1<?php 2 3if (!defined('e107_INIT')) { exit; } 4 5/** 6* Project: MagpieRSS: a simple RSS integration tool 7* File: rss_parse.inc - parse an RSS or Atom feed 8* return as a simple object. 9* 10* Handles RSS 0.9x, RSS 2.0, RSS 1.0, and Atom 0.3 11* 12* The lastest version of MagpieRSS can be obtained from: 13* http://magpierss.sourceforge.net 14* 15* For questions, help, comments, discussion, etc., please join the 16* Magpie mailing list: 17* magpierss-general@lists.sourceforge.net 18* 19* @author Kellan Elliott-McCrea <kellan@protest.net> 20* @version 0.7a 21* @license GPL 22* 23*/ 24 25define('RSS', 'RSS'); 26define('ATOM', 'Atom'); 27 28/** 29* Hybrid parser, and object, takes RSS as a string and returns a simple object. 30* 31* see: rss_fetch.inc for a simpler interface with integrated caching support 32* 33*/ 34class MagpieRSS { 35 var $parser; 36 37 var $current_item = array(); // item currently being parsed 38 var $items = array(); // collection of parsed items 39 var $channel = array(); // hash of channel fields 40 var $textinput = array(); 41 var $image = array(); 42 var $feed_type; 43 var $feed_version; 44 var $encoding = ''; // output encoding of parsed rss 45 46 var $_source_encoding = ''; // only set if we have to parse xml prolog 47 48 var $ERROR = ""; 49 var $WARNING = ""; 50 51 // define some constants 52 53 var $_CONTENT_CONSTRUCTS = array('content', 'summary', 'info', 'title', 'tagline', 'copyright'); 54 var $_KNOWN_ENCODINGS = array('UTF-8', 'US-ASCII', 'ISO-8859-1'); 55 56 // parser variables, useless if you're not a parser, treat as private 57 var $stack = array(); // parser stack 58 var $inchannel = false; 59 var $initem = false; 60 var $incontent = false; // if in Atom <content mode="xml"> field 61 var $intextinput = false; 62 var $inimage = false; 63 var $current_field = ''; 64 var $current_namespace = false; 65 66 67 /** 68 * Set up XML parser, parse source, and return populated RSS object.. 69 * 70 * @param string $source string containing the RSS to be parsed 71 * 72 * NOTE: Probably a good idea to leave the encoding options alone unless 73 * you know what you're doing as PHP's character set support is 74 * a little weird. 75 * 76 * NOTE: A lot of this is unnecessary but harmless with PHP5 77 * 78 * 79 * @param string $output_encoding output the parsed RSS in this character 80 * set defaults to ISO-8859-1 as this is PHP's 81 * default. 82 * 83 * NOTE: might be changed to UTF-8 in future 84 * versions. 85 * 86 * @param string $input_encoding the character set of the incoming RSS source. 87 * Leave blank and Magpie will try to figure it 88 * out. 89 * 90 * 91 * @param bool $detect_encoding if false Magpie won't attempt to detect 92 * source encoding. (caveat emptor) 93 * 94 */ 95 function __construct ($source, $output_encoding = CHARSET, 96 $input_encoding=null, $detect_encoding = true) 97 { 98 # if PHP xml isn't compiled in, die 99 # 100 if (!function_exists('xml_parser_create')) { 101 $this->error( "Failed to load PHP's XML Extension. " . 102 "http://www.php.net/manual/en/ref.xml.php", 103 E_USER_ERROR ); 104 } 105 106 list($parser, $source) = $this->create_parser($source, 107 $output_encoding, $input_encoding, $detect_encoding); 108 109 110 if (!is_resource($parser)) { 111 $this->error( "Failed to create an instance of PHP's XML parser. " . 