path: root/program/lib/Roundcube/rcube_washtml.php

<?php

/**
 +-----------------------------------------------------------------------+
 | This file is part of the Roundcube Webmail client                     |
 | Copyright (C) 2008-2012, The Roundcube Dev Team                       |
 |                                                                       |
 | Licensed under the GNU General Public License version 3 or            |
 | any later version with exceptions for skins & plugins.                |
 | See the README file for a full license statement.                     |
 |                                                                       |
 | PURPOSE:                                                              |
 |   Utility class providing HTML sanityzer (based on Washtml class)     |
 +-----------------------------------------------------------------------+
 | Author: Thomas Bruederli <roundcube@gmail.com>                        |
 | Author: Aleksander Machniak <alec@alec.pl>                            |
 | Author: Frederic Motte <fmotte@ubixis.com>                            |
 +-----------------------------------------------------------------------+
 */

/**
 *                Washtml, a HTML sanityzer.
 *
 * Copyright (c) 2007 Frederic Motte <fmotte@ubixis.com>
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 * OVERVIEW:
 *
 * Wahstml take an untrusted HTML and return a safe html string.
 *
 * SYNOPSIS:
 *
 * $washer = new washtml($config);
 * $washer->wash($html);
 * It return a sanityzed string of the $html parameter without html and head tags.
 * $html is a string containing the html code to wash.
 * $config is an array containing options:
 *   $config['allow_remote'] is a boolean to allow link to remote images.
 *   $config['blocked_src'] string with image-src to be used for blocked remote images
 *   $config['show_washed'] is a boolean to include washed out attributes as x-washed
 *   $config['cid_map'] is an array where cid urls index urls to replace them.
 *   $config['charset'] is a string containing the charset of the HTML document if it is not defined in it.
 * $washer->extlinks is a reference to a boolean that is set to true if remote images were removed. (FE: show remote images link)
 *
 * INTERNALS:
 *
 * Only tags and attributes in the static lists $html_elements and $html_attributes
 * are kept, inline styles are also filtered: all style identifiers matching
 * /[a-z\-]/i are allowed. Values matching colors, sizes, /[a-z\-]/i and safe
 * urls if allowed and cid urls if mapped are kept.
 *
 * Roundcube Changes:
 * - added $block_elements
 * - changed $ignore_elements behaviour
 * - added RFC2397 support
 * - base URL support
 * - invalid HTML comments removal before parsing
 * - "fixing" unitless CSS values for XHTML output
 * - base url resolving
 */

/**
 * Utility class providing HTML sanityzer
 *
 * @package    Framework
 * @subpackage Utils
 */
class rcube_washtml
{
    /* Allowed HTML elements (default) */
    static $html_elements = array('a', 'abbr', 'acronym', 'address', 'area', 'b',
        'basefont', 'bdo', 'big', 'blockquote', 'br', 'caption', 'center',
        'cite', 'code', 'col', 'colgroup', 'dd', 'del', 'dfn', 'dir', 'div', 'dl',
        'dt', 'em', 'fieldset', 'font', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hr', 'i',
        'ins', 'label', 'legend', 'li', 'map', 'menu', 'nobr', 'ol', 'p', 'pre', 'q',
        's', 'samp', 'small', 'span', 'strike', 'strong', 'sub', 'sup', 'table',
        'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'tt', 'u', 'ul', 'var', 'wbr', 'img',
        // form elements
        'button', 'input', 'textarea', 'select', 'option', 'optgroup'
    );

    /* Ignore these HTML tags and their content */
    static $ignore_elements = array('script', 'applet', 'embed', 'object', 'style');

    /* Allowed HTML attributes */
    static $html_attribs = array('name', 'class', 'title', 'alt', 'width', 'height',
        'align', 'nowrap', 'col', 'row', 'id', 'rowspan', 'colspan', 'cellspacing',
        'cellpadding', 'valign', 'bgcolor', 'color', 'border', 'bordercolorlight',
        'bordercolordark', 'face', 'marginwidth', 'marginheight', 'axis', 'border',
        'abbr', 'char', 'charoff', 'clear', 'compact', 'coords', 'vspace', 'hspace',
        'cellborder', 'size', 'lang', 'dir', 'usemap', 'shape', 'media',
        // attributes of form elements
        'type', 'rows', 'cols', 'disabled', 'readonly', 'checked', 'multiple', 'value'
    );

