classTextile.php

<?php

/**
 * Example: get XHTML from a given Textile-markup string ($string)
 *
 * $textile = new Textile();
 * echo $textile->textileThis($string);
 *
 */

/*
_____________
T E X T I L E

A Humane Web Text Generator

Version 2.4.1

Copyright (c) 2003-2004, Dean Allen <dean@textism.com>
All rights reserved.

Thanks to Carlo Zottmann <carlo@g-blog.net> for refactoring
Textile's procedural code into a class framework

Additions and fixes Copyright (c) 2006    Alex Shiels       http://thresholdstate.com/
Additions and fixes Copyright (c) 2010    Stef Dawson       http://stefdawson.com/
Additions and fixes Copyright (c) 2010-12 Netcarver         http://github.com/netcarver
Additions and fixes Copyright (c) 2011    Jeff Soo          http://ipsedixit.net
Additions and fixes Copyright (c) 2012    Robert Wetzlmayr  http://wetzlmayr.com/

_____________
L I C E N S E

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

* Redistributions of source code must retain the above copyright notice,
  this list of conditions and the following disclaimer.

* 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.

* Neither the name Textile nor the names of its contributors may be used to
  endorse or promote products derived from this software without specific
  prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "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 COPYRIGHT OWNER OR CONTRIBUTORS 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.

_________
U S A G E

Block modifier syntax:

    Header: h(1-6).
    Paragraphs beginning with 'hn. ' (where n is 1-6) are wrapped in header tags.
    Example: h1. Header... -> <h1>Header...</h1>

    Paragraph: p. (also applied by default)
    Example: p. Text -> <p>Text</p>

    Blockquote: bq.
    Example: bq. Block quotation... -> <blockquote>Block quotation...</blockquote>

    Blockquote with citation: bq.:http://citation.url
    Example: bq.:http://textism.com/ Text...
    ->    <blockquote cite="http://textism.com">Text...</blockquote>

    Footnote: fn(1-100).
    Example: fn1. Footnote... -> <p id="fn1">Footnote...</p>

    Numeric list: #, ##
    Consecutive paragraphs beginning with # are wrapped in ordered list tags.
    Example: <ol><li>ordered list</li></ol>

    Bulleted list: *, **
    Consecutive paragraphs beginning with * are wrapped in unordered list tags.
    Example: <ul><li>unordered list</li></ul>

    Definition list:
        Terms ;, ;;
        Definitions :, ::
    Consecutive paragraphs beginning with ; or : are wrapped in definition list tags.
    Example: <dl><dt>term</dt><dd>definition</dd></dl>

    Redcloth-style Definition list:
        - Term1 := Definition1
        - Term2 := Extended
          definition =:

Phrase modifier syntax:

           _emphasis_    ->     <em>emphasis</em>
           __italic__    ->     <i>italic</i>
             *strong*    ->     <strong>strong</strong>
             **bold**    ->     <b>bold</b>
         ??citation??    ->     <cite>citation</cite>
       -deleted text-    ->     <del>deleted</del>
      +inserted text+    ->     <ins>inserted</ins>
        ^superscript^    ->     <sup>superscript</sup>
          ~subscript~    ->     <sub>subscript</sub>
               @code@    ->     <code>computer code</code>
          %(bob)span%    ->     <span class="bob">span</span>

        ==notextile==    ->     leave text alone (do not format)

       "linktext":url    ->     <a href="url">linktext</a>
"linktext(title)":url    ->     <a href="url" title="title">linktext</a>
              "$":url    ->     <a href="url">url</a>
       "$(title)":url    ->     <a href="url" title="title">url</a>

           !imageurl!    ->     <img src="imageurl" />
 !imageurl(alt text)!    ->     <img src="imageurl" alt="alt text" />
   !imageurl!:linkurl    ->     <a href="linkurl"><img src="imageurl" /></a>

ABC(Always Be Closing)   ->     <acronym title="Always Be Closing">ABC</acronym>


Linked Notes:
============

    Allows the generation of an automated list of notes with links.

    Linked notes are composed of three parts, a set of named _definitions_, a set of
    _references_ to those definitions and one or more _placeholders_ indicating where
    the consolidated list of notes is to be placed in your document.

    Definitions.
    -----------

    Each note definition must occur in its own paragraph and should look like this...

    note#mynotelabel. Your definition text here.

    You are free to use whatever label you wish after the # as long as it is made up
    of letters, numbers, colon(:) or dash(-).

    References.
    ----------

    Each note reference is marked in your text like this[#mynotelabel] and
    it will be replaced with a superscript reference that links into the list of
    note definitions.

    List Placeholder(s).
    -------------------

    The note list can go anywhere in your document. You have to indicate where
    like this...

    notelist.

    notelist can take attributes (class#id) like this: notelist(class#id).

    By default, the note list will show each definition in the order that they
    are referenced in the text by the _references_. It will show each definition with
    a full list of backlinks to each reference. If you do not want this, you can choose
    to override the backlinks like this...

    notelist(class#id)!.    Produces a list with no backlinks.
    notelist(class#id)^.    Produces a list with only the first backlink.

    Should you wish to have a specific definition display backlinks differently to this
    then you can override the backlink method by appending a link override to the
    _definition_ you wish to customise.

    note#label.    Uses the citelist's setting for backlinks.
    note#label!.   Causes that definition to have no backlinks.
    note#label^.   Causes that definition to have one backlink (to the first ref.)
    note#label*.   Causes that definition to have all backlinks.

    Any unreferenced notes will be left out of the list unless you explicitly state
    you want them by adding a '+'. Like this...

    notelist(class#id)!+. Giving a list of all notes without any backlinks.

    You can mix and match the list backlink control and unreferenced links controls
    but the backlink control (if any) must go first. Like so: notelist^+. , not
    like this: notelist+^.

    Example...
        Scientists say[#lavader] the moon is small.

        note#other. An unreferenced note.

        note#lavader(myliclass). "Proof":url of a small moon.

        notelist(myclass#myid)+.

        Would output (the actual IDs used would be randomised)...

        <p>Scientists say<sup><a href="#def_id_1" id="ref_id_1a">1</sup> the moon is small.</p>

        <ol class="myclass" id="myid">
            <li class="myliclass"><a href="#ref_id_1a"><sup>a</sup></a><span id="def_id_1"> </span><a href="url">Proof</a> of a small moon.</li>
            <li>An unreferenced note.</li>
        </ol>

        The 'a b c' backlink characters can be altered too.
        For example if you wanted the notes to have numeric backlinks starting from 1:

        notelist:1.

Table syntax:

    Simple tables:

        |a|simple|table|row|
        |And|Another|table|row|
        |With an||empty|cell|

        |=. My table caption goes here  (NB. Table captions *must* be the first line of the table else treated as a center-aligned cell.)
        |_. A|_. table|_. header|_.row|
        |A|simple|table|row|

    Tables with attributes:

        table{border:1px solid black}. My table summary here
        {background:#ddd;color:red}. |{}| | | |

    To specify thead / tfoot / tbody groups, add one of these on its own line
    above the row(s) you wish to wrap (you may specify attributes before the dot):

        |^.     # thead
        |-.     # tbody
        |~.     # tfoot

    Column groups:

        |:\3. 100|

        Becomes:
            <colgroup span="3" width="100"></colgroup>

        You can omit either or both of the \N or width values. You may also
        add cells after the colgroup definition to specify col elements with
        span, width, or standard Textile attributes:

        |:. 50|(firstcol). |\2. 250||300|

        Becomes:
            <colgroup width="50">
                <col class="firstcol" />
                <col span="2" width="250" />
                <col />
                <col width="300" />
            </colgroup>

        (Note that, per the HTML specification, you should not add span
        to the colgroup if specifying col elements.)

Applying Attributes:

    Most anywhere Textile code is used, attributes such as arbitrary css style,
    css classes, and ids can be applied. The syntax is fairly consistent.

    The following characters quickly alter the alignment of block elements:

        <  ->  left align     ex. p<. left-aligned para
        >  ->  right align         h3>. right-aligned header 3
        =  ->  centred             h4=. centred header 4
        <> ->  justified         p<>. justified paragraph

    These will change vertical alignment in table cells:

        ^  ->  top           ex. |^. top-aligned table cell|
        -  ->  middle           |-. middle aligned|
        ~  ->  bottom           |~. bottom aligned cell|

    Plain (parentheses) inserted between block syntax and the closing dot-space
    indicate classes and ids:

        p(hector). paragraph -> <p class="hector">paragraph</p>

        p(#fluid). paragraph -> <p id="fluid">paragraph</p>

        (classes and ids can be combined)
        p(hector#fluid). paragraph -> <p class="hector" id="fluid">paragraph</p>

    Curly {brackets} insert arbitrary css style

        p{line-height:18px}. paragraph -> <p style="line-height:18px">paragraph</p>

        h3{color:red}. header 3 -> <h3 style="color:red">header 3</h3>

    Square [brackets] insert language attributes

        p[no]. paragraph -> <p lang="no">paragraph</p>

        %[fr]phrase% -> <span lang="fr">phrase</span>

    Usually Textile block element syntax requires a dot and space before the block
    begins, but since lists don't, they can be styled just using braces

        #{color:blue} one  ->  <ol style="color:blue">
        # big                    <li>one</li>
        # list                    <li>big</li>
                                <li>list</li>
                               </ol>

    Using the span tag to style a phrase

        It goes like this, %{color:red}the fourth the fifth%
              -> It goes like this, <span style="color:red">the fourth the fifth</span>

Ordered List Start & Continuation:

    You can control the start attribute of an ordered list like so;

        #5 Item 5
        # Item 6

    You can resume numbering list items after some intervening anonymous block like so...

