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, $quote_level = 0)
  76 {
  77     if ($quote_level > 0) {
  78         $length = Banana::$msgshow_wrap - $quote_level - 1;
  79         return banana_quote(wordwrap($text, $length), $quote_level);
  80     }
  81     return wordwrap($text, Banana::$msgshow_wrap);
  82 }
  83
  84 function banana_catchFormats($text)
  85 {
  86     $formatting = Array('em' => array('\B\/\b', '\b\/\B'),
  87                         'u' =>  array('\b_', '_\b'),
  88                         'strong' => array('\B\*\b', '\b\*\B'));
  89     $url = Banana::$msgshow_url;
  90     preg_match_all("/$url/ui", $text, $urls);
  91     $text = str_replace($urls[0], "&&&urls&&&", $text);
  92     foreach ($formatting as $mark=>$limit) {
  93         list($ll, $lr) = $limit;
  94         $text = preg_replace('/' . $ll . '(\w+?)' . $lr . '/us',
  95                              "<$mark>\\1</$mark>", $text);
  96     }
  97     return preg_replace('/&&&urls&&&/e', 'array_shift($urls[0])', $text);
  98 }
  99
 100 /** Build a flowed text from plain text
 101  */
 102 function banana_flow($text)
 103 {
 104     $lines = explode("\n", $text);
 105     $text  = '';
 106     while (!is_null($line = array_shift($lines))) {
 107         if ($line != '-- ') {
 108             $text .= rtrim(str_replace("\n", " \n", banana_wordwrap($line))) . "\n";
 109         } else {
 110             $text .= $line . "\n";
 111         }
 112     }
 113     return $text;
 114 }
 115
 116 // {{{ URL Catcher tools
 117
 118 function banana__cutlink($link)
 119 {
 120     $link = banana_html_entity_decode($link, ENT_QUOTES);
 121     if (strlen($link) > Banana::$msgshow_wrap) {
 122         $link = substr($link, 0, Banana::$msgshow_wrap - 3) . "...";
 123     }
 124     return banana_htmlentities($link, ENT_QUOTES);
 125 }
 126
 127 function banana__cleanURL($url)
 128 {
 129     $url = str_replace('@', '%40', $url);
 130     if (strpos($url, '://') === false) {
 131         $url = 'http://' . $url;
 132     }
 133     return '<a href="'.$url.'" title="'.$url.'">' . banana__cutlink($url) . '</a>';
 134 }
 135
 136 function banana__catchMailLink($email)
 137 {
 138     $mid = '<' . $email . '>';
 139     if (isset(Banana::$spool->ids[$mid])) {
 140         return Banana::$page->makeLink(Array('group' => Banana::$group,
 141                                              'artid' => Banana::$spool->ids[$mid],
 142                                              'text'  => $email));
 143     } elseif (strpos($email, '$') !== false) {
 144         return $email;
 145     }
 146     return '<a href="mailto:' . $email . '">' . $email . '</a>';
 147 }
 148
 149 // }}}
 150
 151 function banana_catchURLs($text)
 152 {
 153     $url  = Banana::$msgshow_url;
 154
 155     $res  = preg_replace("/&(lt|gt|quot);/", " &\\1; ", $text);
 156     $res  = preg_replace("/$url/uie", "'\\1'.banana__cleanurl('\\2').'\\3'", $res);
 157     $res  = preg_replace('/(["\[])?(?:mailto:|news:)?([a-z0-9.\-+_\$]+@([\-.+_]?[a-z0-9])+)(["\]])?/ie',
 158                          "'\\1' . banana__catchMailLink('\\2') . '\\4'",
 159                           $res);
 160     $res  = preg_replace("/ &(lt|gt|quot); /", "&\\1;", $res);
 161     return $res;
 162 }
 163
 164 // {{{ Quotes catcher functions