    /* Elements which could be empty and be returned in short form (<tag />) */
    static $void_elements = array('area', 'base', 'br', 'col', 'command', 'embed', 'hr',
        'img', 'input', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
    );

    /* State for linked objects in HTML */
    public $extlinks = false;

    /* Current settings */
    private $config = array();

    /* Registered callback functions for tags */
    private $handlers = array();

    /* Allowed HTML elements */
    private $_html_elements = array();

    /* Ignore these HTML tags but process their content */
    private $_ignore_elements = array();

    /* Elements which could be empty and be returned in short form (<tag />) */
    private $_void_elements = array();

    /* Allowed HTML attributes */
    private $_html_attribs = array();

    /* Max nesting level */
    private $max_nesting_level;


    /**
     * Class constructor
     */
    public function __construct($p = array())
    {
        $this->_html_elements   = array_flip((array)$p['html_elements']) + array_flip(self::$html_elements) ;
        $this->_html_attribs    = array_flip((array)$p['html_attribs']) + array_flip(self::$html_attribs);
        $this->_ignore_elements = array_flip((array)$p['ignore_elements']) + array_flip(self::$ignore_elements);
        $this->_void_elements   = array_flip((array)$p['void_elements']) + array_flip(self::$void_elements);

        unset($p['html_elements'], $p['html_attribs'], $p['ignore_elements'], $p['void_elements']);

        $this->config = $p + array('show_washed' => true, 'allow_remote' => false, 'cid_map' => array());
    }

    /**
     * Register a callback function for a certain tag
     */
    public function add_callback($tagName, $callback)
    {
        $this->handlers[$tagName] = $callback;
    }

    /**
     * Check CSS style
     */
    private function wash_style($style)
    {
        $s = '';

        foreach (explode(';', $style) as $declaration) {
            if (preg_match('/^\s*([a-z\-]+)\s*:\s*(.*)\s*$/i', $declaration, $match)) {
                $cssid = $match[1];
                $str   = $match[2];
                $value = '';

                while (sizeof($str) > 0 &&
                    preg_match('/^(url\(\s*[\'"]?([^\'"\)]*)[\'"]?\s*\)'./*1,2*/
                        '|rgb\(\s*[0-9]+\s*,\s*[0-9]+\s*,\s*[0-9]+\s*\)'.
                        '|-?[0-9.]+\s*(em|ex|px|cm|mm|in|pt|pc|deg|rad|grad|ms|s|hz|khz|%)?'.
                        '|#[0-9a-f]{3,6}'.
                        '|[a-z0-9", -]+'.
                        ')\s*/i', $str, $match)
                ) {
                    if ($match[2]) {
                        if (($src = $this->config['cid_map'][$match[2]])
                            || ($src = $this->config['cid_map'][$this->config['base_url'].$match[2]])
                        ) {
                            $value .= ' url('.htmlspecialchars($src, ENT_QUOTES) . ')';
                        }
                        else if (preg_match('!^(https?:)?//[a-z0-9/._+-]+$!i', $match[2], $url)) {
                            if ($this->config['allow_remote']) {
                                $value .= ' url('.htmlspecialchars($url[0], ENT_QUOTES).')';
                            }
                            else {
                                $this->extlinks = true;
                            }
                        }
                        else if (preg_match('/^data:.+/i', $match[2])) { // RFC2397
                            $value .= ' url('.htmlspecialchars($match[2], ENT_QUOTES).')';
                        }
                    }
                    else {
                        // whitelist ?
                        $value .= ' ' . $match[0];