        #_ Item 7
        # Item 8

*/


/**
 * Class to allow simple assignment to members of the internal data array
 **/
class TextileBag
{
    protected $data;


    public function __construct($initial_data)
    {
        $this->data = (is_array($initial_data)) ? $initial_data : array();
    }


    /**
     * Allows setting of an element in the $data array. eg...
     *
     * $bag->key(value);
     *
     * ...sets $bag's $data['key'] to $value provided $value is not empty.
     * The set can be made forced by following $value with true...
     *
     * $bag->key(value, true);
     *
     * Would force the value into the data array even if it were empty.
     **/
    public function __call($k, $params)
    {
        $allow_empty = isset($params[1]) && is_bool($params[1]) ? $params[1] : false;
        if ($allow_empty || '' != $params[0])
            $this->data[$k] = $params[0];

        return $this;
    }
}


/**
 * Class to allow contruction of HTML tags on conversion of an object to a string
 *
 * Example usage...
 *
 * $img = new TextileTag('img')->class('big blue')->src('images/elephant.jpg');
 * echo $img;
 **/
class TextileTag extends TextileBag
{
    protected $tag;
    protected $selfclose;


    public function __construct($name, $attribs=array(), $selfclosing=true)
    {
        parent::__construct($attribs);
        $this->tag = $name;
        $this->selfclose = $selfclosing;
    }


    public function __toString() {
        $attribs = '';

        if (count($this->data)) {
            ksort($this->data);
            foreach ($this->data as $k=>$v)
                $attribs .= " $k=\"$v\"";
        }

        if ($this->tag)
            $o = '<' . $this->tag . $attribs . (($this->selfclose) ? " />" : '>');
        else
            $o = $attribs;

        return $o;
    }
}



class Textile
{
    protected $hlgn;
    protected $vlgn;
    protected $clas;
    protected $lnge;
    protected $styl;
    protected $cspn;
    protected $rspn;
    protected $a;
    protected $s;
    protected $c;
    protected $pnct;
    protected $rel;
    protected $fn;

    protected $shelf = array();
    protected $restricted = false;
    protected $noimage = false;
    protected $lite = false;
    protected $url_schemes = array();
    protected $glyph_search  = null;
    protected $glyph_replace = null;
    protected $rebuild_glyphs = true;
    protected $relativeImagePrefix = '';
    protected $max_span_depth = 5;

    protected $ver = '2.4.1';

    protected $doc_root;

    protected $doctype;

    protected $symbols;


    /**
     * Constructor for an instance of class Textile.
     *
     * @access public
     * @param  string $doctype  The output document type to target
     * @return void
     **/
    public function __construct($doctype = 'xhtml')
    {
        $doctype_whitelist = array(
            'xhtml',
            'html5',
        );
        $doctype = strtolower($doctype);
        if (!in_array($doctype, $doctype_whitelist))
            $this->doctype = 'xhtml';
        else
            $this->doctype = $doctype;

        // Basic symbols used in textile glyph replacements. To override these, call
        // setSymbol('symbol_name', 'new_string') before calling textileThis() or
        // textileRestricted().
        $this->symbols = array(
            'quote_single_open'  => '&#8216;',
            'quote_single_close' => '&#8217;',
            'quote_double_open'  => '&#8220;',
            'quote_double_close' => '&#8221;',
            'apostrophe'         => '&#8217;',
            'prime'              => '&#8242;',
            'prime_double'       => '&#8243;',
            'ellipsis'           => '&#8230;',
            'emdash'             => '&#8212;',
            'endash'             => '&#8211;',
            'dimension'          => '&#215;',
            'trademark'          => '&#8482;',
            'registered'         => '&#174;',
            'copyright'          => '&#169;',
            'half'               => '&#189;',
            'quarter'            => '&#188;',
            'threequarters'      => '&#190;',
            'degrees'            => '&#176;',
            'plusminus'          => '&#177;',
            'fn_ref_pattern'     => '<sup{atts}>{marker}</sup>',
            'fn_foot_pattern'    => '<sup{atts}>{marker}</sup>',
            'nl_ref_pattern'     => '<sup{atts}>{marker}</sup>',
        );

        $this->hlgn  = "(?:\<(?!>)|&lt;&gt;|&gt;|&lt;|(?<!<)\>|\<\>|\=|[()]+(?! ))";
        $this->vlgn  = "[\-^~]";
        $this->clas  = "(?:\([^)\n]+\))";    // Don't allow classes/ids/languages/styles to span across newlines if used in a dotall regex
        $this->lnge  = "(?:\[[^]\n]+\])";
        $this->styl  = "(?:\{[^}\n]+\})";
        $this->cspn  = "(?:\\\\\d+)";
        $this->rspn  = "(?:\/\d+)";
        $this->a     = "(?:$this->hlgn|$this->vlgn)*";
        $this->s     = "(?:$this->cspn|$this->rspn)*";
        $this->c     = "(?:$this->clas|$this->styl|$this->lnge|$this->hlgn)*";
        $this->lc    = "(?:$this->clas|$this->styl|$this->lnge)*";
        $this->pnct  = '[\!"#\$%&\'()\*\+,\-\./:;<=>\?@\[\\\]\^_`{\|}\~]';
        $this->urlch = '[\w"$\-_.+!*\'(),";\/?:@=&%#{}|\\^~\[\]`]';
        $this->syms  = '¤§µ¶†‡•∗∴◊♠♣♥♦';

        $pnc = '[[:punct:]]';
        $this->mb   = is_callable('mb_strlen');
        $this->cmap = array(0x0080, 0xffff, 0, 0xffff);

        $this->restricted_url_schemes   = array('http','https','ftp','mailto');
        $this->unrestricted_url_schemes = array('http','https','ftp','mailto','file','tel','callto','sftp');

        if (@preg_match('/\pL/u', 'a')) {
            $this->regex_snippets = array(
                'acr' => '\p{Lu}\p{Nd}',
                'abr' => '\p{Lu}',
                'nab' => '\p{Ll}',
                'wrd' => '(?:\p{L}|\p{M}|\p{N}|\p{Pc})',
                'mod' => 'u', // Make sure to mark the unicode patterns as such, Some servers seem to need this.
                'cur' => '\p{Sc}',
            );
        } else {
            $this->regex_snippets = array(
                'acr' => 'A-Z0-9',
                'abr' => 'A-Z',
                'nab' => 'a-z',
                'wrd' => '\w',
                'mod' => '',
                'cur' => '',
            );
        }
        extract($this->regex_snippets);
        $this->urlch = '['.$wrd.'"$\-_.+!*\'(),";\/?:@=&%#{}|\\^~\[\]`]';

        $this->span_tags = array(
            '*'  => 'strong',
            '**' => 'b',
            '??' => 'cite',
            '_'  => 'em',
            '__' => 'i',
            '-'  => 'del',
            '%'  => 'span',
            '+'  => 'ins',
            '~'  => 'sub',
            '^'  => 'sup',
        );

        if (defined('DIRECTORY_SEPARATOR'))
            $this->ds = constant('DIRECTORY_SEPARATOR');
        else
            $this->ds = '/';

        $this->doc_root = @$_SERVER['DOCUMENT_ROOT'];
        if (!$this->doc_root)
            $this->doc_root = @$_SERVER['PATH_TRANSLATED']; // IIS

        $this->doc_root = rtrim($this->doc_root, $this->ds).$this->ds;
    }


    /**
     * Call this (if needed) post constructor call to redefine a substitution symbol to
     * be used when parsing a textile document.
     *
     * @access public
     * @param  string $name  Name of the symbol to assign a new value to.
     * @param  string $value New value for the symbol.
     * @return object $this
     **/
    public function setSymbol($name, $value)
    {
        $this->symbols[$name] = $value;
        $this->rebuild_glyphs = true;
        return $this;
    }


    /**
     * getSymbol() returns an array containing the symbol table or the value of the named symbol
     *
     * @access public
     * @param  string $name  The name of the symbol to access or null if requesting the symbol table
     * @return array|string The symbol table or the requested symbol
     **/
    public function getSymbol($name=null)
    {
        return ($name) ? @$this->symbols['name'] : $this->symbols;
    }


    /**
     * Allows client systems to have textile convert relative image paths to
     * absolute (or prefixed) paths.
     *
     * @access public
     * @param  string $prefix  The string to prefix all relative image paths with
     * @return object $this
     **/
    public function setRelativeImagePrefix($prefix='')
    {
        $this->relativeImagePrefix = $prefix;
        return $this;
    }


    /**
     * Returns the internal version of this instance of textile.
     *
     * @access public
     * @return string Version
     **/
    public function getVersion()
    {
        return $this->ver;
    }


    /**
     * Encodes the given text.
     *
     * @access public
     * @param  string $text The text to be encoded.
     * @return string The encoded text.
     **/
    public function textileEncode($text)
    {
        $text = preg_replace("/&(?![#a-z0-9]+;)/i", "x%x%", $text);
        $text = str_replace("x%x%", "&amp;", $text);
        return $text;
    }


    /**
     * Causes an un-restricted parse of the input textile text to start.
     *
     * @access public
     * @param  string $text      The input document in textile format
     * @param  string $lite      Optional flag to switch the parser into lite mode.
     * @param  string $encode    Optional flag that causes the unput document to be encoded and returned.
     * @param  string $noimage   Optional flag controlling the conversion of images into HTML <img/> tags.
     * @param  string $strict    Optional flag controlling the application of whitespace cleanup prior to parsing the text.
     * @param  string $rel       Relationship to apply to all generated links.
     * @return string The text from the input document
     **/
    public function textileThis($text, $lite = '', $encode = '', $noimage = '', $strict = '', $rel = '')
    {
        $this->prepare($lite, $noimage, $rel);
        $this->url_schemes = $this->unrestricted_url_schemes;

        if ($encode)  // Use of the $encode flag is discouraged. Calling textileEncode() is prefered.
            return $this->textileEncode($text);

        if (!$strict)
            $text = $this->cleanWhiteSpace($text);

        return $this->textileCommon($text, $lite);
    }


    /**
     * Causes a restricted parse of the input textile text. Use this on any untrusted user input.
     *
     * @access public
     * @param  string $text      The input document in textile format
     * @param  string $lite      Optional flag to switch the parser into lite mode. Lite mode is the default.
     * @param  string $noimage   Optional flag controlling the conversion of images into HTML <img/> tags. noimage mode is the default.
     * @param  string $rel       Relationship to apply to all generated links. 'nofollow' is the default
     * @return string The text from the input document
     **/
    public function textileRestricted($text, $lite = 1, $noimage = 1, $rel = 'nofollow')
    {
        $this->prepare($lite, $noimage, $rel);
        $this->url_schemes = $this->restricted_url_schemes;
        $this->restricted = true;

