def filter_format(self, format, regex=False):
"""
Generator including all archive entries with a given format.
Is regex=True, format will be compiled to a regular expression and
matched against the entry formats
"""
if not format:
raise KeyError('You need to provide an format')
if regex is True:
pattern = re.compile(format)
else:
pattern = None
# check format argument against spec
try:
utils.check_format(format)
except exceptions.CombineArchiveFormatException as e:
raise KeyError(
'{format} is no valid format, according to the OMEX specification. {cause}'
.format(format=format, cause=e.message))
for (location, entry) in self.entries.items():
if pattern is not None and pattern.match(entry.format):
yield entry
elif pattern is None and format == entry.format:
yield entry
def filter_format(self, format, regex=False):
"""
Generator including all archive entries with a given format.
Is regex=True, format will be compiled to a regular expression and
matched against the entry formats
"""
if not format:
raise KeyError('You need to provide an format')
if regex is True:
pattern = re.compile(format)
else:
pattern = None
# check format argument against spec
try:
utils.check_format(format)
except exceptions.CombineArchiveFormatException as e:
raise KeyError(
'{format} is no valid format, according to the OMEX specification. {cause}'.format(format=format,
cause=e.message))
for (location, entry) in self.entries.items():
if pattern is not None and pattern.match(entry.format):
yield entry
elif pattern is None and format == entry.format:
yield entry
def _read_manifest(self):
"""
internal function.
Reads the manifest file of a COMBINE Archive
"""
try:
with self._zip.open(self.MANIFEST_LOCATION) as manifest_file:
manifest = ElementTree.fromstring(manifest_file.read())
except KeyError:
# manifest does not exists, probably an empty/new archive
return False
except ElementTree.ParseError as e:
raise exceptions.CombineArchiveException(
'Cannot parse xml manifest. {}'.format(e.msg))
# check for correct root element and namespace
if manifest.tag != utils.extend_tag_name(_XML_ROOT_ELEM, _XML_NS):
raise exceptions.CombineArchiveException(
'manifest has no valid omex root element')
# check entries
for entry in manifest.findall(_XML_CONTENT_TAG, _XML_NS):
try:
location = utils.get_attribute(entry, _XML_CONTENT_LOCATION,
_XML_NS)
entry_format = utils.check_format(utils.get_attribute(
entry, _XML_CONTENT_FORMAT, _XML_NS),
convert=False)
master = True if entry.attrib.get(_XML_CONTENT_MASTER,
False) in ('True', 'true',
True) else False
except KeyError:
raise exceptions.CombineArchiveException(
'location and format field are required. Corrupt manifest.xml'
)
# clean location
location = utils.clean_pathname(location)
# check if file is in zip, if it's not the root element
zipinfo = None
if location not in self.ARCHIVE_REFERENCE:
try:
zipinfo = self._zip.getinfo(location)
except KeyError:
raise exceptions.CombineArchiveException(
'{location} is specified by the manifest, but not contained by the ZIP file'
.format(location=location))
archive_entry = ArchiveEntry(location,
format=entry_format,
master=master,
archive=self,
zipinfo=zipinfo)
self.entries[location] = archive_entry
def add_entry(self,
file,
format,
location=None,
master=False,
replace=False):
"""
adds a file-like object to the COMBINE archive and adds a manifest entry
if file is an instance of unicode or str, the content of this variable is written as content
Returns:
ArchiveEntry
"""
if not file or not format:
raise exceptions.CombineArchiveException(
'both a file and the corresponding format must be provided')
# check format schema
format = utils.check_format(format)
# no location provided. Guess it
if location is None or not location:
location = os.path.basename(file)
# clean location
location = utils.clean_pathname(location)
if location == self.MANIFEST_LOCATION or location in self.ARCHIVE_REFERENCE:
raise exceptions.CombineArchiveException(
'it is not allowed to name a file {loc}'.format(loc=location))
if location in self._zip.namelist():
if replace is False:
raise exceptions.CombineArchiveException(
'{loc} exists already in the COMBINE archive. set replace=True, to override it'
.format(loc=location))
else:
self.remove_entry(location)
# write file to zip
if isinstance(file, (str, unicode)):
# file is actually string
zipinfo = self._zip.writestr(location, file)
else:
zipinfo = self._zip.write(file, location)
entry = ArchiveEntry(location,
format=format,
master=master,
zipinfo=zipinfo,
archive=self)
self.entries[entry.location] = entry
return entry
def _write_manifest(self, zip_file=None):
"""
internal function.
Writes the manifest file of a COMBINE Archive
"""
if zip_file is None:
zip_file = self._zip
# create new DOM object
manifest = ElementTree.Element(
utils.extend_tag_name(_XML_ROOT_ELEM, _XML_NS))
# write first entry for archive itself
content = ElementTree.SubElement(
manifest, utils.extend_tag_name(_XML_CONTENT_TAG, _XML_NS))
content.attrib.update({
utils.extend_tag_name(_XML_CONTENT_LOCATION, _XML_NS):
'.',
utils.extend_tag_name(_XML_CONTENT_FORMAT, _XML_NS):
_XML_CONTENT_ARCHIVE_TYPE,
})
for (location, entry) in self.entries.items():
entry_format = utils.check_format(entry.format)
content = ElementTree.SubElement(
manifest, utils.extend_tag_name(_XML_CONTENT_TAG, _XML_NS))
content.attrib.update({
utils.extend_tag_name(_XML_CONTENT_LOCATION, _XML_NS):
location,
utils.extend_tag_name(_XML_CONTENT_FORMAT, _XML_NS):
entry_format,
})
if entry.master:
content.attrib[utils.extend_tag_name(_XML_CONTENT_MASTER,
_XML_NS)] = 'true'
# write xml to zip
io = StringIO()
ElementTree.ElementTree(manifest).write(io,
xml_declaration=True,
default_namespace=_XML_ROOT_NS)
try:
zip_file.remove(self.MANIFEST_LOCATION)
except KeyError:
pass # Manifest does not exist yet, so removing it will fail
zip_file.writestr(self.MANIFEST_LOCATION, io.getvalue())
io.close()
def _read_manifest(self):
"""
internal function.
