-
Notifications
You must be signed in to change notification settings - Fork 0
/
tool.py
194 lines (177 loc) · 6.42 KB
/
tool.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
import numpy as np
import pandas as pd
import re
from tqdm.notebook import tqdm
def extract_playlist_from(data, dtype="str", dense=True, verbose=True, limit=None):
'''
parameters \\
data : train, val, test \\
dtype : "str"(default) or "list" \\
dense : bool; True(default) \\
verbose : bool; True(default) \\
limit : int or None(default); iteration limit \\
returns \\
dictionaries; ko_title, ko_tag, en_title, en_tag
'''
if dtype == "str":
ko = re.compile("[^ ㄱ-ㅣ가-힣]+")
en = re.compile("[^ a-zA-Z]+")
elif dtype == "list":
ko = re.compile("[ㄱ-ㅣ가-힣]+")
en = re.compile("[a-zA-Z]+")
ko_title, ko_tag = {}, {}
en_title, en_tag = {}, {}
if limit:
_range = tqdm(range(limit)) if verbose else range(limit)
else:
_range = tqdm(data.index) if verbose else data.index
for i in _range:
title = data.plylst_title[i]
tag = " ".join(data.tags[i])
if dtype == "str":
a, b = ko.sub(' ', title).strip(), ko.sub(' ', tag).strip()
c, d = en.sub(' ', title).strip(), en.sub(' ', tag).strip()
elif dtype == "list":
a, b = ko.findall(title), ko.findall(tag)
c, d = en.findall(title), en.findall(tag)
if dense:
if a: ko_title[i] = a
if b: ko_tag[i] = b
if c: en_title[i] = c
if d: en_tag[i] = d
else:
ko_title[i] = a
ko_tag[i] = b
en_title[i] = c
en_tag[i] = d
return ko_title, ko_tag, en_title, en_tag
def extract_song_from(data, dtype="str", dense=True, verbose=True, limit=None):
'''
parameters \\
data : song \\
dtype: "str" or "list" \\
dense : bool; True(default) \\
verbose : bool; True(default) \\
limit : int or None(default); iteration limit \\
returns \\
dictionaries; ko_name, ko_album, ko_artist, en_name, en_album, en_artist
'''
if dtype == "str":
ko = re.compile("[^ ㄱ-ㅣ가-힣]+")
en = re.compile("[^ a-zA-Z]+")
elif dtype == "list":
ko = re.compile("[ㄱ-ㅣ가-힣]+")
en = re.compile("[a-zA-Z]+")
ko_name, ko_album, ko_artist = {}, {}, {}
en_name, en_album, en_artist = {}, {}, {}
if limit:
_range = tqdm(range(limit)) if verbose else range(limit)
else:
_range = tqdm(data.index) if verbose else data.index
for i in _range:
name = data.song_name[i]
album = data.album_name[i]
artist = " ".join(data.artist_name_basket[i])
if album == None: album = "" # 143209 album name == None
if dtype == "str":
a, b, c = ko.sub(' ', name).strip(), ko.sub(' ', album).strip(), ko.sub(' ', artist).strip()
d, e, f = en.sub(' ', name).strip(), en.sub(' ', album).strip(), en.sub(' ', artist).strip()
elif dtype == "list":
a, b, c = ko.findall(name), ko.findall(album), ko.findall(artist)
d, e, f = en.findall(name), en.findall(album), en.findall(artist)
if dense:
if a: ko_name[i] = a
if b: ko_album[i] = b
if c: ko_artist[i] = c
if d: en_name[i] = d
if e: en_album[i] = e
if f: en_artist[i] = f
else:
ko_name[i] = a
ko_album[i] = b
ko_artist[i] = c
en_name[i] = d
en_album[i] = e
en_artist[i] = f
return ko_name, ko_album, ko_artist, en_name, en_album, en_artist
def ko_extract_tag_from(data, api, tag="NNG", verbose=True, limit=None):
'''
parameters \\
data : dictionary \\
api : khaiii.KhaiiiApi (tokenizer) \\
tag : str or list of str \\
verbose : True or False \\
limit: int or None(default); iteration limit \\
returns \\
dictionary
'''
extracted = {}
if limit:
_items = tqdm(list(data.items())[:limit]) if verbose else list(data.items())[:limit]
else:
_items = tqdm(data.items()) if verbose else data.items()
for key, val in _items:
tokens = []
for word in api.analyze(val):
for morph in word.morphs:
if morph.tag in tag:
tokens.append(morph.lex)
if tokens != []:
extracted[key] = tokens
return extracted
def en_extract_tag_from(data, api, tag, verbose=True, limit=None):
pass
def flatten_dict():
pass
def we_numbering(data, start=0, verbose=True, limit=None, return_inverse=False):
'''
data : list of dictionaries
start : int; start numbering index. default is 0.
return_inverse : bool; return dictionary idx -> word
'''
idx = start
words = {}
if return_inverse:
inverse = {}
for i, _dict in enumerate(data):
print(f"{i+1}/{len(data)}") if verbose else None
if limit:
_items = list(_dict.items())[:limit]
else:
_items = tqdm(_dict.items()) if verbose else _dict.items()
for _, vals in _items:
for val in vals:
if not(val in words):
words[val] = idx
if return_inverse:
inverse[idx] = val
idx += 1
if return_inverse:
return words, inverse
else:
return words
def n_gram(n, data):
dct = {}
for key, title in data.items():
bag = []
for pos in range(len(title) - n + 1):
bag.append(title[pos: pos + n])
dct[key] = bag
return dct
def n_gram_sentence(n, sentence):
bag = []
for pos in range(len(sentence) - n + 1):
bag.append(sentence[pos: pos + n])
return bag
if __name__=="__main__":
# train = pd.read_json("res/train.json")[["id", "plylst_title", "songs", "tags", "like_cnt", "updt_date"]]
# a, b, c, d = extract_playlist_from(train, dtype="str", dense=True, limit=100)
# print(a, b, c, d)
# song = pd.read_json("res/song_meta.json")[["id", "song_name", "artist_id_basket", "artist_name_basket",\
# "album_id", "album_name", "song_gn_gnr_basket",\
# "song_gn_dtl_gnr_basket", "issue_date"]]
# a, b, c, d, e, f = extract_song_from(song, limit=100)
# print(a, b, c, d, e, f)
data = [{1:[10,11], 2:[20, 22], 3:[30, 33], -1:[-10, -12]}]
words = we_numbering(data, limit=10)
print(words)