112 "http://www.php.net/manual/en/ref.xml.php", 113 E_USER_ERROR ); 114 } 115 116 117 $this->parser = $parser; 118 119 # pass in parser, and a reference to this object 120 # setup handlers 121 # 122 xml_set_object( $this->parser, $this ); 123 xml_set_element_handler($this->parser, 124 'feed_start_element', 'feed_end_element' ); 125 126 xml_set_character_data_handler( $this->parser, 'feed_cdata' ); 127 128 $status = xml_parse( $this->parser, $source ); 129 130 if (! $status ) { 131 $errorcode = xml_get_error_code( $this->parser ); 132 if ( $errorcode != XML_ERROR_NONE ) { 133 $xml_error = xml_error_string( $errorcode ); 134 $error_line = xml_get_current_line_number($this->parser); 135 $error_col = xml_get_current_column_number($this->parser); 136 $errormsg = "{$xml_error} at line {$error_line}, column {$error_col}"; 137 138 $this->error( $errormsg ); 139 } 140 } 141 142 xml_parser_free( $this->parser ); 143 144 $this->normalize(); 145 } 146 147 function feed_start_element($p, $element, &$attrs) { 148 $el = $element = strtolower($element); 149 $attrs = array_change_key_case($attrs, CASE_LOWER); 150 151 // check for a namespace, and split if found 152 $ns = false; 153 if ( strpos( $element, ':' ) ) { 154 list($ns, $el) = explode( ':', $element, 2); 155 } 156 if ( $ns and $ns != 'rdf' ) { 157 $this->current_namespace = $ns; 158 } 159 160 # if feed type isn't set, then this is first element of feed 161 # identify feed from root element 162 # 163 if (!isset($this->feed_type) ) { 164 if ( $el == 'rdf' ) { 165 $this->feed_type = RSS; 166 $this->feed_version = '1.0'; 167 } 168 elseif ( $el == 'rss' ) { 169 $this->feed_type = RSS; 170 $this->feed_version = $attrs['version']; 171 } 172 elseif ( $el == 'feed' ) { 173 $this->feed_type = ATOM; 174 $this->feed_version = $attrs['version']; 175 $this->inchannel = true; 176 } 177 return; 178 } 179 180 if ( $el == 'channel' ) 181 { 182 $this->inchannel = true; 183 } 184 elseif ($el == 'item' or $el == 'entry' ) 185 { 186 $this->initem = true; 187 if ( isset($attrs['rdf:about']) ) { 188 $this->current_item['about'] = $attrs['rdf:about']; 189 } 190 } 191 192 // if we're in the default namespace of an RSS feed, 193 // record textinput or image fields 194 elseif ( 195 $this->feed_type == RSS and 196 $this->current_namespace == '' and 197 $el == 'textinput' ) 198 { 199 $this->intextinput = true; 200 } 201 202 elseif ( 203 $this->feed_type == RSS and 204 $this->current_namespace == '' and 205 $el == 'image' ) 206 { 207 $this->inimage = true; 208 } 209 210 # handle atom content constructs 211 elseif ( $this->feed_type == ATOM and in_array($el, $this->_CONTENT_CONSTRUCTS) ) 212 { 213 // avoid clashing w/ RSS mod_content 214 if ($el == 'content' ) { 215 $el = 'atom_content'; 216 } 217 218 $this->incontent = $el; 219 220 221 } 222 223 // if inside an Atom content construct (e.g. content or summary) field treat tags as text 224 elseif ($this->feed_type == ATOM and $this->incontent ) 225 { 226 // if tags are inlined, then flatten 227 $attrs_str = join(' ', 228 array_map('map_attrs', 229 array_keys($attrs), 230 array_values($attrs) ) ); 231 232 $this->append_content( "<{$element} {$attrs_str}>" ); 233 234 array_unshift( $this->stack, $el ); 235 } 236 237 // Atom support many links per containging element. 238 // Magpie treats link elements of type rel='alternate' 239 // as being equivalent to RSS's simple link element. 