        // Escape any raw html
        $text = $this->encodeHTML($text, 0);
        $text = $this->cleanWhiteSpace($text);

        return $this->textileCommon($text, $lite);
    }


    /**
     * Perform common parse actions
     *
     * @internal
     **/
    protected function textileCommon($text, $lite)
    {
        if ($lite) {
            $this->btag = array('bq', 'p');
            $text = $this->block($text."\n\n");
        } else {
            $this->btag = array('bq', 'p', 'bc', 'notextile', 'pre', 'h[1-6]', 'fn\d+', '###');
            $text = $this->block($text);
            $text = $this->placeNoteLists($text);
        }

        $text = $this->retrieve($text);
        $text = $this->replaceGlyphs($text);
        $text = $this->retrieveTags($text);
        $text = $this->retrieveURLs($text);

        $text = str_replace("<br />", "<br />\n", $text);

        return $text;
    }


    /**
     * If needed, prepares the glyph find-and-replace patterns from the internal symbol table
     *
     * @internal
     * @return void
     **/
    protected function prepGlyphs()
    {
        if ((null!==$this->glyph_search) && (null!==$this->glyph_replace) && !$this->rebuild_glyphs)
            return;

        extract($this->symbols, EXTR_PREFIX_ALL, 'txt');
        extract($this->regex_snippets );
        $pnc = '[[:punct:]]';

        if ($cur)
            $cur = '(?:['.$cur.']\s*)?';

        $this->glyph_search = array(
            '/([0-9]+[\])]?[\'"]? ?)[xX]( ?[\[(]?)(?=[+-]?'.$cur.'[0-9]*\.?[0-9]+)/'.$mod,   // Dimension sign
            '/('.$wrd.'|\))\'('.$wrd.')/'.$mod,     // I'm an apostrophe
            '/(\s)\'(\d+'.$wrd.'?)\b(?![.]?['.$wrd.']*?\')/'.$mod,    // Back in '88/the '90s but not in his '90s', '1', '1.' '10m' or '5.png'
            "/([([{])'(?=\S)/",                     // Single open following open bracket
            '/(\S)\'(?=\s|'.$pnc.'|<|$)/',          // Single closing
            "/'/",                                  // Default single opening
            '/([([{])"(?=\S)/',                     // Double open following an open bracket. Allows things like Hello ["(Mum) & dad"]
            '/(\S)"(?=\s|'.$pnc.'|<|$)/',           // Double closing
            '/"/',                                  // Default double opening
            '/\b(['.$abr.']['.$acr.']{2,})\b(?:[(]([^)]*)[)])/'.$mod,  // 3+ uppercase acronym
            '/(?<=\s|^|[>(;-])(['.$abr.']{3,})(['.$nab.']*)(?=\s|'.$pnc.'|<|$)(?=[^">]*?(<|$))/'.$mod,  // 3+ uppercase
            '/([^.]?)\.{3}/',                       // Ellipsis
            '/--/',                                 // em dash
            '/ - /',                                // en dash
            '/(\b ?|\s|^)[([]TM[])]/i',             // Trademark
            '/(\b ?|\s|^)[([]R[])]/i',              // Registered
            '/(\b ?|\s|^)[([]C[])]/i',              // Copyright
            '/[([]1\/4[])]/',                       // 1/4
            '/[([]1\/2[])]/',                       // 1/2
            '/[([]3\/4[])]/',                       // 3/4
            '/[([]o[])]/',                          // Degrees -- that's a small 'oh'
            '/[([]\+\/-[])]/',                      // Plus minus
        );

        $this->glyph_replace = array(
            '$1'.$txt_dimension.'$2',               // Dimension sign
            '$1'.$txt_apostrophe.'$2',              // I'm an apostrophe
            '$1'.$txt_apostrophe.'$2',              // Back in '88
            '$1'.$txt_quote_single_open,            // Single open following open bracket
            '$1'.$txt_quote_single_close,           // Single closing
            $txt_quote_single_open,                 // Default single opening
            '$1'.$txt_quote_double_open,            // Double open following open bracket
            '$1'.$txt_quote_double_close,           // Double closing
            $txt_quote_double_open,                 // Default double opening
            (('html5' === $this->doctype) ? '<abbr title="$2">$1</abbr>' : '<acronym title="$2">$1</acronym>'),     // 3+ uppercase acronym
            '<span class="caps">glyph:$1</span>$2', // 3+ uppercase
            '$1'.$txt_ellipsis,                     // Ellipsis
            $txt_emdash,                            // em dash
            ' '.$txt_endash.' ',                    // en dash
            '$1'.$txt_trademark,                    // Trademark
            '$1'.$txt_registered,                   // Registered
            '$1'.$txt_copyright,                    // Copyright
            $txt_quarter,                           // 1/4
            $txt_half,                              // 1/2
            $txt_threequarters,                     // 3/4
            $txt_degrees,                           // Degrees
            $txt_plusminus,                         // Plus minus
        );

        $this->rebuild_glyphs = false; // No need to rebuild next run unless a symbol is redefined
    }


    /**
     * Prepares the transient internal state of the textile parser in preparation for
     * parsing a new document.
     *
     * @internal
     *
     * @param  bool|string $lite    Set to true/non-empty to parse in lite mode. Default ''.
     * @param  bool|string $noimage Disallow images in the generated docuemnt
     * @param  string      $rel     A relationship to be applied to all links. eg. 'nofollow'
     * @return void
     **/
    protected function prepare($lite, $noimage, $rel)
    {
        $this->unreferencedNotes = array();
        $this->notelist_cache    = array();
        $this->notes      = array();
        $this->urlshelf   = array();
        $this->urlrefs    = array();
        $this->shelf      = array();
        $this->fn         = array();
        $this->span_depth = 0;
        $this->tag_index  = 1;
        $this->note_index = 1;
        $this->rel = ($rel) ? ' rel="'.$rel.'"' : '';
        $this->lite       = $lite;
        $this->noimage    = $noimage;
        $this->prepGlyphs();
    }


    /**
     * @internal
     **/
    protected function cleanAttribs($in)
    {
        $tmp    = $in;
        $before = -1;
        $after  =  0;
        $max    =  3;
        $i      =  0;
        while (($after != $before) && ($i < $max))
        {
            $before = strlen($tmp);
            $tmp    = rawurldecode($tmp);
            $after  = strlen($tmp);
            $i++;
        }

        if ($i === $max) // If we hit the max allowed decodes, assume the input is tainted and consume it.
            $out = '';
        else
            $out = str_replace(array('"', "'", '='), '', $tmp);
        return $out;
    }


    /**
     * Helper method that creates a new instance of TextileTag
     *
     * @internal
     *
     * @param  string $name        The type of tag to create. eg. newTag('p',...) for a paragraph tag.
     * @param         $atts        The textile attributes to apply to the tag
     * @param  bool   $selfclosing Determines if the tag should be selfclosing. Default: true
     * @return TextileTag
     **/
    protected function newTag($name, $atts, $selfclosing = true)
    {
        return new TextileTag($name, $atts, $selfclosing);
    }


    /**
     * Parses textile attributes
     *
     * @internal
     *
     * @param   string $in         The textile attribute string to be parsed
     * @param   string $element    Focus the routine to interpret the attributes as applying to a specific HTML tag
     * @param   int    $include_id A value interpreted as a true when cast to bool allows ids to be included in the output
     * @param   string $autoclass  An additional class or classes to be applied to the output
     * @return  array  HTML attributes as key=>value mappings
     **/
    protected function parseAttribs($in, $element = "", $include_id = 1, $autoclass = '')
    {
        $out = '';
        $o = $this->parseAttribsToArray($in, $element, $include_id, $autoclass);
        ksort($o);

        if (count($o))
            foreach ($o as $k=>$v) $out .= " $k=\"$v\"";

        return $out;
    }


    /**
     * @internal
     **/
    protected function parseAttribsToArray($in, $element = "", $include_id = 1, $autoclass = '')
    {
        $style = '';
        $class = '';
        $lang = '';
        $colspan = '';
        $rowspan = '';
        $span = '';
        $width = '';
        $id = '';
        $atts = '';
        $align = '';

