Code Forge

<?php

/**

*

* @package phpBB3

* @version $Id$

* @copyright (c) 2005 phpBB Group

* @license http://opensource.org/licenses/gpl-license.php GNU Public License

*

*/

/**

* @ignore

*/

if (!defined('IN_PHPBB'))

{

        exit;

}

if (!class_exists('bbcode'))

{

        include($phpbb_root_path . 'includes/bbcode.' . $phpEx);

}

/**

* BBCODE FIRSTPASS

* BBCODE first pass class (functions for parsing messages for db storage)

* @package phpBB3

*/

class bbcode_firstpass extends bbcode

{

        var $message = '';

        var $warn_msg = array();

        var $parsed_items = array();

        /**

        * Parse BBCode

        */

        function parse_bbcode()

        {

                if (!$this->bbcodes)

                {

                        $this->bbcode_init();

                }

                global $user;

                $this->bbcode_bitfield = '';

                $bitfield = new bitfield();

                foreach ($this->bbcodes as $bbcode_name => $bbcode_data)

                {

                        if (isset($bbcode_data['disabled']) && $bbcode_data['disabled'])

                        {

                                foreach ($bbcode_data['regexp'] as $regexp => $replacement)

                                {

                                        if (preg_match($regexp, $this->message))

                                        {

                                                $this->warn_msg[] = sprintf($user->lang['UNAUTHORISED_BBCODE'] , '[' . $bbcode_name . ']');

                                                continue;

                                        }

                                }

                        }

                        else

                        {

                                foreach ($bbcode_data['regexp'] as $regexp => $replacement)

                                {

                                        // The pattern gets compiled and cached by the PCRE extension,

                                        // it should not demand recompilation

                                        if (preg_match($regexp, $this->message))

                                        {

                                                $this->message = preg_replace($regexp, $replacement, $this->message);

                                                $bitfield->set($bbcode_data['bbcode_id']);

                                        }

                                }

                        }

                }

                $this->bbcode_bitfield = $bitfield->get_base64();

        }

        /**

        * Prepare some bbcodes for better parsing

        */

        function prepare_bbcodes()

        {

                // Ok, seems like users instead want the no-parsing of urls, smilies, etc. after and before and within quote tags being tagged as "not a bug".

                // Fine by me ;) Will ease our live... but do not come back and cry at us, we won't hear you.

                /* Add newline at the end and in front of each quote block to prevent parsing errors (urls, smilies, etc.)

                if (strpos($this->message, '[quote') !== false && strpos($this->message, '[/quote]') !== false)

                {

                        $this->message = str_replace("\r\n", "\n", $this->message);

                        // We strip newlines and spaces after and before quotes in quotes (trimming) and then add exactly one newline

                        $this->message = preg_replace('#\[quote(=".*?")?\]\s*(.*?)\s*\[/quote\]#siu', '[quote\1]' . "\n" . '\2' ."\n[/quote]", $this->message);

                }

                */

                // Add other checks which needs to be placed before actually parsing anything (be it bbcodes, smilies, urls...)

        }

        /**

        * Init bbcode data for later parsing

        */

        function bbcode_init($allow_custom_bbcode = true)

        {

                static $rowset;

                // This array holds all bbcode data. BBCodes will be processed in this

                // order, so it is important to keep [code] in first position and

                // [quote] in second position.

                // To parse multiline URL we enable dotall option setting only for URL text

                // but not for link itself, thus [url][/url] is not affected.

