banana/message.func.inc.php

   1 <?php
   2 /********************************************************************************
   3  * * banana/message.func.inc.php : function to display messages
   4  * * ------------------------
   5  * *
   6  * * This file is part of the banana distribution
   7  * * Copyright: See COPYING files that comes with this distribution
   8  * ********************************************************************************/
   9
  10 require_once dirname(__FILE__) . '/mimepart.inc.php';
  11 require_once dirname(__FILE__) . '/banana.inc.php';
  12
  13 // {{{ Plain Text Functions
  14
  15 function banana_isFlowed($line)
  16 {
  17     return ctype_space(substr($line, -1)) && $line != '-- ';
  18 }
  19
  20 function banana_removeQuotes($line, &$quote_level, $strict = true)
  21 {
  22     $quote_level = 0;
  23     if (empty($line)) {
  24         return '';
  25     }
  26     while ($line{0} == '>') {
  27         $line = substr($line, 1);
  28         if (!$strict && ctype_space($line{0})) {
  29             $line = substr($line, 1);
  30         }
  31         $quote_level++;
  32     }
  33     if (ctype_space($line{0})) {
  34         $line = substr($line, 1);
  35     }
  36     return $line;
  37 }
  38
  39 function banana_quote($line, $level, $mark = '>')
  40 {
  41     $lines = explode("\n", $line);
  42     $quote = str_repeat($mark, $level);
  43     foreach ($lines as &$line) {
  44         $line = $quote . $line;
  45     }
  46     return implode("\n", $lines);
  47 }
  48
  49 function banana_unflowed($text)
  50 {
  51     $lines = explode("\n", $text);
  52     $text = '';
  53     while (!is_null($line = array_shift($lines))) {
  54         $level = 0;
  55         $line = banana_removeQuotes($line, $level);
  56         while (banana_isFlowed($line)) {
  57             $lvl = 0;
  58             if (empty($lines)) {
  59                 break;
  60             }
  61             $nl  = $lines[0];
  62             $nl = banana_removeQuotes($nl, $lvl);
  63             if ($lvl == $level) {
  64                 $line .= $nl;
  65                 array_shift($lines);
  66             } else {
  67                 break;
  68             }
  69         }
  70         $text .= banana_quote($line, $level) . "\n";
  71     }
  72     return $text;
  73 }
  74
  75 function banana_wordwrap($text, $quote_level)
  76 {
  77     if ($quote_level > 0) {
  78         $length = Banana::$msgshow_wrap - $quote_level - 1;
  79         return banana_quote(wordwrap($text, $length), $quote_level);
  80
  81     }
  82     return wordwrap($text, Banana::$msgshow_wrap);
  83 }
  84
  85 function banana_catchFormats($text)
  86 {
  87     $formatting = Array('/' => 'em', // match / first in order not to match closing markups </...> <> </>
  88                         '_' => 'u',
  89                         '*' => 'strong');
  90     $url = Banana::$msgshow_url;
  91     preg_match_all("/$url/ui", $text, $urls);
  92     $text = str_replace($urls[0], "&&&urls&&&", $text);
  93     foreach ($formatting as $limit=>$mark) {
  94         $limit = preg_quote($limit, '/');
  95         $text = preg_replace('/' . $limit . '(\S+?)' . $limit . '/us',
  96                              "<$mark>\\1</$mark>", $text);
  97     }
  98     return preg_replace('/&&&urls&&&/e', 'array_shift($urls[0])', $text);
  99 }
 100
 101 // {{{ URL Catcher tools
 102
 103 function banana__cutlink($link)
 104 {
 105     $link = banana_html_entity_decode($link, ENT_QUOTES);
 106     if (strlen($link) > Banana::$msgshow_wrap) {
 107         $link = substr($link, 0, Banana::$msgshow_wrap - 3) . "...";
 108     }
 109     return banana_htmlentities($link, ENT_QUOTES);
 110 }
 111
 112 function banana__cleanURL($url)
 113 {
 114     $url = str_replace('@', '%40', $url);
 115     if (strpos($url, '://') === false) {
 116         $url = 'http://' . $url;
 117     }
 118     return '<a href="'.$url.'" title="'.$url.'">' . banana__cutlink($url) . '</a>';
 119 }
 120
 121 function banana__catchMailLink($email)
 122 {
 123     $mid = '<' . $email . '>';
 124     if (isset(Banana::$spool->ids[$mid])) {
 125         return Banana::$page->makeLink(Array('group' => Banana::$group,