        $matched = $in;
        if ($element == 'td') {
            if (preg_match("/\\\\(\d+)/", $matched, $csp))
                $colspan = $csp[1];
            if (preg_match("/\/(\d+)/", $matched, $rsp))
                $rowspan = $rsp[1];
        }

        if ($element == 'td' or $element == 'tr') {
            if (preg_match("/($this->vlgn)/", $matched, $vert))
                $style[] = "vertical-align:" . $this->vAlign($vert[1]);
        }

        if (preg_match("/\{([^}]*)\}/", $matched, $sty)) {
            $style[] = rtrim($sty[1], ';');
            $matched = str_replace($sty[0], '', $matched);
        }

        if (preg_match("/\[([^]]+)\]/U", $matched, $lng)) {
            $matched = str_replace($lng[0], '', $matched);    // Consume entire lang block -- valid or invalid...
            if (preg_match("/\[([a-zA-Z]{2}(?:[\-\_][a-zA-Z]{2})?)\]/U", $lng[0], $lng)) {
                $lang = $lng[1];
            }
        }

        if (preg_match("/\(([^()]+)\)/U", $matched, $cls)) {
            $matched = str_replace($cls[0], '', $matched);    // Consume entire class block -- valid or invalid...
            // Only allow a restricted subset of the CSS standard characters for classes/ids. No encoding markers allowed...
            if (preg_match("/\(([-a-zA-Z 0-9_\.\:\#]+)\)/U", $cls[0], $cls)) {
                $hashpos = strpos($cls[1], '#');
                // If a textile class block attribute was found with a '#' in it
                // split it into the css class and css id...
                if (false !== $hashpos) {
                    if (preg_match("/#([-a-zA-Z0-9_\.\:]*)$/", substr($cls[1], $hashpos), $ids))
                        $id = $ids[1];

                    if (preg_match("/^([-a-zA-Z 0-9_]*)/", substr($cls[1], 0, $hashpos), $ids))
                        $class = $ids[1];
                }
                else {
                    if (preg_match("/^([-a-zA-Z 0-9_]*)$/", $cls[1], $ids))
                        $class = $ids[1];
                }
            }
        }

        if (preg_match("/([(]+)/", $matched, $pl)) {
            $style[] = "padding-left:" . strlen($pl[1]) . "em";
            $matched = str_replace($pl[0], '', $matched);
        }

        if (preg_match("/([)]+)/", $matched, $pr)) {
            $style[] = "padding-right:" . strlen($pr[1]) . "em";
            $matched = str_replace($pr[0], '', $matched);
        }

        if (preg_match("/($this->hlgn)/", $matched, $horiz))
            $style[] = "text-align:" . $this->hAlign($horiz[1]);

        if ($element == 'col') {
            if (preg_match("/(?:\\\\(\d+))?\s*(\d+)?/", $matched, $csp)) {
                $span = isset($csp[1]) ? $csp[1] : '';
                $width = isset($csp[2]) ? $csp[2] : '';
            }
        }

        if ($this->restricted) {
            $o = array();
            $class = trim($autoclass);
            if ($class)
                $o['class'] = $this->cleanAttribs($class);
            if ($lang)
                $o['lang']  = $this->cleanAttribs($lang);
            return $o;
        }
        else
            $class = trim($class . ' ' . $autoclass);

        $o = '';
        if ($style) {
            $tmps = array();
            foreach ($style as $s) {
                $parts = explode(';', $s);
                foreach ($parts as $p)
                    $tmps[] = $p;
            }

            sort($tmps);
            foreach ($tmps as $p) {
                if (!empty($p))
                    $o .= $p.';';
            }
            $style = trim(str_replace(array("\n", ';;'), array('', ';'), $o));
        }

        $o = array();
        if ($class)   $o['class']   = $this->cleanAttribs($class);
        if ($colspan) $o['colspan'] = $this->cleanAttribs($colspan);
        if ($id && $include_id)
                      $o['id']      = $this->cleanAttribs($id);
        if ($lang)    $o['lang']    = $this->cleanAttribs($lang);
        if ($rowspan) $o['rowspan'] = $this->cleanAttribs($rowspan);
        if ($span)    $o['span']    = $this->cleanAttribs($span);
        if ($style)   $o['style']   = $this->cleanAttribs($style);
        if ($width)   $o['width']   = $this->cleanAttribs($width);

        return $o;
    }


    /**
     * Checks whether the text has text not already enclosed by a block tag
	 *
     * @internal
     **/
    protected function hasRawText($text)
    {
        $r = trim(preg_replace('@<(p|blockquote|div|form|table|ul|ol|dl|pre|h\d)[^>]*?'.chr(62).'.*</\1>@s', '', trim($text)));
        $r = trim(preg_replace('@<(hr|br)[^>]*?/>@', '', $r));
        return '' != $r;
    }


    /**
     * Replaces a textile table structure with its HTML version
     *
     * @internal
     *
     * @param  string $text A block of textile
     * @return string Text with tables replaced with HTML tables
     **/
    protected function table($text)
    {
        $text = $text . "\n\n";
        return preg_replace_callback("/^(?:table(_?{$this->s}{$this->a}{$this->c})\.(.*)?\n)?^({$this->a}{$this->c}\.? ?\|.*\|)[\s]*\n\n/smU", array(&$this, "fTable"), $text);
    }


    /**
     * @internal
     **/
    protected function fTable($matches)
    {
        $tatts = $this->parseAttribs($matches[1], 'table');

        $sum = trim($matches[2]) ? ' summary="'.htmlspecialchars(trim($matches[2]), ENT_QUOTES, 'UTF-8').'"' : '';
        $cap       = '';
        $colgrp    = '';
        $last_rgrp = '';
        $c_row     = 1;
        foreach (preg_split("/\|\s*?$/m", $matches[3], -1, PREG_SPLIT_NO_EMPTY) as $row) {

            $row = ltrim($row);

            // Caption -- can only occur on row 1, otherwise treat '|=. foo |...' as a normal center-aligned cell.
            if (($c_row <= 1) && preg_match("/^\|\=($this->s$this->a$this->c)\. ([^\n]*)(.*)/s", ltrim($row), $cmtch)) {
                $capts = $this->parseAttribs($cmtch[1]);
                $cap = "\t<caption".$capts.">".trim($cmtch[2])."</caption>\n";
                $row = ltrim($cmtch[3]);
                if (empty($row))
                    continue;
            }
            $c_row += 1;

            // Colgroup
            if (preg_match("/^\|:($this->s$this->a$this->c\. .*)/m", ltrim($row), $gmtch)) {
                $nl = strpos($row,"\n");    // Is this colgroup def missing a closing pipe? If so, there will be a newline in the middle of $row somewhere.
                $idx=0;
                foreach (explode('|', str_replace('.', '', $gmtch[1])) as $col) {
                    $gatts = $this->parseAttribs(trim($col), 'col');
                    $colgrp .= "\t<col".(($idx==0) ? "group".$gatts.">" : $gatts." />")."\n";
                    $idx++;
                }
                $colgrp .= "\t</colgroup>\n";

                if ($nl === false) {
                    continue;
                }
                else {
                    $row = ltrim(substr($row, $nl));        // Recover from our missing pipe and process the rest of the line...
                }
            }

            preg_match("/(:?^\|($this->vlgn)($this->s$this->a$this->c)\.\s*$\n)?^(.*)/sm", ltrim($row), $grpmatch);

            // Row group
            $rgrp = isset($grpmatch[2]) ? (($grpmatch[2] == '^') ? 'head' : (($grpmatch[2] == '~') ? 'foot' : (($grpmatch[2] == '-') ? 'body' : ''))) : '';
            $rgrpatts = isset($grpmatch[3]) ? $this->parseAttribs($grpmatch[3]) : '';
            $row = $grpmatch[4];

            if (preg_match("/^($this->a$this->c\. )(.*)/m", ltrim($row), $rmtch)) {
                $ratts = $this->parseAttribs($rmtch[1], 'tr');
                $row = $rmtch[2];
            } else $ratts = '';

            $cells = array();
            $cellctr = 0;
            foreach (explode("|", $row) as $cell) {
                $ctyp = "d";
                if (preg_match("/^_(?=[\s[:punct:]])/", $cell)) $ctyp = "h";
                if (preg_match("/^(_?$this->s$this->a$this->c\. )(.*)/", $cell, $cmtch)) {
                    $catts = $this->parseAttribs($cmtch[1], 'td');
                    $cell = $cmtch[2];
                } else $catts = '';

                if (!$this->lite) {
                    $a = array();
                    if (preg_match('/(\s*)(.*)/s', $cell, $a)) {
                        $cell = $this->redclothLists($a[2]);
                        $cell = $this->textileLists($cell);
                        $cell = $a[1] . $cell;
                    }
                }

                if ($cellctr>0) // Ignore first 'cell': it precedes the opening pipe
                    $cells[] = $this->doTagBr("t$ctyp", "\t\t\t<t$ctyp$catts>$cell</t$ctyp>");

                $cellctr++;
            }
            $grp = (($rgrp && $last_rgrp) ? "\t</t".$last_rgrp.">\n" : '') . (($rgrp) ? "\t<t".$rgrp.$rgrpatts.">\n" : '');
            $last_rgrp = ($rgrp) ? $rgrp : $last_rgrp;
            $rows[] = $grp."\t\t<tr$ratts>\n" . join("\n", $cells) . ($cells ? "\n" : "") . "\t\t</tr>";
            unset($cells, $catts);
        }

        return "<table{$tatts}{$sum}>\n" .$cap. $colgrp. join("\n", $rows) . "\n".(($last_rgrp) ? "\t</t".$last_rgrp.">\n" : '')."</table>\n\n";
    }


    /**
     * Parses redcloth-style definition lists and generates corresponding HTML definition lists from them
     *
     * @internal
     *
     * @param  string $text The input textile text to be searched for redcloth-style definition lists
     * @return string The text with tokens inserted at the points definition lists were found
     **/
    protected function redclothLists($text)
    {
        return preg_replace_callback("/^([-]+$this->lc[ .].*:=.*)$(?![^-])/smU", array(&$this, "fRedclothList"), $text);
    }


    /**
     * @internal
     **/
    protected function fRedclothList($m)
    {
        $in = $m[0];
        $out = array();
        $text = preg_split('/\n(?=[-])/m', $in);
        foreach ($text as $nr => $line) {
            $m = array();
            if (preg_match("/^[-]+($this->lc)[ .](.*)$/s", $line, $m)) {
                list(, $atts, $content) = $m;
                $content = trim($content);
                $atts = $this->parseAttribs($atts);

                if (!preg_match("/^(.*?)[\s]*:=(.*?)[\s]*(=:|:=)?[\s]*$/s", $content, $xm)) return $in;

                list(, $term, $def,) = $xm;
                $term = trim($term);
                $def  = trim($def, ' ');

                if (empty($out)) {
                    if (''==$def)
                        $out[] = "<dl$atts>";
                    else
                        $out[] = '<dl>';
                }

                if ('' != $def && '' != $term)
                {
                    $pos = strpos($def, "\n");
                    $def = str_replace("\n", "<br />", trim($def));
                    if (0 === $pos)
                        $def  = '<p>' . $def . '</p>';

                    $term = $this->graf($term);
                    $def  = $this->graf($def);