                        // #1488535: Fix size units, so width:800 would be changed to width:800px
                        if (preg_match('/(left|right|top|bottom|width|height)/i', $cssid)
                            && preg_match('/^[0-9]+$/', $match[0])
                        ) {
                            $value .= 'px';
                        }
                    }

                    $str = substr($str, strlen($match[0]));
                }

                if (isset($value[0])) {
                    $s .= ($s?' ':'') . $cssid . ':' . $value . ';';
                }
            }
        }

        return $s;
    }

    /**
     * Take a node and return allowed attributes and check values
     */
    private function wash_attribs($node)
    {
        $t = '';
        $washed = '';

        foreach ($node->attributes as $key => $plop) {
            $key   = strtolower($key);
            $value = $node->getAttribute($key);

            if (isset($this->_html_attribs[$key]) ||
                ($key == 'href' && ($value = trim($value))
                    && !preg_match('!^(javascript|vbscript|data:text)!i', $value)
                    && preg_match('!^([a-z][a-z0-9.+-]+:|//|#).+!i', $value))
            ) {
                $t .= ' ' . $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"';
            }
            else if ($key == 'style' && ($style = $this->wash_style($value))) {
                $quot = strpos($style, '"') !== false ? "'" : '"';
                $t .= ' style=' . $quot . $style . $quot;
            }
            else if ($key == 'background' || ($key == 'src' && strtolower($node->tagName) == 'img')) { //check tagName anyway
                if (($src = $this->config['cid_map'][$value])
                    || ($src = $this->config['cid_map'][$this->config['base_url'].$value])
                ) {
                    $t .= ' ' . $key . '="' . htmlspecialchars($src, ENT_QUOTES) . '"';
                }
                else if (preg_match('/^(http|https|ftp):.+/i', $value)) {
                    if ($this->config['allow_remote']) {
                        $t .= ' ' . $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"';
                    }
                    else {
                        $this->extlinks = true;
                        if ($this->config['blocked_src']) {
                            $t .= ' ' . $key . '="' . htmlspecialchars($this->config['blocked_src'], ENT_QUOTES) . '"';
                        }
                    }
                }
                else if (preg_match('/^data:.+/i', $value)) { // RFC2397
                    $t .= ' ' . $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"';
                }
            }
            else {
                $washed .= ($washed ? ' ' : '') . $key;
            }
        }

        return $t . ($washed && $this->config['show_washed'] ? ' x-washed="'.$washed.'"' : '');
    }

    /**
     * The main loop that recurse on a node tree.
     * It output only allowed tags with allowed attributes
     * and allowed inline styles
     */
    private function dumpHtml($node, $level = 0)
    {
        if (!$node->hasChildNodes()) {
            return '';
        }

        $level++;

        if ($this->max_nesting_level > 0 && $level == $this->max_nesting_level - 1) {
            // log error message once
            if (!$this->max_nesting_level_error) {
                $this->max_nesting_level_error = true;
                rcube::raise_error(array('code' => 500, 'type' => 'php',
                    'line' => __LINE__, 'file' => __FILE__,
                    'message' => "Maximum nesting level exceeded (xdebug.max_nesting_level={$this->max_nesting_level})"),
                    true, false);
            }
            return '<!-- ignored -->';
        }

        $node = $node->firstChild;
        $dump = '';

        do {
            switch($node->nodeType) {
            case XML_ELEMENT_NODE: //Check element
                $tagName = strtolower($node->tagName);
                if ($callback = $this->handlers[$tagName]) {
                    $dump .= call_user_func($callback, $tagName,
                        $this->wash_attribs($node), $this->dumpHtml($node, $level), $this);
                }
                else if (isset($this->_html_elements[$tagName])) {
                    $content = $this->dumpHtml($node, $level);
                    $dump .= '<' . $tagName . $this->wash_attribs($node) .
                        ($content === '' && isset($this->_void_elements[$tagName]) ? ' />' : ">$content</$tagName>");
                }
                else if (isset($this->_ignore_elements[$tagName])) {
                    $dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' not allowed -->';
                }
                else {
                    $dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' ignored -->';
                    $dump .= $this->dumpHtml($node, $level); // ignore tags not its content
                }
                break;