                $this->bbcodes = array(

                        'code'                        => array('bbcode_id' => 8,        'regexp' => array('#\[code(?:=([a-z]+))?\](.+\[/code\])#uise' => "\$this->bbcode_code('\$1', '\$2')")),

                        'quote'                        => array('bbcode_id' => 0,        'regexp' => array('#\[quote(?:="(.*?)")?\](.+)\[/quote\]#uise' => "\$this->bbcode_quote('\$0')")),

                        'attachment'        => array('bbcode_id' => 12,        'regexp' => array('#\[attachment=([0-9]+)\](.*?)\[/attachment\]#uise' => "\$this->bbcode_attachment('\$1', '\$2')")),

                        'b'                                => array('bbcode_id' => 1,        'regexp' => array('#\[b\](.*?)\[/b\]#uise' => "\$this->bbcode_strong('\$1')")),

                        'i'                                => array('bbcode_id' => 2,        'regexp' => array('#\[i\](.*?)\[/i\]#uise' => "\$this->bbcode_italic('\$1')")),

                        'url'                        => array('bbcode_id' => 3,        'regexp' => array('#\[url(=(.*))?\](?(1)((?s).*(?-s))|(.*))\[/url\]#uiUe' => "\$this->validate_url('\$2', ('\$3') ? '\$3' : '\$4')")),

                        'img'                        => array('bbcode_id' => 4,        'regexp' => array('#\[img\](.*)\[/img\]#uiUe' => "\$this->bbcode_img('\$1')")),

                        'size'                        => array('bbcode_id' => 5,        'regexp' => array('#\[size=([\-\+]?\d+)\](.*?)\[/size\]#uise' => "\$this->bbcode_size('\$1', '\$2')")),

                        'color'                        => array('bbcode_id' => 6,        'regexp' => array('!\[color=(#[0-9a-f]{3}|#[0-9a-f]{6}|[a-z\-]+)\](.*?)\[/color\]!uise' => "\$this->bbcode_color('\$1', '\$2')")),

                        'u'                                => array('bbcode_id' => 7,        'regexp' => array('#\[u\](.*?)\[/u\]#uise' => "\$this->bbcode_underline('\$1')")),

                        'list'                        => array('bbcode_id' => 9,        'regexp' => array('#\[list(?:=(?:[a-z0-9]|disc|circle|square))?].*\[/list]#uise' => "\$this->bbcode_parse_list('\$0')")),

                        'email'                        => array('bbcode_id' => 10,        'regexp' => array('#\[email=?(.*?)?\](.*?)\[/email\]#uise' => "\$this->validate_email('\$1', '\$2')")),

                        'flash'                        => array('bbcode_id' => 11,        'regexp' => array('#\[flash=([0-9]+),([0-9]+)\](.*?)\[/flash\]#uie' => "\$this->bbcode_flash('\$1', '\$2', '\$3')"))

                );

                // Zero the parsed items array

                $this->parsed_items = array();

                foreach ($this->bbcodes as $tag => $bbcode_data)

                {

                        $this->parsed_items[$tag] = 0;

                }

                if (!$allow_custom_bbcode)

                {

                        return;

                }

                if (!is_array($rowset))

                {

                        global $db;

                        $rowset = array();

                        $sql = 'SELECT *

                                FROM ' . BBCODES_TABLE;

                        $result = $db->sql_query($sql);

                        while ($row = $db->sql_fetchrow($result))

                        {

                                $rowset[] = $row;

                        }

                        $db->sql_freeresult($result);

                }

                foreach ($rowset as $row)

                {

                        $this->bbcodes[$row['bbcode_tag']] = array(

                                'bbcode_id'        => (int) $row['bbcode_id'],

                                'regexp'        => array($row['first_pass_match'] => str_replace('$uid', $this->bbcode_uid, $row['first_pass_replace']))

                        );

                }

        }

        /**

        * Making some pre-checks for bbcodes as well as increasing the number of parsed items

        */

        function check_bbcode($bbcode, &$in)

        {

                // when using the /e modifier, preg_replace slashes double-quotes but does not

                // seem to slash anything else

                $in = str_replace("\r\n", "\n", str_replace('\"', '"', $in));

                // Trimming here to make sure no empty bbcodes are parsed accidently

                if (trim($in) == '')

                {

                        return false;

                }

                $this->parsed_items[$bbcode]++;

                return true;

        }

        /**

        * Transform some characters in valid bbcodes

        */

        function bbcode_specialchars($text)

