PluginProbe
FeedWordPress / 2008.1101
FeedWordPress v2008.1101
trunk 0.8 0.9 0.91 0.95 0.96 0.97 0.98 0.981 0.99 0.991 0.992 0.993 2008.1030 2008.1101 2008.1105 2008.1214 2009.0612 2009.0613 2009.0618 2009.0707 2009.1111 2009.1112 2010.0127 2010.0528 All 65 releases
← All changes | feedwordpress.php +1800 -351 2009.07072008.1101 View file →
@@ -2,12 +2,13 @@
2 2 /*
3 3 Plugin Name: FeedWordPress
4 4 Plugin URI: http://projects.radgeek.com/feedwordpress
5 5 Description: simple and flexible Atom/RSS syndication for WordPress
6 -Version: 2009.0707
6 +Version: 2008.1101
7 7 Author: Charles Johnson
8 8 Author URI: http://radgeek.com/
9 9 License: GPL
10 +Last modified: 2008-11-01 1:13pm PDT
10 11 */
11 12
12 13 # This uses code derived from:
13 14 # - wp-rss-aggregate.php by Kellan Elliot-McCrea <kellan@protest.net>
@@ -14,30 +15,24 @@
14 15 # - HTTP Navigator 2 by Keyvan Minoukadeh <keyvan@k1m.com>
15 16 # - Ultra-Liberal Feed Finder by Mark Pilgrim <mark@diveintomark.org>
16 17 # according to the terms of the GNU General Public License.
17 18 #
18 -# INSTALLATION: see readme.txt or <http://projects.radgeek.com/install>
19 +# INSTALLATION: see README.text or <http://projects.radgeek.com/install>
19 20 #
20 21 # USAGE: once FeedWordPress is installed, you manage just about everything from
21 -# the WordPress Dashboard, under the Syndication menu. To ensure that fresh
22 -# content is added as it becomes available, you can convince your contributors
23 -# to put your XML-RPC URI (if WordPress is installed at
22 +# the WordPress Dashboard, under Options --> Syndication or Links --> Syndicated
23 +# To ensure that fresh content is added as it becomes available, get your
24 +# contributors to put your XML-RPC URI (if WordPress is installed at
24 25 # <http://www.zyx.com/blog>, XML-RPC requests should be sent to
25 -# <http://www.zyx.com/blog/xmlrpc.php>), or update manually under the
26 -# Syndication menu, or set up automatic updates under Syndication --> Settings,
27 -# or use a cron job.
26 +# <http://www.zyx.com/blog/xmlrpc.php>), or see `update-feeds.php`
28 27
29 28 # -- Don't change these unless you know what you're doing...
30 29
31 -define ('FEEDWORDPRESS_VERSION', '2009.0707');
30 +define ('FEEDWORDPRESS_VERSION', '2008.1101');
32 31 define ('FEEDWORDPRESS_AUTHOR_CONTACT', 'http://radgeek.com/contact');
33 32 define ('DEFAULT_SYNDICATION_CATEGORY', 'Contributors');
34 33
35 -$feedwordpress_debug = get_option('feedwordpress_debug');
36 -if (is_string($feedwordpress_debug)) :
37 - $feedwordpress_debug = ($feedwordpress_debug == 'yes');
38 -endif;
39 -define ('FEEDWORDPRESS_DEBUG', $feedwordpress_debug);
34 +define ('FEEDWORDPRESS_DEBUG', false);
40 35
41 36 define ('FEEDWORDPRESS_CAT_SEPARATOR_PATTERN', '/[:\n]/');
42 37 define ('FEEDWORDPRESS_CAT_SEPARATOR', "\n");
43 38
@@ -44,16 +39,13 @@
44 39 define ('FEEDVALIDATOR_URI', 'http://feedvalidator.org/check.cgi');
45 40
46 41 define ('FEEDWORDPRESS_FRESHNESS_INTERVAL', 10*60); // Every ten minutes
47 42
48 -define ('FWP_SCHEMA_HAS_USERMETA', 2966);
49 43 define ('FWP_SCHEMA_20', 3308); // Database schema # for WP 2.0
50 44 define ('FWP_SCHEMA_21', 4772); // Database schema # for WP 2.1
51 45 define ('FWP_SCHEMA_23', 5495); // Database schema # for WP 2.3
52 46 define ('FWP_SCHEMA_25', 7558); // Database schema # for WP 2.5
53 47 define ('FWP_SCHEMA_26', 8201); // Database schema # for WP 2.6
54 -define ('FWP_SCHEMA_27', 9872); // Database schema # for WP 2.7
55 -define ('FWP_SCHEMA_28', 11548); // Database schema # for WP 2.8
56 48
57 49 if (FEEDWORDPRESS_DEBUG) :
58 50 // Help us to pick out errors, if any.
59 51 ini_set('error_reporting', E_ALL & ~E_NOTICE);
@@ -92,9 +84,8 @@
92 84 endif;
93 85 endif;
94 86
95 87 require_once(dirname(__FILE__) . '/compatability.php'); // LEGACY API: Replicate or mock up functions for legacy support purposes
96 -require_once(dirname(__FILE__) . '/feedwordpresshtml.class.php');
97 88
98 89 // Magic quotes are just about the stupidest thing ever.
99 90 if (is_array($_POST)) :
100 91 $fwp_post = stripslashes_deep($_POST);
@@ -168,11 +159,9 @@
168 159 remove_filter('pre_link_url', 'wp_filter_kses');
169 160
170 161 # Admin menu
171 162 add_action('admin_menu', 'fwp_add_pages');
172 - add_action('admin_notices', 'fwp_check_debug');
173 - add_action('admin_notices', 'fwp_check_magpie');
174 -
163 +
175 164 # Inbound XML-RPC update methods
176 165 add_filter('xmlrpc_methods', 'feedwordpress_xmlrpc_hook');
177 166
178 167 # Outbound XML-RPC ping reform
@@ -196,12 +185,10 @@
196 185 endif;
197 186
198 187 # Cron-less auto-update. Hooray!
199 188 add_action('init', 'feedwordpress_auto_update');
200 - add_action('init', 'feedwordpress_check_for_magpie_fix');
201 189
202 190 # Default sanitizers
203 - add_filter('syndicated_item_content', array('SyndicatedPost', 'resolve_relative_uris'), 0, 2);
204 191 add_filter('syndicated_item_content', array('SyndicatedPost', 'sanitize_content'), 0, 2);
205 192
206 193 else :
207 194 # Hook in the menus, which will just point to the upgrade interface
@@ -207,33 +194,8 @@
207 194 # Hook in the menus, which will just point to the upgrade interface
208 195 add_action('admin_menu', 'fwp_add_pages');
209 196 endif; // if (!FeedWordPress::needs_upgrade())
210 197
211 -function feedwordpress_check_for_magpie_fix () {
212 - if (isset($_POST['action']) and $_POST['action']=='fix_magpie_version') :
213 - FeedWordPressCompatibility::validate_http_request(/*action=*/ 'feedwordpress_fix_magpie', /*capability=*/ 'edit_files');
214 -
215 - $back_to = $_SERVER['REQUEST_URI'];
216 - if (isset($_POST['ignore'])) :
217 - // kill error message by telling it we ignored the upgrade request for this version
218 - update_option('feedwordpress_magpie_ignored_upgrade_to', EXPECTED_MAGPIE_VERSION);
219 - $ret = 'ignored';
220 - elseif (isset($_POST['upgrade'])) :
221 - $source = dirname(__FILE__)."/MagpieRSS-upgrade/rss.php";
222 - $destination = ABSPATH . WPINC . '/rss.php';
223 - $success = @copy($source, $destination);
224 - $ret = (int) $success;
225 - endif;
226 -
227 - if (strpos($back_to, '?')===false) : $sep = '?';
228 - else : $sep = '&';
229 - endif;
230 -
231 - header("Location: {$back_to}{$sep}feedwordpress_magpie_fix=".$ret);
232 - exit;
233 - endif;
234 -} /* feedwordpress_check_for_magpie_fix() */
235 -
236 198 function feedwordpress_auto_update () {
237 199 if (FeedWordPress::stale()) :
238 200 $feedwordpress =& new FeedWordPress;
239 201 $feedwordpress->update();
@@ -286,80 +248,17 @@
286 248 ################################################################################
287 249
288 250 function is_syndicated () { return (strlen(get_syndication_feed_id()) > 0); }
289 251
290 -function get_syndication_source_link ($original = NULL) {
291 - if (is_null($original)) : $original = FeedWordPress::use_aggregator_source_data();
292 - endif;
252 +function the_syndication_source_link () { echo get_syndication_source_link(); }
253 +function get_syndication_source_link () { list($n) = get_post_custom_values('syndication_source_uri'); return $n; }
293 254
294 - if ($original) : $vals = get_post_custom_values('syndication_source_uri_original');
295 - else : $vals = array();
296 - endif;
297 -
298 - if (count($vals) == 0) : $vals = get_post_custom_values('syndication_source_uri');
299 - endif;
300 -
301 - if (count($vals) > 0) : $ret = $vals[0]; else : $ret = NULL; endif;
255 +function get_syndication_source () { list($n) = get_post_custom_values('syndication_source'); return $n; }
256 +function the_syndication_source () { echo get_syndication_source(); }
302 257
303 - return $ret;
304 -} /* function get_syndication_source_link() */
258 +function get_syndication_feed () { list($u) = get_post_custom_values('syndication_feed'); return $u; }
259 +function the_syndication_feed () { echo get_syndication_feed (); }
305 260
306 -function the_syndication_source_link ($original = NULL) { echo get_syndication_source_link($original); }
307 -
308 -function get_syndication_source ($original = NULL) {
309 - if (is_null($original)) : $original = FeedWordPress::use_aggregator_source_data();
310 - endif;
311 -
312 - if ($original) : $vals = get_post_custom_values('syndication_source_original');
313 - else : $vals = array();
314 - endif;
315 -
316 - if (count($vals) == 0) : $vals = get_post_custom_values('syndication_source');
317 - endif;
318 -
319 - if (count($vals) > 0) : $ret = $vals[0]; else : $ret = NULL; endif;
320 -
321 - return $ret;
322 -} /* function get_syndication_source() */
323 -
324 -function the_syndication_source ($original = NULL) { echo get_syndication_source($original); }
325 -
326 -function get_syndication_feed ($original = NULL) {
327 - if (is_null($original)) : $original = FeedWordPress::use_aggregator_source_data();
328 - endif;
329 -
330 - if ($original) : $vals = get_post_custom_values('syndication_feed_original');
331 - else : $vals = array();
332 - endif;
333 -
334 - if (count($vals) == 0) : $vals = get_post_custom_values('syndication_feed');
335 - endif;
336 -
337 - if (count($vals) > 0) : $ret = $vals[0]; else : $ret = NULL; endif;
338 -
339 - return $ret;
340 -} /* function get_syndication_feed() */
341 -
342 -function the_syndication_feed ($original = NULL) { echo get_syndication_feed ($original); }
343 -
344 -function get_syndication_feed_guid ($original = NULL) {
345 - if (is_null($original)) : $original = FeedWordPress::use_aggregator_source_data();
346 - endif;
347 -
348 - if ($original) : $vals = get_post_custom_values('syndication_source_id_original');
349 - else : $vals = array();
350 - endif;
351 -
352 - if (count($vals) == 0) : $vals = array(get_feed_meta('feed/id'));
353 - endif;
354 -
355 - if (count($vals) > 0) : $ret = $vals[0]; else : $ret = NULL; endif;
356 -
357 - return $ret;
358 -} /* function get_syndication_feed_guid () */
359 -
360 -function the_syndication_feed_guid ($original = NULL) { echo get_syndication_feed_guid($original); }
361 -
362 261 function get_syndication_feed_id () { list($u) = get_post_custom_values('syndication_feed_id'); return $u; }
363 262 function the_syndication_feed_id () { echo get_syndication_feed_id(); }
364 263
365 264 $feedwordpress_linkcache = array (); // only load links from database once
@@ -397,13 +296,9 @@
397 296
398 297 function feedwordpress_preserve_syndicated_content ($text) {
399 298 global $feedwordpress_the_syndicated_content;
400 299
401 - $globalExpose = (get_option('feedwordpress_formatting_filters') == 'yes');
402 - $localExpose = get_post_custom_values('_feedwordpress_formatting_filters');
403 - $expose = ($globalExpose or ((count($localExpose) > 0) and $localExpose[0]));
404 -
405 - if ( is_syndicated() and !$expose ) :
300 + if ( is_syndicated() and get_option('feedwordpress_formatting_filters') != 'yes' ) :
406 301 $feedwordpress_the_syndicated_content = $text;
407 302 else :
408 303 $feedwordpress_the_syndicated_content = NULL;
409 304 endif;
@@ -428,9 +323,9 @@
428 323 <title><?php the_syndication_source(); ?></title>
429 324 <link rel="alternate" type="text/html" href="<?php the_syndication_source_link(); ?>" />
430 325 <link rel="self" href="<?php the_syndication_feed(); ?>" />
431 326 <?php
432 - $id = get_syndication_feed_guid();
327 + $id = get_feed_meta('feed/id');
433 328 if (strlen($id) > 0) :
434 329 ?>
435 330 <id><?php print $id; ?></id>
436 331 <?php
@@ -462,142 +357,13 @@
462 357 function fwp_add_pages () {
463 358 global $fwp_capability;
464 359 global $fwp_path;
465 360
466 - $menu = array('Syndicated Sites', 'Syndication', $fwp_capability['manage_links'], $fwp_path.'/syndication.php', NULL);
467 - if (fwp_test_wp_version(FWP_SCHEMA_27)) :
468 - // add icon parameter
469 - $menu[] = WP_PLUGIN_URL.'/'.$fwp_path.'/feedwordpress-tiny.png';
470 - endif;
361 + add_menu_page('Syndicated Sites', 'Syndication', $fwp_capability['manage_links'], $fwp_path.'/syndication.php');
362 + add_submenu_page($fwp_path.'/syndication.php', 'Syndication Options', 'Options', $fwp_capability['manage_options'], $fwp_path.'/syndication-options.php');
363 + add_options_page('Syndication Options', 'Syndication', $fwp_capability['manage_options'], $fwp_path.'/syndication-options.php');
364 +} // function fwp_add_pages () */
471 365
472 - if (fwp_test_wp_version(FWP_SCHEMA_26)) :
473 - $options = __('Settings');
474 - $longoptions = __('Syndication Settings');
475 - else :
476 - $options = __('Options');
477 - $longoptions = __('Syndication Options');
478 - endif;
479 -
480 - call_user_func_array('add_menu_page', $menu);
481 - add_submenu_page($fwp_path.'/syndication.php', 'Syndicated Posts', 'Posts', $fwp_capability['manage_options'], $fwp_path.'/posts-page.php');
482 - add_submenu_page($fwp_path.'/syndication.php', 'Syndicated Authors', 'Authors', $fwp_capability['manage_options'], $fwp_path.'/authors-page.php');
483 - add_submenu_page($fwp_path.'/syndication.php', 'Categories & Tags', 'Categories & Tags', $fwp_capability['manage_options'], $fwp_path.'/categories-page.php');
484 - add_submenu_page($fwp_path.'/syndication.php', $longoptions, $options, $fwp_capability['manage_options'], $fwp_path.'/syndication-options.php');
485 -
486 - add_options_page($longoptions, 'Syndication', $fwp_capability['manage_options'], $fwp_path.'/syndication-options.php');
487 -} /* function fwp_add_pages () */
488 -
489 -function fwp_check_debug () {
490 - // This is a horrible fucking kludge that I have to do because the
491 - // admin notice code is triggered before the code that updates the
492 - // setting.
