Simple HTML Parser

# Written March 2001 by Nathaniel Hekman
# Comments/bugs to [email protected]
#
# html_parse will parse a chunk of HTML code. 

# string $tagFn  : function to call for each tag.  This function
#                  is called for *every* tag, opening and closing.
#                  It would be easy to change this to call a 
#                  different function for open and close tags.
#                  The tagFn has these arguments:
#   function tagFn($tag, $attribs)
#     string $tag     : the tag
#     string $attribs : all the attributes, as one long string
#                       wouldn't be hard to change this to an assoc array
# string $dataFn : function to call for data (non-tag).  This function
#                  is called for all the data in between tags, including
#                  comments.  The dataFn has these arguments:
#    function dataFn($data)
#      string $data : the text
# string $data   : the html to parse
function html_parse($tagFn, $dataFn, $data) {
    $comment = false;

    # Split on '<', so the beginning of each array entry
    # will be a tag.
    $TagLine = explode('<', $data);

    # Loop through each entry -- each entry is a tag
    # followed by everything up to the next tag.
    foreach ($TagLine as $l) {
        # If we're not in a comment block, then check if
        # one starts here.
        if (!$comment) {
            if (substr(ltrim($l), 0, 3) == "!--") {
                # this is the beginning of a comment, not a tag
                $comment = true;
                $commentline = "";
            }
        }

        # If we're in a comment block, add this entry to
        # the comment block, and check if it ends here.
        if ($comment) {
            $commentline .= "<".$l;
            if (strstr($l, "-->")) {
                $line = $commentline;
                $tag = "";
                $comment = false;
            }
            else {
                $line = "";
                $tag = "";
            }
        }
        # Otherwise, split on '>' to separate the tag from
        # the data.
        else {
            list($tag, $line) = explode('>', $l, 2);
        }

        # If there's a tag, call the tag function
        if (strlen($tag)) {
            $tag = trim($tag);
            list($tag, $attribline) = explode(' ', $tag, 2);

            # I could split $attribline into an
            # associative array, but I don't need
            # that functionality now.
            $tagFn($tag, $attribline);
        }

        # If there's data, call the data function
        if (strlen($line)) {
            $dataFn($line);
        }
    }
}