How to make this Regex more flexible?

Question

I am creating a simple template engine that uses Regex to find special expressions, which are parsed and processed. They are enclosed in Ruby-style opening tags and have the format:

    <% label %>  OR:
    <% function(arg1, arg2, arg3) %>

However the problem is that I cannot have multiple entries on the same line, so I can't have this:

    <p><% name %> - <% age %> - <% gender %></p>

Instead I have to write it like this:

    <p><% name %> - 
       <% age %> -
       <% gender %><p>

How can I fix this? Below I included the code for the function that handles this. Basically what it does is read from a file, split everything line by line, then iterate through each line and do a regex match and return the results.

public function content()
{
    if (!file_exists($this->path))
        return;

    /* 0 = before, 1 = entry #1, 2 = entry #2, 3 = after */
    $content = file_get_contents($this->path); 
    $pattern = '/^(.*)\<\%\s*(.+)\s*\%\>(.*)$/'; 
    $lines = explode("\n", $content); 


    foreach ($lines as $line)
    {
        // Pattern match the line
        preg_match($pattern, $line, $matches); 

        $result = ''; 

        // If there are no matches, then don't parse it and add the raw string to output
        if (count($matches) == 0)
        {
            continue; 
        }   


        // Parse the matches and run the appropriate action         
        $before = $matches[1]; 
        $after = $matches[3]; 
        $entry = $matches[2]; 
        $return = ''; 


        // Process the entry
        $return = $this->processEntry($entry);

        // Add the processed line to the final result if not empty
        $result = $before . $return . $after; 
        if (!empty($result)) $final .= $result . "\n"; 
    }
    return $final; 
}

Answer 1

Your current regex is too greedy.

When running against your single line input, I get this result:

/^(.*)\<\%\s*(.+)\s*\%\>(.*)$/

[0] => <p><% name %> - <% age %> - <% gender %></p>
[1] => <p><% name %> - <% age %> - 
[2] => gender 
[3] => </p>

By upgrading the regex to be lazy and also using preg_match_all, we can get the following result:

/(.*?)\<\%\s*(.+?)\s*\%\>/

[0] =>  [0] => <p><% name %>
        [1] =>  - <% age %>
        [2] =>  - <% gender %>
[1] =>  [0] => <p>
        [1] =>  - 
        [2] =>  - 
[2] =>  [0] => name
        [1] => age
        [2] => gender

The only part missing from that match is the end, the last </p>. I would use another regex match to grab that portion:

/.*\%\>(.*)$/

[0] => <p><% name %> - <% age %> - <% gender %></p>
[1] => </p>

Recommended reading over greedy and lazy: http://www.regular-expressions.info/repeat.html

Answer 2

What to you thing about:

<?=label ?>  OR:
<?=function(arg1, arg2, arg3) ?>

PHP is already a template engine

Just write a wrapper around a include call and your template system is ready to use and more flexible than anything you could write.

---

That said, why to you match single lines? Please have a look at preg_replace_callback.

/\<\%\s*(.+?)\s*\%\>/

If you want to put the whitespace trimming in the callback and as you don't need to escape <>%, you could even use

/<%(.+?)%>/