| commit | author | age | ||
| 4e17e6 | 1 | <?php |
| T | 2 | |
| 3 | /************************************************************************* | |
| 4 | * * | |
| 5 | * class.html2text.inc * | |
| 6 | * * | |
| 7 | ************************************************************************* | |
| 8 | * * | |
| 9 | * Converts HTML to formatted plain text * | |
| 10 | * * | |
| 11 | * Copyright (c) 2005 Jon Abernathy <jon@chuggnutt.com> * | |
| 12 | * All rights reserved. * | |
| 13 | * * | |
| 14 | * This script is free software; you can redistribute it and/or modify * | |
| 15 | * it under the terms of the GNU General Public License as published by * | |
| 16 | * the Free Software Foundation; either version 2 of the License, or * | |
| 17 | * (at your option) any later version. * | |
| 18 | * * | |
| 19 | * The GNU General Public License can be found at * | |
| 20 | * http://www.gnu.org/copyleft/gpl.html. * | |
| 21 | * * | |
| 22 | * This script is distributed in the hope that it will be useful, * | |
| 23 | * but WITHOUT ANY WARRANTY; without even the implied warranty of * | |
| 24 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * | |
| 25 | * GNU General Public License for more details. * | |
| 26 | * * | |
| 27 | * Author(s): Jon Abernathy <jon@chuggnutt.com> * | |
| 28 | * * | |
| 29 | * Last modified: 04/06/05 * | |
| 30 | * Modified: 2004/05/19 (tbr) * | |
| 31 | * * | |
| 32 | *************************************************************************/ | |
| 33 | ||
| 34 | ||
| 35 | /** | |
| 36 | * Takes HTML and converts it to formatted, plain text. | |
| 37 | * | |
| 38 | * Thanks to Alexander Krug (http://www.krugar.de/) to pointing out and | |
| 39 | * correcting an error in the regexp search array. Fixed 7/30/03. | |
| 40 | * | |
| 41 | * Updated set_html() function's file reading mechanism, 9/25/03. | |
| 42 | * | |
| 43 | * Thanks to Joss Sanglier (http://www.dancingbear.co.uk/) for adding | |
| 44 | * several more HTML entity codes to the $search and $replace arrays. | |
| 45 | * Updated 11/7/03. | |
| 46 | * | |
| 47 | * Thanks to Darius Kasperavicius (http://www.dar.dar.lt/) for | |
| 48 | * suggesting the addition of $allowed_tags and its supporting function | |
| 49 | * (which I slightly modified). Updated 3/12/04. | |
| 50 | * | |
| 51 | * Thanks to Justin Dearing for pointing out that a replacement for the | |
| 52 | * <TH> tag was missing, and suggesting an appropriate fix. | |
| 53 | * Updated 8/25/04. | |
| 54 | * | |
| 55 | * Thanks to Mathieu Collas (http://www.myefarm.com/) for finding a | |
| 56 | * display/formatting bug in the _build_link_list() function: email | |
| 57 | * readers would show the left bracket and number ("[1") as part of the | |
| 58 | * rendered email address. | |
| 59 | * Updated 12/16/04. | |
| 60 | * | |
| 61 | * Thanks to Wojciech Bajon (http://histeria.pl/) for submitting code | |
| 62 | * to handle relative links, which I hadn't considered. I modified his | |
| 63 | * code a bit to handle normal HTTP links and MAILTO links. Also for | |
| 64 | * suggesting three additional HTML entity codes to search for. | |
| 65 | * Updated 03/02/05. | |
| 66 | * | |
| 67 | * Thanks to Jacob Chandler for pointing out another link condition | |
| 68 | * for the _build_link_list() function: "https". | |
| 69 | * Updated 04/06/05. | |
| 70 | * | |
| 71 | * @author Jon Abernathy <jon@chuggnutt.com> | |
| 72 | * @version 0.6.1 | |
| 73 | * @since PHP 4.0.2 | |
| 74 | */ | |
| 75 | class html2text | |
| 76 | { | |
| 77 | ||
| 78 | /** | |
| 79 | * Contains the HTML content to convert. | |
| 80 | * | |
| 81 | * @var string $html | |
| 82 | * @access public | |
| 83 | */ | |
| 84 | var $html; | |
| 85 | ||
| 86 | /** | |
| 87 | * Contains the converted, formatted text. | |