493 - if (isset($_POST['feedwordpress_debug'])) :
494 - $feedwordpress_debug = $_POST['feedwordpress_debug'];
495 - else :
496 - $feedwordpress_debug = get_option('feedwordpress_debug');
497 - endif;
498 - if ($feedwordpress_debug==='yes') :
499 -?>
500 - <div class="error">
501 -<p><strong>FeedWordPress warning.</strong> Debugging mode is <strong>ON</strong>.
502 -While it remains on, FeedWordPress displays many diagnostic error messages,
503 -warnings, and notices that are ordinarily suppressed, and also turns off all
504 -caching of feeds. Use with caution: this setting is absolutely inappropriate
505 -for a production server.</p>
506 - </div>
507 -<?php
508 - endif;
509 -} /* function fwp_check_debug () */
510 -
511 -define('EXPECTED_MAGPIE_VERSION', '2009.0618');
512 -function fwp_check_magpie () {
513 - if (isset($_REQUEST['feedwordpress_magpie_fix'])) :
514 - if ($_REQUEST['feedwordpress_magpie_fix']=='ignored') :
515 -?>
516 -<div class="updated fade">
517 -<p>O.K., we'll ignore the problem for now. FeedWordPress will not display any
518 -more error messages.</p>
519 -</div>
520 -<?php
521 - elseif ((bool) $_REQUEST['feedwordpress_magpie_fix']) :
522 -?>
523 -<div class="updated fade">
524 -<p>Congratulations! Your MagpieRSS has been successfully upgraded to the version
525 -shipped with FeedWordPress.</p>
526 -</div>
527 -<?php
528 - else :
529 - $source = dirname(__FILE__)."/MagpieRSS-upgrade/rss.php";
530 - $destination = ABSPATH . WPINC . '/rss.php';
531 - $cmd = "cp '".htmlspecialchars(addslashes($source))."' '".htmlspecialchars(addslashes($destination))."'";
532 - $cmd = wordwrap($cmd, /*width=*/ 75, /*break=*/ " \\\n\t");
533 -
534 -?>
535 -<div class="error">
536 -<p><strong>FeedWordPress was unable to automatically upgrade your copy of MagpieRSS.</strong></p>
537 -<p>It's likely that you need to change the file permissions on <code><?php print htmlspecialchars($source); ?></code>
538 -to allow FeedWordPress to overwrite it.</p>
539 -<p><strong>To perform the upgrade manually,</strong> you can
540 -use a SFTP or FTP client to upload a copy of <code>rss.php</code> from the
541 -<code>MagpieRSS-upgrades/</code> directory of your FeedWordPress archive so that
542 -it overwrites <code><?php print htmlspecialchars($destination); ?></code>. Or,
543 -if your web host provides shell access, you can issue the following command from
544 -a command prompt to perform the upgrade:</p>
545 -<pre>
546 -<samp>$</samp> <kbd><?php print $cmd; ?></kbd>
547 -</pre>
548 -<p><strong>If you've fixed the file permissions,</strong> you can try the
549 -automatic upgrade again.</p>
550 -
551 -<?php feedwordpress_upgrade_old_and_busted_buttons(); ?>
552 -</div>
553 -<?php
554 - endif;
555 - else :
556 - $magpie_version = FeedWordPress::magpie_version();
557 -
558 - $ignored = get_option('feedwordpress_magpie_ignored_upgrade_to');
559 - if (EXPECTED_MAGPIE_VERSION != $magpie_version and EXPECTED_MAGPIE_VERSION != $ignored) :
560 - if (current_user_can('edit_files')) :
561 - $youAre = 'you are';
562 - $itIsRecommendedThatYou = 'It is <strong>strongly recommended</strong> that you';
563 - else :
564 - $youAre = 'this site is';
565 - $itIsRecommendedThatYou = 'You may want to contact the administrator of the site; it is <strong>strongly recommended</strong> that they';
566 - endif;
567 - print '<div class="error">';
568 -?>
569 -<p style="font-style: italic"><strong>FeedWordPress has detected that <?php print $youAre; ?> currently using a version of
570 -MagpieRSS other than the upgraded version that ships with this version of FeedWordPress.</strong></p>
571 -<ul>
572 -<li><strong>Currently running:</strong> MagpieRSS <?php print $magpie_version; ?></li>
573 -<li><strong>Version included with FeedWordPress <?php print FEEDWORDPRESS_VERSION; ?>:</strong> MagpieRSS <?php print EXPECTED_MAGPIE_VERSION; ?></li>
574 -</ul>
575 -<p><?php print $itIsRecommendedThatYou; ?> install the upgraded
576 -version of MagpieRSS supplied with FeedWordPress. The version of
577 -MagpieRSS that ships with WordPress is very old and buggy, and
578 -encounters a number of errors when trying to parse modern Atom
579 -and RSS feeds.</p>
580 -<?php
581 - feedwordpress_upgrade_old_and_busted_buttons();
582 - print '</div>';
583 - endif;
584 - endif;
585 -}
586 -
587 -function feedwordpress_upgrade_old_and_busted_buttons() {
588 - if (current_user_can('edit_files')) :
589 -?>
590 -<form action="" method="post"><div>
591 -<?php FeedWordPressCompatibility::stamp_nonce('feedwordpress_fix_magpie'); ?>
592 -<input type="hidden" name="action" value="fix_magpie_version" />
593 -<input class="button-secondary" type="submit" name="ignore" value="<?php _e('Ignore this problem'); ?>" />
594 -<input class="button-primary" type="submit" name="upgrade" value="<?php _e('Upgrade'); ?>" />
595 -</div></form>
596 -<?php
597 - endif;
598 -}
599 -
600 366 ################################################################################
601 367 ## fwp_hold_pings() and fwp_release_pings(): Outbound XML-RPC ping reform ####
602 368 ## ... 'coz it's rude to send 500 pings the first time your aggregator runs ####
603 369 ################################################################################
@@ -751,9 +517,9 @@
751 517 #
752 518 # * Updates to existing posts since the last poll are mirrored in the
753 519 # WordPress store.
754 520 #
755 - function update ($uri = null, $crash_ts = null) {
521 + function update ($uri = null) {
756 522 global $wpdb;
757 523
758 524 if (FeedWordPress::needs_upgrade()) : // Will make duplicate posts if we don't hold off
759 525 return NULL;
@@ -766,28 +532,11 @@
766 532 endif;
767 533
768 534 do_action('feedwordpress_update', $uri);
769 535
770 - if (is_null($crash_ts)) :
771 - $crash_dt = (int) get_option('feedwordpress_update_time_limit');
772 - if ($crash_dt > 0) :
773 - $crash_ts = time() + $crash_dt;
774 - else :
775 - $crash_ts = NULL;
776 - endif;
777 - endif;
778 -
779 - // Randomize order for load balancing purposes
780 - $feed_set = $this->feeds;
781 - shuffle($feed_set);
782 -
783 536 // Loop through and check for new posts
784 537 $delta = NULL;
785 - foreach ($feed_set as $feed) :
786 - if (!is_null($crash_ts) and (time() > $crash_ts)) : // Check whether we've exceeded the time limit
787 - break;
788 - endif;
789 -
538 + foreach ($this->feeds as $feed) :
790 539 $pinged_that = (is_null($uri) or in_array($uri, array($feed->uri(), $feed->homepage())));
791 540
792 541 if (!is_null($uri)) : // A site-specific ping always updates
793 542 $timely = true;
@@ -800,12 +549,9 @@
800 549 endif;
801 550
802 551 if ($pinged_that and $timely) :
803 552 do_action('feedwordpress_check_feed', $feed->settings);
804 - $start_ts = time();
805 - $added = $feed->poll($crash_ts);
806 - do_action('feedwordpress_check_feed_complete', $feed->settings, $added, time() - $start_ts);
807 -
553 + $added = $feed->poll();
808 554 if (isset($added['new'])) : $delta['new'] += $added['new']; endif;
809 555 if (isset($added['updated'])) : $delta['updated'] += $added['updated']; endif;
810 556 endif;
811 557 endforeach;
@@ -863,9 +609,9 @@
863 609 if (function_exists('wp_insert_link')) { // WordPress 2.x
864 610 $link_id = wp_insert_link(array(
865 611 "link_name" => $name,
866 612 "link_url" => $uri,
867 - "link_category" => (fwp_test_wp_version(0, FWP_SCHEMA_21) ? $cat_id : array($cat_id)),
613 + "link_category" => array($cat_id),
868 614 "link_rss" => $rss
869 615 ));
870 616 } else { // WordPress 1.5.x
871 617 $result = $wpdb->query("
@@ -900,33 +646,8 @@
900 646
901 647 return $ret;
902 648 } // function FeedWordPress::on_unfamiliar()
903 649
904 - function null_email_set () {
905 - $base = get_option('feedwordpress_null_email_set');
906 -
907 - if ($base===false) :
908 - $ret = array('noreply@blogger.com'); // default
909 - else :
910 - $ret = array_map('strtolower',
911 - array_map('trim', explode("\n", $base)));
912 - endif;
913 - $ret = apply_filters('syndicated_item_author_null_email_set', $ret);
914 - return $ret;
915 -
916 - } /* FeedWordPress::null_email_set () */
917 -
918 - function is_null_email ($email) {
919 - $ret = in_array(strtolower(trim($email)), FeedWordPress::null_email_set());
920 - $ret = apply_filters('syndicated_item_author_is_null_email', $ret, $email);
921 - return $ret;
922 - } /* FeedWordPress::is_null_email () */
923 -
924 - function use_aggregator_source_data () {
925 - $ret = get_option('feedwordpress_use_aggregator_source_data');
926 - return apply_filters('syndicated_post_use_aggregator_source_data', ($ret=='yes'));
927 - }
928 -
929 650 function syndicated_links () {
930 651 $contributors = FeedWordPress::link_category_id();
931 652 if (function_exists('get_bookmarks')) :
932 653 $links = get_bookmarks(array("category" => $contributors));
@@ -951,12 +672,12 @@
951 672 // WordPress 2.3 introduces a new taxonomy/term API
952 673 if (function_exists('is_term')) :
953 674 $cat_id = is_term($cat, 'link_category');
954 675 // WordPress 2.1 and 2.2 use a common table for both link and post categories
955 - elseif (fwp_test_wp_version(FWP_SCHEMA_21, FWP_SCHEMA_23)) :
676 + elseif (isset($wp_db_version) and ($wp_db_version >= FWP_SCHEMA_21 and $wp_db_version < FWP_SCHEMA_23) ) :
956 677 $cat_id = $wpdb->get_var("SELECT cat_id FROM {$wpdb->categories} WHERE cat_name='$cat'");
957 678 // WordPress 1.5 and 2.0.x have a separate table for link categories
958 - elseif (fwp_test_wp_version(0, FWP_SCHEMA_21)):
679 + elseif (!isset($wp_db_version) or $wp_db_version < FWP_SCHEMA_21) :
959 680 $cat_id = $wpdb->get_var("SELECT cat_id FROM {$wpdb->linkcategories} WHERE cat_name='$cat'");
960 681 // This should never happen.
961 682 else :
962 683 FeedWordPress::critical_bug('FeedWordPress::link_category_id::wp_db_version', $wp_db_version, __LINE__);
@@ -969,12 +690,12 @@
969 690 if (function_exists('wp_insert_term')) :
970 691 $term = wp_insert_term($cat, 'link_category');
971 692 $cat_id = $term['term_id'];
972 693 // WordPress 2.1, 2.2 category API. By the way, why the fuck is this API function only available in a wp-admin module?
