www/plugins/facteur/lib/markdownify/markdownify_extra.php

   1 <?php
   2 /**
   3  * Class to convert HTML to Markdown with PHP Markdown Extra syntax support.
   4  *
   5  * @version 1.0.0 alpha
   6  * @author Milian Wolff (<mail@milianw.de>, <http://milianw.de>)
   7  * @license LGPL, see LICENSE_LGPL.txt and the summary below
   8  * @copyright (C) 2007  Milian Wolff
   9  *
  10  * This library is free software; you can redistribute it and/or
  11  * modify it under the terms of the GNU Lesser General Public
  12  * License as published by the Free Software Foundation; either
  13  * version 2.1 of the License, or (at your option) any later version.
  14  *
  15  * This library is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18  * Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public
  21  * License along with this library; if not, write to the Free Software
  22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
  23  */
  24
  25 /**
  26  * standard Markdownify class
  27  */
  28 require_once dirname(__FILE__) . '/markdownify.php';
  29
  30 class Markdownify_Extra extends Markdownify {
  31   /**
  32    * table data, including rows with content and the maximum width of each col
  33    *
  34    * @var array
  35    */
  36   var $table = array();
  37   /**
  38    * current col
  39    *
  40    * @var int
  41    */
  42   var $col = -1;
  43   /**
  44    * current row
  45    *
  46    * @var int
  47    */
  48   var $row = 0;
  49   /**
  50    * constructor, see Markdownify::Markdownify() for more information
  51    */
  52   function Markdownify_Extra($linksAfterEachParagraph = MDFY_LINKS_EACH_PARAGRAPH, $bodyWidth = MDFY_BODYWIDTH, $keepHTML = MDFY_KEEPHTML) {
  53     parent::Markdownify($linksAfterEachParagraph, $bodyWidth, $keepHTML);
  54
  55     ### new markdownable tags & attributes
  56     # header ids: # foo {bar}
  57     $this->isMarkdownable['h1']['id'] = 'optional';
  58     $this->isMarkdownable['h2']['id'] = 'optional';
  59     $this->isMarkdownable['h3']['id'] = 'optional';
  60     $this->isMarkdownable['h4']['id'] = 'optional';
  61     $this->isMarkdownable['h5']['id'] = 'optional';
  62     $this->isMarkdownable['h6']['id'] = 'optional';
  63     # tables
  64     $this->isMarkdownable['table'] = array();
  65     $this->isMarkdownable['th'] = array(
  66       'align' => 'optional',
  67     );
  68     $this->isMarkdownable['td'] = array(
  69       'align' => 'optional',
  70     );
  71     $this->isMarkdownable['tr'] = array();
  72     array_push($this->ignore, 'thead');
  73     array_push($this->ignore, 'tbody');
  74     array_push($this->ignore, 'tfoot');
  75     # definition lists
  76     $this->isMarkdownable['dl'] = array();
  77     $this->isMarkdownable['dd'] = array();
  78     $this->isMarkdownable['dt'] = array();
  79     # footnotes
  80     $this->isMarkdownable['fnref'] = array(
  81       'target' => 'required',
  82     );
  83     $this->isMarkdownable['footnotes'] = array();
  84     $this->isMarkdownable['fn'] = array(
  85       'name' => 'required',
  86     );
  87     $this->parser->blockElements['fnref'] = false;
  88     $this->parser->blockElements['fn'] = true;
  89     $this->parser->blockElements['footnotes'] = true;
  90     # abbr
  91     $this->isMarkdownable['abbr'] = array(
  92       'title' => 'required',
  93     );
  94     # build RegEx lookahead to decide wether table can pe parsed or not
  95     $inlineTags = array_keys($this->parser->blockElements, false);
  96     $colContents = '(?:[^<]|<(?:'.implode('|', $inlineTags).'|[^a-z]))+';
  97     $this->tableLookaheadHeader = '{
  98     ^\s*(?:<thead\s*>)?\s*                               # open optional thead
  99       <tr\s*>\s*(?:                                    # start required row with headers
 100         <th(?:\s+align=("|\')(?:left|center|right)\1)?\s*>   # header with optional align
 101         \s*'.$colContents.'\s*                       # contents
 102         </th>\s*                                     # close header
 103       )+</tr>                                          # close row with headers
