逗号分隔的单词正则表达式

Question

I'm having some issues while trying to parse expression such as the following:

word1, word2[a,b,c],   word3, ..., wordN

I'd like to get the following groups:

g1: word1
g2: word2[a,b,c]
g3: word3

Please note that the [.+] is optional, the regular expression must be able to match expressions like the following the following:

word1,word2,word3
word1[a,b,c],word2,word3
word1[a,b,c],word2[e,f,g],word3
word1[a,b,c],word2[e,f,g],word3[i,j,l]

I did some attempts but I can't find the way to correctly separate the groups.

Answer 1

I tried this regex on https://regex101.com , and pasted your expressions into the "test strings" box.

/^([a-zA-Z0-9]+(?:\[.*\])?),([a-zA-Z0-9]+(?:\[.*\])?),([a-zA-Z0-9]+(?:\[.*\])?)$/gm

Each word is separated by a comma, and of the form:

([a-zA-Z0-9]+(?:\[.*\])?)

Explanation:

(
  [a-zA-Z0-9]+ # one or more alphanumeric characters (could use \w)
  (?:\[.*\])? # an optional sequence surrounded by []s. (?: ) means a non-capturing group
)

Answer 2

For the time being this seems to be working:

import re
rgx = re.compile("(\w+(\[.*?\])*).*?,?")
[key for key, val in rgx.findall("word1, word2[a,b,[c,,,]],     word,3")]

# this regex starts by looking for alpha numberic characters with \w+
# then within that it looks if a `[` is present then till we encounter end of bracket ']' consider everything (\[.*?\])*.
# the output of this is a tuple as ('word2[a,b,c]', '[a,b,c]')
# we iterate over the tuple and take only the 1st values in the tuple

output:

['word1', 'word2[a,b,[c,,,]', 'word', '3']

another example

[key for key, val in rgx.findall("word1[bbbb,cccc],word2[bbbb,cccc] ")]

output:

['word1[bbbb,cccc]', 'word2[bbbb,cccc]']

PS: still regexing to improve it.

Answer 3

You can use re.split to split only on commas, that are outside of brackets. This can be determinded by the fact, that those commas will never match a closing bracket before the opening one (using a negative lookahead). This trick is only possible with non-nested brackets.

import re
print(re.split(r',(?![^[]*\])', 'word1[a,b,c],word2[e,f,g],word3'))

outputs ['word1[a,b,c]', 'word2[e,f,g]', 'word3']

http://ideone.com/7vIwFM