973 - elseif (function_exists('wp_insert_category') and !fwp_test_wp_version(FWP_SCHEMA_20, FWP_SCHEMA_21)) :
694 + elseif (function_exists('wp_insert_category')) :
974 695 $cat_id = wp_insert_category(array('cat_name' => $cat));
975 696 // WordPress 1.5 and 2.0.x
976 - elseif (fwp_test_wp_version(0, FWP_SCHEMA_21)) :
697 + elseif (!isset($wp_db_version) or $wp_db_version < FWP_SCHEMA_21) :
977 698 $result = $wpdb->query("
978 699 INSERT INTO $wpdb->linkcategories
979 700 SET
980 701 cat_id = 0,
@@ -1004,9 +725,12 @@
1004 725 $ret = false; // innocent until proven guilty
1005 726 if (!$fwp_db_version or $fwp_db_version < FEEDWORDPRESS_VERSION) :
1006 727 // This is an older version or a fresh install. Does it
1007 728 // require a database upgrade or database initialization?
1008 - if ($fwp_db_version <= 0.96) :
729 + if ($fwp_db_version > 0.96) :
730 + // No. Just brand it with the new version.
731 + update_option('feedwordpress_version', FEEDWORDPRESS_VERSION);
732 + else :
1009 733 // Yes. Check to see whether this is a fresh install or an upgrade.
1010 734 $syn = $wpdb->get_col("
1011 735 SELECT post_id
1012 736 FROM $wpdb->postmeta
@@ -1016,21 +740,8 @@
1016 740 $ret = true;
1017 741 else : // fresh install; brand it as ours
1018 742 update_option('feedwordpress_version', FEEDWORDPRESS_VERSION);
1019 743 endif;
1020 - elseif ($fwp_db_version < 2009.0707) :
1021 - // We need to clear out any busted AJAX crap
1022 - if (fwp_test_wp_version(FWP_SCHEMA_HAS_USERMETA)) :
1023 - $wpdb->query("
1024 - DELETE FROM $wpdb->usermeta
1025 - WHERE LOCATE('feedwordpress', meta_key)
1026 - AND LOCATE('box', meta_key);
1027 - ");
1028 - endif;
1029 - update_option('feedwordpress_version', FEEDWORDPRESS_VERSION);
1030 - else :
1031 - // No. Just brand it with the new version.
1032 - update_option('feedwordpress_version', FEEDWORDPRESS_VERSION);
1033 744 endif;
1034 745 endif;
1035 746 return $ret;
1036 747 }
@@ -1094,23 +805,8 @@
1094 805 endswitch;
1095 806 echo "<p>Upgrade complete. FeedWordPress is now ready to use again.</p>";
1096 807 } /* FeedWordPress::upgrade_database() */
1097 808
1098 - function create_guid_index () {
1099 - global $wpdb;
1100 -
1101 - $wpdb->query("
1102 - CREATE INDEX {$wpdb->posts}_guid_idx ON {$wpdb->posts}(guid)
1103 - ");
1104 - } /* FeedWordPress::create_guid_index () */
1105 -
1106 - function magpie_version () {
1107 - if (!defined('MAGPIE_VERSION')) : $magpie_version = $GLOBALS['wp_version'].'-default';
1108 - else : $magpie_version = MAGPIE_VERSION;
1109 - endif;
1110 - return $magpie_version;
1111 - }
1112 -
1113 809 # Utility functions for handling text settings
1114 810 function negative ($f, $setting) {
1115 811 $nego = array ('n', 'no', 'f', 'false');
1116 812 return (isset($f[$setting]) and in_array(strtolower($f[$setting]), $nego));
@@ -1125,34 +821,1454 @@
1125 821 # Internal debugging functions
1126 822 function critical_bug ($varname, $var, $line) {
1127 823 global $wp_version;
1128 824
1129 - if (defined('MAGPIE_VERSION')) : $mv = MAGPIE_VERSION;
1130 - else : $mv = 'WordPress '.$wp_version.' default.';
1131 - endif;
1132 -
1133 825 echo '<p>There may be a bug in FeedWordPress. Please <a href="'.FEEDWORDPRESS_AUTHOR_CONTACT.'">contact the author</a> and paste the following information into your e-mail:</p>';
1134 826 echo "\n<plaintext>";
1135 827 echo "Triggered at line # ".$line."\n";
1136 828 echo "FeedWordPress version: ".FEEDWORDPRESS_VERSION."\n";
1137 - echo "MagpieRSS version: {$mv}\n";
1138 - echo "WordPress version: {$wp_version}\n";
829 + echo "WordPress version: $wp_version\n";
1139 830 echo "PHP version: ".phpversion()."\n";
1140 831 echo "\n";
1141 832 echo $varname.": "; var_dump($var); echo "\n";
1142 833 die;
1143 834 }
835 +} // class FeedWordPress
836 +
837 +class SyndicatedPost {
838 + var $item = null;
1144 839
1145 - function noncritical_bug ($varname, $var, $line) {
1146 - if (FEEDWORDPRESS_DEBUG) : // halt only when we are doing debugging
1147 - FeedWordPress::critical_bug($varname, $var, $line);
840 + var $link = null;
841 + var $feed = null;
842 + var $feedmeta = null;
843 +
844 + var $post = array ();
845 + var $_base = null;
846 +
847 + var $_freshness = null;
848 + var $_wp_id = null;
849 +
850 + function SyndicatedPost ($item, $link) {
851 + global $wpdb;
852 +
853 + $this->link = $link;
854 + $feedmeta = $link->settings;
855 + $feed = $link->magpie;
856 +
857 + # This is ugly as all hell. I'd like to use apply_filters()'s
858 + # alleged support for a variable argument count, but this seems
859 + # to have been broken in WordPress 1.5. It'll be fixed somehow
860 + # in WP 1.5.1, but I'm aiming at WP 1.5 compatibility across
861 + # the board here.
862 + #
863 + # Cf.: <http://mosquito.wordpress.org/view.php?id=901>
864 + global $fwp_channel, $fwp_feedmeta;
865 + $fwp_channel = $feed; $fwp_feedmeta = $feedmeta;
866 +
867 + $this->feed = $feed;
868 + $this->feedmeta = $feedmeta;
869 +
870 + $this->item = $item;
871 + $this->item = apply_filters('syndicated_item', $this->item, $this);
872 +
873 + # Filters can halt further processing by returning NULL
874 + if (is_null($this->item)) :
875 + $this->post = NULL;
876 + else :
877 + # Note that nothing is run through $wpdb->escape() here.
878 + # That's deliberate. The escaping is done at the point
879 + # of insertion, not here, to avoid double-escaping and
880 + # to avoid screwing with syndicated_post filters
881 +
882 + $this->post['post_title'] = apply_filters('syndicated_item_title', $this->item['title'], $this);
883 +
884 + // This just gives us an alphanumeric representation of
885 + // the author. We will look up (or create) the numeric
886 + // ID for the author in SyndicatedPost::add()
887 + $this->post['named']['author'] = apply_filters('syndicated_item_author', $this->author(), $this);
888 +
889 + # Identify content and sanitize it.
890 + # ---------------------------------
891 + if (isset($this->item['xhtml']['body'])) :
892 + $content = $this->item['xhtml']['body'];
893 + elseif (isset($this->item['xhtml']['div'])) :
894 + $content = $this->item['xhtml']['div'];
895 + elseif (isset($this->item['content']['encoded']) and $this->item['content']['encoded']):
896 + $content = $this->item['content']['encoded'];
897 + else:
898 + $content = $this->item['description'];
899 + endif;
900 + $this->post['post_content'] = apply_filters('syndicated_item_content', $content, $this);
901 +
902 + # Identify and sanitize excerpt
903 + $excerpt = NULL;
904 + if ( isset($this->item['description']) and $this->item['description'] ) :
905 + $excerpt = $this->item['description'];
906 + elseif ( isset($content) and $content ) :
907 + $excerpt = strip_tags($content);
908 + if (strlen($excerpt) > 255) :
909 + $excerpt = substr($excerpt,0,252).'...';
910 + endif;
911 + endif;
912 + $excerpt = apply_filters('syndicated_item_excerpt', $excerpt, $this);
913 +
914 + if (!is_null($excerpt)):
915 + $this->post['post_excerpt'] = $excerpt;
916 + endif;
917 +
918 + // This is unnecessary if we use wp_insert_post
919 + if (!$this->use_api('wp_insert_post')) :
920 + $this->post['post_name'] = sanitize_title($this->post['post_title']);
921 + endif;
922 +
923 + $this->post['epoch']['issued'] = apply_filters('syndicated_item_published', $this->published(), $this);
924 + $this->post['epoch']['created'] = apply_filters('syndicated_item_created', $this->created(), $this);
925 + $this->post['epoch']['modified'] = apply_filters('syndicated_item_updated', $this->updated(), $this);
926 +
927 + // Dealing with timestamps in WordPress is so fucking fucked.
928 + $offset = (int) get_option('gmt_offset') * 60 * 60;
929 + $this->post['post_date'] = gmdate('Y-m-d H:i:s', $this->published() + $offset);
930 + $this->post['post_modified'] = gmdate('Y-m-d H:i:s', $this->updated() + $offset);
931 + $this->post['post_date_gmt'] = gmdate('Y-m-d H:i:s', $this->published());
932 + $this->post['post_modified_gmt'] = gmdate('Y-m-d H:i:s', $this->updated());
933 +
934 + // Use feed-level preferences or the global default.
935 + $this->post['post_status'] = $this->link->syndicated_status('post', 'publish');
936 + $this->post['comment_status'] = $this->link->syndicated_status('comment', 'closed');
937 + $this->post['ping_status'] = $this->link->syndicated_status('ping', 'closed');
938 +
939 + // Unique ID (hopefully a unique tag: URI); failing that, the permalink
940 + $this->post['guid'] = apply_filters('syndicated_item_guid', $this->guid(), $this);
941 +
942 + // RSS 2.0 / Atom 1.0 enclosure support
943 + if ( isset($this->item['enclosure#']) ) :
944 + for ($i = 1; $i <= $this->item['enclosure#']; $i++) :
945 + $eid = (($i > 1) ? "#{$id}" : "");
946 + $this->post['meta']['enclosure'][] =
947 + apply_filters('syndicated_item_enclosure_url', $this->item["enclosure{$eid}@url"], $this)."\n".
948 + apply_filters('syndicated_item_enclosure_length', $this->item["enclosure{$eid}@length"], $this)."\n".
949 + apply_filters('syndicated_item_enclosure_type', $this->item["enclosure{$eid}@type"], $this);
950 + endfor;
951 + endif;
952 +
953 + // In case you want to point back to the blog this was syndicated from
954 + if (isset($this->feed->channel['title'])) :
955 + $this->post['meta']['syndication_source'] = apply_filters('syndicated_item_source_title', $this->feed->channel['title'], $this);
956 + endif;
957 + if (isset($this->feed->channel['link'])) :
958 + $this->post['meta']['syndication_source_uri'] = apply_filters('syndicated_item_source_link', $this->feed->channel['link'], $this);
959 + endif;
960 +
961 + // Store information on human-readable and machine-readable comment URIs
962 + if (isset($this->item['comments'])) :
963 + $this->post['meta']['rss:comments'] = apply_filters('syndicated_item_comments', $this->item['comments']);
964 + endif;
965 + if (isset($this->item['wfw']['commentrss'])) :
966 + $this->post['meta']['wfw:commentRSS'] = apply_filters('syndicated_item_commentrss', $this->item['wfw']['commentrss']);
967 + endif;
968 +
969 + // Store information to identify the feed that this came from
970 + $this->post['meta']['syndication_feed'] = $this->feedmeta['link/uri'];
971 + $this->post['meta']['syndication_feed_id'] = $this->feedmeta['link/id'];
972 +
973 + // In case you want to know the external permalink...
974 + $this->post['meta']['syndication_permalink'] = apply_filters('syndicated_item_link', $this->item['link']);
975 +
976 + // Store a hash of the post content for checking whether something needs to be updated
977 + $this->post['meta']['syndication_item_hash'] = $this->update_hash();
978 +
979 + // Feed-by-feed options for author and category creation
980 + $this->post['named']['unfamiliar']['author'] = $this->feedmeta['unfamiliar author'];
981 + $this->post['named']['unfamiliar']['category'] = $this->feedmeta['unfamiliar category'];
982 +
983 + // Categories: start with default categories, if any
984 + $fc = get_option("feedwordpress_syndication_cats");
985 + if ($fc) :
986 + $this->post['named']['preset/category'] = explode("\n", $fc);
987 + else :
988 + $this->post['named']['preset/category'] = array();
989 + endif;
990 +
991 + if (is_array($this->feedmeta['cats'])) :
992 + $this->post['named']['preset/category'] = array_merge($this->post['named']['preset/category'], $this->feedmeta['cats']);
993 + endif;
994 +
995 + // Now add categories from the post, if we have 'em
996 + $this->post['named']['category'] = array();
997 + if ( isset($this->item['category#']) ) :
998 + for ($i = 1; $i <= $this->item['category#']; $i++) :
999 + $cat_idx = (($i > 1) ? "#{$i}" : "");
1000 + $cat = $this->item["category{$cat_idx}"];
1001 +
1002 + if ( isset($this->feedmeta['cat_split']) and strlen($this->feedmeta['cat_split']) > 0) :
1003 + $pcre = "\007".$this->feedmeta['cat_split']."\007";
1004 + $this->post['named']['category'] = array_merge($this->post['named']['category'], preg_split($pcre, $cat, -1 /*=no limit*/, PREG_SPLIT_NO_EMPTY));
1005 + else :
1006 + $this->post['named']['category'][] = $cat;
1007 + endif;
1008 + endfor;
1009 + endif;
1010 + $this->post['named']['category'] = apply_filters('syndicated_item_categories', $this->post['named']['category'], $this);
1011 +
1012 + // Tags: start with default tags, if any
1013 + $ft = get_option("feedwordpress_syndication_tags");
1014 + if ($ft) :
1015 + $this->post['tags_input'] = explode('FEEDWORDPRESS_CAT_SEPARATOR', $ft);
1016 + else :
1017 + $this->post['tags_input'] = array();
1018 + endif;
1019 +
1020 + if (is_array($this->feedmeta['tags'])) :
1021 + $this->post['tags_input'] = array_merge($this->post['tags_input'], $this->feedmeta['tags']);
1022 + endif;
1023 +
1148 1024 endif;
1025 + } // SyndicatedPost::SyndicatedPost()
1026 +
1027 + function filtered () {
1028 + return is_null($this->post);
1149 1029 }
1150 -} // class FeedWordPress
1151 1030
1152 -require_once(dirname(__FILE__) . '/syndicatedpost.class.php');
1153 -require_once(dirname(__FILE__) . '/syndicatedlink.class.php');
1031 + function freshness () {
1032 + global $wpdb;
1154 1033
1034 + if ($this->filtered()) : // This should never happen.
