def load_id(id, dir, set):
path = txt_path(id, dir, set)
text = load_text(path)
path = spans_path(id, dir, set)
lines = load_lines(path)
spans = list(parse_spans(lines))
path = objects_path(id, dir, set)
lines = load_lines(path)
objects = list(parse_objects(lines, spans))
return FactruMarkup(id, text, objects)
def load_id(id, dir):
path = txt_path(id, dir)
text = load_text(path)
path = ann_path(id, dir)
lines = load_lines(path)
spans = list(parse_spans(lines))
return Ne5Markup(id, text, spans)
def load_id(id, dir, set):
path = part_path(id, dir, set, TXT)
text = load_text(path)
path = part_path(id, dir, set, SPANS)
lines = load_lines(path)
spans = list(parse_spans(lines))
path = part_path(id, dir, set, OBJECTS)
lines = load_lines(path)
objects = list(parse_objects(lines, spans))
path = part_path(id, dir, set, COREF)
lines = load_lines(path)
corefs = list(parse_corefs(lines, objects))
path = part_path(id, dir, set, FACTS)
lines = load_lines(path)
facts = list(parse_facts(lines, corefs, spans))
return FactruMarkup(id, text, objects, corefs, facts)
def load_rudrec(path):
lines = load_lines(path)
items = parse_jsonl(lines)
return parse_rudrec(items)
def load_annotated(records):
for record in records:
lines = load_lines(record.path)
yield parse_annotated(record.name, lines)
def load_ud(path):
lines = load_lines(path)
return parse_ud(lines)
def load_id(id, dir):
path = join_path(dir, '%s.txt.iob' % id)
lines = load_lines(path)
return parse_conll(lines)
def load_russe(path):
lines = load_lines(path)
return parse_russe(lines)
def load_toloka_lrwc(path):
lines = load_lines(path)
return parse_toloka_lrwc(lines)
def load_ruadrect(path):
lines = load_lines(path)
return parse_ruadrect(lines)
def load_simlex(path):
lines = load_lines(path)
return parse_simlex(lines)
def load_mokoron(path):
for line in load_lines(path):
if line.startswith(INSERT):
for match in RECORD.finditer(line):
yield MokoronRecord.from_match(match)
def load_morphoru_corpora(path):
lines = load_lines(path)
return parse_morphoru(lines, parse_morphoru_corpora_sent)
def load_morphoru_rnc(path):
lines = load_lines(path)
return parse_morphoru_rnc(lines)
def load_morphoru_gicrya(path):
lines = load_lines(path)
return parse_morphoru(lines, parse_morphoru_gicrya_sent)