240 // 241 elseif ($this->feed_type == ATOM and $el == 'link' ) 242 { 243 if ( isset($attrs['rel']) and $attrs['rel'] == 'alternate' ) 244 { 245 $link_el = 'link'; 246 } 247 else { 248 $link_el = 'link_' . $attrs['rel']; 249 } 250 251 $this->append($link_el, $attrs['href']); 252 } 253 // set stack[0] to current element 254 else { 255 array_unshift($this->stack, $el); 256 } 257 } 258 259 260 261 function feed_cdata ($p, $text) { 262 263 if ($this->feed_type == ATOM and $this->incontent) 264 { 265 $this->append_content( $text ); 266 } 267 else { 268 $current_el = join('_', array_reverse($this->stack)); 269 $this->append($current_el, $text); 270 } 271 } 272 273 function feed_end_element ($p, $el) { 274 $el = strtolower($el); 275 276 if ( $el == 'item' or $el == 'entry' ) 277 { 278 $this->items[] = $this->current_item; 279 $this->current_item = array(); 280 $this->initem = false; 281 } 282 elseif ($this->feed_type == RSS and $this->current_namespace == '' and $el == 'textinput' ) 283 { 284 $this->intextinput = false; 285 } 286 elseif ($this->feed_type == RSS and $this->current_namespace == '' and $el == 'image' ) 287 { 288 $this->inimage = false; 289 } 290 elseif ($this->feed_type == ATOM and in_array($el, $this->_CONTENT_CONSTRUCTS) ) 291 { 292 $this->incontent = false; 293 } 294 elseif ($el == 'channel' or $el == 'feed' ) 295 { 296 $this->inchannel = false; 297 } 298 elseif ($this->feed_type == ATOM and $this->incontent ) { 299 // balance tags properly 300 // note: i don't think this is actually neccessary 301 if ( $this->stack[0] == $el ) 302 { 303 $this->append_content("</{$el}>"); 304 } 305 else { 306 $this->append_content("<{$el} />"); 307 } 308 309 array_shift( $this->stack ); 310 } 311 else { 312 array_shift( $this->stack ); 313 } 314 315 $this->current_namespace = false; 316 } 317 318 function concat (&$str1, $str2 = "") { 319 if (!isset($str1) ) { 320 $str1 = ""; 321 } 322 $str1 .= $str2; 323 } 324 325 326 327 function append_content($text) { 328 if ( $this->initem ) { 329 $this->concat( $this->current_item[ $this->incontent ], $text ); 330 } 331 elseif ( $this->inchannel ) { 332 $this->concat( $this->channel[ $this->incontent ], $text ); 333 } 334 } 335 336 // smart append - field and namespace aware 337 function append($el, $text) { 338 if (!$el) { 339 return; 340 } 341 if ( $this->current_namespace ) 342 { 343 if ( $this->initem ) { 344 $this->concat( 345 $this->current_item[ $this->current_namespace ][ $el ], $text); 346 } 347 elseif ($this->inchannel) { 348 $this->concat( 349 $this->channel[ $this->current_namespace][ $el ], $text ); 350 } 351 elseif ($this->intextinput) { 352 $this->concat( 353 $this->textinput[ $this->current_namespace][ $el ], $text ); 354 } 355 elseif ($this->inimage) { 356 $this->concat( 357 $this->image[ $this->current_namespace ][ $el ], $text ); 358 } 359 } 360 else { 361 if ( $this->initem ) { 362 $this->concat( 363 $this->current_item[ $el ], $text); 364 } 365 elseif ($this->intextinput) { 366 $this->concat( 367 $this->textinput[ $el ], $text ); 368 } 369 elseif ($this->inimage) { 370 $this->concat( 371 $this->image[ $el ], $text ); 372 } 373 elseif ($this->inchannel) { 374 $this->concat( 375 $this->channel[ $el ], $text ); 376 } 377 378 } 379 } 380 381 function normalize () { 382 // if