekSBlock(first)
|
|
desc = desc_format % (first, previous, previous-first+1)
|
|
size = min(size, property["size"].value*8)
|
|
if name_prefix in ("summary", "doc_summary"):
|
|
yield SummaryFieldSet(self, name, desc, size=size)
|
|
elif name_prefix == "word_doc":
|
|
yield WordDocumentFieldSet(self, name, desc, size=size)
|
|
elif property_index == 1:
|
|
yield CompObj(self, "comp_obj", desc, size=size)
|
|
else:
|
|
yield RawBytes(self, name, size//8, desc)
|
|
if block is None:
|
|
break
|
|
first = block
|
|
previous = block
|
|
size = ole2.sector_size
|
|
|
|
def seekSBlock(self, block):
|
|
self.seekBit(block * self.ole2.ss_size)
|
|
|
|
class FragmentGroup:
|
|
def __init__(self, parser):
|
|
self.items = []
|
|
self.parser = parser
|
|
|
|
def add(self, item):
|
|
self.items.append(item)
|
|
|
|
def createInputStream(self):
|
|
# FIXME: Use lazy stream creation
|
|
data = []
|
|
for item in self.items:
|
|
data.append( item["rawdata"].value )
|
|
data = "".join(data)
|
|
|
|
# FIXME: Use smarter code to send arguments
|
|
args = {"ole2": self.items[0].root}
|
|
tags = {"class": self.parser, "args": args}
|
|
tags = tags.iteritems()
|
|
return StringInputStream(data, "<fragment group>", tags=tags)
|
|
|
|
class CustomFragment(FieldSet):
|
|
def __init__(self, parent, name, size, parser, description=None, group=None):
|
|
FieldSet.__init__(self, parent, name, description, size=size)
|
|
if not group:
|
|
group = FragmentGroup(parser)
|
|
self.group = group
|
|
self.group.add(self)
|
|
|
|
def createFields(self):
|
|
yield RawBytes(self, "rawdata", self.size//8)
|
|
|
|
def _createInputStream(self, **args):
|
|
return self.group.createInputStream()
|
|
|