 104     \s*(?:</thead>)?                                     # close optional thead
 105     }sxi';
 106     $this->tdSubstitute = '\s*'.$colContents.'\s*        # contents
 107           </td>\s*';
 108     $this->tableLookaheadBody = '{
 109       \s*(?:<tbody\s*>)?\s*                            # open optional tbody
 110         (?:<tr\s*>\s*                                # start row
 111           %s                                       # cols to be substituted
 112         </tr>)+                                      # close row
 113       \s*(?:</tbody>)?                                 # close optional tbody
 114     \s*</table>                                          # close table
 115     }sxi';
 116   }
 117   /**
 118    * handle header tags (<h1> - <h6>)
 119    *
 120    * @param int $level 1-6
 121    * @return void
 122    */
 123   function handleHeader($level) {
 124     static $id = null;
 125     if ($this->parser->isStartTag) {
 126       if (isset($this->parser->tagAttributes['id'])) {
 127         $id = $this->parser->tagAttributes['id'];
 128       }
 129     } else {
 130       if (!is_null($id)) {
 131         $this->out(' {#'.$id.'}');
 132         $id = null;
 133       }
 134     }
 135     parent::handleHeader($level);
 136   }
 137   /**
 138    * handle <abbr> tags
 139    *
 140    * @param void
 141    * @return void
 142    */
 143   function handleTag_abbr() {
 144     if ($this->parser->isStartTag) {
 145       $this->stack();
 146       $this->buffer();
 147     } else {
 148       $tag = $this->unstack();
 149       $tag['text'] = $this->unbuffer();
 150       $add = true;
 151       foreach ($this->stack['abbr'] as $stacked) {
 152         if ($stacked['text'] == $tag['text']) {
 153           /** TODO: differing abbr definitions, i.e. different titles for same text **/
 154           $add = false;
 155           break;
 156         }
 157       }
 158       $this->out($tag['text']);
 159       if ($add) {
 160         array_push($this->stack['abbr'], $tag);
 161       }
 162     }
 163   }
 164   /**
 165    * flush stacked abbr tags
 166    *
 167    * @param void
 168    * @return void
 169    */
 170   function flushStacked_abbr() {
 171     $out = array();
 172     foreach ($this->stack['abbr'] as $k => $tag) {
 173       if (!isset($tag['unstacked'])) {
 174         array_push($out, ' *['.$tag['text'].']: '.$tag['title']);
 175         $tag['unstacked'] = true;
 176         $this->stack['abbr'][$k] = $tag;
 177       }
 178     }
 179     if (!empty($out)) {
 180       $this->out("\n\n".implode("\n", $out));
 181     }
 182   }
 183   /**
 184    * handle <table> tags
 185    *
 186    * @param void
 187    * @return void
 188    */
 189   function handleTag_table() {
 190     if ($this->parser->isStartTag) {
 191       # check if upcoming table can be converted
 192       if ($this->keepHTML) {
 193         if (preg_match($this->tableLookaheadHeader, $this->parser->html, $matches)) {
 194           # header seems good, now check body
 195           # get align & number of cols
 196           preg_match_all('#<th(?:\s+align=("|\')(left|right|center)\1)?\s*>#si', $matches[0], $cols);
 197           $regEx = '';
 198           $i = 1;
 199           $aligns = array();
 200           foreach ($cols[2] as $align) {
 201             $align = strtolower($align);
 202             array_push($aligns, $align);
 203             if (empty($align)) {
 204               $align = 'left'; # default value
 205             }
 206             $td = '\s+align=("|\')'.$align.'\\'.$i;
 207             $i++;
 208             if ($align == 'left') {
 209               # look for empty align or left
 210               $td = '(?:'.$td.')?';
 211             }
 212             $td = '<td'.$td.'\s*>';
 213             $regEx .= $td.$this->tdSubstitute;
 214           }
 215           $regEx = sprintf($this->tableLookaheadBody, $regEx);
 216           if (preg_match($regEx, $this->parser->html, $matches, null, strlen($matches[0]))) {
 217             # this is a markdownable table tag!