        {

                $str_from = array('<', '>', '[', ']', '.', ':');

                $str_to = array('<', '>', '[', ']', '.', ':');

                return str_replace($str_from, $str_to, $text);

        }

        /**

        * Parse size tag

        */

        function bbcode_size($stx, $in)

        {

                global $user, $config;

                if (!$this->check_bbcode('size', $in))

                {

                        return $in;

                }

                if ($config['max_' . $this->mode . '_font_size'] && $config['max_' . $this->mode . '_font_size'] < $stx)

                {

                        $this->warn_msg[] = sprintf($user->lang['MAX_FONT_SIZE_EXCEEDED'], $config['max_' . $this->mode . '_font_size']);

                        return '[size=' . $stx . ']' . $in . '[/size]';

                }

                // Do not allow size=0

                if ($stx <= 0)

                {

                        return '[size=' . $stx . ']' . $in . '[/size]';

                }

                return '[size=' . $stx . ':' . $this->bbcode_uid . ']' . $in . '[/size:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse color tag

        */

        function bbcode_color($stx, $in)

        {

                if (!$this->check_bbcode('color', $in))

                {

                        return $in;

                }

                return '[color=' . $stx . ':' . $this->bbcode_uid . ']' . $in . '[/color:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse u tag

        */

        function bbcode_underline($in)

        {

                if (!$this->check_bbcode('u', $in))

                {

                        return $in;

                }

                return '[u:' . $this->bbcode_uid . ']' . $in . '[/u:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse b tag

        */

        function bbcode_strong($in)

        {

                if (!$this->check_bbcode('b', $in))

                {

                        return $in;

                }

                return '[b:' . $this->bbcode_uid . ']' . $in . '[/b:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse i tag

        */

        function bbcode_italic($in)

        {

                if (!$this->check_bbcode('i', $in))

                {

                        return $in;

                }

                return '[i:' . $this->bbcode_uid . ']' . $in . '[/i:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse img tag

        */

        function bbcode_img($in)

        {

                global $user, $config;

                if (!$this->check_bbcode('img', $in))

                {

                        return $in;

                }

                $in = trim($in);

                $error = false;

                $in = str_replace(' ', '%20', $in);

                // Checking urls

                if (!preg_match('#^' . get_preg_expression('url') . '$#i', $in) && !preg_match('#^' . get_preg_expression('www_url') . '$#i', $in))

                {

                        return '[img]' . $in . '[/img]';

                }

                // Try to cope with a common user error... not specifying a protocol but only a subdomain

                if (!preg_match('#^[a-z0-9]+://#i', $in))

                {

                        $in = 'http://' . $in;

                }

                if ($config['max_' . $this->mode . '_img_height'] || $config['max_' . $this->mode . '_img_width'])

                {

                        $stats = @getimagesize(htmlspecialchars_decode($in));

                        if ($stats === false)

                        {

                                $error = true;

                                $this->warn_msg[] = $user->lang['UNABLE_GET_IMAGE_SIZE'];

                        }

                        else

                        {

                                if ($config['max_' . $this->mode . '_img_height'] && $config['max_' . $this->mode . '_img_height'] < $stats[1])

                                {

                                        $error = true;

                                        $this->warn_msg[] = sprintf($user->lang['MAX_IMG_HEIGHT_EXCEEDED'], $config['max_' . $this->mode . '_img_height']);

                                }

                                if ($config['max_' . $this->mode . '_img_width'] && $config['max_' . $this->mode . '_img_width'] < $stats[0])

                                {

                                        $error = true;

                                        $this->warn_msg[] = sprintf($user->lang['MAX_IMG_WIDTH_EXCEEDED'], $config['max_' . $this->mode . '_img_width']);

                                }

                        }

                }

                if ($error || $this->path_in_domain($in))

                {

                        return '[img]' . $in . '[/img]';

                }

                return '[img:' . $this->bbcode_uid . ']' . $this->bbcode_specialchars($in) . '[/img:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse flash tag