 126                                              'artid' => Banana::$spool->ids[$mid],
 127                                              'text'  => $email));
 128     } elseif (strpos($email, '$') !== false) {
 129         return $email;
 130     }
 131     return '<a href="mailto:' . $email . '">' . $email . '</a>';
 132 }
 133
 134 // }}}
 135
 136 function banana_catchURLs($text)
 137 {
 138     $url  = Banana::$msgshow_url;
 139
 140     $res  = preg_replace("/&(lt|gt|quot);/", " &\\1; ", $text);
 141     $res  = preg_replace("/$url/uie", "'\\1'.banana__cleanurl('\\2').'\\3'", $res);
 142     $res  = preg_replace('/(["\[])?(?:mailto:|news:)?([a-z0-9.\-+_\$]+@([\-.+_]?[a-z0-9])+)(["\]])?/ie',
 143                          "'\\1' . banana__catchMailLink('\\2') . '\\4'",
 144                           $res);
 145     $res  = preg_replace("/ &(lt|gt|quot); /", "&\\1;", $res);
 146     return $res;
 147 }
 148
 149 // {{{ Quotes catcher functions
 150
 151 function banana__replaceQuotes($text, $regexp)
 152 {
 153     return stripslashes(preg_replace("@(^|<pre>|\n)$regexp@i", '\1', $text));
 154 }
 155
 156 // }}}
 157
 158 function banana_catchQuotes($res, $strict = true)
 159 {
 160     if ($strict) {
 161         $regexp = "&gt;";
 162     } else {
 163         $regexp = "&gt; *";
 164     }
 165     while (preg_match("/(^|<pre>|\n)$regexp/i", $res)) {
 166         $res  = preg_replace("/(^|<pre>|\n)(($regexp.*(?:\n|$))+)/ie",
 167             "'\\1</pre><blockquote><pre>'"
 168             ." . banana__replaceQuotes('\\2', '$regexp')"
 169             ." . '</pre></blockquote><pre>'",
 170             $res);
 171     }
 172     return $res;
 173 }
 174
 175 function banana_catchSignature($res)
 176 {
 177     $res = preg_replace("@<pre>-- ?\n@", "<pre>\n-- \n", $res);
 178     $parts = preg_split("/\n-- ?\n/", $res);
 179     $sign  = '</pre><hr style="width: 100%; margin: 1em 0em; " /><pre>';
 180     return join($sign, $parts);
 181 }
 182
 183 function banana_plainTextToHtml($text, $strict = true)
 184 {
 185     $text = banana_htmlentities($text);
 186     $text = banana_catchFormats($text);
 187     $text = banana_catchURLs($text);
 188     $text = banana_catchQuotes($text, $strict);
 189     $text = banana_catchSignature($text);
 190     return banana_cleanHtml('<pre>' . $text . '</pre>');
 191 }
 192
 193 function banana_wrap($text, $base_level = 0, $strict = true)
 194 {
 195     $lines  = explode("\n", $text);
 196     $text   = '';
 197     $buffer = array();
 198     $level  = 0;
 199     while (!is_null($line = array_shift($lines))) {
 200         $lvl = 0;
 201         $line = banana_removeQuotes($line, $lvl, $strict);
 202         if($lvl != $level) {
 203             if (!empty($buffer)) {
 204                 $text  .= banana_wordwrap(implode("\n", $buffer), $level + $base_level) . "\n";
 205                 $buffer = array();
 206             }
 207             $level  = $lvl;
 208         }
 209         $buffer[] = $line;
 210     }
 211     if (!empty($buffer)) {
 212         $text .= banana_wordwrap(implode("\n", $buffer), $level + $base_level);
 213     }
 214     return $text;
 215 }
 216
 217 function banana_formatPlainText(BananaMimePart &$part, $base_level = 0)
 218 {
 219     $text = $part->getText();
 220     if ($part->isFlowed()) {
 221         $text = banana_unflowed($text);
 222     }
 223     $text = banana_wrap($text, $base_level, $part->isFlowed());
 224     return banana_plainTextToHtml($text, $part->isFlowed());
 225 }
 226
 227 function banana_quotePlainText(BananaMimePart &$part)
 228 {
 229     $text = $part->getText();
 230     if ($part->isFlowed()) {
 231         $text = banana_unflowed($text);
 232     }
 233     return banana_wrap($text, 1);
 234 }
 235
 236 // }}}
 237 // {{{ HTML Functions
 238
 239 function banana_htmlentities($text, $quote = ENT_COMPAT)
 240 {
 241     return htmlentities($text, $quote, 'UTF-8');
 242 }
 243
 244 function banana_html_entity_decode($text, $quote = ENT_COMPAT)
 245 {
 246     return html_entity_decode($text, $quote, 'UTF-8');
 247 }
 248
 249 function banana_removeEvilAttributes($tagSource)
 250 {
 251     $stripAttrib = 'javascript:|onclick|ondblclick|onmousedown|onmouseup|onmouseover|'.