1035 + FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
1036 + endif;
1037 +
1038 + if (is_null($this->_freshness)) :
1039 + $guid = $wpdb->escape($this->guid());
1040 +
1041 + $result = $wpdb->get_row("
1042 + SELECT id, guid, post_modified_gmt
1043 + FROM $wpdb->posts WHERE guid='$guid'
1044 + ");
1045 +
1046 + if (!$result) :
1047 + $this->_freshness = 2; // New content
1048 + else:
1049 + $stored_update_hashes = get_post_custom_values('syndication_item_hash', $result->id);
1050 + if (count($stored_update_hashes) > 0) :
1051 + $stored_update_hash = $stored_update_hashes[0];
1052 + $update_hash_changed = ($stored_update_hash != $this->update_hash());
1053 + else :
1054 + $update_hash_changed = false;
1055 + endif;
1056 +
1057 + preg_match('/([0-9]+)-([0-9]+)-([0-9]+) ([0-9]+):([0-9]+):([0-9]+)/', $result->post_modified_gmt, $backref);
1058 +
1059 + $last_rev_ts = gmmktime($backref[4], $backref[5], $backref[6], $backref[2], $backref[3], $backref[1]);
1060 + $updated_ts = $this->updated(/*fallback=*/ true, /*default=*/ NULL);
1061 + $updated = ((
1062 + !is_null($updated_ts)
1063 + and ($updated_ts > $last_rev_ts)
1064 + ) or $update_hash_changed);
1065 +
1066 + if ($updated) :
1067 + $this->_freshness = 1; // Updated content
1068 + $this->_wp_id = $result->id;
1069 + else :
1070 + $this->_freshness = 0; // Same old, same old
1071 + $this->_wp_id = $result->id;
1072 + endif;
1073 + endif;
1074 + endif;
1075 + return $this->_freshness;
1076 + }
1077 +
1078 + function wp_id () {
1079 + if ($this->filtered()) : // This should never happen.
1080 + FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
1081 + endif;
1082 +
1083 + if (is_null($this->_wp_id) and is_null($this->_freshness)) :
1084 + $fresh = $this->freshness(); // sets WP DB id in the process
1085 + endif;
1086 + return $this->_wp_id;
1087 + }
1088 +
1089 + function store () {
1090 + global $wpdb;
1091 +
1092 + if ($this->filtered()) : // This should never happen.
1093 + FeedWordPress::critical_bug('SyndicatedPost', $this, __LINE__);
1094 + endif;
1095 +
1096 + $freshness = $this->freshness();
1097 + if ($freshness > 0) :
1098 + # -- Look up, or create, numeric ID for author
1099 + $this->post['post_author'] = $this->author_id (
1100 + FeedWordPress::on_unfamiliar('author', $this->post['named']['unfamiliar']['author'])
1101 + );
1102 +
1103 + if (is_null($this->post['post_author'])) :
1104 + $this->post = NULL;
1105 + endif;
1106 + endif;
1107 +
1108 + if (!$this->filtered() and $freshness > 0) :
1109 + # -- Look up, or create, numeric ID for categories
1110 + list($pcats, $ptags) = $this->category_ids (
1111 + $this->post['named']['category'],
1112 + FeedWordPress::on_unfamiliar('category', $this->post['named']['unfamiliar']['category']),
1113 + /*tags_too=*/ true
1114 + );
1115 +
1116 + $this->post['post_category'] = $pcats;
1117 + $this->post['tags_input'] = array_merge($this->post['tags_input'], $ptags);
1118 +
1119 + if (is_null($this->post['post_category'])) :
1120 + // filter mode on, no matching categories; drop the post
1121 + $this->post = NULL;
1122 + else :
1123 + // filter mode off or at least one match; now add on the feed and global presets
1124 + $this->post['post_category'] = array_merge (
1125 + $this->post['post_category'],
1126 + $this->category_ids (
1127 + $this->post['named']['preset/category'],
1128 + 'default'
1129 + )
1130 + );
1131 +
1132 + if (count($this->post['post_category']) < 1) :
1133 + $this->post['post_category'][] = 1; // Default to category 1 ("Uncategorized" / "General") if nothing else
1134 + endif;
1135 + endif;
1136 + endif;
1137 +
1138 + if (!$this->filtered() and $freshness > 0) :
1139 + unset($this->post['named']);
1140 + $this->post = apply_filters('syndicated_post', $this->post, $this);
1141 + endif;
1142 +
1143 + if (!$this->filtered() and $freshness == 2) :
1144 + // The item has not yet been added. So let's add it.
1145 + $this->insert_new();
1146 + $this->add_rss_meta();
1147 + do_action('post_syndicated_item', $this->wp_id());
1148 +
1149 + $ret = 'new';
1150 + elseif (!$this->filtered() and $freshness == 1) :
1151 + $this->post['ID'] = $this->wp_id();
1152 + $this->update_existing();
1153 + $this->add_rss_meta();
1154 + do_action('update_syndicated_item', $this->wp_id());
1155 +
1156 + $ret = 'updated';
1157 + else :
1158 + $ret = false;
1159 + endif;
1160 +
1161 + return $ret;
1162 + } // function SyndicatedPost::store ()
1163 +
1164 + function insert_new () {
1165 + global $wpdb, $wp_db_version;
1166 +
1167 + // Why the fuck doesn't wp_insert_post already do this?
1168 + foreach ($this->post as $key => $value) :
1169 + if (is_string($value)) :
1170 + $dbpost[$key] = $wpdb->escape($value);
1171 + else :
1172 + $dbpost[$key] = $value;
1173 + endif;
1174 + endforeach;
1175 +
1176 + if (strlen($dbpost['post_title'].$dbpost['post_content'].$dbpost['post_excerpt']) == 0) :
1177 + // FIXME: Option for filtering out empty posts
1178 + endif;
1179 + if (strlen($dbpost['post_title'])==0) :
1180 + $dbpost['post_title'] = $this->post['meta']['syndication_source']
1181 + .' '.gmdate('Y-m-d H:i:s', $this->published() + $offset);
1182 + // FIXME: Option for what to fill a blank title with...
1183 + endif;
1184 +
1185 + if ($this->use_api('wp_insert_post')) :
1186 + $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
1187 +
1188 + // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
1189 + add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1190 +
1191 + // Kludge to prevent kses filters from stripping the
1192 + // content of posts when updating without a logged in
1193 + // user who has `unfiltered_html` capability.
1194 + add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
1195 +
1196 + $this->_wp_id = wp_insert_post($dbpost);
1197 +
1198 + // Turn off ridiculous fucking kludges #1 and #2
1199 + remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1200 + remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
1201 +
1202 + // This should never happen.
1203 + if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
1204 + FeedWordPress::critical_bug('SyndicatedPost (_wp_id problem)', array("dbpost" => $dbpost, "this" => $this), __LINE__);
1205 + endif;
1206 +
1207 + // Unfortunately, as of WordPress 2.3, wp_insert_post()
1208 + // *still* offers no way to use a guid of your choice,
1209 + // and munges your post modified timestamp, too.
1210 + $result = $wpdb->query("
1211 + UPDATE $wpdb->posts
1212 + SET
1213 + guid='{$dbpost['guid']}',
1214 + post_modified='{$dbpost['post_modified']}',
1215 + post_modified_gmt='{$dbpost['post_modified_gmt']}'
1216 + WHERE ID='{$this->_wp_id}'
1217 + ");
1218 + else :
1219 + # The right way to do this is the above. But, alas,
1220 + # in earlier versions of WordPress, wp_insert_post has
1221 + # too much behavior (mainly related to pings) that can't
1222 + # be overridden. In WordPress 1.5, it's enough of a
1223 + # resource hog to make PHP segfault after inserting
1224 + # 50-100 posts. This can get pretty annoying, especially
1225 + # if you are trying to update your feeds for the first
1226 + # time.
1227 +
1228 + $result = $wpdb->query("
1229 + INSERT INTO $wpdb->posts
1230 + SET
1231 + guid = '{$dbpost['guid']}',
1232 + post_author = '{$dbpost['post_author']}',
1233 + post_date = '{$dbpost['post_date']}',
1234 + post_date_gmt = '{$dbpost['post_date_gmt']}',
1235 + post_content = '{$dbpost['post_content']}',"
1236 + .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
1237 + post_title = '{$dbpost['post_title']}',
1238 + post_name = '{$dbpost['post_name']}',
1239 + post_modified = '{$dbpost['post_modified']}',
1240 + post_modified_gmt = '{$dbpost['post_modified_gmt']}',
1241 + comment_status = '{$dbpost['comment_status']}',
1242 + ping_status = '{$dbpost['ping_status']}',
1243 + post_status = '{$dbpost['post_status']}'
1244 + ");
1245 + $this->_wp_id = $wpdb->insert_id;
1246 +
1247 + // This should never happen.
1248 + if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
1249 + FeedWordPress::critical_bug('SyndicatedPost (_wp_id problem)', array("dbpost" => $dbpost, "this" => $this), __LINE__);
1250 + endif;
1251 +
1252 + // WordPress 1.5.x - 2.0.x
1253 + wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
1254 +
1255 + // Since we are not going through official channels, we need to
1256 + // manually tell WordPress that we've published a new post.
1257 + // We need to make sure to do this in order for FeedWordPress
1258 + // to play well with the staticize-reloaded plugin (something
1259 + // that a large aggregator website is going to *want* to be
1260 + // able to use).
1261 + do_action('publish_post', $this->_wp_id);
1262 + endif;
1263 + } /* SyndicatedPost::insert_new() */
1264 +
1265 + function update_existing () {
1266 + global $wpdb;
1267 +
1268 + // Why the fuck doesn't wp_insert_post already do this?
1269 + $dbpost = array();
1270 + foreach ($this->post as $key => $value) :
1271 + if (is_string($value)) :
1272 + $dbpost[$key] = $wpdb->escape($value);
1273 + else :
1274 + $dbpost[$key] = $value;
1275 + endif;
1276 + endforeach;
1277 +
1278 + if ($this->use_api('wp_insert_post')) :
1279 + $dbpost['post_pingback'] = false; // Tell WP 2.1 and 2.2 not to process for pingbacks
1280 +
1281 + // This is a ridiculous fucking kludge necessitated by WordPress 2.6 munging authorship meta-data
1282 + add_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1283 +
1284 + // Kludge to prevent kses filters from stripping the
1285 + // content of posts when updating without a logged in
1286 + // user who has `unfiltered_html` capability.
1287 + add_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
1288 +
1289 + $this->_wp_id = wp_insert_post($dbpost);
1290 +
1291 + // Turn off ridiculous fucking kludges #1 and #2
1292 + remove_action('_wp_put_post_revision', array($this, 'fix_revision_meta'));
1293 + remove_filter('content_save_pre', array($this, 'avoid_kses_munge'), 11);
1294 +
1295 + // This should never happen.
1296 + if (!is_numeric($this->_wp_id) or ($this->_wp_id == 0)) :
1297 + FeedWordPress::critical_bug('SyndicatedPost (_wp_id problem)', array("dbpost" => $dbpost, "this" => $this), __LINE__);
1298 + endif;
1299 +
1300 + // Unfortunately, as of WordPress 2.3, wp_insert_post()
1301 + // munges your post modified timestamp.
1302 + $result = $wpdb->query("
1303 + UPDATE $wpdb->posts
1304 + SET
1305 + post_modified='{$dbpost['post_modified']}',
1306 + post_modified_gmt='{$dbpost['post_modified_gmt']}'
1307 + WHERE ID='{$this->_wp_id}'
1308 + ");
1309 + else :
1310 +
1311 + $result = $wpdb->query("
1312 + UPDATE $wpdb->posts
1313 + SET
1314 + post_author = '{$dbpost['post_author']}',
1315 + post_content = '{$dbpost['post_content']}',"
1316 + .(isset($dbpost['post_excerpt']) ? "post_excerpt = '{$dbpost['post_excerpt']}'," : "")."
1317 + post_title = '{$dbpost['post_title']}',
1318 + post_name = '{$dbpost['post_name']}',
1319 + post_modified = '{$dbpost['post_modified']}',
1320 + post_modified_gmt = '{$dbpost['post_modified_gmt']}'
1321 + WHERE guid='{$dbpost['guid']}'
1322 + ");
1323 +
1324 + // WordPress 2.1.x and up
1325 + if (function_exists('wp_set_post_categories')) :
1326 + wp_set_post_categories($this->wp_id(), $this->post['post_category']);
1327 + // WordPress 1.5.x - 2.0.x
1328 + elseif (function_exists('wp_set_post_cats')) :
1329 + wp_set_post_cats('1', $this->wp_id(), $this->post['post_category']);
1330 + // This should never happen.
1331 + else :
1332 + FeedWordPress::critical_bug('SyndicatedPost (_wp_id problem)', array("dbpost" => $dbpost, "this" => $this), __LINE__);
1333 + endif;
1334 +
1335 + // Since we are not going through official channels, we need to
1336 + // manually tell WordPress that we've published a new post.
1337 + // We need to make sure to do this in order for FeedWordPress
1338 + // to play well with the staticize-reloaded plugin (something
1339 + // that a large aggregator website is going to *want* to be
1340 + // able to use).