                    $out[] = "\t<dt$atts>$term</dt>";
                    $out[] = "\t<dd>$def</dd>";
                }
            }
        }
        $out[] = '</dl>';
        return implode("\n", $out);
    }


    /**
     * Searches for ordered, un-ordered and definition lists in the textile input and generates HTML lists for them.
     * The lists are stashed in a cache and a token representing that list is injected into the document, overwriting the textile
     * source for the list.
     *
     * @internal
     *
     * @param  string $text  Input textile source to parse for lists
     * @return string Text with lists replaced with tokens
     **/
    protected function textileLists($text)
    {
        return preg_replace_callback("/^((?:[*;:]+|[*;:#]*#(?:_|\d+)?)$this->lc[ .].*)$(?![^#*;:])/smU", array(&$this, "fTextileList"), $text);
    }


    /**
     * @internal
     **/
    protected function fTextileList($m)
    {
        $text = preg_split('/\n(?=[*#;:])/m', $m[0]);
        $pt = '';
        foreach ($text as $nr => $line) {
            $nextline = isset($text[$nr+1]) ? $text[$nr+1] : false;
            if (preg_match("/^([#*;:]+)(_|\d+)?($this->lc)[ .](.*)$/s", $line, $m)) {
                list(, $tl, $st, $atts, $content) = $m;
                $content = trim($content);
                $nl = '';
                $ltype = $this->liType($tl);
                $litem = (strpos($tl, ';') !== false) ? 'dt' : ((strpos($tl, ':') !== false) ? 'dd' : 'li');
                $showitem = (strlen($content) > 0);

                if ('o' === $ltype) {                           // Handle list continuation/start attribute on ordered lists...
                    if (!isset($this->olstarts[$tl]))
                        $this->olstarts[$tl] = 1;

                    if (strlen($tl) > strlen($pt)) {            // First line of this level of ol -- has a start attribute?
                        if ('' == $st)
                            $this->olstarts[$tl] = 1;           // No => reset count to 1.
                        elseif ('_' !== $st)
                            $this->olstarts[$tl] = (int)$st;    // Yes, and numeric => reset to given.
                                                                // TRICKY: the '_' continuation marker just means
                                                                // output the count so don't need to do anything
                                                                // here.
                    }

                    if ((strlen($tl) > strlen($pt)) && '' !== $st)        // Output the start attribute if needed...
                        $st = ' start="' . $this->olstarts[$tl] . '"';

                    if ($showitem)                              // TRICKY: Only increment the count for list items; not when a list definition line is encountered.
                        $this->olstarts[$tl] += 1;
                }

                if (preg_match("/^([#*;:]+)(_|[\d]+)?($this->lc)[ .].*/", $nextline, $nm))
                    $nl = $nm[1];

                if ((strpos($pt, ';') !== false) && (strpos($tl, ':') !== false)) {
                    $lists[$tl] = 2; // We're already in a <dl> so flag not to start another
                }

                $tabs = str_repeat("\t", strlen($tl));
                $atts = $this->parseAttribs($atts);
                if (!isset($lists[$tl])) {
                    $lists[$tl] = 1;
                    $line = "$tabs<" . $ltype . "l$atts$st>" . (($showitem) ? "\n$tabs\t<$litem>" . $content : '');
                } else {
                    $line = ($showitem) ? "$tabs\t<$litem$atts>" . $content : '';
                }

                if ((strlen($nl) <= strlen($tl)))
                    $line .= (($showitem) ? "</$litem>" : '');

                foreach (array_reverse($lists) as $k => $v) {
                    if (strlen($k) > strlen($nl)) {
                        $line .= ($v==2) ? '' : "\n$tabs</" . $this->liType($k) . "l>";
                        if ((strlen($k) > 1) && ($v != 2))
                            $line .= "</".$litem.">";
                        unset($lists[$k]);
                    }
                }
                $pt = $tl; // Remember the current Textile tag
            }
            else {
                $line .= "\n";
            }
            $out[] = $line;
        }

        $out = implode("\n", $out);
        return $this->doTagBr($litem, $out);
    }


    /**
     * Determine the list type from the textile input symbol
     *
     * @internal
     *
     * @param  string $in Textile input containing possible list marker
     * @return string The type of list (missing the final 'l' character)
     **/
    protected function liType($in)
    {
        $m = array();
        $type = 'd';
        if (preg_match('/^([#*]+)/', $in, $m))
            $type = ('#' === substr($m[1], -1)) ? 'o' : 'u';
        return $type;
    }


    /**
     * @internal
     **/
    protected function doTagBr($tag, $in)
    {
        return preg_replace_callback('@<('.preg_quote($tag).')([^>]*?)>(.*)(</\1>)@s', array(&$this, 'fBr'), $in);
    }


    /**
     * @internal
     **/
    protected function doPBr($in)
    {
        return preg_replace_callback('@<(p)([^>]*?)>(.*)(</\1>)@s', array(&$this, 'fPBr'), $in);
    }


    /**
     * Less restrictive version of fBr() ... used only in paragraphs where the next
     * row may start with a smiley or perhaps something like '#8 bolt...' or '*** stars...'
     * @internal
     **/
    protected function fPBr($m)
    {
        $content = preg_replace("@((?!<br>\n|<br />\n))+\n(?![\s|])@", '$1<br />', $m[3]);
        return '<'.$m[1].$m[2].'>'.$content.$m[4];
    }


    /**
     * @internal
     **/
    protected function fBr($m)
    {
        $content = preg_replace("@(.+)(?<!<br>|<br />)\n(?![#*;:\s|])@", '$1<br />', $m[3]);
        return '<'.$m[1].$m[2].'>'.$content.$m[4];
    }


    /**
     * Splits the input textile text into blocks on double line-break boundaries and then
     * processes each block.
     *
     * @internal
     *
     * @param  string $text Textile source text
     * @return string Input text with blocks processed
     **/
    protected function block($text)
    {
        $find = $this->btag;
        $tre = join('|', $find);

        $text = explode("\n\n", $text);

        $tag  = 'p';
        $atts = '';
        $cite = '';
        $graf = '';
        $ext  = '';
        $eat  = false;

        $out  = array();

        foreach ($text as $line) {
            $anon = 0;
            if (preg_match("/^($tre)($this->a$this->c)\.(\.?)(?::(\S+))? (.*)$/s", $line, $m)) {
                // Last block was extended, so close it
                if ($ext)
                    $out[count($out)-1] .= $c1;
                // New block
                list(,$tag,$atts,$ext,$cite,$graf) = $m;
                list($o1, $o2, $content, $c2, $c1, $eat) = $this->fBlock(array(0,$tag,$atts,$ext,$cite,$graf));

                // Leave off c1 if this block is extended, we'll close it at the start of the next block
                if ($ext)
                    $line = $o1.$o2.$content.$c2;
                else
                    $line = $o1.$o2.$content.$c2.$c1;
            }
            else {
                // Anonymous block
                $anon = 1;
                if ($ext or !preg_match('/^ /', $line)) {
                    list($o1, $o2, $content, $c2, $c1, $eat) = $this->fBlock(array(0,$tag,$atts,$ext,$cite,$line));
                    // Skip $o1/$c1 because this is part of a continuing extended block
                    if ($tag == 'p' and !$this->hasRawText($content)) {
                        $line = $content;
                    }
                    else {
                        $line = $o2.$content.$c2;
                    }
                }
                else {
                    $line = $this->graf($line);
                }
            }

            $line = $this->doPBr($line);
            $line = preg_replace('/<br>/', '<br />', $line);

            if ($ext and $anon)
                $out[count($out)-1] .= "\n".$line;
            elseif (!$eat)
                $out[] = $line;

            if (!$ext) {
                $tag  = 'p';
                $atts = '';
                $cite = '';
                $graf = '';
                $eat  = false;
            }
        }
        if ($ext) $out[count($out)-1] .= $c1;
        return join("\n\n", $out);
    }


    /**
     * @internal
     **/
    protected function fBlock($m)
    {
        list(, $tag, $att, $ext, $cite, $content) = $m;
        $atts = $this->parseAttribs($att);

        $o1  = '';
        $o2  = '';
        $c2  = '';
        $c1  = '';
        $eat = false;

        if ($tag === 'p') {
            // Is this an anonymous block with a note definition?
            $notedef = preg_replace_callback("/
                    ^note\#               #  start of note def marker
                    ([^%<*!@#^([{ \s.]+)  # !label
                    ([*!^]?)              # !link
                    ({$this->c})          # !att
                    \.?                   #  optional period.
                    [\s]+                 #  whitespace ends def marker
                    (.*)$                 # !content
                /x".$this->regex_snippets['mod'], array(&$this, "fParseNoteDefs"), $content);

            if ('' === $notedef) // It will be empty if the regex matched and ate it.
                return array($o1, $o2, $notedef, $c2, $c1, true);
            }

        if (preg_match("/fn(\d+)/", $tag, $fns)) {
            $tag = 'p';
            $fnid = empty($this->fn[$fns[1]]) ? $fns[1] : $this->fn[$fns[1]];

            // If there is an author-specified ID goes on the wrapper & the auto-id gets pushed to the <sup>
            $supp_id = '';
            if (strpos($atts, 'class=') === false)
                $atts .= ' class="footnote"';

            if (strpos($atts, ' id=') === false)
                $atts .= ' id="fn' . $fnid . '"';
            else
                $supp_id = ' id="fn' . $fnid . '"';


            $sup = (strpos($att, '^') === false) ? $this->formatFootnote($fns[1], $supp_id) : $this->formatFootnote('<a href="#fnrev' . $fnid . '">'.$fns[1] .'</a>', $supp_id);

