<?php
// File name:	page2rel.inc.php
// Description:	PHP subroutine to convert URLs in a page to relative
// Date:	2003-09-09
// Author:	imacat <imacat@pristine.com.tw>
// Copyright:	Copyright (C) 2003-2007 Pristine Communications

// Set the include path
if (!defined("INCPATH_SET")) {
    require_once dirname(__FILE__) . "/incpath.inc.php";
}
// Referenced subroutines
require_once "monica/htmlchar.inc.php";
require_once "monica/rel2abs.inc.php";

// Settings
define("_PAGE2REL_2REL", true);
define("_PAGE2REL_2ABS", false);

// page2rel: Convert URLs in a page to relative
function page2rel($html, $base)
{
    return _page2rel_cnvturl($html, $base, _PAGE2REL_2REL, null);
}

// page2abs: Convert URLs in a page to absolute
function page2abs($html, $base, $rdiff = REL2ABS_RDIFF_KEEP)
{
    return _page2rel_cnvturl($html, $base, _PAGE2REL_2ABS, $rdiff);
}

// _page2rel_cnvturl: Convert URLs in a page to relative/absolute
function _page2rel_cnvturl($source, $base, $is_rel, $rdiff)
{
    $result = "";
    while (preg_match("/^(.*?)(<!--.*?-->|<[a-z]+\d?(?:\s+[a-z\-]+=(?:\"[^\"]*\"|'[^']*'|[^\"'\s<>]+))*(?:\s+\/)?>|<\?xml-stylesheet(?:\s+[a-z\-]+=(?:\"[^\"]*\"|'[^']*'|[^\"'\s<>]+))*\s*\?>)(.*)$/is", $source, $m)) {
        $result .= $m[1];
        $ele = $m[2];
        $source = $m[3];
        // Skip if not a relevant element
        if (    !preg_match("/^<((a|applet|area|base|embed|form|frame|img|input|link|object|param|script|style)\s+)(.+?)>$/is", $ele, $m)
                && !preg_match("/^<((\?xml-stylesheet)\s+)(.+?)\?>$/s", $ele, $m)) {
            $result .= $ele;
            continue;
        }
        $elepref = $m[1];
        $elename = $m[2];
        $elebody = $m[3];
        switch (strtolower($elename)) {
        case "a":
        case "area":
        case "base":
        case "link":
            $elebody = _page2rel_cnvtatt($elebody, "href", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . ">";
            break;
        
        case "img":
        case "input":
        case "embed":
        case "frame":
            $elebody = _page2rel_cnvtatt($elebody, "src", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . ">";
            break;
        
        case "form":
            $elebody = _page2rel_cnvtatt($elebody, "action", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . ">";
            break;
        
        case "applet":
            $elebody = _page2rel_cnvtatt($elebody, "codebase", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . ">";
            break;
        
        case "object":
            $elebody = _page2rel_cnvtatt($elebody, "classid", $base, $is_rel, $rdiff);
            $elebody = _page2rel_cnvtatt($elebody, "codebase", $base, $is_rel, $rdiff);
            $elebody = _page2rel_cnvtatt($elebody, "data", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . ">";
            break;
        
        case "script":
            $elebody = _page2rel_cnvtatt($elebody, "src", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . ">";
            $type = strtolower(_page2rel_attval($elebody, "type"));
            // Compatibility with the old "language" attribute
            if (is_null($type)) {
                $scptlang = strtolower(_page2rel_attval($elebody, "language"));
                switch ($scptlang) {
                case "javascript":
                    $type = "text/javascript";
                    break;
                case "vbscript":
                case "vbs":
                    $type = "text/vbscript";
                    break;
                case "perlscript":
                    $type = "text/perlscript";
                    break;
                }
            }
            switch ($type) {
            case "text/javascript":
            default:
                $m = _page2rel_foreign_js($source);
                break;
            case "text/vbscript":
            case "text/vbs":
                $m = _page2rel_foreign_vbs($source);
                break;
            case "text/perlscript":
                $m = _page2rel_foreign_pls($source);
                break;
            }
            $result .= $m[0];
            $source = $m[1];
            break;
        
