计算字符串中的字符匹配数（仅Regex）？

Question

So I recently had an interview question asking me to determine if a string was balanced or not. A string is balanced under these two conditions:

1. The sum of the occurrences of 'a' and 'c' is even.
2. The sum of the occurrences of 'b' and 'd' is even.

So given the example " cccddbba " these conditions hold true. Occurrences of c (3) + occurrences of a (1) is even (1+3=4) and occurrences of d (2) + occurrences of b (2) is even (2+2=4) . The strings can only contain the characters 'a', 'b', 'c', and 'd'. The regex should work for any string, not just this example.

My task was only to write the regex for this. I assume in the background they count the number of matches of 'a' and 'c' found using my regex and sum them up. The same goes for 'b' and 'd'.

I haven't touched regexes in a while, so I did horrible and only got as far as ^([ac])*?[db]*?([ac])*?$ , which is obviously incorrect.

Answer 1

Use separate look aheads for each assertion:

^(?=(([^ac]*[ac]){2})*[^ac]*$)(?=(([^bd]*[bd]){2})*[^bd]*$).*$

See live demo .

This works basically because ([^ac]*[ac]){2}) matches pairs of [ac] . The rest is relatively simple.

Answer 2

Using positive lookahead, you can check whether a and c occur an even number of times, and ditto for b and d:

^(?=[^ac]*([ac][^ac]*[ac][^ac]*)*$)(?=[^bd]*([bd][^bd]*[bd][^bd]*)*$).*$

Each of the two lookahead groups check that the whole string contains an even number of their respective characters. Note that each lookahead is anchored at the end ( $ ), to force it to scan until the end of the string.