Parsing nested curly brace with php regex - php

Question

I want to parse raw string with nested curly braces to multidimensional arrays. Below I added sample code which works. But main problem my regex captures only first matched group and ignores another occurences.
Any help highly appreciated.
Code:
$regex = '/(?ims)(group [a-z0-9\s\,\.\:#_\-#]+)\{([^\}]*)\}/';
preg_match_all( $regex, file_get_contents('data.txt', FILE_USE_INCLUDE_PATH), $arr);
$result = array();
foreach ($arr[0] as $i => $x)
{
$selector = trim($arr[1][$i]);
$rules = explode(';', trim($arr[2][$i]));
$result[$selector] = array();
foreach ($rules as $strRule)
{
if (!empty($strRule))
{
$rule = explode(" = ", $strRule);
$result[$selector][][trim($rule[0])] = trim($rule[1]);
}
}
}
Raw string (data.txt):
group A { T1 { X = 44; }
T2 { Y = 33; } }
group B { T1 { X = 555; } }
Code Output:
Array (
[group A] => Array (
[0] => Array (
[T1 { X] => 44
)
)
[group B] => Array (
[0] => Array (
[T1 { X] => 555
)
)
)
But excepted Output:
Array(
[group A] => Array(
[T1] => Array(
[X] => 44
)
[T2] => Array (
[Y] => 33
)
)
[group B] => Array(
[T1] => Array(
[X] => 555
)
)
)

Answer 1

A better solutions might exists, but it works.
Ensure your input cannot have deeper values, or you have to improve/rewrite the entiere rule !
$input = <<<REG
group A { T1 { X = 44; }
T2 { Y = 33; } }
group B { T1 { X = 555; } }
REG;
$out = [];
$cbuff = null;
$rule = "/^(?<group>[^\{\}]*)?\{|((?<set>[\w\s]+)\{(?<var>[\w\s]+)=(?<val>[\d\s]+)[;\s]+\})/im";
preg_replace_callback($rule, function($m)
use(&$out, &$cbuff) // import locally our working requirements
{
$m = array_filter($m, "is_string", ARRAY_FILTER_USE_KEY); // cleaning
$m = array_map("trim", array_filter($m)); // cleaning and trim
if(key($m)=="group") // get & create the current root group
$out[($cbuff = reset($m))] = [];
if(isset($m["var"])) // assuming 'var' exists, you'll get 'set/var' too
{
if(is_null($cbuff)) return; // prevent no group found
$out[$cbuff][$m["set"]] = [$m["var"] => $m["val"]];
}
}, $input);
var_dump($out);
Will output:
array(2) {
["group A"]=>
array(2) {
["T1"]=>
array(1) {
["X"]=>
string(2) "44"
}
["T2"]=>
array(1) {
["Y"]=>
string(2) "33"
}
}
["group B"]=>
array(1) {
["T1"]=>
array(1) {
["X"]=>
string(3) "555"
}
}
}
Hint
I think you are ok with PHP,
but i put a little bit of explaination about regexes:
first i've used "named group" for each parts 'set/var/val' pattern: ?<name>
That confer an own group key under the matching output
(?<set>...)|(?<var>...) gives --> ["set" => ...], ["var" => "..."]
Finally, to separate the sub properties to root group,
i'm just matching any part does not contains brackets
Bench
You can retrive my bench on regex101
3 matches, 60 steps (~0ms)
5 matches, 107 steps (~0ms)
yes it can be optimized!