Advice for implementing simple regex (for bbcode/geshi parsing)

I had made a personal note software in PHP so I can store and organize my notes and wished for a nice simple format to write them in.

I had done it in Markdown but found it was a little confusing and there was no simple syntax highlighting, so I did bbcode before and wished to implement that.

Now for GeSHi which I really wish to implement (the syntax highlighter), it requires the most simple code like this:

$geshi = new GeSHi($sourcecode, $language);
$geshi->parse_code();

Now this is the easy part , but what I wish to do is allow my bbcode to call it.

My current regular expression to match a made up [syntax=cpp][/syntax] bbcode is the following:

preg_replace('#\[syntax=(.*?)\](.*?)\[/syntax\]#si' , 'geshi(\\2,\\1)????', text);

You will notice I capture the language and the content, how on earth would I connect it to the GeSHi code?

preg_replace seems to just be able to replace it with a string not an 'expr开发者_如何学编程ession', I am not sure how to use those two lines of code for GeSHi up there with the captured data..

I really am excited about this project and wish to overcome this.

---

I wrote this class a while back, the reason for the class was to allow easy customization / parsing. Maybe a little overkill, but works well and I needed it overkill for my application. The usage is pretty simple:

$geshiH = new Geshi_Helper();
$text = $geshiH->geshi($text); // this assumes that the text should be parsed (ie inline syntaxes)

---- OR ----

$geshiH = new Geshi_Helper();
$text = $geshiH->geshi($text, $lang);  // assumes that you have the language, good for a snippets deal

I had to do some chopping from other custom items I had, but pending no syntax errors from the chopping it should work. Feel free to use it.

<?php

require_once 'Geshi/geshi.php';

class Geshi_Helper 
{
    /**
     * @var array Array of matches from the code block.
     */
    private $_codeMatches = array();

    private $_token = "";

    private $_count = 1;

    public function __construct()
    {
        /* Generate a unique hash token for replacement) */
        $this->_token = md5(time() . rand(9999,9999999));
    }

    /**
     * Performs syntax highlights using geshi library to the content.
     *
     * @param string $content - The context to parse
     * @return string Syntax Highlighted content
     */
    public function geshi($content, $lang=null)
    {
        if (!is_null($lang)) {
            /* Given the returned results 0 is not set, adding the "" should make this compatible */
            $content = $this->_highlightSyntax(array("", strtolower($lang), $content));
        }else {
            /* Need to replace this prior to the code replace for nobbc */
            $content = preg_replace('~\[nobbc\](.+?)\[/nobbc\]~ie', '\'[nobbc]\' . strtr(\'$1\', array(\'[\' => \'&#91;\', \']\' => \'&#93;\', \':\' => \'&#58;\', \'@\' => \'&#64;\')) . \'[/nobbc]\'', $content);

            /* For multiple content we have to handle the br's, hence the replacement filters */
            $content = $this->_preFilter($content);

            /* Reverse the nobbc markup */
            $content = preg_replace('~\[nobbc\](.+?)\[/nobbc\]~ie', 'strtr(\'$1\', array(\'&amp;#91;\' => \'[\', \'&amp;#93;\' => \']\', \'&amp;#58;\' => \':\', \'&amp;#64;\' => \'@\'))', $content);

            $content = $this->_postFilter($content);
        }

        return $content;
    }

    /**
     * Performs syntax highlights using geshi library to the content.
     * If it is unknown the number of blocks, use highlightContent
     * instead.
     *
     * @param string $content - The code block to parse
     * @param string $language - The language to highlight with
     * @return string Syntax Highlighted content
     * @todo Add any extra / customization styling here.
     */
    private function _highlightSyntax($contentArray)
    {
        $codeCount = $contentArray[1];

        /* If the count is 2 we are working with the filter */
        if (count($contentArray) == 2) {
            $contentArray = $this->_codeMatches[$contentArray[1]];
        }

        /* for default [syntax] */
        if ($contentArray[1] == "")
            $contentArray[1] = "php";

        /* Grab the language */
        $language = (isset($contentArray[1]))?$contentArray[1]:'text';

        /* Remove leading spaces to avoid problems */
        $content = ltrim($contentArray[2]);

        /* Parse the code to be highlighted */
        $geshi = new GeSHi($content, strtolower($language));
        return $geshi->parse_code();
    }

    /**
     * Substitute the code blocks for formatting to be done without
     * messing up the code.
     *
     * @param array $match - Referenced array of items to substitute
     * @return string Substituted content
     */
    private function _substitute(&$match)
    {
        $index = sprintf("%02d", $this->_count++);
        $this->_codeMatches[$index] = $match;
        return "----" . $this->_token . $index . "----";
    }

    /**
     * Removes the code from the rest of the content to apply other filters.
     *
     * @param string $content - The content to filter out the code lines
     * @return string Content with code removed.
     */
    private function _preFilter($content)
    {
        return preg_replace_callback("#\s*\[syntax=(.*?)\](.*?)\[/syntax\]\s*#siU", array($this, "_substitute"), $content);
    }

    /**
     * Replaces the code after the filters have been ran.
     *
     * @param string $content - The content to replace the code lines
     * @return string Content with code re-applied.
     */
    private function _postFilter($content)
    {
        /* using dashes to prevent the old filtered tag being escaped */
        return preg_replace_callback("/----\s*" . $this->_token . "(\d{2})\s*----/si", array($this, "_highlightSyntax"), $content);
    }
}
?>

---

It looks to me like you already got the regex right. Your problem lies in the invocation, so I suggest making a wrapper function:

function geshi($src, $l) {
    $geshi = new GeSHi($sourcecode, $language);
    $geshi->parse_code();
    return $geshi->how_do_I_get_the_results();
}

Now this would normally suffice, but the source code is likely to contain single or dobule quotes itself. Therefore you cannot write preg_replace(".../e", "geshi('$2','$1')", ...) as you would need. (Note that '$1' and '$2' need quotes because preg_replace just substitutes the $1,$2 placeholders, but this needs to be valid php inline code).

That's why you need to use preg_replace_callback to avoid escaping issues in the /e exec replacement code. So for example:

preg_replace_callback('#\[syntax=(.*?)\](.*?)\[/syntax\]#si' , 'geshi_replace', $text);

And I'd make a second wrapper, but you can combine it with the original code:

function geshi_replace($uu) {
    return geshi($uu[2], $uu[1]);
}

---

Use preg_match:

$match = preg_match('#\[syntax=(.*?)\](.*?)\[/syntax\]#si', $text);
$geshi = new GeSHi($match[2], $match[1]);