            case XML_CDATA_SECTION_NODE:
                $dump .= $node->nodeValue;
                break;

            case XML_TEXT_NODE:
                $dump .= htmlspecialchars($node->nodeValue);
                break;

            case XML_HTML_DOCUMENT_NODE:
                $dump .= $this->dumpHtml($node, $level);
                break;

            case XML_DOCUMENT_TYPE_NODE:
                break;

            default:
                $dump .= '<!-- node type ' . $node->nodeType . ' -->';
            }
        } while($node = $node->nextSibling);

        return $dump;
    }

    /**
     * Main function, give it untrusted HTML, tell it if you allow loading
     * remote images and give it a map to convert "cid:" urls.
     */
    public function wash($html)
    {
        // Charset seems to be ignored (probably if defined in the HTML document)
        $node = new DOMDocument('1.0', $this->config['charset']);
        $this->extlinks = false;

        $html = $this->cleanup($html);

        // Find base URL for images
        if (preg_match('/<base\s+href=[\'"]*([^\'"]+)/is', $html, $matches)) {
            $this->config['base_url'] = $matches[1];
        }
        else {
            $this->config['base_url'] = '';
        }

        // Detect max nesting level (for dumpHTML) (#1489110)
        $this->max_nesting_level = (int) @ini_get('xdebug.max_nesting_level');

        // Use optimizations if supported
        if (version_compare(PHP_VERSION, '5.4.0', '>=')) {
            @$node->loadHTML($html, LIBXML_PARSEHUGE | LIBXML_COMPACT);
        }
        else {
            @$node->loadHTML($html);
        }

        return $this->dumpHtml($node);
    }

    /**
     * Getter for config parameters
     */
    public function get_config($prop)
    {
        return $this->config[$prop];
    }

    /**
     * Clean HTML input
     */
    private function cleanup($html)
    {
        // special replacements (not properly handled by washtml class)
        $html_search = array(
            '/(<\/nobr>)(\s+)(<nobr>)/i',       // space(s) between <NOBR>
            '/<title[^>]*>[^<]*<\/title>/i',    // PHP bug #32547 workaround: remove title tag
            '/^(\0\0\xFE\xFF|\xFF\xFE\0\0|\xFE\xFF|\xFF\xFE|\xEF\xBB\xBF)/',    // byte-order mark (only outlook?)
            '/<html\s[^>]+>/i',                 // washtml/DOMDocument cannot handle xml namespaces
        );

        $html_replace = array(
            '\\1'.' &nbsp; '.'\\3',
            '',
            '',
            '<html>',
        );
        $html = preg_replace($html_search, $html_replace, trim($html));

        //-> Replace all of those weird MS Word quotes and other high characters
        $badwordchars = array(
            "\xe2\x80\x98", // left single quote
            "\xe2\x80\x99", // right single quote
            "\xe2\x80\x9c", // left double quote
            "\xe2\x80\x9d", // right double quote
            "\xe2\x80\x94", // em dash
            "\xe2\x80\xa6" // elipses
        );
        $fixedwordchars = array(
            "'",
            "'",
            '"',
            '"',
            '&mdash;',
            '...'
        );
        $html = str_replace($badwordchars, $fixedwordchars, $html);

        // PCRE errors handling (#1486856), should we use something like for every preg_* use?
        if ($html === null && ($preg_error = preg_last_error()) != PREG_NO_ERROR) {
            $errstr = "Could not clean up HTML message! PCRE Error: $preg_error.";

            if ($preg_error == PREG_BACKTRACK_LIMIT_ERROR) {
                $errstr .= " Consider raising pcre.backtrack_limit!";
            }
            if ($preg_error == PREG_RECURSION_LIMIT_ERROR) {
                $errstr .= " Consider raising pcre.recursion_limit!";
            }

            rcube::raise_error(array('code' => 620, 'type' => 'php',
                'line' => __LINE__, 'file' => __FILE__,
                'message' => $errstr), true, false);

            return '';
        }

        // fix (unknown/malformed) HTML tags before "wash"
        $html = preg_replace_callback('/(<(?!\!)[\/]*)([^\s>]+)([^>]*)/', array($this, 'html_tag_callback'), $html);