 165
 166 function banana__replaceQuotes($text, $regexp)
 167 {
 168     return stripslashes(preg_replace("@(^|<pre>|\n)$regexp@i", '\1', $text));
 169 }
 170
 171 // }}}
 172
 173 function banana_catchQuotes($res, $strict = true)
 174 {
 175     if ($strict) {
 176         $regexp = "&gt;";
 177     } else {
 178         $regexp = "&gt; *";
 179     }
 180     while (preg_match("/(^|<pre>|\n)$regexp/i", $res)) {
 181         $res  = preg_replace("/(^|<pre>|\n)(($regexp.*(?:\n|$))+)/ie",
 182             "'\\1</pre><blockquote><pre>'"
 183             ." . banana__replaceQuotes('\\2', '$regexp')"
 184             ." . '</pre></blockquote><pre>'",
 185             $res);
 186     }
 187     return $res;
 188 }
 189
 190 function banana_catchSignature($res)
 191 {
 192     $res = preg_replace("@<pre>-- ?\n@", "<pre>\n-- \n", $res);
 193     $parts = preg_split("/\n-- ?\n/", $res);
 194     $sign  = '</pre><hr style="width: 100%; margin: 1em 0em; " /><pre>';
 195     return join($sign, $parts);
 196 }
 197
 198 function banana_plainTextToHtml($text, $strict = true)
 199 {
 200     $text = banana_htmlentities($text);
 201     $text = banana_catchFormats($text);
 202     $text = banana_catchURLs($text);
 203     $text = banana_catchQuotes($text, $strict);
 204     $text = banana_catchSignature($text);
 205     return banana_cleanHtml('<pre>' . $text . '</pre>');
 206 }
 207
 208 function banana_wrap($text, $base_level = 0, $strict = true)
 209 {
 210     $lines  = explode("\n", $text);
 211     $text   = '';
 212     $buffer = array();
 213     $level  = 0;
 214     while (!is_null($line = array_shift($lines))) {
 215         $lvl = 0;
 216         $line = banana_removeQuotes($line, $lvl, $strict);
 217         if($lvl != $level) {
 218             if (!empty($buffer)) {
 219                 $text  .= banana_wordwrap(implode("\n", $buffer), $level + $base_level) . "\n";
 220                 $buffer = array();
 221             }
 222             $level  = $lvl;
 223         }
 224         $buffer[] = $line;
 225     }
 226     if (!empty($buffer)) {
 227         $text .= banana_wordwrap(implode("\n", $buffer), $level + $base_level);
 228     }
 229     return $text;
 230 }
 231
 232 function banana_formatPlainText(BananaMimePart &$part, $base_level = 0)
 233 {
 234     $text = $part->getText();
 235     if ($part->isFlowed()) {
 236         $text = banana_unflowed($text);
 237     }
 238     $text = banana_wrap($text, $base_level, $part->isFlowed());
 239     return banana_plainTextToHtml($text, $part->isFlowed());
 240 }
 241
 242 function banana_quotePlainText(BananaMimePart &$part)
 243 {
 244     $text = $part->getText();
 245     if ($part->isFlowed()) {
 246         $text = banana_unflowed($text);
 247     }
 248     return banana_wrap($text, 1);
 249 }
 250
 251 // }}}
 252 // {{{ HTML Functions
 253
 254 function banana_htmlentities($text, $quote = ENT_COMPAT)
 255 {
 256     return htmlentities($text, $quote, 'UTF-8');
 257 }
 258
 259 function banana_html_entity_decode($text, $quote = ENT_COMPAT)
 260 {
 261     return html_entity_decode($text, $quote, 'UTF-8');
 262 }
 263
 264 function banana_removeEvilAttributes($tagSource)
 265 {
 266     $stripAttrib = 'javascript:|onclick|ondblclick|onmousedown|onmouseup|onmouseover|'.