        case "style":
            $result .= $ele;
            $type = strtolower(_page2rel_attval($elebody, "type"));
            switch ($type) {
            case "text/css":
            default:
                $m = _page2rel_foreign_css($source, $base, $is_rel, $rdiff);
                break;
            case "text/javascript":
                $m = _page2rel_foreign_jsss($source);
                break;
            }
            $result .= $m[0];
            $source = $m[1];
            break;
        
        case "param":
            $name = strtolower(_page2rel_attval($elebody, "name"));
            if (in_array($name, array("src", "movie", "filename"))) {
                $elebody = _page2rel_cnvtatt($elebody, "value", $base, $is_rel, $rdiff);
            }
            $result .= "<" . $elepref . $elebody . ">";
            break;
        
        case "?xml-stylesheet":
            $elebody = _page2rel_cnvtatt($elebody, "href", $base, $is_rel, $rdiff);
            $result .= "<" . $elepref . $elebody . "?>";
            break;
        }
    }
    // Append the remains
    $result .= $source;
    return $result;
}

// _page2rel_cnvtatt: Convert URLs in HTML abbitutes to relative/absolute
function _page2rel_cnvtatt($source, $att, $base, $is_rel, $rdiff)
{
    $result = "";
    while ($source != "" && preg_match("/^([^\"'\s]*(?:(?:\"[^\"]*\"|'[^']*')[^\"'\s]*)*)(\s*)(.*?)$/s", $source, $m)) {
        $piece = $m[1];
        $sep = $m[2];
        $source = $m[3];
        // Skip if not in the attribute="value" format
        if (    !(preg_match("/^($att)=(\")([^\"]*)\"$/i", $piece, $m)
                || preg_match("/^($att)=(')([^']*)'$/i", $piece, $m)
                || preg_match("/^($att)=()([^\"']+)$/i", $piece, $m))) {
            $result .= $piece . $sep;
            continue;
        }
        $thisatt = $m[1];
        $quote = $m[2];
        $url = $m[3];
        // $url is a variable to be replaces dynamically, as in error pages
        if ($url != "\$url") {
            $url = $is_rel? abs2rel($url, $base):
                rel2abs($url, $base, REL2ABS_SKIP_FRAGMENT, REL2ABS_NO_HOST, $rdiff);
        }
        $result .= $thisatt . "=" . $quote . $url . $quote. $sep;
    }
    return $result;
}

// _page2rel_attval: Obtain the value of a attribute
function _page2rel_attval($elebody, $att)
{
    $att = strtolower($att);
    $val = null;
    while ($elebody != "" && preg_match("/^([^\"'\s]*(?:(?:\"[^\"]*\"|'[^']*')[^\"'\s]*)*)(\s*)(.*?)$/s", $elebody, $m)) {
        $piece = $m[1];
        $elebody = $m[3];
        // Skip if not in the attribute="value" format
        if (    !(preg_match("/^($att)=(\")([^\"]*)\"$/i", $piece, $m)
                || preg_match("/^($att)=(')([^']*)'$/i", $piece, $m)
                || preg_match("/^($att)=()([^\"']+)$/i", $piece, $m))) {
            continue;
        }
        $thisatt = $m[1];
        $thisval = dh($m[3]);
        if (strtolower($thisatt) == $att) {
            $val = $thisval;
        }
    }
    return $val;
}

// _page2rel_foreign_js: Skip the next javascript block
function _page2rel_foreign_js($html)
{
    $block = "";
    $ended = false;
    while (preg_match("/^(.*?)(\"(?:[^\\\\\"]|\\\\.)*\"|'(?:[^\\\\']|\\\\.)*'|\/\*.*?\*\/|\/\/[^\n]*|<\/script>)(.*)$/is", $html, $m)) {
        // The end of the block
        if (strtolower($m[2]) == "</script>") {
            $block .= $m[1];
            $html = $m[2] . $m[3];
            $ended = true;
            break;
        }
        // Add this block
        $block .= $m[1] . $m[2];
        $html = $m[3];
    }
    // Not ended at last
    if (!$ended) {
        $block .= $html;
        $html = "";
    }
    return array($block, $html);
}

