def gzip_handler(self, entity):
""" Handles .gz files """
self.log.debug("gzip_handler(%s)" % entity)
entity[u'format'] = u'gz'
p = subprocess.Popen('gunzip "%s"' % entity[u'filename'],
shell=True, stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
cwd=dirname)
out, err = p.communicate()
if err:
self.log.error("Error unzipping: " + err)
return
magic = utils.get_magic(extracted)
self.call_magic_handler(extracted, child)
def call_magic_handler(self, filename, entity):
""" Determine the file magic, and call the appropriate handler """
magic = utils.get_magic(filename)
if magic in self.magic_types:
self.log.info('Calling %r for %s magic' % (
self.magic_types[magic], magic))
self.magic_types[magic](self, entity)
else:
# Try to match any regex magic patterns
for pattern in self.magic_types:
if isinstance(pattern, type(re.compile(r'foo'))):
if re.match(pattern, magic):
self.log.info('Calling %r for %s magic' % (
self.magic_types[pattern], magic))
self.magic_types[pattern](self, entity)
break
else:
self.log.error('No handler for magic: %s' % magic)
return magic
def zip_exe_handler(self, entity):
""" Handles self-extracting zip files """
self.log.debug("zip_exe_handler(%s)" % entity)
entity[u'format'] = u'zip'
dirname = os.path.dirname(entity[u'filename'])
p = subprocess.Popen('unzip -o "%s"' % entity[u'filename'],
shell=True, stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
cwd=dirname)
out, err = p.communicate()
if err:
self.log.error("Error unzipping: " + err)
else:
# Delete compressed data after extracting
os.unlink(entity[u'filename'])
for line in out.split('\n'):
if line.strip().startswith('inflating'):
extracted = os.path.join(dirname, line.strip().split()[-1])
self.log.debug("extracted " + extracted)
magic = utils.get_magic(extracted)
# Create a new child Entity for each extracted file
extracted = to_unicode(extracted)
child = Entity.by_name(extracted)
if not child:
child = Entity(name=os.path.basename(extracted))
child[u'filename'] = extracted
DBSession.add(child)
child.parent = entity
child[u'magic'] = to_unicode(magic)
self.log.debug("Created %s" % child)
else:
child.parent = entity
DBSession.flush()
self.call_magic_handler(extracted, child)