|
4 | 4 | import random |
5 | 5 | import re |
6 | 6 | import time |
| 7 | +from collections import defaultdict |
7 | 8 | from contextlib import closing |
8 | 9 | from functools import partial |
| 10 | +from itertools import islice |
9 | 11 | from multiprocessing import RawArray |
10 | 12 | from multiprocessing.pool import Pool |
11 | 13 | from typing import Any, Iterable, List, NamedTuple, Optional, Tuple |
12 | 14 |
|
13 | 15 | import cython |
14 | 16 | import joblib |
15 | 17 | import numpy as np |
16 | | -from collections import defaultdict |
17 | | -from itertools import islice |
18 | | -from marisa_trie import Trie, RecordTrie |
| 18 | +from marisa_trie import RecordTrie, Trie |
19 | 19 | from tqdm import tqdm |
20 | | -from cython.cimports.libcpp.random import mt19937 |
21 | | -from cython.cimports.scipy.linalg.cython_blas import saxpy, sdot |
| 20 | + |
| 21 | +from cython.cimports.libcpp.random cimport mt19937 |
| 22 | +from cython.cimports.scipy.linalg.cython_blas cimport saxpy, sdot |
22 | 23 |
|
23 | 24 | from .dictionary import Dictionary |
24 | 25 | from .dump_db import DumpDB |
25 | 26 | from .link_graph import LinkGraph |
26 | 27 | from .mention_db import MentionDB |
27 | 28 | from .utils.sentence_detector.base_sentence_detector import BaseSentenceDetector |
28 | 29 | from .utils.tokenizer.base_tokenizer import BaseTokenizer |
29 | | -from cython.cimports.wikipedia2vec.dictionary import Item, Word |
30 | | -from cython.cimports.wikipedia2vec.dump_db import Paragraph, WikiLink |
31 | | -from cython.cimports.wikipedia2vec.mention_db import Mention |
32 | | -from cython.cimports.wikipedia2vec.utils.tokenizer.token import Token |
33 | | -from cython.cimports.wikipedia2vec.utils.sentence_detector.sentence import Sentence |
| 30 | + |
| 31 | +from cython.cimports.wikipedia2vec.dictionary cimport Item, Word |
| 32 | +from cython.cimports.wikipedia2vec.dump_db cimport Paragraph, WikiLink |
| 33 | +from cython.cimports.wikipedia2vec.mention_db cimport Mention |
| 34 | +from cython.cimports.wikipedia2vec.utils.sentence_detector.sentence cimport Sentence |
| 35 | +from cython.cimports.wikipedia2vec.utils.tokenizer.token cimport Token |
34 | 36 |
|
35 | 37 | MAX_EXP = cython.declare(cython.float, 6.0) |
36 | 38 | EXP_TABLE_SIZE = cython.declare(cython.int, 1000) |
|
0 commit comments