Reads the manifest file of a COMBINE Archive
"""
try:
with self._zip.open(self.MANIFEST_LOCATION) as manifest_file:
manifest = ElementTree.fromstring(manifest_file.read())
except KeyError:
# manifest does not exists, probably an empty/new archive
return False
except ElementTree.ParseError as e:
raise exceptions.CombineArchiveException('Cannot parse xml manifest. {}'.format(e.msg))
# check for correct root element and namespace
if manifest.tag != utils.extend_tag_name(_XML_ROOT_ELEM, _XML_NS):
raise exceptions.CombineArchiveException('manifest has no valid omex root element')
# check entries
for entry in manifest.findall(_XML_CONTENT_TAG, _XML_NS):
try:
location = utils.get_attribute(entry, _XML_CONTENT_LOCATION, _XML_NS)
entry_format = utils.check_format(utils.get_attribute(entry, _XML_CONTENT_FORMAT, _XML_NS), convert=False)
master = True if entry.attrib.get(_XML_CONTENT_MASTER, False) in ('True', 'true', True) else False
except KeyError:
raise exceptions.CombineArchiveException('location and format field are required. Corrupt manifest.xml')
# clean location
location = utils.clean_pathname(location)
# check if file is in zip, if it's not the root element
zipinfo = None
if location not in self.ARCHIVE_REFERENCE:
try:
zipinfo = self._zip.getinfo(location)
except KeyError:
raise exceptions.CombineArchiveException(
'{location} is specified by the manifest, but not contained by the ZIP file'.format(location=location))
archive_entry = ArchiveEntry(location, format=entry_format, master=master, archive=self, zipinfo=zipinfo)
self.entries[location] = archive_entry
def add_entry(self, file, format, location=None, master=False, replace=False):
"""
adds a file-like object to the COMBINE archive and adds a manifest entry
if file is an instance of unicode or str, the content of this variable is written as content
Returns:
ArchiveEntry
"""
if not file or not format:
raise exceptions.CombineArchiveException('both a file and the corresponding format must be provided')
# check format schema
format = utils.check_format(format)
# no location provided. Guess it
if location is None or not location:
location = os.path.basename(file)
# clean location
location = utils.clean_pathname(location)
if location == self.MANIFEST_LOCATION or location in self.ARCHIVE_REFERENCE:
raise exceptions.CombineArchiveException('it is not allowed to name a file {loc}'.format(loc=location))
if location in self._zip.namelist():
if replace is False:
raise exceptions.CombineArchiveException('{loc} exists already in the COMBINE archive. set replace=True, to override it'.format(loc=location))
else:
self.remove_entry(location)
# write file to zip
if isinstance(file, (str, unicode)):
# file is actually string
zipinfo = self._zip.writestr(location, file)
else:
zipinfo = self._zip.write(file, location)
entry = ArchiveEntry(location, format=format, master=master, zipinfo=zipinfo, archive=self)
self.entries[entry.location] = entry
return entry
def _write_manifest(self, zip_file=None):
"""
internal function.
Writes the manifest file of a COMBINE Archive
"""
if zip_file is None:
zip_file = self._zip
# create new DOM object
manifest = ElementTree.Element(utils.extend_tag_name(_XML_ROOT_ELEM, _XML_NS))
# write first entry for archive itself
content = ElementTree.SubElement(manifest, utils.extend_tag_name(_XML_CONTENT_TAG, _XML_NS))
content.attrib.update({
utils.extend_tag_name(_XML_CONTENT_LOCATION, _XML_NS): '.',
utils.extend_tag_name(_XML_CONTENT_FORMAT, _XML_NS): _XML_CONTENT_ARCHIVE_TYPE,
})
for (location, entry) in self.entries.items():
entry_format = utils.check_format(entry.format)
content = ElementTree.SubElement(manifest, utils.extend_tag_name(_XML_CONTENT_TAG, _XML_NS))
content.attrib.update({
utils.extend_tag_name(_XML_CONTENT_LOCATION, _XML_NS): location,
utils.extend_tag_name(_XML_CONTENT_FORMAT, _XML_NS): entry_format,
})
if entry.master:
content.attrib[utils.extend_tag_name(_XML_CONTENT_MASTER, _XML_NS)] = 'true'
# write xml to zip
io = StringIO()
ElementTree.ElementTree(manifest).write(io, xml_declaration=True, default_namespace=_XML_ROOT_NS)
try:
zip_file.remove(self.MANIFEST_LOCATION)
except KeyError:
pass # Manifest does not exist yet, so removing it will fail
zip_file.writestr(self.MANIFEST_LOCATION, io.getvalue())
io.close()
