def _process_meta(self, elem):
"""
Process the <meta> element of an LGR XML file.
"""
metadata = Metadata(self.rfc7940_checks)
reference_manager = ReferenceManager()
MAPPER = {
DATE_TAG:
lambda d: metadata.set_date(d, force=self.force_mode),
VALIDITY_START_TAG:
lambda d: metadata.set_validity_start(d, force=self.force_mode),
VALIDITY_END_TAG:
lambda d: metadata.set_validity_end(d, force=self.force_mode),
UNICODE_VERSION_TAG:
lambda d: metadata.set_unicode_version(d, force=self.force_mode),
}
unicode_version_tag_found = False
for child in elem:
tag = child.tag
logger.debug("Got '%s' element", tag)
if tag in MAPPER:
MAPPER[tag](child.text)
if tag == UNICODE_VERSION_TAG:
unicode_version_tag_found = True
elif tag == VERSION_TAG:
metadata.version = Version(child.text,
child.get('comment', None))
elif tag == LANGUAGE_TAG:
metadata.add_language(child.text, force=self.force_mode)
elif tag == SCOPE_TAG:
metadata.scopes.append(
Scope(child.text, child.get('type', None)))
elif tag == DESCRIPTION_TAG:
# Seems to be an issue with CDATA/iterparse: https://bugs.launchpad.net/lxml/+bug/1788449
# For now, manually replace CRLF with LF
metadata.description = Description(
child.text.replace('\r\n', '\n'), child.get('type', None))
elif tag == REFERENCES_TAG:
for reference in child:
value = reference.text
# Don't convert it to an int since ref_id may be a string
ref_id = reference.get('id')
comment = reference.get('comment', None)
reference_manager.add_reference(value,
comment=comment,
ref_id=ref_id)
# Since we have processed <reference> elements here, let's clean-up
child.clear()
else:
logger.warning("Unhandled '%s' element in <meta> section", tag)
self.rfc7940_checks.error('parse_xml')
child.clear()
self.rfc7940_checks.add_test_result('explicit_unicode_version',
unicode_version_tag_found)
self._lgr = LGR(name=self.filename,
metadata=metadata,
reference_manager=reference_manager,
unicode_database=self._unicode_database)
def union_reference_manager(first, second):
"""
Union two reference managers.
:param first: The first object to union.
:param second: The other object to union.
:return: A new reference manager object.
"""
first_ref = {r['value'] for r in first.values()}
second_ref = {r['value'] for r in second.values()}
output = ReferenceManager()
for ref in set.union(first_ref, second_ref):
output.add_reference(ref)
return output
def setUp(self):
self.manager = ReferenceManager()
class TestReferenceManager(unittest.TestCase):
def setUp(self):
self.manager = ReferenceManager()
def test_add_reference(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1',
comment='A pretty old version')
self.assertEqual('0', ref_id)
self.assertIn('0', self.manager)
reference = self.manager[ref_id]
self.assertIsInstance(reference, dict)
self.assertEqual('The Unicode Standard 1.1', reference['value'])
self.assertEqual('A pretty old version', reference['comment'])
def test_add_existing_reference(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1',
comment='A pretty old version')
self.assertEqual(ref_id, '0')
self.assertRaises(ReferenceAlreadyExists,
self.manager.add_reference,
'The Unicode Standard 6.3',
ref_id=ref_id)
def test_add_existing_reference_other_type(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1',
comment='A pretty old version')
self.assertEqual(ref_id, '0')
self.assertRaises(ReferenceAlreadyExists,
self.manager.add_reference,
'The Unicode Standard 6.3',
ref_id=0)
def test_del_existing_reference(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1',
comment='A pretty old version')
self.manager.del_reference(ref_id)
self.assertEqual(0, len(self.manager))
def test_del_inexisting_reference(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1',
comment='A pretty old version')
self.assertEqual(ref_id, '0')
self.assertRaises(ReferenceNotDefined, self.manager.del_reference, 1)
self.assertRaises(ReferenceNotDefined, self.manager.del_reference, '1')
def test_update_reference(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1')
self.manager.update_reference(ref_id, value='The Unicode Standard 1.2')
self.assertEqual(self.manager[ref_id]['value'],
'The Unicode Standard 1.2')
self.assertNotIn('comment', self.manager[ref_id])
self.manager.update_reference(ref_id, comment='A recent version')
self.assertEqual(self.manager[ref_id]['value'],
'The Unicode Standard 1.2')
self.assertEqual(self.manager[ref_id]['comment'], 'A recent version')
def test_update_inexsting_reference(self):
ref_id = self.manager.add_reference('The Unicode Standard 1.1',
comment='A pretty old version')
self.assertEqual(ref_id, '0')
self.assertRaises(ReferenceNotDefined, self.manager.update_reference,
1, **{'value': 'The Unicode Standard 1.2'})