def FileInputStream(filename, real_filename=None, **args):
"""
Create an input stream of a file. filename must be unicode.
real_filename is an optional argument used to specify the real filename,
its type can be 'str' or 'unicode'. Use real_filename when you are
not able to convert filename to real unicode string (ie. you have to
use unicode(name, 'replace') or unicode(name, 'ignore')).
"""
assert isinstance(filename, unicode)
if not real_filename:
real_filename = filename
try:
inputio = FileOpener(real_filename, 'rb')
except IOError as err:
charset = getTerminalCharset()
errmsg = unicode(str(err), charset)
raise InputStreamError(
_("Unable to open file %s: %s") % (filename, errmsg))
source = "file:" + filename
offset = args.pop("offset", 0)
size = args.pop("size", None)
if offset or size:
if size:
size = 8 * size
stream = InputIOStream(inputio, source=source, **args)
return InputSubStream(stream, 8 * offset, size, **args)
else:
args.setdefault("tags", []).append(("filename", filename))
return InputIOStream(inputio, source=source, **args)
def get_creation_date_from_metadata(media_file):
with open(media_file) as f:
f.seek(0)
parser = guessParser(InputIOStream(f, None, tags=[]))
metadata = extractMetadata(parser)
return time.mktime(metadata.get('creation_date').timetuple())
def createInputStream(cis, source=None, **args):
if field._parent:
stream = cis(source=source)
args.setdefault("tags", []).extend(stream.tags)
else:
stream = field.stream
input = CompressedStream(stream, decompressor)
if source is None:
source = "Compressed source: '%s' (offset=%s)" % (
stream.source, field.absolute_address)
return InputIOStream(input, source=source, **args)
def metadata_for_filelike(filelike):
try:
filelike.seek(0)
except (AttributeError, IOError):
return None
stream = InputIOStream(filelike, None, tags=[])
parser = guessParser(stream)
if not parser:
return None
try:
metadata = extractMetadata(parser)
except HachoirError:
return None
return metadata._Metadata__data
def extract_metadata(self, file):
config.MAX_STR_LENGTH = float("inf")
try:
filename = file.name
if not isinstance(filename, unicode):
filename = unicodeFilename(filename)
stream = InputIOStream(file,
source="file:%s" % filename,
tags=[],
filename=filename)
parser = guessParser(stream)
return extractMetadata(parser)
except (HachoirError, TypeError) as e:
raise MetadataException(e)
def META_OLECF(s, buff):
META_DICT = { }
try:
stream = InputIOStream(StringIO(buff))
parser = guessParser(stream)
meta = extractMetadata(parser)
except:
return META_DICT
for data in sorted(meta):
if data.values:
if len(data.values) == 1:
META_DICT['%s' % data.key] = data.values[0].text
else:
values = []
for value in data.values:
values.append(value.text)
META_DICT['%s' % data.key] = values
return META_DICT
not able to convert filename to real unicode string (ie. you have to
use unicode(name, 'replace') or unicode(name, 'ignore')).
"""
assert isinstance(filename, unicode)
if not real_filename:
real_filename = filename
try:
inputio = FileOpener(real_filename, 'rb')
except IOError, err:
charset = getTerminalCharset()
errmsg = unicode(str(err), charset)
raise InputStreamError(
_("Unable to open file %s: %s") % (filename, errmsg))
source = "file:" + filename
offset = args.pop("offset", 0)
size = args.pop("size", None)
if offset or size:
if size:
size = 8 * size
stream = InputIOStream(inputio, source=source, **args)
return InputSubStream(stream, 8 * offset, size, **args)
else:
args.setdefault("tags", []).append(("filename", filename))
return InputIOStream(inputio, source=source, **args)
def guessStreamCharset(stream, address, size, default=None):
size = min(size, 1024 * 8)
bytes = stream.readBytes(address, size // 8)
return guessBytesCharset(bytes, default)
def parse_raw(raw):
stream = InputIOStream(raw)
parser = guessParser(stream)
return extract_metadata(parser)