1341 + do_action('edit_post', $this->post['ID']);
1342 + endif;
1343 + } /* SyndicatedPost::update_existing() */
1344 +
1345 + /**
1346 + * SyndicatedPost::fix_revision_meta() - Fixes the way WP 2.6+ fucks up
1347 + * meta-data (authorship, etc.) when storing revisions of an updated
1348 + * syndicated post.
1349 + *
1350 + * In their infinite wisdom, the WordPress coders have made it completely
1351 + * impossible for a plugin that uses wp_insert_post() to set certain
1352 + * meta-data (such as the author) when you store an old revision of an
1353 + * updated post. Instead, it uses the WordPress defaults (= currently
1354 + * active user ID if the process is running with a user logged in, or
1355 + * = #0 if there is no user logged in). This results in bogus authorship
1356 + * data for revisions that are syndicated from off the feed, unless we
1357 + * use a ridiculous kludge like this to end-run the munging of meta-data
1358 + * by _wp_put_post_revision.
1359 + *
1360 + * @param int $revision_id The revision ID to fix up meta-data
1361 + */
1362 + function fix_revision_meta ($revision_id) {
1363 + global $wpdb;
1364 +
1365 + $post_author = (int) $this->post['post_author'];
1366 +
1367 + $revision_id = (int) $revision_id;
1368 + $wpdb->query("
1369 + UPDATE $wpdb->posts
1370 + SET post_author={$this->post['post_author']}
1371 + WHERE post_type = 'revision' AND ID='$revision_id'
1372 + ");
1373 + } /* SyndicatedPost::fix_revision_meta () */
1374 +
1375 + /**
1376 + * SyndicatedPost::avoid_kses_munge() -- If FeedWordPress is processing
1377 + * an automatic update, that generally means that wp_insert_post() is
1378 + * being called under the user credentials of whoever is viewing the
1379 + * blog at the time -- usually meaning no user at all. But if WordPress
1380 + * gets a wp_insert_post() when current_user_can('unfiltered_html') is
1381 + * false, it will run the content of the post through a kses function
1382 + * that strips out lots of HTML tags -- notably <object> and some others.
1383 + * This causes problems for syndicating (for example) feeds that contain
1384 + * YouTube videos. It also produces an unexpected asymmetry between
1385 + * automatically-initiated updates and updates initiated manually from
1386 + * the WordPress Dashboard (which are usually initiated under the
1387 + * credentials of a logged-in admin, and so don't get run through the
1388 + * kses function). So, to avoid the whole mess, what we do here is
1389 + * just forcibly disable the kses munging for a single syndicated post,
1390 + * by restoring the contents of the `post_content` field.
1391 + *
1392 + * @param string $content The content of the post, after other filters have gotten to it
1393 + * @return string The original content of the post, before other filters had a chance to munge it.
1394 + */
1395 + function avoid_kses_munge ($content) {
1396 + global $wpdb;
1397 + return $wpdb->escape($this->post['post_content']);
1398 + }
1399 +
1400 + // SyndicatedPost::add_rss_meta: adds interesting meta-data to each entry
1401 + // using the space for custom keys. The set of keys and values to add is
1402 + // specified by the keys and values of $post['meta']. This is used to
1403 + // store anything that the WordPress user might want to access from a
1404 + // template concerning the post's original source that isn't provided
1405 + // for by standard WP meta-data (i.e., any interesting data about the
1406 + // syndicated post other than author, title, timestamp, categories, and
1407 + // guid). It's also used to hook into WordPress's support for
1408 + // enclosures.
1409 + function add_rss_meta () {
1410 + global $wpdb;
1411 + if ( is_array($this->post) and isset($this->post['meta']) and is_array($this->post['meta']) ) :
1412 + $postId = $this->wp_id();
1413 +
1414 + // Aggregated posts should NOT send out pingbacks.
1415 + // WordPress 2.1-2.2 claim you can tell them not to
1416 + // using $post_pingback, but they don't listen, so we
1417 + // make sure here.
1418 + $result = $wpdb->query("
1419 + DELETE FROM $wpdb->postmeta
1420 + WHERE post_id='$postId' AND meta_key='_pingme'
1421 + ");
1422 +
1423 + foreach ( $this->post['meta'] as $key => $values ) :
1424 +
1425 + $key = $wpdb->escape($key);
1426 +
1427 + // If this is an update, clear out the old
1428 + // values to avoid duplication.
1429 + $result = $wpdb->query("
1430 + DELETE FROM $wpdb->postmeta
1431 + WHERE post_id='$postId' AND meta_key='$key'
1432 + ");
1433 +
1434 + // Allow for either a single value or an array
1435 + if (!is_array($values)) $values = array($values);
1436 + foreach ( $values as $value ) :
1437 + $value = $wpdb->escape($value);
1438 + $result = $wpdb->query("
1439 + INSERT INTO $wpdb->postmeta
1440 + SET
1441 + post_id='$postId',
1442 + meta_key='$key',
1443 + meta_value='$value'
1444 + ");
1445 + endforeach;
1446 + endforeach;
1447 + endif;
1448 + } /* SyndicatedPost::add_rss_meta () */
1449 +
1450 + // SyndicatedPost::author_id (): get the ID for an author name from
1451 + // the feed. Create the author if necessary.
1452 + function author_id ($unfamiliar_author = 'create') {
1453 + global $wpdb, $wp_db_version; // test for WordPress 2.0 database schema
1454 +
1455 + $a = $this->author();
1456 + $author = $a['name'];
1457 + $email = $a['email'];
1458 + $url = $a['uri'];
1459 +
1460 + // Never can be too careful...
1461 + $login = sanitize_user($author, /*strict=*/ true);
1462 + $login = apply_filters('pre_user_login', $login);
1463 +
1464 + $nice_author = sanitize_title($author);
1465 + $nice_author = apply_filters('pre_user_nicename', $nice_author);
1466 +
1467 + $reg_author = $wpdb->escape(preg_quote($author));
1468 + $author = $wpdb->escape($author);
1469 + $email = $wpdb->escape($email);
1470 + $url = $wpdb->escape($url);
1471 +
1472 + // Check for an existing author rule....
1473 + if (isset($this->link->settings['map authors']['name'][strtolower(trim($author))])) :
1474 + $author_rule = $this->link->settings['map authors']['name'][strtolower(trim($author))];
1475 + else :
1476 + $author_rule = NULL;
1477 + endif;
1478 +
1479 + // User name is mapped to a particular author. If that author ID exists, use it.
1480 + if (is_numeric($author_rule) and get_userdata((int) $author_rule)) :
1481 + $id = (int) $author_rule;
1482 +
1483 + // User name is filtered out
1484 + elseif ('filter' == $author_rule) :
1485 + $id = NULL;
1486 +
1487 + else :
1488 + // Check the database for an existing author record that might fit
1489 +
1490 + #-- WordPress 1.5.x
1491 + if (!isset($wp_db_version)) :
1492 + $id = $wpdb->get_var(
1493 + "SELECT ID from $wpdb->users
1494 + WHERE
1495 + TRIM(LCASE(user_login)) = TRIM(LCASE('$login')) OR
1496 + (
1497 + LENGTH(TRIM(LCASE(user_email))) > 0
1498 + AND TRIM(LCASE(user_email)) = TRIM(LCASE('$email'))
1499 + ) OR
1500 + TRIM(LCASE(user_firstname)) = TRIM(LCASE('$author')) OR
1501 + TRIM(LCASE(user_nickname)) = TRIM(LCASE('$author')) OR
1502 + TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author')) OR
1503 + TRIM(LCASE(user_description)) = TRIM(LCASE('$author')) OR
1504 + (
1505 + LOWER(user_description)
1506 + RLIKE CONCAT(
1507 + '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
1508 + LCASE('$reg_author'),
1509 + '( |\\t|\\r)*(\\n|\$)'
1510 + )
1511 + )
1512 + ");
1513 +
1514 + #-- WordPress 2.0+
1515 + elseif ($wp_db_version >= 2966) :
1516 +
1517 + // First try the user core data table.
1518 + $id = $wpdb->get_var(
1519 + "SELECT ID FROM $wpdb->users
1520 + WHERE
1521 + TRIM(LCASE(user_login)) = TRIM(LCASE('$login'))
1522 + OR (
1523 + LENGTH(TRIM(LCASE(user_email))) > 0
1524 + AND TRIM(LCASE(user_email)) = TRIM(LCASE('$email'))
1525 + )
1526 + OR TRIM(LCASE(user_nicename)) = TRIM(LCASE('$nice_author'))
1527 + ");
1528 +
1529 + // If that fails, look for aliases in the user meta data table
1530 + if (is_null($id)) :
1531 + $id = $wpdb->get_var(
1532 + "SELECT user_id FROM $wpdb->usermeta
1533 + WHERE
1534 + (meta_key = 'description' AND TRIM(LCASE(meta_value)) = TRIM(LCASE('$author')))
1535 + OR (
1536 + meta_key = 'description'
1537 + AND TRIM(LCASE(meta_value))
1538 + RLIKE CONCAT(
1539 + '(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*',
1540 + TRIM(LCASE('$reg_author')),
1541 + '( |\\t|\\r)*(\\n|\$)'
1542 + )
1543 + )
1544 + ");
1545 + endif;
1546 + endif;
1547 +
1548 + // ... if you don't find one, then do what you need to do
1549 + if (is_null($id)) :
1550 + if ($unfamiliar_author === 'create') :
1551 + $userdata = array();
1552 +
1553 + #-- user table data
1554 + $userdata['ID'] = NULL; // new user
1555 + $userdata['user_login'] = $login;
1556 + $userdata['user_nicename'] = $nice_author;
1557 + $userdata['user_pass'] = substr(md5(uniqid(microtime())), 0, 6); // just something random to lock it up
1558 + $userdata['user_email'] = $email;
1559 + $userdata['user_url'] = $url;
1560 + $userdata['display_name'] = $author;
1561 +
1562 + $id = wp_insert_user($userdata);
1563 + elseif (is_numeric($unfamiliar_author) and get_userdata((int) $unfamiliar_author)) :
1564 + $id = (int) $unfamiliar_author;
1565 + elseif ($unfamiliar_author === 'default') :
1566 + $id = 1;
1567 + endif;
1568 + endif;
1569 + endif;
1570 +
1571 + if ($id) :
1572 + $this->link->settings['map authors']['name'][strtolower(trim($author))] = $id;
1573 + endif;
1574 + return $id;
1575 + } // function SyndicatedPost::author_id ()
1576 +
1577 + // look up (and create) category ids from a list of categories
1578 + function category_ids ($cats, $unfamiliar_category = 'create', $tags_too = false) {
1579 + global $wpdb;
1580 +
1581 + // We need to normalize whitespace because (1) trailing
1582 + // whitespace can cause PHP and MySQL not to see eye to eye on
1583 + // VARCHAR comparisons for some versions of MySQL (cf.
1584 + // <http://dev.mysql.com/doc/mysql/en/char.html>), and (2)
1585 + // because I doubt most people want to make a semantic
1586 + // distinction between 'Computers' and 'Computers '
1587 + $cats = array_map('trim', $cats);
1588 +
1589 + $tags = array();
1590 +
1591 + $cat_ids = array ();
1592 + foreach ($cats as $cat_name) :
1593 + if (preg_match('/^{#([0-9]+)}$/', $cat_name, $backref)) :
1594 + $cat_id = (int) $backref[1];
1595 + if (function_exists('is_term') and is_term($cat_id, 'category')) :
1596 + $cat_ids[] = $cat_id;
1597 + elseif (get_category($cat_id)) :
1598 + $cat_ids[] = $cat_id;
1599 + endif;
1600 + else :
1601 + $esc = $wpdb->escape($cat_name);
1602 + $resc = $wpdb->escape(preg_quote($cat_name));
1603 +
1604 + // WordPress 2.3+
1605 + if (function_exists('is_term')) :
1606 + $cat_id = is_term($cat_name, 'category');
1607 + if ($cat_id) :
1608 + $cat_ids[] = $cat_id['term_id'];
1609 + // There must be a better way to do this...
1610 + elseif ($results = $wpdb->get_results(
1611 + "SELECT term_id
1612 + FROM $wpdb->term_taxonomy
1613 + WHERE
1614 + LOWER(description) RLIKE
1615 + CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)')"
1616 + )) :
1617 + foreach ($results AS $term) :
1618 + $cat_ids[] = (int) $term->term_id;
1619 + endforeach;
1620 + elseif ('tag'==$unfamiliar_category) :
1621 + $tags[] = $cat_name;
1622 + elseif ('create'===$unfamiliar_category) :
1623 + $term = wp_insert_term($cat_name, 'category');
1624 + $cat_ids[] = $term['term_id'];
1625 + endif;
1626 +
1627 + // WordPress 1.5.x - 2.2.x
1628 + else :
1629 + $results = $wpdb->get_results(
1630 + "SELECT cat_ID
1631 + FROM $wpdb->categories
1632 + WHERE
1633 + (LOWER(cat_name) = LOWER('$esc'))
1634 + OR (LOWER(category_description)
1635 + RLIKE CONCAT('(^|\\n)a\\.?k\\.?a\\.?( |\\t)*:?( |\\t)*', LOWER('{$resc}'), '( |\\t|\\r)*(\\n|\$)'))
1636 + ");
1637 + if ($results) :
1638 + foreach ($results as $term) :
1639 + $cat_ids[] = (int) $term->cat_ID;
1640 + endforeach;
1641 + elseif ('create'===$unfamiliar_category) :
1642 + if (function_exists('wp_insert_category')) :
1643 + $cat_id = wp_insert_category(array('cat_name' => $cat_name));
1644 + // And into the database we go.