 218             $this->table = array(
 219               'rows' => array(),
 220               'col_widths' => array(),
 221               'aligns' => $aligns,
 222             );
 223             $this->row = 0;
 224           } else {
 225             # non markdownable table
 226             $this->handleTagToText();
 227           }
 228         } else {
 229           # non markdownable table
 230           $this->handleTagToText();
 231         }
 232       } else {
 233         $this->table = array(
 234           'rows' => array(),
 235           'col_widths' => array(),
 236           'aligns' => array(),
 237         );
 238         $this->row = 0;
 239       }
 240     } else {
 241       # finally build the table in Markdown Extra syntax
 242       $separator = array();
 243       # seperator with correct align identifikators
 244       foreach($this->table['aligns'] as $col => $align) {
 245         if (!$this->keepHTML && !isset($this->table['col_widths'][$col])) {
 246           break;
 247         }
 248         $left = ' ';
 249         $right = ' ';
 250         switch ($align) {
 251           case 'left':
 252             $left = ':';
 253             break;
 254           case 'center':
 255             $right = ':';
 256             $left = ':';
 257           case 'right':
 258             $right = ':';
 259             break;
 260         }
 261         array_push($separator, $left.str_repeat('-', $this->table['col_widths'][$col]).$right);
 262       }
 263       $separator = '|'.implode('|', $separator).'|';
 264
 265       $rows = array();
 266       # add padding
 267       array_walk_recursive($this->table['rows'], array(&$this, 'alignTdContent'));
 268       $header = array_shift($this->table['rows']);
 269       array_push($rows, '| '.implode(' | ', $header).' |');
 270       array_push($rows, $separator);
 271       foreach ($this->table['rows'] as $row) {
 272         array_push($rows, '| '.implode(' | ', $row).' |');
 273       }
 274       $this->out(implode("\n".$this->indent, $rows));
 275       $this->table = array();
 276       $this->setLineBreaks(2);
 277     }
 278   }
 279   /**
 280    * properly pad content so it is aligned as whished
 281    * should be used with array_walk_recursive on $this->table['rows']
 282    *
 283    * @param string &$content
 284    * @param int $col
 285    * @return void
 286    */
 287   function alignTdContent(&$content, $col) {
 288     switch ($this->table['aligns'][$col]) {
 289       default:
 290       case 'left':
 291         $content .= str_repeat(' ', $this->table['col_widths'][$col] - $this->strlen($content));
 292         break;
 293       case 'right':
 294         $content = str_repeat(' ', $this->table['col_widths'][$col] - $this->strlen($content)).$content;
 295         break;
 296       case 'center':
 297         $paddingNeeded = $this->table['col_widths'][$col] - $this->strlen($content);
 298         $left = floor($paddingNeeded / 2);
 299         $right = $paddingNeeded - $left;
 300         $content = str_repeat(' ', $left).$content.str_repeat(' ', $right);
 301         break;
 302     }
 303   }
 304   /**
 305    * handle <tr> tags
 306    *
 307    * @param void
 308    * @return void
 309    */
 310   function handleTag_tr() {
 311     if ($this->parser->isStartTag) {
 312       $this->col = -1;
 313     } else {
 314       $this->row++;
 315     }
 316   }
 317   /**
 318    * handle <td> tags
 319    *
 320    * @param void
 321    * @return void
 322    */
 323   function handleTag_td() {
 324     if ($this->parser->isStartTag) {
 325       $this->col++;
 326       if (!isset($this->table['col_widths'][$this->col])) {
 327         $this->table['col_widths'][$this->col] = 0;
 328       }
 329       $this->buffer();
 330     } else {
 331       $buffer = trim($this->unbuffer());
 332       $this->table['col_widths'][$this->col] = max($this->table['col_widths'][$this->col], $this->strlen($buffer));
 333       $this->table['rows'][$this->row][$this->col] = $buffer;
 334     }
 335   }
 336   /**
 337    * handle <th> tags
 338    *
 339    * @param void
 340    * @return void
 341    */
 342   function handleTag_th() {
 343     if (!$this->keepHTML && !isset($this->table['rows'][1]) && !isset($this->table['aligns'][$this->col+1])) {
 344       if (isset($this->parser->tagAttributes['align'])) {
 345         $this->table['aligns'][$this->col+1] = $this->parser->tagAttributes['align'];
 346       } else {
 347         $this->table['aligns'][$this->col+1] = '';
 348       }
 349     }