| 88 | * | |
| 89 | * @var string $text | |
| 90 | * @access public | |
| 91 | */ | |
| 92 | var $text; | |
| 93 | ||
| 94 | /** | |
| 95 | * Maximum width of the formatted text, in columns. | |
| 96 | * | |
| 97 | * @var integer $width | |
| 98 | * @access public | |
| 99 | */ | |
| 100 | var $width = 70; | |
| 101 | ||
| 102 | /** | |
| 103 | * List of preg* regular expression patterns to search for, | |
| 104 | * used in conjunction with $replace. | |
| 105 | * | |
| 106 | * @var array $search | |
| 107 | * @access public | |
| 108 | * @see $replace | |
| 109 | */ | |
| 110 | var $search = array( | |
| 111 | "/\r/", // Non-legal carriage return | |
| 112 | "/[\n\t]+/", // Newlines and tabs | |
| 113 | '/<script[^>]*>.*?<\/script>/i', // <script>s -- which strip_tags supposedly has problems with | |
| 114 | //'/<!-- .* -->/', // Comments -- which strip_tags might have problem a with | |
| 115 | '/<a href="([^"]+)"[^>]*>(.+?)<\/a>/ie', // <a href=""> | |
| 116 | '/<h[123][^>]*>(.+?)<\/h[123]>/ie', // H1 - H3 | |
| 117 | '/<h[456][^>]*>(.+?)<\/h[456]>/ie', // H4 - H6 | |
| 118 | '/<p[^>]*>/i', // <P> | |
| 119 | '/<br[^>]*>/i', // <br> | |
| 120 | '/<b[^>]*>(.+?)<\/b>/ie', // <b> | |
| 121 | '/<i[^>]*>(.+?)<\/i>/i', // <i> | |
| 122 | '/(<ul[^>]*>|<\/ul>)/i', // <ul> and </ul> | |
| 123 | '/(<ol[^>]*>|<\/ol>)/i', // <ol> and </ol> | |
| 124 | '/<li[^>]*>/i', // <li> | |
| 125 | '/<hr[^>]*>/i', // <hr> | |
| 126 | '/(<table[^>]*>|<\/table>)/i', // <table> and </table> | |
| 127 | '/(<tr[^>]*>|<\/tr>)/i', // <tr> and </tr> | |
| 128 | '/<td[^>]*>(.+?)<\/td>/i', // <td> and </td> | |
| 40ed9b | 129 | '/<th[^>]*>(.+?)<\/th>/ie', // <th> and </th> |
| 4e17e6 | 130 | '/ /i', |
| T | 131 | '/"/i', |
| 132 | '/>/i', | |
| 133 | '/</i', | |
| 40ed9b | 134 | '/&(amp|#38);/i', |
| 4e17e6 | 135 | '/©/i', |
| T | 136 | '/™/i', |
| 137 | '/“/', | |
| 138 | '/”/', | |
| 139 | '/–/', | |
| 40ed9b | 140 | '/&#(8217|39);/', |
| 4e17e6 | 141 | '/©/', |
| T | 142 | '/™/', |
| 143 | '/—/', | |
| 144 | '/“/', | |
| 145 | '/”/', | |
| 146 | '/•/', | |
| 147 | '/®/i', | |
| 148 | '/•/i', | |
| 149 | '/&[&;]+;/i' | |
| 150 | ); | |
| 151 | ||
| 152 | /** | |
| 153 | * List of pattern replacements corresponding to patterns searched. | |
| 154 | * | |
| 155 | * @var array $replace | |
| 156 | * @access public | |
| 157 | * @see $search | |
| 158 | */ | |
| 159 | var $replace = array( | |
| 160 | '', // Non-legal carriage return | |
| 161 | ' ', // Newlines and tabs | |
| 162 | '', // <script>s -- which strip_tags supposedly has problems with | |
| 163 | //'', // Comments -- which strip_tags might have problem a with | |
| 164 | '$this->_build_link_list("\\1", "\\2")', // <a href=""> | |
| 165 | "strtoupper(\"\n\n\\1\n\n\")", // H1 - H3 | |
| 166 | "ucwords(\"\n\n\\1\n\n\")", // H4 - H6 | |
| 167 | "\n\n\t", // <P> | |
| 168 | "\n", // <br> | |
| 169 | 'strtoupper("\\1")', // <b> | |
| 170 | '_\\1_', // <i> | |
| 171 | "\n\n", // <ul> and </ul> | |
| 172 | "\n\n", // <ol> and </ol> | |
| 173 | "\t*", // <li> | |
| 174 | "\n-------------------------\n", // <hr> | |
| 175 | "\n\n", // <table> and </table> | |
| 176 | "\n", // <tr> and </tr> | |
| 177 | "\t\t\\1\n", // <td> and </td> | |
| 178 | "strtoupper(\"\t\t\\1\n\")", // <th> and </th> | |
| 179 | ' ', | |
| 180 | '"', | |
| 181 | '>', | |
| 182 | '<', | |
| 183 | '&', | |
| 184 | '(c)', | |
| 185 | '(tm)', | |
| 186 | '"', | |
| 187 | '"', | |
| 188 | '-', | |
| 189 | "'", | |
| 190 | '(c)', | |
| 191 | '(tm)', | |
| 192 | '--', | |
| 193 | '"', | |
| 194 | '"', | |
| 195 | '*', | |
| 196 | '(R)', | |
| 197 | '*', | |
| 198 | '' | |
| 199 | ); | |
| 200 | ||
| 201 | /** | |
| 202 | * Contains a list of HTML tags to allow in the resulting text. | |
| 203 | * | |
| 204 | * @var string $allowed_tags | |