1645 + else :
1646 + $nice_kitty = sanitize_title($cat_name);
1647 + $wpdb->query(sprintf("
1648 + INSERT INTO $wpdb->categories
1649 + SET
1650 + cat_name='%s',
1651 + category_nicename='%s'
1652 + ", $wpdb->escape($cat_name), $nice_kitty
1653 + ));
1654 + $cat_id = $wpdb->insert_id;
1655 + endif;
1656 + $cat_ids[] = $cat_id;
1657 + endif;
1658 + endif;
1659 + endif;
1660 + endforeach;
1661 +
1662 + if ((count($cat_ids) == 0) and ($unfamiliar_category === 'filter')) :
1663 + $cat_ids = NULL; // Drop the post
1664 + else :
1665 + $cat_ids = array_unique($cat_ids);
1666 + endif;
1667 +
1668 + if ($tags_too) : $ret = array($cat_ids, $tags);
1669 + else : $ret = $cat_ids;
1670 + endif;
1671 +
1672 + return $ret;
1673 + } // function SyndicatedPost::category_ids ()
1674 +
1675 + function use_api ($tag) {
1676 + global $wp_db_version;
1677 + switch ($tag) :
1678 + case 'wp_insert_post':
1679 + // Before 2.2, wp_insert_post does too much of the wrong stuff to use it
1680 + // In 1.5 it was such a resource hog it would make PHP segfault on big updates
1681 + $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_21);
1682 + break;
1683 + case 'post_status_pending':
1684 + $ret = (isset($wp_db_version) and $wp_db_version > FWP_SCHEMA_23);
1685 + break;
1686 + endswitch;
1687 + return $ret;
1688 + } // function SyndicatedPost::use_api ()
1689 +
1690 + #### EXTRACT DATA FROM FEED ITEM ####
1691 +
1692 + function created () {
1693 + $epoch = null;
1694 + if (isset($this->item['dc']['created'])) :
1695 + $epoch = @parse_w3cdtf($this->item['dc']['created']);
1696 + elseif (isset($this->item['dcterms']['created'])) :
1697 + $epoch = @parse_w3cdtf($this->item['dcterms']['created']);
1698 + elseif (isset($this->item['created'])): // Atom 0.3
1699 + $epoch = @parse_w3cdtf($this->item['created']);
1700 + endif;
1701 + return $epoch;
1702 + }
1703 + function published ($fallback = true) {
1704 + $epoch = null;
1705 +
1706 + # RSS is a fucking mess. Figure out whether we have a date in
1707 + # <dc:date>, <issued>, <pubDate>, etc., and get it into Unix
1708 + # epoch format for reformatting. If we can't find anything,
1709 + # we'll use the last-updated time.
1710 + if (isset($this->item['dc']['date'])): // Dublin Core
1711 + $epoch = @parse_w3cdtf($this->item['dc']['date']);
1712 + elseif (isset($this->item['dcterms']['issued'])) : // Dublin Core extensions
1713 + $epoch = @parse_w3cdtf($this->item['dcterms']['issued']);
1714 + elseif (isset($this->item['published'])) : // Atom 1.0
1715 + $epoch = @parse_w3cdtf($this->item['published']);
1716 + elseif (isset($this->item['issued'])): // Atom 0.3
1717 + $epoch = @parse_w3cdtf($this->item['issued']);
1718 + elseif (isset($this->item['pubdate'])): // RSS 2.0
1719 + $epoch = strtotime($this->item['pubdate']);
1720 + elseif ($fallback) : // Fall back to <updated> / <modified> if present
1721 + $epoch = $this->updated(/*fallback=*/ false);
1722 + endif;
1723 +
1724 + # If everything failed, then default to the current time.
1725 + if (is_null($epoch)) :
1726 + if (-1 == $default) :
1727 + $epoch = time();
1728 + else :
1729 + $epoch = $default;
1730 + endif;
1731 + endif;
1732 +
1733 + return $epoch;
1734 + }
1735 + function updated ($fallback = true, $default = -1) {
1736 + $epoch = null;
1737 +
1738 + # As far as I know, only dcterms and Atom have reliable ways to
1739 + # specify when something was *modified* last. If neither is
1740 + # available, then we'll try to get the time of publication.
1741 + if (isset($this->item['dc']['modified'])) : // Not really correct
1742 + $epoch = @parse_w3cdtf($this->item['dc']['modified']);
1743 + elseif (isset($this->item['dcterms']['modified'])) : // Dublin Core extensions
1744 + $epoch = @parse_w3cdtf($this->item['dcterms']['modified']);
1745 + elseif (isset($this->item['modified'])): // Atom 0.3
1746 + $epoch = @parse_w3cdtf($this->item['modified']);
1747 + elseif (isset($this->item['updated'])): // Atom 1.0
1748 + $epoch = @parse_w3cdtf($this->item['updated']);
1749 + elseif ($fallback) : // Fall back to issued / dc:date
1750 + $epoch = $this->published(/*fallback=*/ false, /*default=*/ $default);
1751 + endif;
1752 +
1753 + # If everything failed, then default to the current time.
1754 + if (is_null($epoch)) :
1755 + if (-1 == $default) :
1756 + $epoch = time();
1757 + else :
1758 + $epoch = $default;
1759 + endif;
1760 + endif;
1761 +
1762 + return $epoch;
1763 + }
1764 +
1765 + function update_hash () {
1766 + return md5(serialize($this->item));
1767 + }
1768 +
1769 + function guid () {
1770 + $guid = null;
1771 + if (isset($this->item['id'])): // Atom 0.3 / 1.0
1772 + $guid = $this->item['id'];
1773 + elseif (isset($this->item['atom']['id'])) : // Namespaced Atom
1774 + $guid = $this->item['atom']['id'];
1775 + elseif (isset($this->item['guid'])) : // RSS 2.0
1776 + $guid = $this->item['guid'];
1777 + elseif (isset($this->item['dc']['identifier'])) :// yeah, right
1778 + $guid = $this->item['dc']['identifier'];
1779 + else :
1780 + // The feed does not seem to have provided us with a
1781 + // unique identifier, so we'll have to cobble together
1782 + // a tag: URI that might work for us. The base of the
1783 + // URI will be the host name of the feed source ...
1784 + $bits = parse_url($this->feedmeta['link/uri']);
1785 + $guid = 'tag:'.$bits['host'];
1786 +
1787 + // If we have a date of creation, then we can use that
1788 + // to uniquely identify the item. (On the other hand, if
1789 + // the feed producer was consicentious enough to
1790 + // generate dates of creation, she probably also was
1791 + // conscientious enough to generate unique identifiers.)
1792 + if (!is_null($this->created())) :
1793 + $guid .= '://post.'.date('YmdHis', $this->created());
1794 +
1795 + // Otherwise, use both the URI of the item, *and* the
1796 + // item's title. We have to use both because titles are
1797 + // often not unique, and sometimes links aren't unique
1798 + // either (e.g. Bitch (S)HITLIST, Mozilla Dot Org news,
1799 + // some podcasts). But it's rare to have *both* the same
1800 + // title *and* the same link for two different items. So
1801 + // this is about the best we can do.
1802 + else :
1803 + $guid .= '://'.md5($this->item['link'].'/'.$this->item['title']);
1804 + endif;
1805 + endif;
1806 + return $guid;
1807 + }
1808 +
1809 + function author () {
1810 + $author = array ();
1811 +
1812 + if (isset($this->item['author_name'])):
1813 + $author['name'] = $this->item['author_name'];
1814 + elseif (isset($this->item['dc']['creator'])):
1815 + $author['name'] = $this->item['dc']['creator'];
1816 + elseif (isset($this->item['dc']['contributor'])):
1817 + $author['name'] = $this->item['dc']['contributor'];
1818 + elseif (isset($this->feed->channel['dc']['creator'])) :
1819 + $author['name'] = $this->feed->channel['dc']['creator'];
1820 + elseif (isset($this->feed->channel['dc']['contributor'])) :
1821 + $author['name'] = $this->feed->channel['dc']['contributor'];
1822 + elseif (isset($this->feed->channel['author_name'])) :
1823 + $author['name'] = $this->feed->channel['author_name'];
1824 + elseif ($this->feed->is_rss() and isset($this->item['author'])) :
1825 + // The author element in RSS is allegedly an
1826 + // e-mail address, but lots of people don't use
1827 + // it that way. So let's make of it what we can.
1828 + $author = parse_email_with_realname($this->item['author']);
1829 +
1830 + if (!isset($author['name'])) :
1831 + if (isset($author['email'])) :
1832 + $author['name'] = $author['email'];
1833 + else :
1834 + $author['name'] = $this->feed->channel['title'];
1835 + endif;
1836 + endif;
1837 + else :
1838 + $author['name'] = $this->feed->channel['title'];
1839 + endif;
1840 +
1841 + if (isset($this->item['author_email'])):
1842 + $author['email'] = $this->item['author_email'];
1843 + elseif (isset($this->feed->channel['author_email'])) :
1844 + $author['email'] = $this->feed->channel['author_email'];
1845 + endif;
1846 +
1847 + if (isset($this->item['author_url'])):
1848 + $author['uri'] = $this->item['author_url'];
1849 + elseif (isset($this->feed->channel['author_url'])) :
1850 + $author['uri'] = $this->item['author_url'];
1851 + else:
1852 + $author['uri'] = $this->feed->channel['link'];
1853 + endif;
1854 +
1855 + return $author;
1856 + } // SyndicatedPost::author()
1857 +
1858 + var $strip_attrs = array (
1859 + array('[a-z]+', 'style'),
1860 + array('[a-z]+', 'target'),
1861 + );
1862 + function sanitize_content ($content, $obj) {
1863 + # FeedWordPress used to resolve URIs relative to the
1864 + # feed URI. It now relies on the xml:base support
1865 + # baked in to the MagpieRSS upgrade. So all we do here
1866 + # now is to sanitize problematic attributes.
1867 + #
1868 + # This kind of sucks. I intend to replace it with
1869 + # lib_filter sometime soon.
1870 + foreach ($obj->strip_attrs as $pair):
1871 + list($tag,$attr) = $pair;
1872 + $content = preg_replace (
1873 + ":(<$tag [^>]*)($attr=(\"[^\">]*\"|[^>\\s]+))([^>]*>):i",
1874 + "\\1\\4",
1875 + $content
1876 + );
1877 + endforeach;
1878 + return $content;
1879 + }
1880 +} // class SyndicatedPost
1881 +
1882 +# class SyndicatedLink: represents a syndication feed stored within the
1883 +# WordPress database
1884 +#
1885 +# To keep things compact and editable from within WordPress, we use all the
1886 +# links under a particular category in the WordPress "Blogroll" for the list of
1887 +# feeds to syndicate. "Contributors" is the category used by default; you can
1888 +# configure that under Options --> Syndication.
1889 +#
1890 +# Fields used are:
1891 +#
1892 +# * link_rss: the URI of the Atom/RSS feed to syndicate
1893 +#
1894 +# * link_notes: user-configurable options, with keys and values
1895 +# like so:
1896 +#
1897 +# key: value
1898 +# cats: computers\nweb
1899 +# feed/key: value
1900 +#
1901 +# Keys that start with "feed/" are gleaned from the data supplied
1902 +# by the feed itself, and will be overwritten with each update.
1903 +#
1904 +# Values have linebreak characters escaped with C-style
1905 +# backslashes (so, for example, a newline becomes "\n").
1906 +#
1907 +# The value of `cats` is used as a newline-separated list of
1908 +# default categories for any post coming from a particular feed.
1909 +# (In the example above, any posts from this feed will be placed
1910 +# in the "computers" and "web" categories--*in addition to* any
1911 +# categories that may already be applied to the posts.)
1912 +#
1913 +# Values of keys in link_notes are accessible from templates using
1914 +# the function `get_feed_meta($key)` if this plugin is activated.
1915 +
1916 +class SyndicatedLink {
1917 + var $id = null;
1918 + var $link = null;
1919 + var $settings = array ();
1920 + var $magpie = null;
1921 +
1922 + function SyndicatedLink ($link) {
1923 + global $wpdb;
1924 +
1925 + if (is_object($link)) :
1926 + $this->link = $link;
1927 + $this->id = $link->link_id;
1928 + else :
1929 + $this->id = $link;
1930 + if (function_exists('get_bookmark')) : // WP 2.1+
1931 + $this->link = get_bookmark($link);
1932 + else :
1933 + $this->link = $wpdb->get_row("
1934 + SELECT * FROM $wpdb->links
1935 + WHERE (link_id = '".$wpdb->escape($link)."')"
1936 + );
1937 + endif;
1938 + endif;
1939 +
1940 + if (strlen($this->link->link_rss) > 0) :
1941 + // Read off feed settings from link_notes
1942 + $notes = explode("\n", $this->link->link_notes);
1943 + foreach ($notes as $note):
1944 + list($key, $value) = explode(": ", $note, 2);
1945 +
1946 + if (strlen($key) > 0) :
1947 + // Unescape and trim() off the whitespace.
1948 + // Thanks to Ray Lischner for pointing out the
1949 + // need to trim off whitespace.