        */

        function bbcode_flash($width, $height, $in)

        {

                global $user, $config;

                if (!$this->check_bbcode('flash', $in))

                {

                        return $in;

                }

                $in = trim($in);

                $error = false;

                // Do not allow 0-sizes generally being entered

                if ($width <= 0 || $height <= 0)

                {

                        return '[flash=' . $width . ',' . $height . ']' . $in . '[/flash]';

                }

                $in = str_replace(' ', '%20', $in);

                // Make sure $in is a URL.

                if (!preg_match('#^' . get_preg_expression('url') . '$#i', $in) &&

                        !preg_match('#^' . get_preg_expression('www_url') . '$#i', $in))

                {

                        return '[flash=' . $width . ',' . $height . ']' . $in . '[/flash]';

                }

                // Apply the same size checks on flash files as on images

                if ($config['max_' . $this->mode . '_img_height'] || $config['max_' . $this->mode . '_img_width'])

                {

                        if ($config['max_' . $this->mode . '_img_height'] && $config['max_' . $this->mode . '_img_height'] < $height)

                        {

                                $error = true;

                                $this->warn_msg[] = sprintf($user->lang['MAX_FLASH_HEIGHT_EXCEEDED'], $config['max_' . $this->mode . '_img_height']);

                        }

                        if ($config['max_' . $this->mode . '_img_width'] && $config['max_' . $this->mode . '_img_width'] < $width)

                        {

                                $error = true;

                                $this->warn_msg[] = sprintf($user->lang['MAX_FLASH_WIDTH_EXCEEDED'], $config['max_' . $this->mode . '_img_width']);

                        }

                }

                if ($error || $this->path_in_domain($in))

                {

                        return '[flash=' . $width . ',' . $height . ']' . $in . '[/flash]';

                }

                return '[flash=' . $width . ',' . $height . ':' . $this->bbcode_uid . ']' . $this->bbcode_specialchars($in) . '[/flash:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse inline attachments [ia]

        */

        function bbcode_attachment($stx, $in)

        {

                if (!$this->check_bbcode('attachment', $in))

                {

                        return $in;

                }

                return '[attachment=' . $stx . ':' . $this->bbcode_uid . ']<!-- ia' . $stx . ' -->' . trim($in) . '<!-- ia' . $stx . ' -->[/attachment:' . $this->bbcode_uid . ']';

        }

        /**

        * Parse code text from code tag

        * @access private

        */

        function bbcode_parse_code($stx, &$code)

        {

                switch (strtolower($stx))

                {

                        case 'php':

                                $remove_tags = false;

                                $str_from = array('<', '>', '[', ']', '.', ':', ':');

                                $str_to = array('<', '>', '[', ']', '.', ':', ':');

                                $code = str_replace($str_from, $str_to, $code);

                                if (!preg_match('/\<\?.*?\?\>/is', $code))

                                {

                                        $remove_tags = true;

                                        $code = "<?php $code ?>";

                                }

                                $conf = array('highlight.bg', 'highlight.comment', 'highlight.default', 'highlight.html', 'highlight.keyword', 'highlight.string');

                                foreach ($conf as $ini_var)

                                {

                                        @ini_set($ini_var, str_replace('highlight.', 'syntax', $ini_var));

                                }

                                // Because highlight_string is specialcharing the text (but we already did this before), we have to reverse this in order to get correct results

                                $code = htmlspecialchars_decode($code);

                                $code = highlight_string($code, true);

                                $str_from = array('<span style="color: ', '<font color="syntax', '</font>', '<code>', '</code>','[', ']', '.', ':');

                                $str_to = array('<span class="', '<span class="syntax', '</span>', '', '', '&#91;', '&#93;', '&#46;', '&#58;');

                                if ($remove_tags)

                                {

                                        $str_from[] = '<span class="syntaxdefault">&lt;?php </span>';

                                        $str_to[] = '';

