php正则表达式替换大括号中的空格，包括大括号

Question

I wonder if it's possible to solve such kind of template parsing using RegEx-es I have text similar to:

'Hello, {{ Na me|{{Surname }}}}. Would you like some drink? {{CoffeePreferred? We have a special discount for coffee}} {{Tea Preferred? We have a special discount for tea}}'

First, I need to fix errors in variable names (variables must always come right after braces to special symbol like "?" or "|" or something else could be comma or dot - it means rules can be added in future).

Na me , Tea Preferred Surname are spelling errors of variable names. Also variable names should start with right after curly braces contain no spaces after - otherwise it's also an error.

Then I need to keep text as is.

But Dear friend , We have a special discount for coffee and We have a special discount for tea are simple text.

So my question is how can I replace unnecessary spaces withing underscores in variables and keep them in text? Also I want to make it in as fewer steps as possible (due to performance impact and my texts can take megabytes).

I expect my text to be:

'Hello, {{Na_me|{{Surname}}}}. Would you like some drink? {{CoffeePreferred? We have a special discount for coffee}} {{Tea_Preferred? We have a special discount for tea}}'

My first step was to extract content of braces:

preg_replace_callback(
    '/(\{\{)(.*?)(\}\})/ui',
    function ($matches) {
        echo $matches[2];
});

Then I tried to replace spaces by rules described above with no success. I tried to use assertions like /[\\s](?!<=\\|)/ and expressions like (\\s(?!\\|))|((?!\\|)\\s) Does anybody have any idea?

Answer 1

Here is a way to do the job:

$str = preg_replace_callback(
    '/\{\{([^|?{}]+)/',
    function ($matches) {
        $ret = trim($matches[1]);
        return preg_replace('/ /', '_', $ret);
    },
    $str
);

After that you can remove all the braces:

$str = preg_replace('/[{}]+/', '', $str);