            $content = $sup . ' ' . $content;
        }

        if ($tag == "bq") {
            $cite = $this->shelveURL($cite);
            $cite = ($cite != '') ? ' cite="' . $cite . '"' : '';
            $o1 = "<blockquote$cite$atts>\n";
            $o2 = "\t\t<p".$this->parseAttribs($att, '', 0).">";
            $c2 = "</p>";
            $c1 = "\n</blockquote>";
        }
        elseif ($tag == 'bc') {
            $o1 = "<pre$atts>";
            $o2 = "<code>";
            $c2 = "</code>";
            $c1 = "</pre>";
            $content = $this->shelve($this->rEncodeHTML(rtrim($content, "\n")."\n"));
        }
        elseif ($tag == 'notextile') {
            $content = $this->shelve($content);
            $o1 = '';
            $o2 = '';
            $c1 = '';
            $c2 = '';
        }
        elseif ($tag == 'pre') {
            $content = $this->shelve($this->rEncodeHTML(rtrim($content, "\n")."\n"));
            $o1 = "<pre$atts>";
            $o2 = '';
            $c2 = '';
            $c1 = "</pre>";
        }
        elseif ($tag == '###') {
            $eat = true;
        }
        else {
            $o2 = "<$tag$atts>";
            $c2 = "</$tag>";
        }

        $content = (!$eat) ? $this->graf($content) : '';

        return array($o1, $o2, $content, $c2, $c1, $eat);
    }


    /**
     * @internal
     **/
    protected function formatFootnote($marker, $atts='', $anchor=true)
    {
        $pattern = ($anchor) ? $this->symbols['fn_foot_pattern'] : $this->symbols['fn_ref_pattern'];
        return $this->replaceMarkers($pattern, array('atts' => $atts, 'marker' => $marker));
    }


    /**
     * @internal
     **/
    protected function replaceMarkers($text, $replacements)
    {
        if (!empty($replacements))
            foreach ($replacements as $k => $r)
                $text = str_replace('{'.$k.'}', $r, $text);
        return $text;
    }


    /**
     * @internal
     **/
    protected function getHTMLComments($text)
    {
        $text = preg_replace_callback("/
            \<!--    #  start
            (.*?)    # !content *not* greedy
            -->      #  end
        /sx", array(&$this, "fParseHTMLComments"), $text);
        return $text;
    }


    /**
     * @internal
     **/
    protected function fParseHTMLComments($m)
    {
        list(, $content) = $m;
        if ($this->restricted)
            $content = $this->shelve($this->rEncodeHTML($content));
        else
            $content = $this->shelve($content);
        return "<!--$content-->";
    }


    /**
     * @internal
     **/
    protected function graf($text)
    {
        // Handle normal paragraph text
        if (!$this->lite) {
            $text = $this->noTextile($text);       // Notextile blocks and inlines
            $text = $this->code($text);            // Handle code
        }

        $text = $this->getHTMLComments($text);     // HTML comments --
        $text = $this->getRefs($text);             // Consume link aliases
        $text = $this->links($text);               // Generate links

        if (!$this->noimage)
            $text = $this->image($text);           // Handle images (if permitted)

        if (!$this->lite) {
            $text = $this->table($text);           // Handle tables
            $text = $this->redclothLists($text);   // Handle redcloth-style definition lists
            $text = $this->textileLists($text);    // Handle ordered & unordered lists plus txp-style definition lists
        }

        $text = $this->span($text);                // Inline markup (em, strong, sup, sub, del etc)

        if (!$this->lite) {
            // Turn footnote references into supers or links. As footnote blocks are banned in lite mode there is no point generating links for them
            $text = $this->footnoteRef($text);

            // Turn note references into links
            $text = $this->noteRef($text);
        }

        $text = $this->glyphs($text);              // Glyph level substitutions (mainly typographic -- " & ' => curly quotes, -- => em-dash etc.

        return rtrim($text, "\n");
    }


    /**
     * Replaces textile spans with their equivalent HTML inline tags.
     *
     * @internal
     *
     * @param  string $text The textile document to perform the replacements in.
     * @return string       The textile document with spans replaced by their HTML inline equivalents
     **/
    protected function span($text)
    {
        $span_tags = array_keys($this->span_tags);
        $pnct = ".,\"'?!;:‹›«»„“”‚‘’";
        $this->span_depth++;

        if ($this->span_depth <= $this->max_span_depth)
        {
            foreach ($span_tags as $f)
            {
                $f = preg_quote($f);
                $text = preg_replace_callback("/
                    (^|(?<=[\s>$pnct\(])|[{[])            # pre
                    ($f)(?!$f)                            # tag
                    ({$this->c})                          # atts
                    (?::(\S+))?                           # cite
                    ([^\s$f]+|\S.*?[^\s$f\n])             # content
                    ([$pnct]*)                            # end
                    $f
                    ($|[\[\]}<]|(?=[$pnct]{1,2}[^0-9]|\s|\)))  # tail
                /x".$this->regex_snippets['mod'], array(&$this, "fSpan"), $text);
            }
        }
        $this->span_depth--;
        return $text;
    }


    /**
     * @internal
     **/
    protected function fSpan($m)
    {
        list(, $pre, $tag, $atts, $cite, $content, $end, $tail) = $m;

        $tag  = $this->span_tags[$tag];
        $atts = $this->parseAttribs($atts);
        $atts .= ($cite != '') ? 'cite="' . $cite . '"' : '';

        $content = $this->span($content);

        $opentag = '<'.$tag.$atts.'>';
        $closetag = '</'.$tag.'>';
        $tags = $this->storeTags($opentag, $closetag);
        $out = "{$tags['open']}{$content}{$end}{$tags['close']}";

        if (($pre and !$tail) or ($tail and !$pre))
            $out = $pre.$out.$tail;

        return $out;
    }


    /**
     * @internal
     **/
    protected function storeTags($opentag,$closetag='')
    {
        $key = ($this->tag_index++);

        $key = str_pad((string)$key, 10, '0', STR_PAD_LEFT).'z'; // $key must be of fixed length to allow proper matching in retrieveTags
        $this->tagCache[$key] = array('open'=>$opentag, 'close'=>$closetag);
        $tags = array(
            'open'  => "textileopentag{$key} ",
            'close' => " textileclosetag{$key}",
        );
        return $tags;
    }


    /**
     * @internal
     **/
    protected function retrieveTags($text)
    {
        $text = preg_replace_callback('/textileopentag([\d]{10}z) /' , array(&$this, 'fRetrieveOpenTags'),  $text);
        $text = preg_replace_callback('/ textileclosetag([\d]{10}z)/', array(&$this, 'fRetrieveCloseTags'), $text);
        return $text;
    }


    /**
     * @internal
     **/
    protected function fRetrieveOpenTags($m)
    {
        return $this->tagCache[$m[1]]['open'];
    }


    /**
     * @internal
     **/
    protected function fRetrieveCloseTags($m)
    {
        return $this->tagCache[$m[1]]['close'];
    }


    /**
     * @internal
     **/
    protected function placeNoteLists($text)
    {
        extract($this->regex_snippets);

        // Sequence all referenced definitions...
        if (!empty($this->notes)) {
            $o = array();
            foreach ($this->notes as $label=>$info) {
                $i = @$info['seq'];
                if (!empty($i)) {
                    $info['seq'] = $label;
                    $o[$i] = $info;
                } else {
                    $this->unreferencedNotes[] = $info;    // Unreferenced definitions go here for possible future use.
                }
            }
            if (!empty($o)) ksort($o);
            $this->notes = $o;
        }

        // Replace list markers...
        $text = preg_replace_callback("@<p>notelist({$this->c})(?:\:([$wrd|{$this->syms}]))?([\^!]?)(\+?)\.?[\s]*</p>@U$mod", array(&$this, "fNoteLists"), $text);

        return $text;
    }


    /**
     * @internal
     **/
    protected function fNoteLists($m)
    {
        list(, $att, $start_char, $g_links, $extras) = $m;
        if (!$start_char) $start_char = 'a';
        $index = $g_links.$extras.$start_char;

        if (empty($this->notelist_cache[$index])) { // If not in cache, build the entry...
            $o = array();

            if (!empty($this->notes)) {
                foreach ($this->notes as $seq=>$info) {
                    $links = $this->makeBackrefLink($info, $g_links, $start_char);
                    $atts = '';
                    if (!empty($info['def'])) {
                        $id = $info['id'];
                        extract($info['def']);
                        $o[] = "\t".'<li'.$atts.'>'.$links.'<span id="note'.$id.'"> </span>'.$content.'</li>';
                    } else {
                        $o[] = "\t".'<li'.$atts.'>'.$links.' Undefined Note [#'.$info['seq'].'].</li>';
                    }
                }
            }
            if ('+' == $extras && !empty($this->unreferencedNotes)) {
                foreach ($this->unreferencedNotes as $seq=>$info) {
                    if (!empty($info['def'])) {
                        extract($info['def']);
                        $o[] = "\t".'<li'.$atts.'>'.$content.'</li>';
                    }
                }
            }

            $this->notelist_cache[$index] = join("\n",$o);
        }

        $_ = ($this->notelist_cache[$index]) ? $this->notelist_cache[$index] : '';

        if (!empty($_)) {
            $list_atts = $this->parseAttribs($att);
            $_ = "<ol$list_atts>\n$_\n</ol>";
        }

        return $_;
    }


    /**
     * @internal
     **/
    protected function makeBackrefLink(&$info, $g_links, $i)
    {
        $link    = '';
        $atts    = '';
        $content = '';
        $id      = '';
        @extract($info['def']);
        $backlink_type = ($link) ? $link : $g_links;
        $allow_inc = (false === strpos($this->syms, $i));

        $i_ = str_replace(array('&', ';', '#'), '', $this->encodeHigh($i));
        $decode = (strlen($i) !== strlen($i_));

        if ($backlink_type === '!')
            return '';
        elseif ($backlink_type === '^')
            return '<sup><a href="#noteref'.$info['refids'][0].'">'.$i.'</a></sup>';
        else {
            $_ = array();
            foreach ($info['refids'] as $id) {
                $_[] = '<sup><a href="#noteref'.$id.'">'. (($decode) ? $this->decodeHigh($i_) : $i_) .'</a></sup>';
                if ($allow_inc)
                    $i_++;
            }
            $_ = join(' ', $_);
            return $_;
        }
    }