                                        $str_from[] = '<span class="syntaxdefault">&lt;?php&nbsp;';

                                        $str_to[] = '<span class="syntaxdefault">';

                                }

                                $code = str_replace($str_from, $str_to, $code);

                                $code = preg_replace('#^(<span class="[a-z_]+">)\n?(.*?)\n?(</span>)$#is', '$1$2$3', $code);

                                if ($remove_tags)

                                {

                                        $code = preg_replace('#(<span class="[a-z]+">)?\?&gt;(</span>)#', '$1&nbsp;$2', $code);

                                }

                                $code = preg_replace('#^<span class="[a-z]+"><span class="([a-z]+)">(.*)</span></span>#s', '<span class="$1">$2</span>', $code);

                                $code = preg_replace('#(?:\s++|&nbsp;)*+</span>$#u', '</span>', $code);

                                // remove newline at the end

                                if (!empty($code) && substr($code, -1) == "\n")

                                {

                                        $code = substr($code, 0, -1);

                                }

                                return "[code=$stx:" . $this->bbcode_uid . ']' . $code . '[/code:' . $this->bbcode_uid . ']';

                        break;

                        default:

                                return '[code:' . $this->bbcode_uid . ']' . $this->bbcode_specialchars($code) . '[/code:' . $this->bbcode_uid . ']';

                        break;

                }

        }

        /**

        * Parse code tag

        * Expects the argument to start right after the opening [code] tag and to end with [/code]

        */

        function bbcode_code($stx, $in)

        {

                if (!$this->check_bbcode('code', $in))

                {

                        return $in;

                }

                // We remove the hardcoded elements from the code block here because it is not used in code blocks

                // Having it here saves us one preg_replace per message containing [code] blocks

                // Additionally, magic url parsing should go after parsing bbcodes, but for safety those are stripped out too...

                $htm_match = get_preg_expression('bbcode_htm');

                unset($htm_match[4], $htm_match[5]);

                $htm_replace = array('\1', '\1', '\2', '\1');

                $out = $code_block = '';

                $open = 1;

                while ($in)

                {

                        // Determine position and tag length of next code block

                        preg_match('#(.*?)(\[code(?:=([a-z]+))?\])(.+)#is', $in, $buffer);

                        $pos = (isset($buffer[1])) ? strlen($buffer[1]) : false;

                        $tag_length = (isset($buffer[2])) ? strlen($buffer[2]) : false;

                        // Determine position of ending code tag

                        $pos2 = stripos($in, '[/code]');

                        // Which is the next block, ending code or code block

                        if ($pos !== false && $pos < $pos2)

                        {

                                // Open new block

                                if (!$open)

                                {

                                        $out .= substr($in, 0, $pos);

                                        $in = substr($in, $pos);

                                        $stx = (isset($buffer[3])) ? $buffer[3] : '';

                                        $code_block = '';

                                }

                                else

                                {

                                        // Already opened block, just append to the current block

                                        $code_block .= substr($in, 0, $pos) . ((isset($buffer[2])) ? $buffer[2] : '');

                                        $in = substr($in, $pos);

                                }

                                $in = substr($in, $tag_length);

                                $open++;

                        }

                        else

                        {

                                // Close the block

                                if ($open == 1)

                                {

                                        $code_block .= substr($in, 0, $pos2);

                                        $code_block = preg_replace($htm_match, $htm_replace, $code_block);

                                        // Parse this code block

                                        $out .= $this->bbcode_parse_code($stx, $code_block);

                                        $code_block = '';

                                        $open--;

                                }

                                else if ($open)

                                {

                                        // Close one open tag... add to the current code block

                                        $code_block .= substr($in, 0, $pos2 + 7);

                                        $open--;

                                }

                                else

                                {

                                        // end code without opening code... will be always outside code block

                                        $out .= substr($in, 0, $pos2 + 7);

                                }

                                $in = substr($in, $pos2 + 7);

                        }

                }

                // if now $code_block has contents we need to parse the remaining code while removing the last closing tag to match up.

