Spaces:
Running
Running
ORI-Muchim
commited on
Commit
•
49ebe40
1
Parent(s):
c1d9b48
Update text/cleaners.py
Browse files- text/cleaners.py +0 -170
text/cleaners.py
CHANGED
@@ -1,25 +1,5 @@
|
|
1 |
import re
|
2 |
-
from text.japanese import japanese_to_romaji_with_accent, japanese_to_ipa, japanese_to_ipa2, japanese_to_ipa3
|
3 |
from text.korean import latin_to_hangul, number_to_hangul, divide_hangul, korean_to_lazy_ipa, korean_to_ipa
|
4 |
-
from text.mandarin import number_to_chinese, chinese_to_bopomofo, latin_to_bopomofo, chinese_to_romaji, chinese_to_lazy_ipa, chinese_to_ipa, chinese_to_ipa2
|
5 |
-
#from text.sanskrit import devanagari_to_ipa
|
6 |
-
from text.english import english_to_lazy_ipa, english_to_ipa2, english_to_lazy_ipa2
|
7 |
-
#from text.thai import num_to_thai, latin_to_thai
|
8 |
-
#from text.shanghainese import shanghainese_to_ipa
|
9 |
-
#from text.cantonese import cantonese_to_ipa
|
10 |
-
#from text.ngu_dialect import ngu_dialect_to_ipa
|
11 |
-
|
12 |
-
|
13 |
-
def japanese_cleaners(text):
|
14 |
-
text = japanese_to_romaji_with_accent(text)
|
15 |
-
if re.match('[A-Za-z]', text[-1]):
|
16 |
-
text += '.'
|
17 |
-
return text
|
18 |
-
|
19 |
-
|
20 |
-
def japanese_cleaners2(text):
|
21 |
-
return japanese_cleaners(text).replace('ts', 'ʦ').replace('...', '…')
|
22 |
-
|
23 |
|
24 |
def korean_cleaners(text):
|
25 |
'''Pipeline for Korean text'''
|
@@ -29,153 +9,3 @@ def korean_cleaners(text):
|
|
29 |
if re.match('[\u3131-\u3163]', text[-1]):
|
30 |
text += '.'
|
31 |
return text
|
32 |
-
|
33 |
-
|
34 |
-
def korean_cleaners2(text):
|
35 |
-
text = korean_to_ipa(text)
|
36 |
-
return text
|
37 |
-
|
38 |
-
|
39 |
-
def chinese_cleaners(text):
|
40 |
-
'''Pipeline for Chinese text'''
|
41 |
-
text = number_to_chinese(text)
|
42 |
-
text = chinese_to_bopomofo(text)
|
43 |
-
text = latin_to_bopomofo(text)
|
44 |
-
if re.match('[ˉˊˇˋ˙]', text[-1]):
|
45 |
-
text += '。'
|
46 |
-
return text
|
47 |
-
|
48 |
-
|
49 |
-
def zh_ja_mixture_cleaners(text):
|
50 |
-
chinese_texts = re.findall(r'\[ZH\].*?\[ZH\]', text)
|
51 |
-
japanese_texts = re.findall(r'\[JA\].*?\[JA\]', text)
|
52 |
-
for chinese_text in chinese_texts:
|
53 |
-
cleaned_text = chinese_to_romaji(chinese_text[4:-4])
|
54 |
-
text = text.replace(chinese_text, cleaned_text+' ', 1)
|
55 |
-
for japanese_text in japanese_texts:
|
56 |
-
cleaned_text = japanese_to_romaji_with_accent(
|
57 |
-
japanese_text[4:-4]).replace('ts', 'ʦ').replace('u', 'ɯ').replace('...', '…')
|
58 |
-
text = text.replace(japanese_text, cleaned_text+' ', 1)
|
59 |
-
text = text[:-1]
|
60 |
-
if re.match('[A-Za-zɯɹəɥ→↓↑]', text[-1]):
|
61 |
-
text += '.'