 252                    'onmousemove|onmouseout|onkeypress|onkeydown|onkeyup';
 253     return stripslashes(preg_replace("/$stripAttrib/i", '', $tagSource));
 254 }
 255
 256 function banana_cleanStyles($tag, $attributes)
 257 {
 258     static $td_style, $conv, $size_conv;
 259     if (!isset($td_style)) {
 260         $conv = array('style' => 'style', 'width' => 'width', 'height' => 'height', 'border' => 'border-size',
 261                       'size' => 'font-size', 'align' => 'text-align', 'valign' => 'vertical-align', 'face' => 'font',
 262                       'bgcolor' => 'background-color', 'color' => 'color', 'style' => 'style',
 263                       'cellpadding' => 'padding', 'cellspacing' => 'border-spacing');
 264         $size_conv = array(1 => 'xx-small', 2 => 'x-small', 3 => 'small', 4 => 'medium', 5 => 'large',
 265                            6 => 'x-large',  7 => 'xx-large',
 266                            '-2' => 'xx-small', '-1' => 'x-small', '+1' => 'medium', '+2' => 'large',
 267                            '+3' => 'x-large', '+4' => 'xx-large');
 268         $td_style = array();
 269     }
 270     if ($tag == 'table') {
 271         array_unshift($td_style, '');
 272     }
 273     if ($tag == '/table') {
 274         array_shift($td_style);
 275     }
 276     if ($tag{0} == '/') {
 277         return '';
 278     }
 279     if ($tag == 'td') {
 280         $style = $td_style[0];
 281     } else {
 282         $style = '';
 283     }
 284     $attributes = str_replace("\n", ' ', stripslashes($attributes));
 285     $attributes = str_replace('= "', '="', $attributes);
 286     foreach ($conv as $att=>$stl) {
 287         $pattern = '/\b' . preg_quote($att, '/') . '="(.+?)"/i';
 288         if (preg_match($pattern, $attributes, $matches)) {
 289             $attributes = preg_replace($pattern, '', $attributes);
 290             $val = $matches[1];
 291             if ($att == 'cellspacing' && strpos($style, 'border-collapse') === false) {
 292                 $style .= "border-collapse: separate; border-spacing: $val $val; ";
 293             } elseif ($att == 'cellpadding' && $tag == 'table') {
 294                 $td_style[0] = "$stl: {$val}px; ";
 295             } elseif ($att == 'style') {
 296                 $val = rtrim($val, ' ;');
 297                 $style .= "$val; ";
 298             } elseif ($att == 'size') {
 299                 $val = $size_conv[$val];
 300                 $style .= "$stl: $val; ";
 301             } elseif (is_numeric($val)) {
 302                 $style .= "$stl: {$val}px; ";
 303             } else {
 304                 $style .= "$stl: $val; ";
 305             }
 306         }
 307     }
 308     if (!empty($style)) {
 309         $style = 'style="' . $style . '" ';
 310     }
 311     return ' ' . $style . trim($attributes);
 312 }
 313
 314 /**
 315  * @return string
 316  * @param string
 317  * @desc Strip forbidden tags and delegate tag-source check to removeEvilAttributes()
 318  */
 319 function banana_cleanHtml($source, $to_xhtml = false)
 320 {
 321     if (function_exists('tidy_repair_string')) {
 322         $tidy_on = Array(
 323             'drop-empty-paras', 'drop-proprietary-attributes',
 324             'hide-comments', 'logical-emphasis', 'output-xhtml',
 325             'replace-color',
 326         );
 327         $tidy_off = Array('join-classes', 'clean', 'show-body-only'); // 'clean' may be a good idea, but it is too aggressive
 328
 329         foreach($tidy_on as $opt) {
 330             tidy_setopt($opt, true);
 331         }
 332         foreach($tidy_off as $opt) {
