def infer(self):
noad_map = {}
dst = Distiller(dictName='ode')
dst.load_distilled_file()
for e in dst.entries:
noad_id = None
for p in e.wordclass_blocks:
if p.complement is not None:
noad_id = p.complement
if noad_id is not None:
noad_map[e.lexid] = noad_id
tree = etree.parse(self.in_file)
entries = tree.findall('./link')
for e in entries:
e.attrib.pop('targetHref')
oed_id = e.get('sourceID')
lexid = e.get('targetID')
if lexid in noad_map:
e.set('targetID', noad_map[lexid])
else:
e.getparent().remove(e)
tree.getroot().set('type', 'noad')
with open(self.out_file, 'w') as filehandle:
filehandle.write(etree.tostring(tree,
pretty_print=True,
encoding='unicode'))
def __init__(self, **kwargs):
self.dict_name = kwargs.get('dictName')
self.oed_in = kwargs.get('oedIn', None)
self.oed_out = kwargs.get('oedOut', None)
self.odo_in = kwargs.get('odoIn', None)
self.odo_out = kwargs.get('odoOut', None)
self.oed_index = VitalStatisticsCache()
self.odo_index = Distiller(dictName=self.dict_name)
self.odo_index.load_distilled_file()
def parse_distilled_file(self):
distilled = Distiller(dictName=self.dict_name)
distilled.load_distilled_file()
for entry in distilled.entries:
self.entry_content[entry.lexid] = entry
class LinkUpdater(object):
error_message = '!ERROR entry not found'
def __init__(self, **kwargs):
self.dict_name = kwargs.get('dictName')
self.oed_in = kwargs.get('oedIn', None)
self.oed_out = kwargs.get('oedOut', None)
self.odo_in = kwargs.get('odoIn', None)
self.odo_out = kwargs.get('odoOut', None)
self.oed_index = VitalStatisticsCache()
self.odo_index = Distiller(dictName=self.dict_name)
self.odo_index.load_distilled_file()
def update_odo(self, **kwargs):
valid_links_only = kwargs.get('validLinksOnly', False)
tree = etree.parse(self.odo_in)
for entry in tree.findall('./e'):
lexid = entry.get('lexid', None)
odo_label = entry.find('./label')
odo_label_text = self.odo_index.headword_by_id(lexid) or LinkUpdater.error_message
etree.strip_tags(odo_label, 'i', 'sup', 'sub', 'hm')
odo_label.text = odo_label_text
link = entry.find('./linkSet/link')
if link is not None:
refentry = link.get('refentry', '0')
refid = link.get('refid', '0')
oed_label_text = self.oed_index.find(refentry, field='label') or LinkUpdater.error_message
etree.strip_tags(link, 'i', 'sup', 'sub', 'hm')
link.text = oed_label_text
if (valid_links_only and
(link is None or
link.text == LinkUpdater.error_message or
odo_label.text == LinkUpdater.error_message or
not check_match(link.text, odo_label.text))):
entry.getparent().remove(entry)
with open(self.odo_out, 'w') as filehandle:
filehandle.write(etree.tostring(tree,
pretty_print=True,
encoding='unicode'))
def update_oed(self, **kwargs):
valid_links_only = kwargs.get('validLinksOnly', False)
tree = etree.parse(self.oed_in)
for entry in tree.findall('./link'):
oed_id = entry.get('sourceID', None)
oed_label_text = self.oed_index.find(oed_id, field='label') or LinkUpdater.error_message
source_label = entry.find('./sourceLabel')
etree.strip_tags(source_label, 'i', 'sup', 'sub', 'hm')
source_label.text = oed_label_text
lexid = entry.get('targetID', None)
ode_label_text = self.odo_index.headword_by_id(lexid) or LinkUpdater.error_message
target_label = entry.find('./targetLabel')
etree.strip_tags(target_label, 'i', 'sup', 'sub', 'hm')
target_label.text = ode_label_text
if (valid_links_only and
(oed_id is None or
lexid is None or
source_label.text == LinkUpdater.error_message or
target_label.text == LinkUpdater.error_message or
not check_match(source_label.text, target_label.text))):
entry.getparent().remove(entry)
with open(self.oed_out, 'w') as filehandle:
filehandle.write(etree.tostring(tree,
pretty_print=True,
encoding='unicode'))