    /**
     * @internal
     **/
    protected function fParseNoteDefs($m)
    {
        list(, $label, $link, $att, $content) = $m;
        // Assign an id if the note reference parse hasn't found the label yet.
        $id = @$this->notes[$label]['id'];
        if (!$id)
            $this->notes[$label]['id'] = uniqid(rand());

        if (empty($this->notes[$label]['def'])) // Ignores subsequent defs using the same label
        {
            $this->notes[$label]['def'] = array(
                'atts'    => $this->parseAttribs($att),
                'content' => $this->graf($content),
                'link'    => $link,
            );
        }
        return '';
    }


    /**
     * @internal
     **/
    protected function noteRef($text)
    {
        $text = preg_replace_callback("/
            \[                   #  start
            ({$this->c})         # !atts
            \#
            ([^\]!]+?)           # !label
            ([!]?)               # !nolink
            \]
        /Ux", array(&$this, "fParseNoteRefs"), $text);
        return $text;
    }


    /**
     * @internal
     **/
    protected function fParseNoteRefs($m)
    {
        //   By the time this function is called, all the defs will have been processed
        // into the notes array. So now we can resolve the link numbers in the order
        // we process the refs...

        list(, $atts, $label, $nolink) = $m;
        $atts = $this->parseAttribs($atts);
        $nolink = ($nolink === '!');

        // Assign a sequence number to this reference if there isn't one already...
        $num = @$this->notes[$label]['seq'];
        if (!$num)
            $num = $this->notes[$label]['seq'] = ($this->note_index++);

        // Make our anchor point & stash it for possible use in backlinks when the
        // note list is generated later...
        $refid = uniqid(rand());
        $this->notes[$label]['refids'][] = $refid;

        // If we are referencing a note that hasn't had the definition parsed yet, then assign it an ID...
        $id = @$this->notes[$label]['id'];
        if (!$id)
            $id = $this->notes[$label]['id'] = uniqid(rand());

        // Build the link (if any)...
        $_ = '<span id="noteref'.$refid.'">'.$num.'</span>';
        if (!$nolink)
            $_ = '<a href="#note'.$id.'">'.$_.'</a>';

        // Build the reference...
        $_ = $this->replaceMarkers($this->symbols['nl_ref_pattern'], array('atts' => $atts, 'marker' => $_));

        return $_;
    }


    /**
     * Parse URI pulls a URI-like string apart into component parts
     * Regex taken from the RFC at http://tools.ietf.org/html/rfc3986#appendix-B
     *
     * @internal
     *
     * @param  string  $uri     The string to pick apart (if possible)
     * @param  array   $m       Reference to an array to receive the parts the URI
     * @return bool             Match/nomatch of the URI pattern.
     **/
    protected function parseURI($uri, &$m)
    {
        $r = "@^((?P<scheme>[^:/?#]+):)?(//(?P<authority>[^/?#]*))?(?P<path>[^?#]*)(\?(?P<query>[^#]*))?(#(?P<fragment>.*))?@";
        //      12                      3  4                       5               6  7                 8 9
        //
        //    scheme    = $2
        //    authority = $4
        //    path      = $5
        //    query     = $7
        //    fragment  = $9

        $ok = preg_match($r, $uri, $m);
        return $ok;
    }


    /**
     * @internal
     **/
    protected function addPart(&$mask, $name, &$parts) {
        return (in_array($name, $mask) && isset($parts[$name]) && '' !== $parts[$name]);
    }


    /**
     * Rebuild a URI from parsed parts and a mask.
     * Algorithm based on example from http://tools.ietf.org/html/rfc3986#section-5.3
     *
     * @internal
     *
     * @param  array  $parts  Full array of URI parts
     * @param  string $mask   Comma separated list of URI parts to include in the rebuilt URI
     * @param  bool   $encode Flag to control encoding of the path part of the rebuilt URI
     * @return string         The rebuilt URI
     **/
    protected function rebuildURI($parts, $mask='scheme,authority,path,query,fragment', $encode=true)
    {
        $mask = explode(',', $mask);
        $out  = '';

        if ($this->addPart($mask, 'scheme', $parts)) {
            $out .= $parts['scheme'] . ':';
        }

        if ($this->addPart($mask, 'authority', $parts)) {
            $out .= '//' . $parts['authority'];
        }

        if ($this->addPart($mask, 'path', $parts)) {
            if (!$encode)
                $out .= $parts['path'];
            else {
                $pp = explode('/', $parts['path']);
                foreach ($pp as &$p) {
                    $p = str_replace(array('%25', '%40'), array('%', '@'), rawurlencode($p));
                    if (!in_array($parts['scheme'], array('tel','mailto'))) $p = str_replace('%2B', '+', $p);
                }

                $pp = implode('/', $pp);
                $out .= $pp;
            }
        }

        if ($this->addPart($mask, 'query', $parts)) {
            $out .= '?' . $parts['query'];
        }

        if ($this->addPart($mask, 'fragment', $parts)) {
            $out .= '#' . $parts['fragment'];
        }

        return $out;
    }


    /**
     * Initiates recognition and conversion of links in the source document.
     *
     * @internal
     *
     * @param  string $text  The source textile text
     * @return string        The document with tokens inserted representing the links
     **/
    protected function links($text)
    {
        return preg_replace_callback('/
            (^|(?<=[\s>.\(\|])|[{[])   # $pre
            "                          #  start
            (' . $this->c . ')         # $atts
            ([^"]+?)                   # $text
            (?:\(([^)]+?)\)(?="))?     # $title
            ":
            ('.$this->urlch.'+?)       # $url
            (\/)?                      # $slash
            ([^'.$this->regex_snippets['wrd'].'\/]*?)  # $post
            ([\]}]|(?=\s|$|\)|\|))     # $tail
            /x'.$this->regex_snippets['mod'], array(&$this, "fLink"), $text);
    }


    /**
     * @internal
     **/
    protected function fLink($m)
    {
        list(, $pre, $atts, $text, $title, $url, $slash, $post, $tail) = $m;

        // Strip any ':' or '?' characters from the end of the url and return them to $post. This seems to be needed
        // when using the unicode version of the word character class in the regex.
        $a = array();
        if (preg_match('/^(.*)([?:]+)$/', $url, $a)) {
            $url   = $a[1];
            $post .= $a[2];
        }

        $uri_parts = array();
        $this->parseURI($url, $uri_parts);

        $scheme         = $uri_parts['scheme'];
        $scheme_in_list = in_array($scheme, $this->url_schemes);
        $scheme_ok      = ('' === $scheme) || $scheme_in_list;

        if (!$scheme_ok)
            return $m[0];

        if ('$' === $text) {
            if ($scheme_in_list)
                $text = ltrim($this->rebuildURI($uri_parts, 'authority,path,query,fragment', false), '/');
            else {
                if (isset($this->urlrefs[$url]))
                    $url = urldecode($this->urlrefs[$url]);
                $text = $url;
            }
        }

        $text  = trim($text);
        $title = $this->encodeHTML($title);

        // If the text was in parenthesis and there was no title then the regex will have an empty $text and a non-empty $title so...
        if (empty($text) && !empty($title)) {
            $text  = "($title)";
            $title = '';
        }

        if (!$this->noimage)
            $text = $this->image($text);

        $text = $this->span($text);
        $text = $this->glyphs($text);
        $url  = $this->shelveURL($this->rebuildURI($uri_parts) . $slash);

        $a    = $this->newTag('a', $this->parseAttribsToArray($atts), false)->title($title)->href($url, true)->rel($this->rel);
        $tags = $this->storeTags((string)$a, '</a>');
        $out  = $tags['open'].$text.$tags['close'];

        if (($pre and !$tail) or ($tail and !$pre))
        {
            $out = $pre.$out.$post.$tail;
            $post = '';
        }

        return $this->shelve($out).$post;
    }


     /**
      * Finds URI aliases within the textile source text and pulls them out of it and stores them in an internal
      * link alias cache so that they can be referenced from any links in the document.
      *
      * This parse happens before the link parsing takes place.
      *
      * @internal
      *
      * @param  string $text textile document
      * @return string The textile document with any URI aliases/refs removed.
      **/
    protected function getRefs($text)
    {
        if ($this->restricted)
            $pattern = "/^\[(.+)\]((?:https?:\/\/|\/)\S+)(?=\s|$)/Um";
        else
            $pattern = "/^\[(.+)\]((?:https?:\/\/|tel:|file:|ftp:\/\/|sftp:\/\/|mailto:|callto:|\/)\S+)(?=\s|$)/Um";
        return preg_replace_callback($pattern, array(&$this, "refs"), $text);
    }


    /**
     * @internal
     **/
    protected function refs($m)
    {
        list(, $flag, $url) = $m;
        $uri_parts = array();
        $this->parseURI($url, $uri_parts);
        $url = ltrim($this->rebuildURI($uri_parts)); // Encodes URL if needed.
        $this->urlrefs[$flag] = $url;
        return '';
    }


    /**
     * @internal
     **/
    protected function shelveURL($text)
    {
        if ('' === $text) return '';
        $ref = md5($text).'z';
        $this->urlshelf[$ref] = $text;
        return 'urlref:'.$ref;
    }


    /**
     * @internal
     **/
    protected function retrieveURLs($text)
    {
        return preg_replace_callback('/urlref:(\w{32}z)/', array(&$this, "retrieveURL"), $text);
    }


    /**
     * @internal
     **/
    protected function retrieveURL($m)
    {
        $ref = $m[1];
        if (!isset($this->urlshelf[$ref])) return $ref;

        $url = $this->urlshelf[$ref];
        if (isset($this->urlrefs[$url]))
            $url = $this->urlrefs[$url];
        return $this->rEncodeHTML($this->relURL($url));
    }