                if ($code_block)

                {

                        $code_block = substr($code_block, 0, -7);

                        $code_block = preg_replace($htm_match, $htm_replace, $code_block);

                        $out .= $this->bbcode_parse_code($stx, $code_block);

                }

                return $out;

        }

        /**

        * Parse list bbcode

        * Expects the argument to start with a tag

        */

        function bbcode_parse_list($in)

        {

                if (!$this->check_bbcode('list', $in))

                {

                        return $in;

                }

                // $tok holds characters to stop at. Since the string starts with a '[' we'll get everything up to the first ']' which should be the opening [list] tag

                $tok = ']';

                $out = '[';

                // First character is [

                $in = substr($in, 1);

                $list_end_tags = $item_end_tags = array();

                do

                {

                        $pos = strlen($in);

                        for ($i = 0, $tok_len = strlen($tok); $i < $tok_len; ++$i)

                        {

                                $tmp_pos = strpos($in, $tok[$i]);

                                if ($tmp_pos !== false && $tmp_pos < $pos)

                                {

                                        $pos = $tmp_pos;

                                }

                        }

                        $buffer = substr($in, 0, $pos);

                        $tok = $in[$pos];

                        $in = substr($in, $pos + 1);

                        if ($tok == ']')

                        {

                                // if $tok is ']' the buffer holds a tag

                                if (strtolower($buffer) == '/list' && sizeof($list_end_tags))

                                {

                                        // valid [/list] tag, check nesting so that we don't hit false positives

                                        if (sizeof($item_end_tags) && sizeof($item_end_tags) >= sizeof($list_end_tags))

                                        {

                                                // current li tag has not been closed

                                                $out = preg_replace('/\n?\[$/', '[', $out) . array_pop($item_end_tags) . '][';

                                        }

                                        $out .= array_pop($list_end_tags) . ']';

                                        $tok = '[';

                                }

                                else if (preg_match('#^list(=[0-9a-z]+)?$#i', $buffer, $m))

                                {

                                        // sub-list, add a closing tag

                                        if (empty($m[1]) || preg_match('/^=(?:disc|square|circle)$/i', $m[1]))

                                        {

                                                array_push($list_end_tags, '/list:u:' . $this->bbcode_uid);

                                        }

                                        else

                                        {

                                                array_push($list_end_tags, '/list:o:' . $this->bbcode_uid);

                                        }

                                        $out .= 'list' . substr($buffer, 4) . ':' . $this->bbcode_uid . ']';

                                        $tok = '[';

                                }

                                else

                                {

                                        if (($buffer == '*' || substr($buffer, -2) == '[*') && sizeof($list_end_tags))

                                        {

                                                // the buffer holds a bullet tag and we have a [list] tag open

                                                if (sizeof($item_end_tags) >= sizeof($list_end_tags))

                                                {

                                                        if (substr($buffer, -2) == '[*')

                                                        {

                                                                $out .= substr($buffer, 0, -2) . '[';

                                                        }

                                                        // current li tag has not been closed

                                                        if (preg_match('/\n\[$/', $out, $m))

                                                        {

                                                                $out = preg_replace('/\n\[$/', '[', $out);

                                                                $buffer = array_pop($item_end_tags) . "]\n[*:" . $this->bbcode_uid;

                                                        }

                                                        else

                                                        {

                                                                $buffer = array_pop($item_end_tags) . '][*:' . $this->bbcode_uid;

                                                        }

                                                }

                                                else

                                                {

                                                        $buffer = '*:' . $this->bbcode_uid;

                                                }

                                                $item_end_tags[] = '/*:m:' . $this->bbcode_uid;

                                        }

                                        else if ($buffer == '/*')

                                        {

                                                array_pop($item_end_tags);

                                                $buffer = '/*:' . $this->bbcode_uid;

                                        }

                                        $out .= $buffer . $tok;

                                        $tok = '[]';

                                }

                        }

                        else

                        {