Adam Adam - 8 months ago 26
Python Question

Regular expression to capture groups delimited by an expression

How to capture

[0-9]+
groups delimited by
(\], \[)
? For example, in the case of

[[[[u'1', u'2'], u'3'], u'4'], [[[u'1', u'2'], u'4'], [[u'1', u'5'], u'4']]]


I would like to capture three groups,
1 2 3 4
,
1 2 4
and
1 5 4
.

Answer Source

Assuming that you don't have sub-patterns like [[u'1', u'2'], [u'3',u'5']] (multiple nested sub-groups at the same level, in which case you need to use a stack and parse like pushdown automata) you could do this with regular expressions in two steps:

(1) split the expression with regex \]\s*,\s*\[ to get the groups first, you will get 3 groups for the example provided.

(2) within each group use the regex [^0-9u]*u'([0-9]+)'[^0-9u]* to extract the digits.

For example, in R, the code will be:

str <- "[[[[u'1', u'2'], u'3'], u'4'], [[[u'1', u'2'], u'4'], [[u'1', u'5'], u'4']]]"
groups <- unlist(strsplit(str, split='\\]\\s*,\\s*\\['))
pattern <- "[^0-9u]*u'([0-9]+)'[^0-9u]*"
lapply(groups, function(str) gsub(pattern, "\\1", regmatches(str,gregexpr(pattern,str))[[1]]))

#[[1]]
#[1] "1" "2" "3" "4"

#[[2]]
#[1] "1" "2" "4"

#[[3]]
#[1] "1" "5" "4"

In python:

import re
str = "[[[[u'1', u'2'], u'3'], u'4'], [[[u'1', u'2'], u'4'], [[u'1', u'5'], u'4']]]"
groups = re.split('\]\s*,\s*\[', str)
pattern = "[^0-9u]*u'([0-9]+)'[^0-9u]*"
print map(lambda x: re.findall(pattern, x), groups)
# [['1', '2', '3', '4'], ['1', '2', '4'], ['1', '5', '4']]

you could map the digits to integers if required.