Gget all substring inside potentially nested curly braces

Question

I'm trying to parse the following format with PHP:

// This is a comment
{
this is an entry
}
{
this is another entry
}
{
entry
{entry within entry}
{entry within entry}
}

Maybe is just the lack of caffeine, but I can't think of a decent way of getting the contents of the curly braces.

Answer 1

This is quite a common parsing task, basically you need to keep track of the various states you can be in and use a combination of constants and function calls to maintain them.

Here is some rather inelegant code that does just that:

<?php

$input = file_get_contents('input.txt');

define('STATE_CDATA', 0);
define('STATE_COMMENT', 1);

function parseBrace($input, &$i)
{
    $parsed = array(
        'cdata' => '',
        'children' => array()
    );
    $length = strlen($input);
    $state = STATE_CDATA;
    for(++$i; $i < $length; ++$i) {
        switch($input[$i]) {
            case '/':
                if ('/' === $input[$i+1]) {
                    $state = STATE_COMMENT;
                    ++$i;
                } if (STATE_CDATA === $state) {
                    $parsed['cdata'] .= $input[$i];
                }
                break;
            case '{':
                if (STATE_CDATA === $state) {
                    $parsed['children'][] = parseBrace($input, $i);
                }
                break;
            case '}':
                if (STATE_CDATA === $state) {
                    break 2; // for
                }
                break;
            case "\n":
                if (STATE_CDATA === $state) {
                    $parsed['cdata'] .= $input[$i];
                }
                $state = STATE_CDATA;
                break;
            default:
                if (STATE_CDATA === $state) {
                    $parsed['cdata'] .= $input[$i];
                }
        }
    }
    return $parsed;
}

function parseInput($input)
{
    $parsed = array(
        'cdata' => '',
        'children' => array()
    );
    $state = STATE_CDATA;
    $length = strlen($input);
    for($i = 0; $i < $length; ++$i) {
        switch($input[$i]) {
            case '/':
                if ('/' === $input[$i+1]) {
                    $state = STATE_COMMENT;
                    ++$i;
                } if (STATE_CDATA === $state) {
                    $parsed['cdata'] .= $input[$i];
                }
                break;
            case '{':
                if (STATE_CDATA === $state) {
                    $parsed['children'][] = parseBrace($input, $i);
                }
                break;
            case "\n":
                if (STATE_CDATA === $state) {
                    $parsed['cdata'] .= $input[$i];
                }
                $state = STATE_CDATA;
                break;
            default:
                if (STATE_CDATA === $state) {
                    $parsed['cdata'] .= $input[$i];
                }
        }
    }
    return $parsed;
}

print_r(parseInput($input));

This produces the following output:

Array
(
    [cdata] =>




    [children] => Array
    (
        [0] => Array
        (
            [cdata] =>
this is an entry

            [children] => Array
            (
            )

        )

        [1] => Array
        (
            [cdata] =>
this is another entry

            [children] => Array
            (
            )   

        )

        [2] => Array
        (
            [cdata] => 
entry



            [children] => Array
            (
                [0] => Array
                (
                    [cdata] => entry within entry
                    [children] => Array
                    (
                    )


                )

                [1] => Array
                (
                    [cdata] => entry within entry
                    [children] => Array
                    (
                    )

                )

            )

        )

    )

)

You'll probably want to clean up all the whitespace but some well placed trim's will sort that for you.