|
62 |
-
return text
|
63 |
-
|
64 |
-
|
65 |
-
def sanskrit_cleaners(text):
|
66 |
-
text = text.replace('॥', '।').replace('ॐ', 'ओम्')
|
67 |
-
if text[-1] != '।':
|
68 |
-
text += ' ।'
|
69 |
-
return text
|
70 |
-
|
71 |
-
|
72 |
-
def cjks_cleaners(text):
|
73 |
-
chinese_texts = re.findall(r'\[ZH\].*?\[ZH\]', text)
|
74 |
-
japanese_texts = re.findall(r'\[JA\].*?\[JA\]', text)
|
75 |
-
korean_texts = re.findall(r'\[KO\].*?\[KO\]', text)
|
76 |
-
sanskrit_texts = re.findall(r'\[SA\].*?\[SA\]', text)
|
77 |
-
english_texts = re.findall(r'\[EN\].*?\[EN\]', text)
|
78 |
-
for chinese_text in chinese_texts:
|
79 |
-
cleaned_text = chinese_to_lazy_ipa(chinese_text[4:-4])
|
80 |
-
text = text.replace(chinese_text, cleaned_text+' ', 1)
|
81 |
-
for japanese_text in japanese_texts:
|
82 |
-
cleaned_text = japanese_to_ipa(japanese_text[4:-4])
|
83 |
-
text = text.replace(japanese_text, cleaned_text+' ', 1)
|
84 |
-
for korean_text in korean_texts:
|
85 |
-
cleaned_text = korean_to_lazy_ipa(korean_text[4:-4])
|
86 |
-
text = text.replace(korean_text, cleaned_text+' ', 1)
|
87 |
-
for sanskrit_text in sanskrit_texts:
|
88 |
-
cleaned_text = devanagari_to_ipa(sanskrit_text[4:-4])
|
89 |
-
text = text.replace(sanskrit_text, cleaned_text+' ', 1)
|
90 |
-
for english_text in english_texts:
|
91 |
-
cleaned_text = english_to_lazy_ipa(english_text[4:-4])
|
92 |
-
text = text.replace(english_text, cleaned_text+' ', 1)
|
93 |
-
text = text[:-1]
|
94 |
-
if re.match(r'[^\.,!\?\-…~]', text[-1]):
|
95 |
-
text += '.'
|
96 |
-
return text
|
97 |
-
|
98 |
-
|
99 |
-
def cjke_cleaners(text):
|
100 |
-
chinese_texts = re.findall(r'\[ZH\].*?\[ZH\]', text)
|
101 |
-
japanese_texts = re.findall(r'\[JA\].*?\[JA\]', text)
|
102 |
-
korean_texts = re.findall(r'\[KO\].*?\[KO\]', text)
|
103 |
-
#english_texts = re.findall(r'\[EN\].*?\[EN\]', text)
|
104 |
-
for chinese_text in chinese_texts:
|
105 |
-
cleaned_text = chinese_to_lazy_ipa(chinese_text[4:-4])
|
106 |
-
cleaned_text = cleaned_text.replace(
|
107 |
-
'ʧ', 'tʃ').replace('ʦ', 'ts').replace('ɥan', 'ɥæn')
|
108 |
-
text = text.replace(chinese_text, cleaned_text+' ', 1)
|
109 |
-
for japanese_text in japanese_texts:
|
110 |
-
cleaned_text = japanese_to_ipa(japanese_text[4:-4])
|
111 |
-
cleaned_text = cleaned_text.replace('ʧ', 'tʃ').replace(
|
112 |
-
'ʦ', 'ts').replace('ɥan', 'ɥæn').replace('ʥ', 'dz')
|
113 |
-
text = text.replace(japanese_text, cleaned_text+' ', 1)
|
114 |
-
for korean_text in korean_texts:
|
115 |
-
cleaned_text = korean_to_ipa(korean_text[4:-4])
|
116 |
-
text = text.replace(korean_text, cleaned_text+' ', 1)
|
117 |
-
#for english_text in english_texts:
|
118 |
-
# cleaned_text = english_to_ipa2(english_text[4:-4])
|
119 |
-
# cleaned_text = cleaned_text.replace('ɑ', 'a').replace(
|
120 |
-
# 'ɔ', 'o').replace('ɛ', 'e').replace('ɪ', 'i').replace('ʊ', 'u')
|
121 |
-
# text = text.replace(english_text, cleaned_text+' ', 1)
|
122 |
-
text = text[:-1]
|
123 |
-
if re.match(r'[^\.,!\?\-…~]', text[-1]):
|
124 |
-
text += '.'