 333             tidy_setopt($opt, false);
 334         }
 335         tidy_setopt('alt-text', '[ inserted by TIDY ]');
 336         tidy_setopt('wrap', '120');
 337         tidy_set_encoding('utf8');
 338         $source = tidy_repair_string($source);
 339     }
 340
 341     // To XHTML
 342     if ($to_xhtml) {
 343         // catch inline CSS
 344         $css = null;
 345         if (preg_match('/<head.*?>(.*?)<\/head>/is', $source, $matches)) {
 346             $source = preg_replace('/<head.*?>.*?<\/head>/is', '', $source);
 347             preg_match_all('/<style.*?type="text\/css".*?>(.*?)<\/style>/is', $matches[1], $matches);
 348             foreach ($matches[1] as &$match) {
 349                 $css .= $match;
 350             }
 351             $css = preg_replace("/(^|\n|,)\s*(\w+[^\{\}\<]+\{)/s", '\1.banana .message .body .html \2', $css);
 352             $css = preg_replace('/ body\b/i', '', $css);
 353             Banana::$page->addCssInline($css);
 354         }
 355
 356         // clean DTD
 357         $source = str_replace('<font', '<span', $source);
 358         $source = preg_replace('/<u\b/', '<span style="text-decoration: underline"', $source);
 359         $source = preg_replace('/<\/(font|u)>/', '</span>', $source);
 360         $source = str_replace('<body', $css ? '<div class="html"' : '<div class="html default"', $source);
 361         $source = str_replace('</body>', '</div>', $source);
 362     }
 363     $allowedTags = '<h1><h2><h3><b><i><a><ul><li><pre><hr><blockquote><img><br><div><span>'
 364                  . '<p><small><big><sup><sub><code><em><strong><table><tr><td><th>';
 365     $source = strip_tags($source, $allowedTags);
 366
 367     // Use inlined style instead of old html attributes
 368     if ($to_xhtml) {
 369         $source = preg_replace('/<(\/?\w+)(.*?)(\/?>)/ise', "'<\\1' . banana_cleanStyles('\\1', '\\2') . '\\3'", $source);
 370     }
 371     return preg_replace('/<(.*?)>/ie', "'<'.banana_removeEvilAttributes('\\1').'>'", $source);
 372 }
 373
 374 function banana_catchHtmlSignature($res)
 375 {
 376     $res = preg_replace("@(</p>)\n?-- ?\n?(<p[^>]*>|<br[^>]*>)@", "\\1<br/>-- \\2", $res);
 377     $res = preg_replace("@<br[^>]*>\n?-- ?\n?(<p[^>]*>)@", "<br/>-- <br/>\\2", $res);
 378     $res = preg_replace("@(<pre[^>]*>)\n?-- ?\n@", "<br/>-- <br/>\\1", $res);
 379     $parts = preg_split("@(:?<p[^>]*>\n?-- ?\n?</p>|<br[^>]*>\n?-- ?\n?<br[^>]*>)@", $res);
 380     $sign  = '<hr style="width: 100%; margin: 1em 0em; " />';
 381     return join($sign, $parts);
 382 }
 383
 384 // {{{ Link to part catcher tools
 385
 386 function banana__linkAttachment($cid)
 387 {
 388     return banana_htmlentities(
 389         Banana::$page->makeUrl(Array('group' => Banana::$group,
 390                                      'artid' => Banana::$artid,
 391                                      'part'  => $cid)));
 392 }
 393
 394 // }}}
 395
 396 function banana_hideExternalImages($text)
 397 {
 398     return preg_replace("/<img([^>]*?)src=['\"](?!cid).*?['\"](.*?)>/i",
 399                         '<img\1src="invalid"\2>',
 400                         $text);
 401 }
 402
 403 function banana_catchPartLinks($text)
 404 {
 405     $article = Banana::$page->makeURL(array('group' => Banana::$group, 'artid' => Banana::$artid, 'part' => Banana::$part));
 406     $article = banana_htmlentities($article);
 407     $text = preg_replace('/cid:([^\'" ]+)/e', "banana__linkAttachment('\\1')", $text);
 408     $text = preg_replace('/href="(#.*?)"/i', 'href="' . $article . '\1"', $text);
 409     return $text;
 410 }