 267                    'onmousemove|onmouseout|onkeypress|onkeydown|onkeyup';
 268     return stripslashes(preg_replace("/$stripAttrib/i", '', $tagSource));
 269 }
 270
 271 function banana_cleanStyles($tag, $attributes)
 272 {
 273     static $td_style, $conv, $size_conv;
 274     if (!isset($td_style)) {
 275         $conv = array('style' => 'style', 'width' => 'width', 'height' => 'height', 'border' => 'border-size',
 276                       'size' => 'font-size', 'align' => 'text-align', 'valign' => 'vertical-align', 'face' => 'font',
 277                       'bgcolor' => 'background-color', 'color' => 'color', 'style' => 'style',
 278                       'cellpadding' => 'padding', 'cellspacing' => 'border-spacing');
 279         $size_conv = array(1 => 'xx-small', 2 => 'x-small', 3 => 'small', 4 => 'medium', 5 => 'large',
 280                            6 => 'x-large',  7 => 'xx-large',
 281                            '-2' => 'xx-small', '-1' => 'x-small', '+1' => 'medium', '+2' => 'large',
 282                            '+3' => 'x-large', '+4' => 'xx-large');
 283         $td_style = array();
 284     }
 285     if ($tag == 'table') {
 286         array_unshift($td_style, '');
 287     }
 288     if ($tag == '/table') {
 289         array_shift($td_style);
 290     }
 291     if ($tag{0} == '/') {
 292         return '';
 293     }
 294     if ($tag == 'td') {
 295         $style = $td_style[0];
 296     } else {
 297         $style = '';
 298     }
 299     $attributes = str_replace("\n", ' ', stripslashes($attributes));
 300     $attributes = str_replace('= "', '="', $attributes);
 301     foreach ($conv as $att=>$stl) {
 302         $pattern = '/\b' . preg_quote($att, '/') . '=([\'"])?(.+?)(?(1)\1|(?:$| ))/i';
 303         if (preg_match($pattern, $attributes, $matches)) {
 304             $attributes = preg_replace($pattern, '', $attributes);
 305             $val = $matches[2];
 306             if ($att == 'cellspacing' && strpos($style, 'border-collapse') === false) {
 307                 $style .= "border-collapse: separate; border-spacing: $val $val; ";
 308             } elseif ($att == 'cellpadding' && $tag == 'table') {
 309                 $td_style[0] = "$stl: {$val}px; ";
 310             } elseif ($att == 'style') {
 311                 $val = rtrim($val, ' ;');
 312                 $style .= "$val; ";
 313             } elseif ($att == 'size') {
 314                 $val = $size_conv[$val];
 315                 $style .= "$stl: $val; ";
 316             } elseif (is_numeric($val)) {
 317                 $style .= "$stl: {$val}px; ";
 318             } else {
 319                 $style .= "$stl: $val; ";
 320             }
 321         }
 322     }
 323     if (!empty($style)) {
 324         $style = 'style="' . $style . '" ';
 325     }
 326     return ' ' . $style . trim($attributes);
 327 }
 328
 329 function banana__filterCss($text)
 330 {
 331     $text = preg_replace("/(,[\s\n\r]*)/s", '\1 .banana .message .body .html ', $text);
 332     return '.banana .message .body .html ' . $text;
 333 }
 334
 335 function banana_filterCss($css)
 336 {
 337     preg_match_all("/(^|\n|,\s*)\s*([\#\.@\w][^;\{\}\<]*?[\{])/s", $css, $matches);
 338     $css = preg_replace("/(^|\n)\s*([\#\.@\w][^;\{\}\<]*?)([\{])/se", '"\1" . banana__filterCss("\2") . "\3"', $css);
 339     $css = preg_replace('/ body\b/i', '', $css);
 340     if (!Banana::$msgshow_externalimages) {
 341         if (preg_match('!url\([^:\)]+:(//|\\\).*?\)!i', $css)) {
 342             $css = preg_replace('!url\([^:\)]+:(//|\\\).*?\)!i', 'url(invalid-image.png)', $css);
 343             Banana::$msgshow_hasextimages = true;
 344         }
 345     }
 346     return $css;
 347 }
 348
 349 /**
 350  * @return string
 351  * @param string
 352  * @desc Strip forbidden tags and delegate tag-source check to removeEvilAttributes()
 353  */
 354 function banana_cleanHtml($source, $to_xhtml = false)
 355 {
 356     if (function_exists('tidy_repair_string')) {
 357         $tidy_config = array('drop-empty-paras' => true,