 350     $this->handleTag_td();
 351   }
 352   /**
 353    * handle <dl> tags
 354    *
 355    * @param void
 356    * @return void
 357    */
 358   function handleTag_dl() {
 359     if (!$this->parser->isStartTag) {
 360       $this->setLineBreaks(2);
 361     }
 362   }
 363   /**
 364    * handle <dt> tags
 365    *
 366    * @param void
 367    * @return void
 368    **/
 369   function handleTag_dt() {
 370     if (!$this->parser->isStartTag) {
 371       $this->setLineBreaks(1);
 372     }
 373   }
 374   /**
 375    * handle <dd> tags
 376    *
 377    * @param void
 378    * @return void
 379    */
 380   function handleTag_dd() {
 381     if ($this->parser->isStartTag) {
 382       if (substr(ltrim($this->parser->html), 0, 3) == '<p>') {
 383         # next comes a paragraph, so we'll need an extra line
 384         $this->out("\n".$this->indent);
 385       } elseif (substr($this->output, -2) == "\n\n") {
 386         $this->output = substr($this->output, 0, -1);
 387       }
 388       $this->out(':   ');
 389       $this->indent('    ', false);
 390     } else {
 391       # lookahead for next dt
 392       if (substr(ltrim($this->parser->html), 0, 4) == '<dt>') {
 393         $this->setLineBreaks(2);
 394       } else {
 395         $this->setLineBreaks(1);
 396       }
 397       $this->indent('    ');
 398     }
 399   }
 400   /**
 401    * handle <fnref /> tags (custom footnote references, see markdownify_extra::parseString())
 402    *
 403    * @param void
 404    * @return void
 405    */
 406   function handleTag_fnref() {
 407     $this->out('[^'.$this->parser->tagAttributes['target'].']');
 408   }
 409   /**
 410    * handle <fn> tags (custom footnotes, see markdownify_extra::parseString()
 411    * and markdownify_extra::_makeFootnotes())
 412    *
 413    * @param void
 414    * @return void
 415    */
 416   function handleTag_fn() {
 417     if ($this->parser->isStartTag) {
 418       $this->out('[^'.$this->parser->tagAttributes['name'].']:');
 419       $this->setLineBreaks(1);
 420     } else {
 421       $this->setLineBreaks(2);
 422     }
 423     $this->indent('    ');
 424   }
 425   /**
 426    * handle <footnotes> tag (custom footnotes, see markdownify_extra::parseString()
 427    *  and markdownify_extra::_makeFootnotes())
 428    *
 429    *  @param void
 430    *  @return void
 431    */
 432   function handleTag_footnotes() {
 433     if (!$this->parser->isStartTag) {
 434       $this->setLineBreaks(2);
 435     }
 436   }
 437   /**
 438    * parse a HTML string, clean up footnotes prior
 439    *
 440    * @param string $HTML input
 441    * @return string Markdown formatted output
 442    */
 443   function parseString($html) {
 444     /** TODO: custom markdown-extra options, e.g. titles & classes **/
 445     # <sup id="fnref:..."><a href"#fn..." rel="footnote">...</a></sup>
 446     # => <fnref target="..." />
 447     $html = preg_replace('@<sup id="fnref:([^"]+)">\s*<a href="#fn:\1" rel="footnote">\s*\d+\s*</a>\s*</sup>@Us', '<fnref target="$1" />', $html);
 448     # <div class="footnotes">
 449     # <hr />
 450     # <ol>
 451     #
 452     # <li id="fn:...">...</li>
 453     # ...
 454     #
 455     # </ol>
 456     # </div>
 457     # =>
 458     # <footnotes>
 459     #   <fn name="...">...</fn>
 460     #   ...
 461     # </footnotes>
 462     $html = preg_replace_callback('#<div class="footnotes">\s*<hr />\s*<ol>\s*(.+)\s*</ol>\s*</div>#Us', array(&$this, '_makeFootnotes'), $html);
 463     return parent::parseString($html);
 464   }
 465   /**
 466    * replace HTML representation of footnotes with something more easily parsable
 467    *
 468    * @note this is a callback to be used in parseString()
 469    *
 470    * @param array $matches
 471    * @return string
 472    */
 473   function _makeFootnotes($matches) {
 474     # <li id="fn:1">
 475     #   ...
 476     #   <a href="#fnref:block" rev="footnote">&#8617;</a></p>
 477     # </li>
 478     # => <fn name="1">...</fn>
 479     # remove footnote link
 480     $fns = preg_replace('@\s*(&#160;\s*)?<a href="#fnref:[^"]+" rev="footnote"[^>]*>&#8617;</a>\s*@s', '', $matches[1]);
 481     # remove empty paragraph
 482     $fns = preg_replace('@<p>\s*</p>@s', '', $fns);
 483     # <li id="fn:1">...</li> -> <footnote nr="1">...</footnote>
 484     $fns = str_replace('<li id="fn:', '<fn name="', $fns);
 485
 486     $fns = '<footnotes>'.$fns.'</footnotes>';
 487     return preg_replace('#</li>\s*(?=(?:<fn|</footnotes>))#s', '</fn>$1', $fns);
 488   }
 489 }