| 205 | * @access public | |
| 206 | * @see set_allowed_tags() | |
| 207 | */ | |
| 208 | var $allowed_tags = ''; | |
| 209 | ||
| 210 | /** | |
| 211 | * Contains the base URL that relative links should resolve to. | |
| 212 | * | |
| 213 | * @var string $url | |
| 214 | * @access public | |
| 215 | */ | |
| 216 | var $url; | |
| 217 | ||
| 218 | /** | |
| 219 | * Indicates whether content in the $html variable has been converted yet. | |
| 220 | * | |
| 221 | * @var boolean $converted | |
| 222 | * @access private | |
| 223 | * @see $html, $text | |
| 224 | */ | |
| 225 | var $_converted = false; | |
| 226 | ||
| 227 | /** | |
| 228 | * Contains URL addresses from links to be rendered in plain text. | |
| 229 | * | |
| 230 | * @var string $link_list | |
| 231 | * @access private | |
| 232 | * @see _build_link_list() | |
| 233 | */ | |
| 234 | var $_link_list = array(); | |
| 235 | ||
| 236 | /** | |
| 237 | * Constructor. | |
| 238 | * | |
| 239 | * If the HTML source string (or file) is supplied, the class | |
| 240 | * will instantiate with that source propagated, all that has | |
| 241 | * to be done it to call get_text(). | |
| 242 | * | |
| 243 | * @param string $source HTML content | |
| 244 | * @param boolean $from_file Indicates $source is a file to pull content from | |
| 245 | * @access public | |
| 246 | * @return void | |
| 247 | */ | |
| 248 | function html2text( $source = '', $from_file = false ) | |
| 249 | { | |
| 250 | if ( !empty($source) ) { | |
| 251 | $this->set_html($source, $from_file); | |
| 252 | } | |
| 253 | $this->set_base_url(); | |
| 254 | } | |
| 255 | ||
| 256 | /** | |
| 257 | * Loads source HTML into memory, either from $source string or a file. | |
| 258 | * | |
| 259 | * @param string $source HTML content | |
| 260 | * @param boolean $from_file Indicates $source is a file to pull content from | |
| 261 | * @access public | |
| 262 | * @return void | |
| 263 | */ | |
| 264 | function set_html( $source, $from_file = false ) | |
| 265 | { | |
| 266 | $this->html = $source; | |
| 267 | ||
| 268 | if ( $from_file && file_exists($source) ) { | |
| 269 | $fp = fopen($source, 'r'); | |
| 270 | $this->html = fread($fp, filesize($source)); | |
| 271 | fclose($fp); | |
| 272 | } | |
| 273 | ||
| 274 | $this->_converted = false; | |
| 275 | } | |
| 276 | ||
| 277 | /** | |
| 278 | * Returns the text, converted from HTML. | |
| 279 | * | |
| 280 | * @access public | |
| 281 | * @return string | |
| 282 | */ | |
| 283 | function get_text() | |
| 284 | { | |
| 285 | if ( !$this->_converted ) { | |
| 286 | $this->_convert(); | |
| 287 | } | |
| 288 | ||
| 289 | return $this->text; | |
| 290 | } | |
| 291 | ||
| 292 | /** | |
| 293 | * Prints the text, converted from HTML. | |
| 294 | * | |
| 295 | * @access public | |
| 296 | * @return void | |
| 297 | */ | |
| 298 | function print_text() | |
| 299 | { | |
| 300 | print $this->get_text(); | |
| 301 | } | |
| 302 | ||
| 303 | /** | |
| 304 | * Alias to print_text(), operates identically. | |
| 305 | * | |
| 306 | * @access public | |
| 307 | * @return void | |
| 308 | * @see print_text() | |
| 309 | */ | |
| 310 | function p() | |
| 311 | { | |
| 312 | print $this->get_text(); | |
| 313 | } | |
| 314 | ||
| 315 | /** | |
| 316 | * Sets the allowed HTML tags to pass through to the resulting text. | |
| 317 | * | |
| 318 | * Tags should be in the form "<p>", with no corresponding closing tag. | |
| 319 | * | |
| 320 | * @access public | |
| 321 | * @return void | |
| 322 | */ | |
| 323 | function set_allowed_tags( $allowed_tags = '' ) | |
| 324 | { | |
| 325 | if ( !empty($allowed_tags) ) { | |
| 326 | $this->allowed_tags = $allowed_tags; | |
| 327 | } | |
| 328 | } | |
| 329 | ||
| 330 | /** | |
| 331 | * Sets a base URL to handle relative links. | |