 358                              'drop-proprietary-attributes' => true,
 359                              'hide-comments' => true,
 360                              'logical-emphasis' => true,
 361                              'output-xhtml' => true,
 362                              'replace-color' => true,
 363                              'join-classes'  => false,
 364                              'clean' => false,
 365                              'show-body-only' => false,
 366                              'alt-text' => '[ inserted by TIDY ]',
 367                              'wrap' => 120);
 368         if (function_exists('tidy_setopt')) { // Tidy 1.0
 369             foreach ($tidy_config as $field=>$value) {
 370                 tidy_setopt($field, $value);
 371             }
 372             tidy_set_encoding('utf8');
 373             $source = tidy_repair_string($source);
 374         } else { // Tidy 2.0
 375             $source = tidy_repair_string($source, $tidy_config, 'utf8');
 376         }
 377     }
 378
 379     // To XHTML
 380     if ($to_xhtml) {
 381         // catch inline CSS
 382         $css = null;
 383         if (preg_match('/<head.*?>(.*?)<\/head>/is', $source, $matches)) {
 384             $source = preg_replace('/<head.*?>.*?<\/head>/is', '', $source);
 385             preg_match_all('/<style(?:.*?type="text\/css".*?)?>(.*?)<\/style>/is', $matches[1], $matches);
 386             foreach ($matches[1] as &$match) {
 387                 $css .= $match;
 388             }
 389             $css = banana_filterCss($css);
 390             Banana::$page->addCssInline($css);
 391         }
 392
 393         // clean DTD
 394         $source = str_replace('<font', '<span', $source);
 395         $source = preg_replace('/<u\b/', '<span style="text-decoration: underline"', $source);
 396         $source = preg_replace('/<\/(font|u)>/', '</span>', $source);
 397         $source = str_replace('<body', $css ? '<div class="html"' : '<div class="html default"', $source);
 398         $source = str_replace('</body>', '</div>', $source);
 399     }
 400     $allowedTags = '<h1><h2><h3><b><i><a><ul><li><pre><hr><blockquote><img><br><div><span>'
 401                  . '<p><small><big><sup><sub><code><em><strong><table><tr><td><th>';
 402     $source = strip_tags($source, $allowedTags);
 403
 404     // Use inlined style instead of old html attributes
 405     if ($to_xhtml) {
 406         $source = preg_replace('/<(\/?\w+)(.*?)(\/?>)/uise', "'<\\1' . banana_cleanStyles('\\1', '\\2') . '\\3'", $source);
 407     }
 408     return preg_replace('/<(.*?)>/ie', "'<'.banana_removeEvilAttributes('\\1').'>'", $source);
 409 }
 410
 411 function banana_catchHtmlSignature($res)
 412 {
 413     $res = preg_replace("@(</p>)\n?-- ?\n?(<p[^>]*>|<br[^>]*>)@", "\\1<br/>-- \\2", $res);
 414     $res = preg_replace("@<br[^>]*>\n?-- ?\n?(<p[^>]*>)@", "<br/>-- <br/>\\2", $res);
 415     $res = preg_replace("@(<pre[^>]*>)\n?-- ?\n@", "<br/>-- <br/>\\1", $res);
 416     $parts = preg_split("@(:?<p[^>]*>\n?-- ?\n?</p>|<br[^>]*>\n?-- ?\n?<br[^>]*>)@", $res);
 417     $sign  = '<hr style="width: 100%; margin: 1em 0em; " />';
 418     return join($sign, $parts);
 419 }
 420
 421 // {{{ Link to part catcher tools
 422
 423 function banana__linkAttachment($cid)
 424 {
 425     return banana_htmlentities(
 426         Banana::$page->makeUrl(Array('group' => Banana::$group,
 427                                      'artid' => Banana::$artid,
 428                                      'part'  => $cid)));
 429 }
 430
 431 // }}}
 432
 433 function banana_hideExternalImages($text)
 434 {
 435     if (preg_match("/<img([^>]*?)src=['\"](?!cid).*?['\"](.*?)>/i", $text)) {
 436         Banana::$msgshow_hasextimages = true;
 437         return preg_replace("/<img([^>]*?)src=['\"](?!cid).*?['\"](.*?)>/i",
 438                             '<img\1src="invalid"\2>',
 439                             $text);
 440     }
 441     return $text;
 442 }
 443