atom populate rss fields 383 if ( $this->is_atom() ) { 384 $this->channel['description'] = $this->channel['tagline']; 385 for ( $i = 0; $i < count($this->items); $i++) { 386 $item = $this->items[$i]; 387 if ( isset($item['summary']) ) 388 $item['description'] = $item['summary']; 389 if ( isset($item['atom_content'])) 390 $item['content']['encoded'] = $item['atom_content']; 391 392 $atom_date = (isset($item['issued']) ) ? $item['issued'] : $item['modified']; 393 if ( $atom_date ) { 394 $epoch = @parse_w3cdtf($atom_date); 395 if ($epoch and $epoch > 0) { 396 $item['date_timestamp'] = $epoch; 397 } 398 } 399 400 $this->items[$i] = $item; 401 } 402 } 403 elseif ( $this->is_rss() ) { 404 $this->channel['tagline'] = $this->channel['description']; 405 for ( $i = 0; $i < count($this->items); $i++) { 406 $item = $this->items[$i]; 407 if ( isset($item['description'])) 408 $item['summary'] = $item['description']; 409 if ( isset($item['content']['encoded'] ) ) 410 $item['atom_content'] = $item['content']['encoded']; 411 412 if ( $this->is_rss() == '1.0' and isset($item['dc']['date']) ) { 413 $epoch = @parse_w3cdtf($item['dc']['date']); 414 if ($epoch and $epoch > 0) { 415 $item['date_timestamp'] = $epoch; 416 } 417 } 418 elseif ( isset($item['pubdate']) ) { 419 $epoch = @strtotime($item['pubdate']); 420 if ($epoch > 0) { 421 $item['date_timestamp'] = $epoch; 422 } 423 } 424 425 $this->items[$i] = $item; 426 } 427 } 428 } 429 430 431 function is_rss () { 432 if ( $this->feed_type == RSS ) { 433 return $this->feed_version; 434 } 435 else { 436 return false; 437 } 438 } 439 440 function is_atom() { 441 if ( $this->feed_type == ATOM ) { 442 return $this->feed_version; 443 } 444 else { 445 return false; 446 } 447 } 448 449 /** 450 * return XML parser, and possibly re-encoded source 451 * 452 */ 453 function create_parser($source, $out_enc, $in_enc, $detect) { 454 if ( substr(phpversion(), 0, 1) == 5) { 455 $parser = $this->php5_create_parser($in_enc, $detect); 456 } 457 else { 458 list($parser, $source) = $this->php4_create_parser($source, $in_enc, $detect); 459 } 460 if ($out_enc) { 461 $this->encoding = $out_enc; 462 xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, $out_enc); 463 } 464 465 return array($parser, $source); 466 } 467 468 /** 469 * Instantiate an XML parser under PHP5 470 * 471 * PHP5 will do a fine job of detecting input encoding 472 * if passed an empty string as the encoding. 473 * 474 * All hail libxml2! 475 * 476 */ 477 function php5_create_parser($in_enc, $detect) { 478 // by default php5 does a fine job of detecting input encodings 479 if(!$detect && $in_enc) { 480 return xml_parser_create($in_enc); 481 } 482 else { 483 return xml_parser_create(''); 484 } 485 } 486 487 /** 488 * Instaniate an XML parser under PHP4 489 * 490 * Unfortunately PHP4's support for character encodings 491 * and especially XML and character encodings sucks. As 492 * long as the documents you parse only contain characters 493 * from the ISO-8859-1 character set (a superset of ASCII, 494 * and a subset of UTF-8) you're fine. However once you 495 * step out of that comfy little world things get mad, bad, 496 * and dangerous to know. 497 * 498 * The following code is based on SJM's work with FoF 499 * @see http://minutillo.com/steve/weblog/2004/6/17/php-xml-and-character-encodings-a-tale-of-sadness-rage-and-data-loss 500 * 501 */ 502 function php4_create_parser($source, $in_enc, $detect) { 503 if ( !