// _page2rel_foreign_vbs: Skip the next vbscript block
function _page2rel_foreign_vbs($html)
{
    $block = "";
    $ended = false;
    while (preg_match("/^(.*?)(\"(?:[^\"]|\"\")*\"|[:\n]\s*Rem\b[^\n]*|'[^\n]*|<\/script>)(.*)$/is", $html, $m)) {
        // The end of the block
        if (strtolower($m[2]) == "</script>") {
            $block .= $m[1];
            $html = $m[2] . $m[3];
            $ended = true;
            break;
        }
        // Add this block
        $block .= $m[1] . $m[2];
        $html = $m[3];
    }
    // Not ended at last
    if (!$ended) {
        $block .= $html;
        $html = "";
    }
    return array($block, $html);
}

// _page2rel_foreign_css: Skip the next cascading stylesheet block
function _page2rel_foreign_css($html, $base, $is_rel, $rdiff)
{
    $block = "";
    $ended = false;
    while (preg_match("/^(.*?)(\"(?:[^\\\\\"]|\\\\.)*\"|'(?:[^\\\\']|\\\\.)*'|\/\*.*?\*\/|<\/style>)(.*)$/is", $html, $m)) {
        // The end of the block
        if (strtolower($m[2]) == "</style>") {
            $block .= $m[1];
            $html = $m[2] . $m[3];
            $ended = true;
            break;
        }
        // Add this block
        $block .= $m[1] . $m[2];
        $html = $m[3];
    }
    // Not ended at last
    if (!$ended) {
        $block .= $html;
        $html = "";
    }
    // Convert URLs in the CSS block
    $block = _page2rel_foreign_cssurls($block, $base, $is_rel, $rdiff);
    return array($block, $html);
}

// _page2rel_foreign_cssurls: Convert URLs in a CSS block
function _page2rel_foreign_cssurls($source, $base, $is_rel, $rdiff)
{
    $result = "";
    while (preg_match("/^(.*?)(\burl\([^\"']*\)|\burl\(\"(?:[^\\\\\"]|\\\\.)*\"\)|\burl\('(?:[^\\\\']|\\\\.)*'\)|\"(?:[^\\\\\"]|\\\\.)*\"|'(?:[^\\\\']|\\\\.)*'|\/\*.*?\*\/)(.*)$/is", $source, $m)) {
        $result .= $m[1];
        $piece = $m[2];
        $source = $m[3];
        // An URL
        if (    preg_match("/^(url\()(\")((?:[^\\\\\"]|\\\\.)*)\"(\))$/i", $piece, $m)
                || preg_match("/^(url\()(')((?:[^\\\\']|\\\\.)*)'(\))$/i", $piece, $m)
                || preg_match("/^(url\()()([^\"']*)(\))$/i", $piece, $m)) {
            $m[3] = $is_rel? abs2rel($m[3], $base):
                rel2abs($m[3], $base, REL2ABS_SKIP_FRAGMENT, REL2ABS_NO_HOST, $rdiff);
            $piece = $m[1] . $m[2] . $m[3] . $m[2] . $m[4];
        }
        $result .= $piece;
    }
    // Append the remains
    $result .= $source;
    return $result;
}

// _page2rel_foreign_jsss: Skip the next javascript stylesheet block
function _page2rel_foreign_jsss($html)
{
    $block = "";
    $ended = false;
    while (preg_match("/^(.*?)(\"(?:[^\\\\\"]|\\\\.)*\"|'(?:[^\\\\']|\\\\.)*'|\/\*.*?\*\/|\/\/[^\n]*|<\/script>)(.*)$/is", $html, $m)) {
        // The end of the block
        if (strtolower($m[2]) == "</style>") {
            $block .= $m[1];
            $html = $m[2] . $m[3];
            $ended = true;
            break;
        }
        // Add this block
        $block .= $m[1] . $m[2];
        $html = $m[3];
    }
    // Not ended at last
    if (!$ended) {
        $block .= $html;
        $html = "";
    }
    return array($block, $html);
}

?>