|
125 |
-
return text
|
126 |
-
|
127 |
-
|
128 |
-
def cjke_cleaners2(text):
|
129 |
-
chinese_texts = re.findall(r'\[ZH\].*?\[ZH\]', text)
|
130 |
-
japanese_texts = re.findall(r'\[JA\].*?\[JA\]', text)
|
131 |
-
korean_texts = re.findall(r'\[KO\].*?\[KO\]', text)
|
132 |
-
english_texts = re.findall(r'\[EN\].*?\[EN\]', text)
|
133 |
-
for chinese_text in chinese_texts:
|
134 |
-
cleaned_text = chinese_to_ipa(chinese_text[4:-4])
|
135 |
-
text = text.replace(chinese_text, cleaned_text+' ', 1)
|
136 |
-
for japanese_text in japanese_texts:
|
137 |
-
cleaned_text = japanese_to_ipa2(japanese_text[4:-4])
|
138 |
-
text = text.replace(japanese_text, cleaned_text+' ', 1)
|
139 |
-
for korean_text in korean_texts:
|
140 |
-
cleaned_text = korean_to_ipa(korean_text[4:-4])
|
141 |
-
text = text.replace(korean_text, cleaned_text+' ', 1)
|
142 |
-
for english_text in english_texts:
|
143 |
-
cleaned_text = english_to_ipa2(english_text[4:-4])
|
144 |
-
text = text.replace(english_text, cleaned_text+' ', 1)
|
145 |
-
#text = text[:-1]
|
146 |
-
#if re.match(r'[^\.,!\?\-…~]', text[-1]):
|
147 |
-
text += '.'
|
148 |
-
return text
|
149 |
-
|
150 |
-
|
151 |
-
def thai_cleaners(text):
|
152 |
-
text = num_to_thai(text)
|
153 |
-
text = latin_to_thai(text)
|
154 |
-
return text
|
155 |
-
|
156 |
-
|
157 |
-
def shanghainese_cleaners(text):
|
158 |
-
text = shanghainese_to_ipa(text)
|
159 |
-
if re.match(r'[^\.,!\?\-…~]', text[-1]):
|
160 |
-
text += '.'
|
161 |
-
return text
|
162 |
-
|
163 |
-
|
164 |
-
def chinese_dialect_cleaners(text):
|
165 |
-
text = re.sub(r'\[MD\](.*?)\[MD\]',
|
166 |
-
lambda x: chinese_to_ipa2(x.group(1))+' ', text)
|
167 |
-
text = re.sub(r'\[TW\](.*?)\[TW\]',
|
168 |
-
lambda x: chinese_to_ipa2(x.group(1), True)+' ', text)
|
169 |
-
text = re.sub(r'\[JA\](.*?)\[JA\]',
|
170 |
-
lambda x: japanese_to_ipa3(x.group(1)).replace('Q', 'ʔ')+' ', text)
|
171 |
-
text = re.sub(r'\[SH\](.*?)\[SH\]', lambda x: shanghainese_to_ipa(x.group(1)).replace('1', '˥˧').replace('5',
|
172 |
-
'˧˧˦').replace('6', '˩˩˧').replace('7', '˥').replace('8', '˩˨').replace('ᴀ', 'ɐ').replace('ᴇ', 'e')+' ', text)
|
173 |
-
text = re.sub(r'\[GD\](.*?)\[GD\]',
|
174 |
-
lambda x: cantonese_to_ipa(x.group(1))+' ', text)
|
175 |
-
text = re.sub(r'\[EN\](.*?)\[EN\]',
|
176 |
-
lambda x: english_to_lazy_ipa2(x.group(1))+' ', text)
|
177 |
-
text = re.sub(r'\[([A-Z]{2})\](.*?)\[\1\]', lambda x: ngu_dialect_to_ipa(x.group(2), x.group(
|
178 |
-
1)).replace('ʣ', 'dz').replace('ʥ', 'dʑ').replace('ʦ', 'ts').replace('ʨ', 'tɕ')+' ', text)
|
179 |
-
text = re.sub(r'\s+$', '', text)
|
180 |
-
text = re.sub(r'([^\.,!\?\-…~])$', r'\1.', text)
|
181 |
-
return text
|
|
|
1 |
import re
|
|
|
2 |
from text.korean import latin_to_hangul, number_to_hangul, divide_hangul, korean_to_lazy_ipa, korean_to_ipa
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
|
4 |
def korean_cleaners(text):
|
5 |
'''Pipeline for Korean text'''
|
|
|
9 |
if re.match('[\u3131-\u3163]', text[-1]):
|
10 |
text += '.'
|
11 |
return text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|