1950 + $this->settings[$key] = stripcslashes (trim($value));
1951 + endif;
1952 + endforeach;
1953 +
1954 + // "Magic" feed settings
1955 + $this->settings['link/uri'] = $this->link->link_rss;
1956 + $this->settings['link/name'] = $this->link->link_name;
1957 + $this->settings['link/id'] = $this->link->link_id;
1958 +
1959 + // `hardcode categories` and `unfamiliar categories` are deprecated in favor of `unfamiliar category`
1960 + if (
1961 + isset($this->settings['unfamiliar categories'])
1962 + and !isset($this->settings['unfamiliar category'])
1963 + ) :
1964 + $this->settings['unfamiliar category'] = $this->settings['unfamiliar categories'];
1965 + endif;
1966 + if (
1967 + FeedWordPress::affirmative($this->settings, 'hardcode categories')
1968 + and !isset($this->settings['unfamiliar category'])
1969 + ) :
1970 + $this->settings['unfamiliar category'] = 'default';
1971 + endif;
1972 +
1973 + // Set this up automagically for del.icio.us
1974 + $bits = parse_url($this->link->link_rss);
1975 + $tagspacers = array('del.icio.us', 'feeds.delicious.com');
1976 + if (!isset($this->settings['cat_split']) and in_array($bits['host'], $tagspacers)) :
1977 + $this->settings['cat_split'] = '\s'; // Whitespace separates multiple tags in del.icio.us RSS feeds
1978 + endif;
1979 +
1980 + if (isset($this->settings['cats'])):
1981 + $this->settings['cats'] = preg_split(FEEDWORDPRESS_CAT_SEPARATOR_PATTERN, $this->settings['cats']);
1982 + endif;
1983 + if (isset($this->settings['tags'])):
1984 + $this->settings['tags'] = preg_split(FEEDWORDPRESS_CAT_SEPARATOR_PATTERN, $this->settings['tags']);
1985 + endif;
1986 +
1987 + if (isset($this->settings['map authors'])) :
1988 + $author_rules = explode("\n\n", $this->settings['map authors']);
1989 + $ma = array();
1990 + foreach ($author_rules as $rule) :
1991 + list($rule_type, $author_name, $author_action) = explode("\n", $rule);
1992 +
1993 + // Normalize for case and whitespace
1994 + $rule_type = strtolower(trim($rule_type));
1995 + $author_name = strtolower(trim($author_name));
1996 + $author_action = strtolower(trim($author_action));
1997 +
1998 + $ma[$rule_type][$author_name] = $author_action;
1999 + endforeach;
2000 + $this->settings['map authors'] = $ma;
2001 + endif;
2002 + endif;
2003 + } // SyndicatedLink::SyndicatedLink ()
2004 +
2005 + function found () {
2006 + return is_object($this->link);
2007 + }
2008 +
2009 + function stale () {
2010 + $stale = true;
2011 + if (isset($this->settings['update/hold']) and ($this->settings['update/hold']=='ping')) :
2012 + $stale = false; // don't update on any timed updates; pings only
2013 + elseif (isset($this->settings['update/hold']) and ($this->settings['update/hold']=='next')) :
2014 + $stale = true; // update on the next timed update
2015 + elseif (!isset($this->settings['update/ttl']) or !isset($this->settings['update/last'])) :
2016 + $stale = true; // initial update
2017 + else :
2018 + $after = ((int) $this->settings['update/last'])
2019 + +((int) $this->settings['update/ttl'] * 60);
2020 + $stale = (time() >= $after);
2021 + endif;
2022 + return $stale;
2023 + }
2024 +
2025 + function poll () {
2026 + global $wpdb;
2027 +
2028 + $this->magpie = fetch_rss($this->link->link_rss);
2029 + $new_count = NULL;
2030 +
2031 + if (is_object($this->magpie)) :
2032 + $new_count = array('new' => 0, 'updated' => 0);
2033 +
2034 + # -- Update Link metadata live from feed
2035 + $channel = $this->magpie->channel;
2036 +
2037 + if (!isset($channel['id'])) :
2038 + $channel['id'] = $this->link->link_rss;
2039 + endif;
2040 +
2041 + $update = array();
2042 + if (!$this->hardcode('url') and isset($channel['link'])) :
2043 + $update[] = "link_url = '".$wpdb->escape($channel['link'])."'";
2044 + endif;
2045 +
2046 + if (!$this->hardcode('name') and isset($channel['title'])) :
2047 + $update[] = "link_name = '".$wpdb->escape($channel['title'])."'";
2048 + endif;
2049 +
2050 + if (!$this->hardcode('description')) :
2051 + if (isset($channel['tagline'])) :
2052 + $update[] = "link_description = '".$wpdb->escape($channel['tagline'])."'";
2053 + elseif (isset($channel['description'])) :
2054 + $update[] = "link_description = '".$wpdb->escape($channel['description'])."'";
2055 + endif;
2056 + endif;
2057 +
2058 + $this->settings = array_merge($this->settings, $this->flatten_array($channel));
2059 +
2060 + $this->settings['update/last'] = time(); $ttl = $this->ttl();
2061 + if (!is_null($ttl)) :
2062 + $this->settings['update/ttl'] = $ttl;
2063 + $this->settings['update/timed'] = 'feed';
2064 + else :
2065 + $this->settings['update/ttl'] = rand(30, 120); // spread over time interval for staggered updates
2066 + $this->settings['update/timed'] = 'automatically';
2067 + endif;
2068 +
2069 + if (!isset($this->settings['update/hold']) or $this->settings['update/hold']!='ping') :
2070 + $this->settings['update/hold'] = 'scheduled';
2071 + endif;
2072 +
2073 + // Copy back without a few things that we don't want to save in the notes
2074 + $to_notes = $this->settings;
2075 +
2076 + if (is_array($to_notes['cats'])) :
2077 + $to_notes['cats'] = implode(FEEDWORDPRESS_CAT_SEPARATOR, $to_notes['cats']);
2078 + endif;
2079 + if (is_array($to_notes['tags'])) :
2080 + $to_notes['tags'] = implode(FEEDWORDPRESS_CAT_SEPARATOR, $to_notes['tags']);
2081 + endif;
2082 +
2083 + if (isset($to_notes['map authors'])) :
2084 + $ma = array();
2085 + foreach ($to_notes['map authors'] as $rule_type => $author_rules) :
2086 + foreach ($author_rules as $author_name => $author_action) :
2087 + $ma[] = $rule_type."\n".$author_name."\n".$author_action;
2088 + endforeach;
2089 + endforeach;
2090 + $to_notes['map authors'] = implode("\n\n", $ma);
2091 + endif;
2092 +
2093 + unset($to_notes['link/id']); unset($to_notes['link/uri']);
2094 + unset($to_notes['link/name']);
2095 + unset($to_notes['hardcode categories']); // Deprecated
2096 + unset($to_notes['unfamiliar categories']); // Deprecated
2097 +
2098 + $notes = '';
2099 + foreach ($to_notes as $key => $value) :
2100 + $notes .= $key . ": ". addcslashes($value, "\0..\37".'\\') . "\n";
2101 + endforeach;
2102 + $update[] = "link_notes = '".$wpdb->escape($notes)."'";
2103 +
2104 + $update_set = implode(',', $update);
2105 +
2106 + // Update the properties of the link from the feed information
2107 + $result = $wpdb->query("
2108 + UPDATE $wpdb->links
2109 + SET $update_set
2110 + WHERE link_id='$this->id'
2111 + ");
2112 +
2113 + # -- Add new posts from feed and update any updated posts
2114 + if (is_array($this->magpie->items)) :
2115 + foreach ($this->magpie->items as $item) :
2116 + $post =& new SyndicatedPost($item, $this);
2117 + if (!$post->filtered()) :
2118 + $new = $post->store();
2119 + if ( $new !== false ) $new_count[$new]++;
2120 + endif;
2121 + endforeach;
2122 + endif;
2123 +
2124 + // Copy back any changes to feed settings made in the course of updating (e.g. new author rules)
2125 + $to_notes = $this->settings;
2126 +
2127 + if (is_array($to_notes['cats'])) :
2128 + $to_notes['cats'] = implode(FEEDWORDPRESS_CAT_SEPARATOR, $to_notes['cats']);
2129 + endif;
2130 + if (is_array($to_notes['tags'])) :
2131 + $to_notes['tags'] = implode(FEEDWORDPRESS_CAT_SEPARATOR, $to_notes['tags']);
2132 + endif;
2133 +
2134 + if (isset($to_notes['map authors'])) :
2135 + $ma = array();
2136 + foreach ($to_notes['map authors'] as $rule_type => $author_rules) :
2137 + foreach ($author_rules as $author_name => $author_action) :
2138 + $ma[] = $rule_type."\n".$author_name."\n".$author_action;
2139 + endforeach;
2140 + endforeach;
2141 + $to_notes['map authors'] = implode("\n\n", $ma);
2142 + endif;
2143 +
2144 + unset($to_notes['link/id']); unset($to_notes['link/uri']);
2145 + unset($to_notes['link/name']);
2146 + unset($to_notes['hardcode categories']); // Deprecated
2147 + unset($to_notes['unfamiliar categories']); // Deprecated
2148 +
2149 + $notes = '';
2150 + foreach ($to_notes as $key => $value) :
2151 + $notes .= $key . ": ". addcslashes($value, "\0..\37".'\\') . "\n";
2152 + endforeach;
2153 +
2154 + $update_set = "link_notes = '".$wpdb->escape($notes)."'";
2155 +
2156 + // Update the properties of the link from the feed information
2157 + $result = $wpdb->query("
2158 + UPDATE $wpdb->links
2159 + SET $update_set
2160 + WHERE link_id='$this->id'
2161 + ");
2162 + endif;
2163 + return $new_count;
2164 + } /* SyndicatedLink::poll() */
2165 +
2166 + function uri () {
2167 + return (is_object($this->link) ? $this->link->link_rss : NULL);
2168 + }
2169 + function homepage () {
2170 + return (isset($this->settings['feed/link']) ? $this->settings['feed/link'] : NULL);
2171 + }
2172 +
2173 + function ttl () {
2174 + if (is_object($this->magpie)) :
2175 + $channel = $this->magpie->channel;
2176 + else :
2177 + $channel = array();
2178 + endif;
2179 +
2180 + if (isset($channel['ttl'])) :
2181 + // "ttl stands for time to live. It's a number of
2182 + // minutes that indicates how long a channel can be
2183 + // cached before refreshing from the source."
2184 + // <http://blogs.law.harvard.edu/tech/rss#ltttlgtSubelementOfLtchannelgt>
2185 + $ret = $channel['ttl'];
2186 + elseif (isset($channel['sy']['updatefrequency']) or isset($channel['sy']['updateperiod'])) :
2187 + $period_minutes = array (
2188 + 'hourly' => 60, /* minutes in an hour */
2189 + 'daily' => 1440, /* minutes in a day */
2190 + 'weekly' => 10080, /* minutes in a week */
2191 + 'monthly' => 43200, /* minutes in a month */
2192 + 'yearly' => 525600, /* minutes in a year */
2193 + );
2194 +
2195 + // "sy:updatePeriod: Describes the period over which the
2196 + // channel format is updated. Acceptable values are:
2197 + // hourly, daily, weekly, monthly, yearly. If omitted,
2198 + // daily is assumed." <http://web.resource.org/rss/1.0/modules/syndication/>
2199 + if (isset($channel['sy']['updateperiod'])) : $period = $channel['sy']['updateperiod'];
2200 + else : $period = 'daily';
2201 + endif;
2202 +
2203 + // "sy:updateFrequency: Used to describe the frequency
2204 + // of updates in relation to the update period. A
2205 + // positive integer indicates how many times in that
2206 + // period the channel is updated. ... If omitted a value
2207 + // of 1 is assumed." <http://web.resource.org/rss/1.0/modules/syndication/>
2208 + if (isset($channel['sy']['updatefrequency'])) : $freq = (int) $channel['sy']['updatefrequency'];
2209 + else : $freq = 1;
2210 + endif;
2211 +
2212 + $ret = (int) ($period_minutes[$period] / $freq);
2213 + else :
2214 + $ret = NULL;
2215 + endif;
2216 + return $ret;
2217 + } /* SyndicatedLink::ttl() */
2218 +
2219 + // SyndicatedLink::flatten_array (): flatten an array. Useful for
2220 + // hierarchical and namespaced elements.
2221 + //
2222 + // Given an array which may contain array or object elements in it,
2223 + // return a "flattened" array: a one-dimensional array of scalars
2224 + // containing each of the scalar elements contained within the array
2225 + // structure. Thus, for example, if $a['b']['c']['d'] == 'e', then the
2226 + // returned array for FeedWordPress::flatten_array($a) will contain a key
2227 + // $a['feed/b/c/d'] with value 'e'.
2228 + function flatten_array ($arr, $prefix = 'feed/', $separator = '/') {
2229 + $ret = array ();
2230 + if (is_array($arr)) :
2231 + foreach ($arr as $key => $value) :
2232 + if (is_scalar($value)) :
2233 + $ret[$prefix.$key] = $value;
2234 + else :
2235 + $ret = array_merge($ret, $this->flatten_array($value, $prefix.$key.$separator, $separator));
2236 + endif;
2237 + endforeach;
2238 + endif;
2239 + return $ret;
2240 + } // function SyndicatedLink::flatten_array ()
2241 +
2242 + function hardcode ($what) {
2243 + $default = get_option("feedwordpress_hardcode_$what");
2244 + if ( $default === 'yes' ) :
2245 + // If the default is to hardcode, then we want the
2246 + // negation of negative(): TRUE by default and FALSE if
2247 + // the setting is explicitly "no"
2248 + $ret = !FeedWordPress::negative($this->settings, "hardcode $what");
2249 + else :
2250 + // If the default is NOT to hardcode, then we want
2251 + // affirmative(): FALSE by default and TRUE if the
2252 + // setting is explicitly "yes"
2253 + $ret = FeedWordPress::affirmative($this->settings, "hardcode $what");
2254 + endif;
2255 + return $ret;
2256 + } // function SyndicatedLink::hardcode ()
2257 +
2258 + function syndicated_status ($what, $default) {
2259 + global $wpdb;
2260 +
2261 + $ret = get_option("feedwordpress_syndicated_{$what}_status");
2262 + if ( isset($this->settings["$what status"]) ) :
2263 + $ret = $this->settings["$what status"];
2264 + elseif (!$ret) :
2265 + $ret = $default;
2266 + endif;
2267 + return $wpdb->escape(trim(strtolower($ret)));
2268 + } // function SyndicatedLink:syndicated_status ()
2269 +} // class SyndicatedLink
2270 +
1155 2271 ################################################################################
1156 2272 ## XML-RPC HOOKS: accept XML-RPC update pings from Contributors ################
1157 2273 ################################################################################
1158 2274
@@ -1174,12 +2290,345 @@
1174 2290 return array('flerror' => false, 'message' => "Thanks for the ping.".implode(' and', $mesg));
1175 2291 endif;
1176 2292 }
1177 2293
2294 +################################################################################
2295 +## class FeedFinder: find likely feeds using autodetection and/or guesswork ####
2296 +################################################################################
2297 +
2298 +class FeedFinder {
2299 + var $uri = NULL;
2300 + var $_cache_uri = NULL;
2301 +
2302 + var $verify = FALSE;
2303 +
2304 + var $_data = NULL;
2305 + var $_head = NULL;
2306 +
2307 + # -- Recognition patterns
2308 + var $_feed_types = array(
2309 + 'application/rss+xml',
2310 + 'text/xml',
2311 + 'application/atom+xml',
2312 + 'application/x.atom+xml',
2313 + 'application/x-atom+xml'
2314 + );
2315 + var $_feed_markers = array('\\<feed', '\\<rss', 'xmlns="http://purl.org/rss/1.0');
2316 + var $_html_markers = array('\\<html');
2317 + var $_obvious_feed_url = array('[./]rss', '[./]rdf', '[./]atom', '[./]feed', '\.xml');
2318 + var $_maybe_feed_url = array ('rss', 'rdf', 'atom', 'feed', 'xml');
2319 +
2320 + function FeedFinder ($uri = NULL, $verify = TRUE) {
2321 + $this->uri = $uri; $this->verify = $verify;
2322 + } /* FeedFinder::FeedFinder () */
2323 +
2324 + function find ($uri = NULL) {
2325 + $ret = array ();
2326 + if (!is_null($this->data($uri))) {
2327 + if ($this->is_feed($uri)) {
2328 + $ret = array($this->uri);
2329 + } else {
2330 + // Assume that we have HTML or XHTML (even if we don't, who's it gonna hurt?)