 444 function banana_catchPartLinks($text)
 445 {
 446     $article = Banana::$page->makeURL(array('group' => Banana::$group, 'artid' => Banana::$artid, 'part' => Banana::$part));
 447     $article = banana_htmlentities($article);
 448     $text = preg_replace('/cid:([^\'" ]+)/e', "banana__linkAttachment('\\1')", $text);
 449     $text = preg_replace('/href="(#.*?)"/i', 'href="' . $article . '\1"', $text);
 450     return $text;
 451 }
 452
 453 // {{{ HTML to Plain Text tools
 454
 455 function banana__convertFormats($res)
 456 {
 457     $table = array('em|i'     => '/',
 458                    'strong|b' => '*',
 459                    'u'        => '_');
 460     foreach ($table as $tags=>$format) {
 461         $res = preg_replace("!</?($tags)( .*?)?>!is", $format, $res);
 462     }
 463     return $res;
 464 }
 465
 466 function banana__convertQuotes($res)
 467 {
 468     return preg_replace('!<blockquote.*?>([^<]*)</blockquote>!ies',
 469                         "\"\n\" . banana_quote(banana__convertQuotes('\\1' . \"\n\"), 1, '&gt;')",
 470                         $res);
 471 }
 472
 473 // }}}
 474
 475 function banana_htmlToPlainText($res)
 476 {
 477     $res = str_replace("\n", '', $res);
 478     $res = banana__convertFormats($res);
 479     $res = trim(strip_tags($res, '<div><br><p><blockquote>'));
 480     $res = preg_replace("@</?(br|p|div).*?>@si", "\n", $res);
 481     $res = banana__convertQuotes($res);
 482     return banana_html_entity_decode($res);
 483 }
 484
 485 function banana_formatHtml(BananaMimePart &$part)
 486 {
 487     $text = $part->getText();
 488     $text = banana_catchHtmlSignature($text);
 489     if (!Banana::$msgshow_externalimages) {
 490         $text = banana_hideExternalImages($text);
 491     }
 492     $text = banana_catchPartLinks($text);
 493     return banana_cleanHtml($text, true);
 494 }
 495
 496 function banana_quoteHtml(BananaMimePart &$part)
 497 {
 498     $text = $part->getText();
 499     $text = banana_htmlToPlainText($text);
 500     return banana_wrap($text, 1);
 501 }
 502
 503 // }}}
 504 // {{{ Richtext Functions
 505
 506 /** Convert richtext to html
 507  */
 508 function banana_richtextToHtml($source)
 509 {
 510     $tags = Array('bold'        => 'b',
 511                   'italic'      => 'i',
 512                   'smaller'     => 'small',
 513                   'bigger'      => 'big',
 514                   'underline'   => 'u',
 515                   'subscript'   => 'sub',
 516                   'superscript' => 'sup',
 517                   'excerpt'     => 'blockquote',
 518                   'paragraph'   => 'p',
 519                   'nl'          => 'br'
 520     );
 521
 522     // clean unsupported tags
 523     $protectedTags = '<signature><lt><comment><'.join('><', array_keys($tags)).'>';
 524     $source = strip_tags($source, $protectedTags);
 525
 526     // convert richtext tags to html
 527     foreach (array_keys($tags) as $tag) {
 528         $source = preg_replace('@(</?)'.$tag.'([^>]*>)@i', '\1'.$tags[$tag].'\2', $source);
 529     }
 530
 531     // some special cases
 532     $source = preg_replace('@<signature>@i', '<br>-- <br>', $source);
 533     $source = preg_replace('@</signature>@i', '', $source);
 534     $source = preg_replace('@<lt>@i', '&lt;', $source);
 535     $source = preg_replace('@<comment[^>]*>((?:[^<]|<(?!/comment>))*)</comment>@i', '<!-- \1 -->', $source);
 536     return banana_cleanHtml($source);
 537 }
 538
 539 function banana_formatRichText(BananaMimePart &$part)
 540 {
 541     $text = $part->getText();
 542     $text = banana_richtextToHtml($text);
 543     $text = banana_catchHtmlSignature($text);
 544     return banana_cleanHtml($text);
 545 }
 546
 547 function banana_quoteRichtText(BananaMimePart &$part)
 548 {
 549     $text = $part->getText();
 550     $text = banana_richtextToHtml($text);
 551     $text = banana_htmlToPlainText($text);
 552     return banana_wrap($text, 1);
 553 }
 554
 555 // }}}
 556
 557 // vim:set et sw=4 sts=4 ts=4 enc=utf-8:
 558 ?>