re 正则从文本中提取全球电话和邮箱

FOAF-lambda

已于 2022-08-27 13:00:55 修改

阅读量868

点赞数

文章标签：正则表达式

于 2022-08-22 15:09:18 首次发布

本文链接：https://blog.csdn.net/lwdfzr/article/details/126466412

版权

1. 提取电话

text = '''
"(+91)966-352-2720", "☎ +91 9663522720"
 9663522720
+79082343434   8(912)2342554,  +7 982 342 sdfdsf 34 34 fsf 8-923-132-34-23 +7 982 342 34 34! sdfsd
13045542316
01727 830357
Tlf. 788 96 888
8553664787;3
'(866) 997-3523(615) 488-4607'
+TollFree:1-888-412-3160
914-345-0800 1
'+?911235410HorariodeLV:09:0018:00'
"Telephone:4697021130"
CALL US: +91- 421- 435 00 02, 224 84 72
"+57x018000423 670"
"416-998-8438 | 416-630-3053"
"Whatsapp + 44 75 88755 173"
(852)95788980 | Tel: (852) 23428887"
phone: +974 4411-0147  |  : +974 5512-3374
fax: 517/253-7366
¡llámanos! (55) 71 00 35 00
+dubai:043804009|abudhabi:024146688
monfri:9:00am12:30pm/2:00pm5:30pm(cet)33983400075
https:wa-me/15551234567
216-812-1615 ext 115
wsp +56 9 5713 2204
call us +91 96888 78333
+contact: +91 70953 23456
15551234569/15551234567
port perry - 289-485-brew (2739) newmarket - 289-803-9111
(518) 858u200b-u200b9589
+971 58505 8980
waes footwear, u2028suite 2, duke st chambers, bridge street, tq7 1hx, united kingdom (+44) 020 3287 7957
contact-alexy-metals-216-410-8661
toll free +1 (888) 815-6150
call us: +91- 421- 435 00 02, 224 84 72
+61 405 385 704
(0044) (0)1782 611 599
0123-456-7890
55 6930 9729
+1 (800) 123 456 789
'''

numbers0 = re.findall("[\(\+\)]*[\+\d\)]*[\( -]?\d{3}[\) -]?\d{3}[ -]?\d{2}[ -]?\d{2}|[\(]?[0-9]{3}[\)]?[- \.]?[0-9]{3}[- \.]?[0-9]{4,6}|[\(\+\)/\d+]* *[\-\d]{8,}|[\+\d ]*[\d ]{5,} ?[\d]{5,}", text)
numbers0 = [t[1:] if t.startswith(')') or t.startswith('/') else t.strip() for t in numbers0]
print('numbers0:',numbers0)

numbers1 = re.findall(r"[\(\+\)]*[1-9]* ?[0-9.\-\(\)]{8,} ?[0-9]", text,re.S)
print('numbers1:',numbers1)

numbers2 = re.findall(r"[\(\+\d\)]* ?[\(]?[0-9]{3}[\)]?[- \.]?[0-9]{3}[- \.]?[0-9]{4,6}", text,re.S)
print('numbers2:',numbers2)

numbers3 = re.findall(r"[\+\d\(\) ]+[\d -]{2,} ?[\d-]{2,}", text)
numbers3 = [t[1:].strip() if t.strip()[0] not in ['+','('] and not t.strip()[0].isdigit() else t.strip() for t in numbers3 if len(re.sub('\D','',t))>7]

print('numbers3:',numbers3)

2.提取邮箱

emails="""
geon@ihateregex.io
test@gmail.com mail@test.org
mail@testing.com
hello@
@test
email@gmail
theproblem@test@gmail.com
"""
emails_0 =  re.findall(r'[a-zA-Z]?[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,15}', emails,re.S)
print('emails_0:',emails_0)