        // Remove invalid HTML comments (#1487759)
        // Don't remove valid conditional comments
        // Don't remove MSOutlook (<!-->) conditional comments (#1489004)
        $html = preg_replace('/<!--[^->\[\n]+>/', '', $html);

        // fix broken nested lists
        self::fix_broken_lists($html);

        // turn relative into absolute urls
        $html = self::resolve_base($html);

        return $html;
    }

    /**
     * Callback function for HTML tags fixing
     */
    public static function html_tag_callback($matches)
    {
        $tagname = $matches[2];
        $tagname = preg_replace(array(
            '/:.*$/',               // Microsoft's Smart Tags <st1:xxxx>
            '/[^a-z0-9_\[\]\!-]/i', // forbidden characters
        ), '', $tagname);

        // fix invalid closing tags - remove any attributes (#1489446)
        if ($matches[1] == '</') {
            $matches[3] = '';
        }

        return $matches[1] . $tagname . $matches[3];
    }

    /**
     * Convert all relative URLs according to a <base> in HTML
     */
    public static function resolve_base($body)
    {
        // check for <base href=...>
        if (preg_match('!(<base.*href=["\']?)([hftps]{3,5}://[a-z0-9/.%-]+)!i', $body, $regs)) {
            $replacer = new rcube_base_replacer($regs[2]);
            $body     = $replacer->replace($body);
        }

        return $body;
    }

    /**
     * Fix broken nested lists, they are not handled properly by DOMDocument (#1488768)
     */
    public static function fix_broken_lists(&$html)
    {
        // do two rounds, one for <ol>, one for <ul>
        foreach (array('ol', 'ul') as $tag) {
            $pos = 0;
            while (($pos = stripos($html, '<' . $tag, $pos)) !== false) {
                $pos++;

                // make sure this is an ol/ul tag
                if (!in_array($html[$pos+2], array(' ', '>'))) {
                    continue;
                }

                $p      = $pos;
                $in_li  = false;
                $li_pos = 0;

                while (($p = strpos($html, '<', $p)) !== false) {
                    $tt = strtolower(substr($html, $p, 4));

                    // li open tag
                    if ($tt == '<li>' || $tt == '<li ') {
                        $in_li = true;
                        $p += 4;
                    }
                    // li close tag
                    else if ($tt == '</li' && in_array($html[$p+4], array(' ', '>'))) {
                        $li_pos = $p;
                        $p += 4;
                        $in_li = false;
                    }
                    // ul/ol closing tag
                    else if ($tt == '</' . $tag && in_array($html[$p+4], array(' ', '>'))) {
                        break;
                    }
                    // nested ol/ul element out of li
                    else if (!$in_li && $li_pos && ($tt == '<ol>' || $tt == '<ol ' || $tt == '<ul>' || $tt == '<ul ')) {
                        // find closing tag of this ul/ol element
                        $element = substr($tt, 1, 2);
                        $cpos    = $p;
                        do {
                            $tpos = stripos($html, '<' . $element, $cpos+1);
                            $cpos = stripos($html, '</' . $element, $cpos+1);
                        }
                        while ($tpos !== false && $cpos !== false && $cpos > $tpos);

                        // not found, this is invalid HTML, skip it
                        if ($cpos === false) {
                            break;
                        }

                        // get element content
                        $end     = strpos($html, '>', $cpos);
                        $len     = $end - $p + 1;
                        $element = substr($html, $p, $len);

                        // move element to the end of the last li
                        $html    = substr_replace($html, '', $p, $len);
                        $html    = substr_replace($html, $element, $li_pos, 0);

                        $p = $end;
                    }
                    else {
                        $p++;
                    }
                }
            }
        }
    }
}