| 332 | * | |
| 333 | * @access public | |
| 334 | * @return void | |
| 335 | */ | |
| 336 | function set_base_url( $url = '' ) | |
| 337 | { | |
| 338 | if ( empty($url) ) { | |
| 339 | $this->url = 'http://' . $_SERVER['HTTP_HOST']; | |
| 340 | } else { | |
| 341 | // Strip any trailing slashes for consistency (relative | |
| 342 | // URLs may already start with a slash like "/file.html") | |
| 343 | if ( substr($url, -1) == '/' ) { | |
| 344 | $url = substr($url, 0, -1); | |
| 345 | } | |
| 346 | $this->url = $url; | |
| 347 | } | |
| 348 | } | |
| 349 | ||
| 350 | /** | |
| 351 | * Workhorse function that does actual conversion. | |
| 352 | * | |
| 353 | * First performs custom tag replacement specified by $search and | |
| 354 | * $replace arrays. Then strips any remaining HTML tags, reduces whitespace | |
| 355 | * and newlines to a readable format, and word wraps the text to | |
| 356 | * $width characters. | |
| 357 | * | |
| 358 | * @access private | |
| 359 | * @return void | |
| 360 | */ | |
| 361 | function _convert() | |
| 362 | { | |
| 363 | // Variables used for building the link list | |
| 364 | //$link_count = 1; | |
| 365 | //$this->_link_list = ''; | |
| 366 | ||
| 367 | $text = trim(stripslashes($this->html)); | |
| 368 | ||
| 369 | // Run our defined search-and-replace | |
| 370 | $text = preg_replace($this->search, $this->replace, $text); | |
| 371 | ||
| 372 | // Strip any other HTML tags | |
| 373 | $text = strip_tags($text, $this->allowed_tags); | |
| 374 | ||
| 375 | // Bring down number of empty lines to 2 max | |
| 376 | $text = preg_replace("/\n\s+\n/", "\n", $text); | |
| 377 | $text = preg_replace("/[\n]{3,}/", "\n\n", $text); | |
| 378 | ||
| 379 | // Add link list | |
| 380 | if ( sizeof($this->_link_list) ) { | |
| 381 | $text .= "\n\nLinks:\n------\n"; | |
| 382 | foreach ($this->_link_list as $id => $link) { | |
| 383 | $text .= '[' . ($id+1) . '] ' . $link . "\n"; | |
| 384 | } | |
| 385 | } | |
| 386 | ||
| 387 | // Wrap the text to a readable format | |
| 388 | // for PHP versions >= 4.0.2. Default width is 75 | |
| 389 | $text = wordwrap($text, $this->width); | |
| 390 | ||
| 391 | $this->text = $text; | |
| 392 | ||
| 393 | $this->_converted = true; | |
| 394 | } | |
| 395 | ||
| 396 | /** | |
| 397 | * Helper function called by preg_replace() on link replacement. | |
| 398 | * | |
| 399 | * Maintains an internal list of links to be displayed at the end of the | |
| 400 | * text, with numeric indices to the original point in the text they | |
| 401 | * appeared. Also makes an effort at identifying and handling absolute | |
| 402 | * and relative links. | |
| 403 | * | |
| 404 | * @param integer $link_count Counter tracking current link number | |
| 405 | * @param string $link URL of the link | |
| 406 | * @param string $display Part of the text to associate number with | |
| 407 | * @access private | |
| 408 | * @return string | |
| 409 | */ | |
| 410 | function _build_link_list($link, $display) | |
| 411 | { | |
| 412 | $link_lc = strtolower($link); | |
| 413 | ||
| 414 | if (substr($link_lc, 0, 7) == 'http://' || substr($link_lc, 0, 8) == 'https://' || substr($link_lc, 0, 7) == 'mailto:') | |
| 415 | { | |
| 416 | $url = $link; | |
| 417 | } | |
| 418 | else | |
| 419 | { | |
| 420 | $url = $this->url; | |
| 421 | if ($link{0} != '/') { | |
| 422 | $url .= '/'; | |
| 423 | } | |
| 424 | $url .= $link; | |
| 425 | } | |
| 426 | ||
| 427 | $index = array_search($url, $this->_link_list); | |
| 428 | if ($index===FALSE) | |
| 429 | { | |
| 430 | $index = sizeof($this->_link_list); | |
| 431 | $this->_link_list[$index] = $url; | |
| 432 | } | |
| 433 | ||
| 434 | return $display . ' [' . ($index+1) . ']'; | |
| 435 | } | |
| 436 | } | |
| 437 | ||
| 438 | ?> | |
