问题
Is there a way for PCRE regular expressions to count how many occurrences of a character it encounters (n), and to stop searching after it has found n occurrences of another character (specifically { and }).
This is to grab code blocks (which may or may not have code blocks nested inside them).
If it makes it simpler, the input will be a single-line string, with the only characters other than braces are digits, colons and commas. The input must pass the following criteria before code blocks are even attempted to be extracted:
$regex = '%^(\\d|\\:|\\{|\\}|,)*$%';
All braces will have a matching pair, and nested correctly.
I would like to know if this can be achieved before I start writing a script to check every character in the string and count each occurrence of a brace. Regular expressions would be much more memory friendly as these strings can be several kilobytes in size!
Thanks, mniz.
Solution
PCRE: Lazy and Greedy at the same time (Possessive Quantifiers)
回答1:
pcre has recursive patterns, so you can do something like this
$code_is_valid = preg_match('~^({ ( (?>[^{}]+) | (?1) )* })$~x', '{' . $code .'}');
the other thing, i don't think this will be faster or less memory consuming than simple counter, especially on large strings.
and this is how to find all (valid) codeblocks in a string
preg_match_all('~ { ( (?>[^{}]+) | (?R) )* } ~x', $input, $blocks);
print_r($blocks);
回答2:
This is exactly what regular expressions are not good for. It's the classic example.
You should just iterate over the string character by character, and keep a count of the nesting level.
回答3:
$regex='%^(\\d|\\:|\\{|\\}|,){0,25)$%';
preg_match($regex,$target,$matches);
where: 25 on first line indicates maximum number of occurrences. then check:
$n=count($matches);
回答4:
It is impossible since the language you are describing is not a regular language.
Use a parser instead.
来源：https://stackoverflow.com/questions/2348547/pcre-find-matching-brace-for-code-block