def mangleData(self, data, index):
self.sha1_offset = 208
self.md5_offset = 256
self.header_offset = 360
self.filedata_offset = 3170
data = MangleFile.mangleData(self, data, index)
if USE_HACHOIR:
#data.tofile(open('/tmp/oops', 'wb'))
hachoir_config.quiet = True
data_str = data.tostring()
parser = guessParser(StringInputStream(data_str))
if parser:
self.useHachoirParser(parser)
summary_data = data[self.header_offset:].tostring()
checksum = md5(summary_data).digest()
data[self.md5_offset:self.md5_offset + 16] = array('B', checksum)
summary_data = data[self.header_offset:self.filedata_offset].tostring()
checksum = sha(summary_data).hexdigest()
data[self.sha1_offset:self.sha1_offset + 40] = array('B', checksum)
return data
def __init__(self, project, *args, **kw):
MangleFile.__init__(self, project, args[0], int(kw['nb_file']))
self.hard_max_op = 10000
self.hard_min_op = 0
self.aggressivity = None
self.fixed_size_factor = 1.0
if kw.has_key('ext'):
if kw['ext'] in ('.xml', '.svg', '.rdf'):
if kw.has_key('nofile') and kw['nofile']:
self.xml = MangleXML(nofile=True, xmltype=MangleXML.SVG11)
else:
self.xml = MangleXML(xmltype=MangleXML.SVG11)
elif kw['ext'] == ".html":
if kw.has_key('nofile'):
self.xml = MangleXML(nofile=kw['nofile'], xmltype=MangleXML.XHTML1)
else:
self.xml = MangleXML(xmltype=MangleXML.XHTML1)
else:
self.xml = None
def __init__(self, project, *args, **kw):
MangleFile.__init__(self, project, args[0], int(kw['nb_file']))
self.hard_max_op = 10000
self.hard_min_op = 0
self.aggressivity = None
self.fixed_size_factor = 1.0
if kw.has_key('ext'):
if kw['ext'] in ('.xml', '.svg', '.rdf'):
if kw.has_key('nofile') and kw['nofile']:
self.xml = MangleXML(nofile=True, xmltype=MangleXML.SVG11)
else:
self.xml = MangleXML(xmltype=MangleXML.SVG11)
elif kw['ext'] == ".html":
if kw.has_key('nofile'):
self.xml = MangleXML(nofile=kw['nofile'],
xmltype=MangleXML.XHTML1)
else:
self.xml = MangleXML(xmltype=MangleXML.XHTML1)
else:
self.xml = None
def setupProject(project):
USE_STDOUT = True
time = ProcessTimeWatch(
project,
too_slow=3.0,
too_slow_score=0.10,
too_fast=0.100,
too_fast_score=-0.80,
)
orig_filename = project.application().getInputFilename("PDF document")
if AUTO_MANGLE:
mangle = AutoMangle(project, orig_filename)
mangle.hard_max_op = 1000
else:
mangle = MangleFile(project, orig_filename)
mangle.config.max_op = 1000
options = {'timeout': 5.0}
if not USE_STDOUT:
options['stdout'] = 'null'
process = PopplerProcess(project, ['pdftotext'], **options)
WatchProcess(process, exitcode_score=-0.10)
if USE_STDOUT:
stdout = WatchStdout(process)
def cleanupLine(line):
match = re.match(r"Error(?: \([0-9]+\))?: (.*)", line)
if match:
line = match.group(1)
return line
stdout.cleanup_func = cleanupLine
del stdout.words['unknown']
# stdout.show_not_matching = True
# stdout.ignoreRegex(r"Unknown operator 'allocate'$")
# stdout.ignoreRegex(r" operator is wrong type \(error\)$")
# stdout.ignoreRegex(r'^No current point in lineto$')
# stdout.ignoreRegex(r'^No current point in lineto')
# stdout.ignoreRegex(r'^Unknown operator ')
# stdout.ignoreRegex(r"^Couldn't open 'nameToUnicode' file ")
# stdout.ignoreRegex(r"^Illegal character ")
# stdout.ignoreRegex(r"^No font in show$")
# stdout.ignoreRegex(r"^Element of show/space array must be number or string$")
# stdout.ignoreRegex(r"^No current point in curveto$")
# stdout.ignoreRegex(r"^Badly formatted number$")
# stdout.ignoreRegex(r"^Dictionary key must be a name object$")
# stdout.ignoreRegex(r"^End of file inside array$")
# stdout.ignoreRegex(r"^Too few \([0-9]+\) args to .* operator$")
# stdout.ignoreRegex(r"Too many args in content stream")
stdout.max_nb_line = (100, 0.20)
def mangleData(self, data, file_index):
self.setupConf(data)
if self.xml:
return array('B', self.xml.mangleData(data.tostring()))
return MangleFile.mangleData(self, data, file_index)
def mangleData(self, data, file_index):
self.setupConf(data)
if self.xml:
return array('B', self.xml.mangleData(data.tostring()))
return MangleFile.mangleData(self, data, file_index)
def __init__(self, project, *args, **kw):
MangleFile.__init__(self, project, *args, **kw)
self.hard_max_op = 10000
self.hard_min_op = 0
self.aggressivity = None
self.fixed_size_factor = 1.0
def mangleData(self, data, file_index):
self.setupConf(data)
return MangleFile.mangleData(self, data, file_index)