$detect ) { 504 return array(xml_parser_create($in_enc), $source); 505 } 506 507 if (!$in_enc) { 508 if (preg_match('/<?xml.*encoding=[\'"](.*?)[\'"].*?>/m', $source, $m)) { 509 $in_enc = strtoupper($m[1]); 510 $this->source_encoding = $in_enc; 511 } 512 else { 513 $in_enc = 'UTF-8'; 514 } 515 } 516 517 if ($this->known_encoding($in_enc)) { 518 return array(xml_parser_create($in_enc), $source); 519 } 520 521 // the dectected encoding is not one of the simple encodings PHP knows 522 523 // attempt to use the iconv extension to 524 // cast the XML to a known encoding 525 // @see http://php.net/iconv 526 527 if (function_exists('iconv')) { 528 $encoded_source = iconv($in_enc,'UTF-8', $source); 529 if ($encoded_source) { 530 return array(xml_parser_create('UTF-8'), $encoded_source); 531 } 532 } 533 534 // iconv didn't work, try mb_convert_encoding 535 // @see http://php.net/mbstring 536 if(function_exists('mb_convert_encoding')) { 537 $encoded_source = mb_convert_encoding($source, 'UTF-8', $in_enc ); 538 if ($encoded_source) { 539 return array(xml_parser_create('UTF-8'), $encoded_source); 540 } 541 } 542 543 // else 544 $this->error("Feed is in an unsupported character encoding. ({$in_enc}) " . 545 "You may see strange artifacts, and mangled characters.", 546 E_USER_NOTICE); 547 548 return array(xml_parser_create(), $source); 549 } 550 551 function known_encoding($enc) { 552 $enc = strtoupper($enc); 553 if ( in_array($enc, $this->_KNOWN_ENCODINGS) ) { 554 return $enc; 555 } 556 else { 557 return false; 558 } 559 } 560 561 function error ($errormsg, $lvl=E_USER_WARNING) { 562 // append PHP's error message if track_errors enabled 563 if ( $php_errormsg ) { 564 $errormsg .= " ({$php_errormsg})"; 565 } 566 if ( MAGPIE_DEBUG ) { 567 trigger_error( $errormsg, $lvl); 568 } 569 else { 570 error_log( $errormsg, 0); 571 } 572 573 $notices = E_USER_NOTICE|E_NOTICE; 574 if ( $lvl&$notices ) { 575 $this->WARNING = $errormsg; 576 } else { 577 $this->ERROR = $errormsg; 578 } 579 } 580 581 582} // end class RSS 583 584function map_attrs($k, $v) { 585 return "$k=\"$v\""; 586} 587 588function parse_w3cdtf ( $date_str ) { 589 590 # regex to match wc3dtf 591 $pat = "/(\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2})(:(\d{2}))?(?:([-+])(\d{2}):?(\d{2})|(Z))?/"; 592 593 if ( preg_match( $pat, $date_str, $match ) ) { 594 list( $year, $month, $day, $hours, $minutes, $seconds) = 595 array( $match[1], $match[2], $match[3], $match[4], $match[5], $match[6]); 596 597 # calc epoch for current date assuming GMT 598 $epoch = gmmktime( $hours, $minutes, $seconds, $month, $day, $year); 599 600 $offset = 0; 601 if ( $match[10] == 'Z' ) { 602 # zulu time, aka GMT 603 } 604 else { 605 list( $tz_mod, $tz_hour, $tz_min ) = 606 array( $match[8], $match[9], $match[10]); 607 608 # zero out the variables 609 if ( ! $tz_hour ) { $tz_hour = 0; } 610 if ( ! $tz_min ) { $tz_min = 0; } 611 612 $offset_secs = (($tz_hour * 60) + $tz_min) * 60; 613 614 # is timezone ahead of GMT? then subtract offset 615 # 616 if ( $tz_mod == '+' ) { 617 $offset_secs = $offset_secs * -1; 618 } 619 620 $offset = $offset_secs; 621 } 622 $epoch = $epoch + $offset; 623 return $epoch; 624 } 625 else { 626 return -1; 627 } 628} 629 630 631