 411
 412 // {{{ HTML to Plain Text tools
 413
 414 function banana__convertFormats($res)
 415 {
 416     $table = array('em|i'     => '/',
 417                    'strong|b' => '*',
 418                    'u'        => '_');
 419     foreach ($table as $tags=>$format) {
 420         $res = preg_replace("!</?($tags)( .*?)?>!is", $format, $res);
 421     }
 422     return $res;
 423 }
 424
 425 function banana__convertQuotes($res)
 426 {
 427     return preg_replace('!<blockquote.*?>([^<]*)</blockquote>!ies',
 428                         "\"\n\" . banana_quote(banana__convertQuotes('\\1' . \"\n\"), 1, '&gt;')",
 429                         $res);
 430 }
 431
 432 // }}}
 433
 434 function banana_htmlToPlainText($res)
 435 {
 436     $res = str_replace("\n", '', $res);
 437     $res = banana__convertFormats($res);
 438     $res = trim(strip_tags($res, '<div><br><p><blockquote>'));
 439     $res = preg_replace("@</?(br|p|div).*?>@si", "\n", $res);
 440     $res = banana__convertQuotes($res);
 441     return banana_html_entity_decode($res);
 442 }
 443
 444 function banana_formatHtml(BananaMimePart &$part)
 445 {
 446     $text = $part->getText();
 447     $text = banana_catchHtmlSignature($text);
 448     if (!Banana::$msgshow_externalimages) {
 449         $text = banana_hideExternalImages($text);
 450     }
 451     $text = banana_catchPartLinks($text);
 452     return banana_cleanHtml($text, true);
 453 }
 454
 455 function banana_quoteHtml(BananaMimePart &$part)
 456 {
 457     $text = $part->getText();
 458     $text = banana_htmlToPlainText($text);
 459     return banana_wrap($text, 1);
 460 }
 461
 462 // }}}
 463 // {{{ Richtext Functions
 464
 465 /** Convert richtext to html
 466  */
 467 function banana_richtextToHtml($source)
 468 {
 469     $tags = Array('bold'        => 'b',
 470                   'italic'      => 'i',
 471                   'smaller'     => 'small',
 472                   'bigger'      => 'big',
 473                   'underline'   => 'u',
 474                   'subscript'   => 'sub',
 475                   'superscript' => 'sup',
 476                   'excerpt'     => 'blockquote',
 477                   'paragraph'   => 'p',
 478                   'nl'          => 'br'
 479     );
 480
 481     // clean unsupported tags
 482     $protectedTags = '<signature><lt><comment><'.join('><', array_keys($tags)).'>';
 483     $source = strip_tags($source, $protectedTags);
 484
 485     // convert richtext tags to html
 486     foreach (array_keys($tags) as $tag) {
 487         $source = preg_replace('@(</?)'.$tag.'([^>]*>)@i', '\1'.$tags[$tag].'\2', $source);
 488     }
 489
 490     // some special cases
 491     $source = preg_replace('@<signature>@i', '<br>-- <br>', $source);
 492     $source = preg_replace('@</signature>@i', '', $source);
 493     $source = preg_replace('@<lt>@i', '&lt;', $source);
 494     $source = preg_replace('@<comment[^>]*>((?:[^<]|<(?!/comment>))*)</comment>@i', '<!-- \1 -->', $source);
 495     return banana_cleanHtml($source);
 496 }
 497
 498 function banana_formatRichText(BananaMimePart &$part)
 499 {
 500     $text = $part->getText();
 501     $text = banana_richtextToHtml($text);
 502     $text = banana_catchHtmlSignature($text);
 503     return banana_cleanHtml($text);
 504 }
 505
 506 function banana_quoteRichtText(BananaMimePart &$part)
 507 {
 508     $text = $part->getText();
 509     $text = banana_richtextToHtml($text);
 510     $text = banana_htmlToPlainText($text);
 511     return banana_wrap($text, 1);
 512 }
 513
 514 // }}}
 515
 516 // vim:set et sw=4 sts=4 ts=4 enc=utf-8:
 517 ?>