spaCy/spacy/lang/el/get_pos_from_wiktionary.py

65 lines
2.0 KiB
Python

def get_pos_from_wiktionary():
import re
from gensim.corpora.wikicorpus import extract_pages
regex = re.compile(r"==={{(\w+)\|el}}===")
regex2 = re.compile(r"==={{(\w+ \w+)\|el}}===")
# get words based on the Wiktionary dump
# check only for specific parts
# ==={{κύριο όνομα|el}}===
expected_parts = [
"μετοχή",
"ρήμα",
"επίθετο",
"επίρρημα",
"ουσιαστικό",
"κύριο όνομα",
"άρθρο",
]
wiktionary_file_path = (
"/data/gsoc2018-spacy/spacy/lang/el/res/elwiktionary-latest-pages-articles.xml"
)
proper_names_dict = {
"ουσιαστικό": "nouns",
"επίθετο": "adjectives",
"άρθρο": "dets",
"επίρρημα": "adverbs",
"κύριο όνομα": "proper_names",
"μετοχή": "participles",
"ρήμα": "verbs",
}
expected_parts_dict = {}
for expected_part in expected_parts:
expected_parts_dict[expected_part] = []
for title, text, pageid in extract_pages(wiktionary_file_path):
if text.startswith("#REDIRECT"):
continue
title = title.lower()
all_regex = regex.findall(text)
all_regex.extend(regex2.findall(text))
for a in all_regex:
if a in expected_parts:
expected_parts_dict[a].append(title)
for i in expected_parts_dict:
with open("_{0}.py".format(proper_names_dict[i]), "w") as f:
f.write("from __future__ import unicode_literals\n")
f.write('{} = set("""\n'.format(proper_names_dict[i].upper()))
words = sorted(expected_parts_dict[i])
line = ""
to_write = []
for word in words:
if len(line + " " + word) > 79:
to_write.append(line)
line = ""
else:
line = line + " " + word
f.write("\n".join(to_write))
f.write('\n""".split())')