Spaces:
Runtime error
Runtime error
Update preprocessing.py
Browse files- preprocessing.py +73 -0
preprocessing.py
CHANGED
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import hazm
|
2 |
+
from cleantext import clean
|
3 |
+
import regex as re
|
4 |
+
|
5 |
+
def cleanhtml(raw_html):
|
6 |
+
cleanr = re.compile('<.*?>')
|
7 |
+
cleantext = re.sub(cleanr, '', raw_html)
|
8 |
+
return cleantext
|
9 |
+
|
10 |
+
def cleaning(text):
|
11 |
+
text = text.strip()
|
12 |
+
|
13 |
+
# regular cleaning
|
14 |
+
# https://pypi.org/project/clean-text/ >> works well for eng and de languages
|
15 |
+
text = clean(text,
|
16 |
+
fix_unicode=True,
|
17 |
+
to_ascii=False,
|
18 |
+
lower=True,
|
19 |
+
no_line_breaks=True,
|
20 |
+
no_urls=True,
|
21 |
+
no_emails=True,
|
22 |
+
no_phone_numbers=True,
|
23 |
+
no_numbers=False,
|
24 |
+
no_digits=False,
|
25 |
+
no_currency_symbols=True,
|
26 |
+
no_punct=False, #Keep the punc
|
27 |
+
replace_with_url="",
|
28 |
+
replace_with_email="",
|
29 |
+
replace_with_phone_number="",
|
30 |
+
replace_with_number="",
|
31 |
+
replace_with_digit="0",
|
32 |
+
replace_with_currency_symbol="",
|
33 |
+
)
|
34 |
+
|
35 |
+
# cleaning htmls
|
36 |
+
text = cleanhtml(text)
|
37 |
+
|
38 |
+
# normalizing > https://github.com/sobhe/hazm
|
39 |
+
normalizer = hazm.Normalizer()
|
40 |
+
text = normalizer.normalize(text)
|
41 |
+
|
42 |
+
# removing wierd patterns
|
43 |
+
wierd_pattern = re.compile("["
|
44 |
+
u"\U0001F600-\U0001F64F" # emoticons
|
45 |
+
u"\U0001F300-\U0001F5FF" # symbols & pictographs
|
46 |
+
u"\U0001F680-\U0001F6FF" # transport & map symbols
|
47 |
+
u"\U0001F1E0-\U0001F1FF" # flags (iOS)
|
48 |
+
u"\U00002702-\U000027B0"
|
49 |
+
u"\U000024C2-\U0001F251"
|
50 |
+
u"\U0001f926-\U0001f937"
|
51 |
+
u'\U00010000-\U0010ffff'
|
52 |
+
u"\u200d"
|
53 |
+
u"\u2640-\u2642"
|
54 |
+
u"\u2600-\u2B55"
|
55 |
+
u"\u23cf"
|
56 |
+
u"\u23e9"
|
57 |
+
u"\u231a"
|
58 |
+
u"\u3030"
|
59 |
+
u"\ufe0f"
|
60 |
+
u"\u2069"
|
61 |
+
u"\u2066"
|
62 |
+
# u"\u200c"
|
63 |
+
u"\u2068"
|
64 |
+
u"\u2067"
|
65 |
+
"]+", flags=re.UNICODE)
|
66 |
+
|
67 |
+
text = wierd_pattern.sub(r'', text)
|
68 |
+
|
69 |
+
# removing extra spaces, hashtags
|
70 |
+
text = re.sub("#", "", text)
|
71 |
+
text = re.sub("\s+", " ", text)
|
72 |
+
|
73 |
+
return text
|