Python Regex：从字符串中解析键/值对

Question

I am trying to disect a string like the following into a list of key/value pairs:

line1 = "keyword1: value1 keyword2: value2 keyword1: value3 keyword3: value4"

I wrote the following code using regex to achieve that goal:

import re

line1 = "keyword1: value1 keyword2: value2 keyword1: value3 keyword3: value4"

keywords = [ re.escape(k) for k in ['keyword1', 'keyword2', 'keyword3'] ]

any_keyword = '|'.join(keywords)
regex = "(" + any_keyword + "):(.+?)(?:" + any_keyword + "|$)"

print(line1)
print(regex)

for m in re.finditer(regex, line1):
  print(m)

The matches I get are

<re.Match object; span=(0, 25), match='keyword1: value1 keyword2'>
<re.Match object; span=(34, 59), match='keyword1: value3 keyword3'>

and, of course, they include keyword2 and keyword3 at the end of the string so that I don't get additional Match objects for those keywords.

How can I receive 4 matches, one for each keyword in the line?

Answer 1

You may extract the matches using a lookahead rather than a non-capturing group as the last pattern in the regex since a non-capturing group pattern still consumes characters:

import re
line1 = "keyword1: value1 keyword2: value2 keyword1: value3 keyword3: value4"
keywords = ['keyword1', 'keyword2', 'keyword3']
any_keyword = '|'.join(map(re.escape, keywords))
regex = "(" + any_keyword + "):(.+?)(?=(?:" + any_keyword + "):|$)"
print([m.group() for m in re.finditer(regex, line1)])
# => ['keyword1: value1 ', 'keyword2: value2 ', 'keyword1: value3 ', 'keyword3: value4']

See the Python demo

If your keys can contain whitespace, make sure to sort the keys before defining any_keyword pattern by sorting them by length in descending order, eg sorted(keywords,key=len,reverse=True) .

It might be a good idea to match keywords as whole words, too:

regex = r"\b(" + any_keyword + r"):(.+?)(?=\b(?:" + any_keyword + "):|$)"

See the regex demo . Details:

• \\b - a word boundary
• (keyword1|keyword2|keyword3) - Group 1: keyword alternatives
• : - a : char
• (.+?) - Group 2: any one or more chars other than line break chars as few as possible
• (?=\\b(?:keyword1|keyword2|keyword3):|$) - a positive lookahead that makes sure that, immediately to the right of the current position, there is either
  • \\b(?:keyword1|keyword2|keyword3): - any keyword from the list followed with :
  • | - or
  • $ - end of string.

Answer 2

One way using re.split :

line1 = "keyword1: value1 keyword2: value2 keyword1: value3 keyword3: value4"
l = re.split('(keyword\d+):', line1)[1:]

[(k,v.strip()) for k, v in zip(l[::2], l[1::2])]

Output:

[('keyword1', 'value1'),
 ('keyword2', 'value2'),
 ('keyword1', 'value3'),
 ('keyword3', 'value4')]

Answer 3

another way

>>> regex = "(" + any_keyword + "):\s(\w+)"
>>> pattern = re.compile(regex)
>>> pattern.search(line1)
<_sre.SRE_Match object; span=(0, 16), match='keyword1: value1'>
>>> pattern.findall(line1)
[('keyword1', 'value1'), ('keyword2', 'value2'), ('keyword1', 'value3'), ('keyword3', 'value4')]
>>>