    /**
     * @internal
     **/
    protected function relURL($url)
    {
        $parts = @parse_url(urldecode($url));
        if ((empty($parts['scheme']) or @$parts['scheme'] == 'http') and
             empty($parts['host']) and
             preg_match('/^\w/', @$parts['path']))
            $url = $this->relativeImagePrefix.$url;
        if ($this->restricted and !empty($parts['scheme']) and
                !in_array($parts['scheme'], $this->url_schemes))
            return '#';
        return $url;
    }


    /**
     * @internal
     **/
    protected function isRelURL($url)
    {
        $parts = @parse_url($url);
        return (empty($parts['scheme']) and empty($parts['host']));
    }


    /**
     * Parses the input textile document for images and generates img HTML tags for each one found, caching the
     * generated img tag internally and replacing the textile image with a token to the cached tag.
     *
     * @internal
     *
     * @param  string $text Textile source text
     * @return string The input document with images pulled out and replaced with tokens
     **/
    protected function image($text)
    {
        return preg_replace_callback('/
            (?:[[{])?                  # pre
            \!                         # opening !
            (\<|\=|\>)?                # optional alignment              $algn
            ('.$this->lc.')            # optional style,class atts       $atts
            (?:\.\s)?                  # optional dot-space
            ([^\s(!]+)                 # presume this is the src         $url
            \s?                        # optional space
            (?:\(([^\)]+)\))?          # optional title                  $title
            \!                         # closing
            (?::(\S+)(?<![\]).,]))?    # optional href sans final punct. $href
            (?:[\]}]|(?=[.,\s)|]|$))   # lookahead: space , . ) | or end of string ... "|" needed if image in table cell
        /x', array(&$this, "fImage"), $text);
    }


    /**
     * @internal
     **/
    protected function fImage($m)
    {
        $extras = '';
        $align  = '';
        $alt    = '';
        $size   = false;

        list(, $algn, $atts, $url, $title, $href) = array_pad($m, 6, null);
        $url = htmlspecialchars($url, ENT_QUOTES, 'UTF-8');

        if ('' !== $algn) {
            $vals = array('<' => 'left', '=' => 'center', '>' => 'right');
            if (isset($vals[$algn])) {
                if ('html5' === $this->doctype)
                    $extras = "align-{$vals[$algn]}";
                else
                    $align = $vals[$algn];
            }
        }

        if ($title) {
            $title = $this->encodeHTML($title);
            $alt   = $title;
        }

        if ($this->isRelUrl($url))
            $size = @getimagesize(realpath($this->doc_root.ltrim($url, $this->ds)));
        if ($size)
            $atts .= " $size[3]";

        $href = ($href) ? $this->shelveURL($href) : '';
        $img  = $this->newTag('img', $this->parseAttribsToArray($atts, '', 1, $extras))->align($align)->alt($alt, true)->src($this->shelveURL($url), true)->title($title);

        $out  = ($href) ? "<a href=\"$href\"{$this->rel}>$img</a>" : (string)$img;
        return $this->shelve($out);
    }


    /**
     * @internal
     **/
    protected function code($text)
    {
        $text = $this->doSpecial($text, '<code>', '</code>', 'fCode');
        $text = $this->doSpecial($text, '@', '@', 'fCode');
        $text = $this->doSpecial($text, '<pre>', '</pre>', 'fPre');
        return $text;
    }


    /**
     * @internal
     **/
    protected function fCode($m)
    {
        list(, $before, $text, $after) = array_pad($m, 4, '');
        return $before.$this->shelve('<code>'.$this->rEncodeHTML($text).'</code>').$after;
    }


    /**
     * @internal
     **/
    protected function fPre($m)
    {
        list(, $before, $text, $after) = array_pad($m, 4, '');
        return $before.'<pre>'.$this->shelve($this->rEncodeHTML($text)).'</pre>'.$after;
    }


    /**
     * @internal
     **/
    protected function shelve($val)
    {
        $i = uniqid(rand()).'z';
        $this->shelf[$i] = $val;
        return $i;
    }


    /**
     * @internal
     **/
    protected function retrieve($text)
    {
        if (is_array($this->shelf))
            do {
                $old = $text;
                $text = str_replace(array_keys($this->shelf), $this->shelf, $text);
             } while ($text != $old);

        return $text;
    }


    /**
     * Cleans up the textile input text, removing BOM and unifying line endings etc.
     *
     * @internal
     *
     * @param  string $text Input textile text
     * @return string Cleaned version of the input
     **/
    protected function cleanWhiteSpace($text)
    {
        $out = preg_replace("/^\xEF\xBB\xBF|\x1A/", '', $text); // Byte order mark (if present)
        $out = preg_replace("/\r\n?/", "\n", $out);             // DOS and MAC line endings to *NIX style endings
        $out = preg_replace("/^[ \t]*\n/m", "\n", $out);        // lines containing only whitespace
        $out = preg_replace("/\n{3,}/", "\n\n", $out);          // 3 or more line ends
        $out = preg_replace("/^\n*/", "", $out);                // leading blank lines
        return $out;
    }


    /**
     * @internal
     **/
    protected function doSpecial($text, $start, $end, $method='fSpecial')
    {
        return preg_replace_callback('/(^|\s|[|[({>])'.preg_quote($start, '/').'(.*?)'.preg_quote($end, '/').'(\s|$|[\])}|])?/ms', array(&$this, $method), $text);
    }


    /**
     * @internal
     **/
    protected function fSpecial($m)
    {
        // A special block like notextile or code
        list(, $before, $text, $after) = array_pad($m, 4, '');
        return $before.$this->shelve($this->encodeHTML($text)).$after;
    }


    /**
     * @internal
     **/
    protected function noTextile($text)
    {
         $text = $this->doSpecial($text, '<notextile>', '</notextile>', 'fTextile');
         return $this->doSpecial($text, '==', '==', 'fTextile');
    }


    /**
     * @internal
     **/
    protected function fTextile($m)
    {
        list(, $before, $notextile, $after) = array_pad($m, 4, '');
        return $before.$this->shelve($notextile).$after;
    }


    /**
     * @internal
     **/
    protected function footnoteRef($text)
    {
        return preg_replace_callback('/(?<=\S)\[(\d+)(!?)\]\s?/U', array(&$this, 'footnoteID'), $text);
    }


    /**
     * @internal
     **/
    protected function footnoteID($m)
    {
        list(, $id, $nolink) = array_pad($m, 3, '');
        $backref = ' class="footnote"';

        if (empty($this->fn[$id])) {
            $this->fn[$id] = $a = uniqid(rand());
            $backref .= " id=\"fnrev$a\"";
        }

        $fnid = $this->fn[$id];
        $footref = ('!' == $nolink) ? $id : '<a href="#fn'.$fnid.'">'.$id.'</a>';
        $footref = $this->formatFootnote($footref, $backref, false);

        return $footref;
    }


    /**
     * Perfoms typographical glyph replacements within the input textile text.
     * The source text is split across HTML-like tags in order to avoid attempting glyph
     * replacements within tags.
     *
     * @internal
     *
     * @param  string $text Input textile source text
     * @return string
     **/
    protected function glyphs($text)
    {
        // Fix: hackish -- adds a space if final char of text is a double quote.
        $text = preg_replace('/"\z/', "\" ", $text);

        $text = preg_split("@(<[\w/!?].*>)@Us", $text, -1, PREG_SPLIT_DELIM_CAPTURE);
        $i = 0;
        foreach ($text as $line) {
            // Text tag text tag text ...
            if (++$i % 2) {
                // Raw < > & chars are already entity encoded in restricted mode
                if (!$this->restricted) {
                    $line = preg_replace('/&(?!#?[a-z0-9]+;)/i', '&amp;', $line);
                    $line = str_replace(array('<', '>'), array('&lt;', '&gt;'), $line);
                }
                $line = preg_replace($this->glyph_search, $this->glyph_replace, $line);
            }
            $glyph_out[] = $line;
        }
        return join('', $glyph_out);
    }


    /**
     * @internal
     **/
    protected function replaceGlyphs($text)
    {
        return preg_replace('/glyph:([^<]+)/','$1',$text);
    }


    /**
     * @internal
     **/
    protected function hAlign($in)
    {
        $vals = array(
            '&lt;'     => 'left',
            '&gt;'     => 'right',
            '&lt;&gt;' => 'justify',
            '<'        => 'left',
            '='        => 'center',
            '>'        => 'right',
            '<>'       => 'justify');
        return (isset($vals[$in])) ? $vals[$in] : '';
    }


    /**
     * @internal
     **/
    protected function vAlign($in)
    {
        $vals = array(
            '^' => 'top',
            '-' => 'middle',
            '~' => 'bottom');
        return (isset($vals[$in])) ? $vals[$in] : '';
    }


    /**
     * @internal
     **/
    protected function encodeHigh($text, $charset = "UTF-8")
    {
        return ($this->mb) ? mb_encode_numericentity($text, $this->cmap, $charset) : htmlentities($text, ENT_NOQUOTES, $charset);
    }


    /**
     * @internal
     **/
    protected function decodeHigh($text, $charset = "UTF-8")
    {
        $text = (ctype_digit($text)) ? "&#$text;" : "&$text;" ;
        return ($this->mb) ? mb_decode_numericentity($text, $this->cmap, $charset) : html_entity_decode($text, ENT_NOQUOTES, $charset);
    }


    /**
     * @internal
     **/
    protected function encodeHTML($str, $quotes=1)
    {
        $a = array(
            '&' => '&amp;',
            '<' => '&lt;',
            '>' => '&gt;',
        );
        if ($quotes) $a = $a + array(
            "'" => '&#39;', // Numeric, as in htmlspecialchars
            '"' => '&quot;',
        );

        return str_replace(array_keys($a), $a, $str);
    }


    /**
     * @internal
     **/
    protected function rEncodeHTML($str, $quotes=1)
    {
        // In restricted mode, all input but quotes has already been escaped
        if ($this->restricted)
            return str_replace('"', '&quot;', $str);
        return $this->encodeHTML($str, $quotes);
    }
}