2331 + // Autodiscovery is the preferred method
2332 + $href = $this->_link_rel_feeds();
2333 +
2334 + // ... but we'll also take the little orange buttons
2335 + $href = array_merge($href, $this->_a_href_feeds(TRUE));
2336 +
2337 + // If all that failed, look harder
2338 + if (count($href) == 0) $href = $this->_a_href_feeds(FALSE);
2339 +
2340 + // Verify feeds and resolve relative URIs
2341 + foreach ($href as $u) {
2342 + $the_uri = Relative_URI::resolve($u, $this->uri);
2343 + if ($this->verify) {
2344 + $feed =& new FeedFinder($the_uri);
2345 + if ($feed->is_feed()) $ret[] = $the_uri;
2346 + $feed = NULL;
2347 + } else {
2348 + $ret[] = $the_uri;
2349 + }
2350 + } /* foreach */
2351 + } /* if */
2352 + } /* if */
2353 + return array_unique($ret);
2354 + } /* FeedFinder::find () */
2355 +
2356 + function data ($uri = NULL) {
2357 + $this->_get($uri);
2358 + return $this->_data;
2359 + }
2360 +
2361 + function is_feed ($uri = NULL) {
2362 + $data = $this->data($uri);
2363 +
2364 + return (
2365 + preg_match (
2366 + "\007(".implode('|',$this->_feed_markers).")\007i",
2367 + $data
2368 + ) and !preg_match (
2369 + "\007(".implode('|',$this->_html_markers).")\007i",
2370 + $data
2371 + )
2372 + );
2373 + } /* FeedFinder::is_feed () */
2374 +
2375 + # --- Private methods ---
2376 + function _get ($uri = NULL) {
2377 + if ($uri) $this->uri = $uri;
2378 +
2379 + // Is the result not yet cached?
2380 + if ($this->_cache_uri !== $this->uri) :
2381 + // Snoopy is an HTTP client in PHP
2382 + $client = new Snoopy();
2383 +
2384 + // Prepare headers and internal settings
2385 + $client->rawheaders['Connection'] = 'close';
2386 + $client->accept = 'application/atom+xml application/rdf+xml application/rss+xml application/xml text/html */*';
2387 + $client->agent = 'feedfinder/1.2 (compatible; PHP FeedFinder) +http://projects.radgeek.com/feedwordpress';
2388 + $client->read_timeout = 25;
2389 +
2390 + // Fetch the HTML or feed
2391 + @$client->fetch($this->uri);
2392 + $this->_data = $client->results;
2393 +
2394 + // Kilroy was here
2395 + $this->_cache_uri = $this->uri;
2396 + endif;
2397 + } /* FeedFinder::_get () */
2398 +
2399 + function _link_rel_feeds () {
2400 + $links = $this->_tags('link');
2401 + $link_count = count($links);
2402 +
2403 + // now figure out which one points to the RSS file
2404 + $href = array ();
2405 + for ($n=0; $n<$link_count; $n++) {
2406 + if (strtolower($links[$n]['rel']) == 'alternate') {
2407 + if (in_array(strtolower($links[$n]['type']), $this->_feed_types)) {
2408 + $href[] = $links[$n]['href'];
2409 + } /* if */
2410 + } /* if */
2411 + } /* for */
2412 + return $href;
2413 + }
2414 +
2415 + function _a_href_feeds ($obvious = TRUE) {
2416 + $pattern = ($obvious ? $this->_obvious_feed_url : $this->_maybe_feed_url);
2417 +
2418 + $links = $this->_tags('a');
2419 + $link_count = count($links);
2420 +
2421 + // now figure out which one points to the RSS file
2422 + $href = array ();
2423 + for ($n=0; $n<$link_count; $n++) {
2424 + if (preg_match("\007(".implode('|',$pattern).")\007i", $links[$n]['href'])) {
2425 + $href[] = $links[$n]['href'];
2426 + } /* if */
2427 + } /* for */
2428 + return $href;
2429 + }
2430 +
2431 + function _tags ($tag) {
2432 + $html = $this->data();
2433 +
2434 + // search through the HTML, save all <link> tags
2435 + // and store each link's attributes in an associative array
2436 + preg_match_all('/<'.$tag.'\s+(.*?)\s*\/?>/si', $html, $matches);
2437 + $links = $matches[1];
2438 + $ret = array();
2439 + $link_count = count($links);
2440 + for ($n=0; $n<$link_count; $n++) {
2441 + $attributes = preg_split('/\s+/s', $links[$n]);
2442 + foreach($attributes as $attribute) {
2443 + $att = preg_split('/\s*=\s*/s', $attribute, 2);
2444 + if (isset($att[1])) {
2445 + $att[1] = preg_replace('/([\'"]?)(.*)\1/', '$2', $att[1]);
2446 + $final_link[strtolower($att[0])] = $att[1];
2447 + } /* if */
2448 + } /* foreach */
2449 + $ret[$n] = $final_link;
2450 + } /* for */
2451 + return $ret;
2452 + }
2453 +} /* class FeedFinder */
2454 +
2455 +# Relative URI static class: PHP class for resolving relative URLs
2456 +#
2457 +# This class is derived (under the terms of the GPL) from URL Class 0.3 by
2458 +# Keyvan Minoukadeh <keyvan@k1m.com>, which is great but more than we need
2459 +# for FeedWordPress's purposes. The class has been stripped down to a single
2460 +# public method: Relative_URI::resolve($url, $base), which resolves the URI in
2461 +# $url relative to the URI in $base
2462 +#
1178 2463 # The upgraded MagpieRSS also uses this class. So if we have it loaded
1179 2464 # in, don't load it again.
1180 2465 if (!class_exists('Relative_URI')) {
1181 - require_once(dirname(__FILE__) . '/relative_uri.class.php');
2466 +
2467 + class Relative_URI
2468 + {
2469 + // Resolve relative URI in $url against the base URI in $base. If $base
2470 + // is not supplied, then we use the REQUEST_URI of this script.
2471 + //
2472 + // I'm hoping this method reflects RFC 2396 Section 5.2
2473 + function resolve ($url, $base = NULL)
2474 + {
2475 + if (is_null($base)):
2476 + $base = 'http://'.$_SERVER['HTTP_HOST'].$_SERVER['REQUEST_URI'];
2477 + endif;
2478 +
2479 + $base = Relative_URI::_encode(trim($base));
2480 + $uri_parts = Relative_URI::_parse_url($base);
2481 +
2482 + $url = Relative_URI::_encode(trim($url));
2483 + $parts = Relative_URI::_parse_url($url);
2484 +
2485 + $uri_parts['fragment'] = (isset($parts['fragment']) ? $parts['fragment'] : null);
2486 + $uri_parts['query'] = (isset($parts['query']) ? $parts['query'] : null);
2487 +
2488 + // if path is empty, and scheme, host, and query are undefined,
2489 + // the URL is referring the base URL
2490 +
2491 + if (($parts['path'] == '') && !isset($parts['scheme']) && !isset($parts['host']) && !isset($parts['query'])) {
2492 + // If the URI is empty or only a fragment, return the base URI
2493 + return $base . (isset($parts['fragment']) ? '#'.$parts['fragment'] : '');
2494 + } elseif (isset($parts['scheme'])) {
2495 + // If the scheme is set, then the URI is absolute.
2496 + return $url;
2497 + } elseif (isset($parts['host'])) {
2498 + $uri_parts['host'] = $parts['host'];
2499 + $uri_parts['path'] = $parts['path'];
2500 + } else {
2501 + // We have a relative path but not a host.
2502 +
2503 + // start ugly fix:
2504 + // prepend slash to path if base host is set, base path is not set, and url path is not absolute
2505 + if ($uri_parts['host'] && ($uri_parts['path'] == '')
2506 + && (strlen($parts['path']) > 0)
2507 + && (substr($parts['path'], 0, 1) != '/')) {
2508 + $parts['path'] = '/'.$parts['path'];
2509 + } // end ugly fix
2510 +
2511 + if (substr($parts['path'], 0, 1) == '/') {
2512 + $uri_parts['path'] = $parts['path'];
2513 + } else {
2514 + // copy base path excluding any characters after the last (right-most) slash character
2515 + $buffer = substr($uri_parts['path'], 0, (int)strrpos($uri_parts['path'], '/')+1);
2516 + // append relative path
2517 + $buffer .= $parts['path'];
2518 + // remove "./" where "." is a complete path segment.
2519 + $buffer = str_replace('/./', '/', $buffer);
2520 + if (substr($buffer, 0, 2) == './') {
2521 + $buffer = substr($buffer, 2);
2522 + }
2523 + // if buffer ends with "." as a complete path segment, remove it
2524 + if (substr($buffer, -2) == '/.') {
2525 + $buffer = substr($buffer, 0, -1);
2526 + }
2527 + // remove "<segment>/../" where <segment> is a complete path segment not equal to ".."
2528 + $search_finished = false;
2529 + $segment = explode('/', $buffer);
2530 + while (!$search_finished) {
2531 + for ($x=0; $x+1 < count($segment);) {
2532 + if (($segment[$x] != '') && ($segment[$x] != '..') && ($segment[$x+1] == '..')) {
2533 + if ($x+2 == count($segment)) $segment[] = '';
2534 + unset($segment[$x], $segment[$x+1]);
2535 + $segment = array_values($segment);
2536 + continue 2;
2537 + } else {
2538 + $x++;
2539 + }
2540 + }
2541 + $search_finished = true;
2542 + }
2543 + $buffer = (count($segment) == 1) ? '/' : implode('/', $segment);
2544 + $uri_parts['path'] = $buffer;
2545 +
2546 + }
2547 + }
2548 +
2549 + // If we've gotten to this point, we can try to put the pieces
2550 + // back together.
2551 + $ret = '';
2552 + if (isset($uri_parts['scheme'])) $ret .= $uri_parts['scheme'].':';
2553 + if (isset($uri_parts['user'])) {
2554 + $ret .= $uri_parts['user'];
2555 + if (isset($uri_parts['pass'])) $ret .= ':'.$uri_parts['parts'];
2556 + $ret .= '@';
2557 + }
2558 + if (isset($uri_parts['host'])) {
2559 + $ret .= '//'.$uri_parts['host'];
2560 + if (isset($uri_parts['port'])) $ret .= ':'.$uri_parts['port'];
2561 + }
2562 + $ret .= $uri_parts['path'];
2563 + if (isset($uri_parts['query'])) $ret .= '?'.$uri_parts['query'];
2564 + if (isset($uri_parts['fragment'])) $ret .= '#'.$uri_parts['fragment'];
2565 +
2566 + return $ret;
2567 + }
2568 +
2569 + /**
2570 + * Parse URL
2571 + *
2572 + * Regular expression grabbed from RFC 2396 Appendix B.
2573 + * This is a replacement for PHPs builtin parse_url().
2574 + * @param string $url
2575 + * @access private
2576 + * @return array
2577 + */
2578 + function _parse_url($url)
2579 + {
2580 + // I'm using this pattern instead of parse_url() as there's a few strings where parse_url()
2581 + // generates a warning.
2582 + if (preg_match('!^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?!', $url, $match)) {
2583 + $parts = array();
2584 + if ($match[1] != '') $parts['scheme'] = $match[2];
2585 + if ($match[3] != '') $parts['auth'] = $match[4];
2586 + // parse auth
2587 + if (isset($parts['auth'])) {
2588 + // store user info
2589 + if (($at_pos = strpos($parts['auth'], '@')) !== false) {
2590 + $userinfo = explode(':', substr($parts['auth'], 0, $at_pos), 2);
2591 + $parts['user'] = $userinfo[0];
2592 + if (isset($userinfo[1])) $parts['pass'] = $userinfo[1];
2593 + $parts['auth'] = substr($parts['auth'], $at_pos+1);
2594 + }
2595 + // get port number
2596 + if ($port_pos = strrpos($parts['auth'], ':')) {
2597 + $parts['host'] = substr($parts['auth'], 0, $port_pos);
2598 + $parts['port'] = (int)substr($parts['auth'], $port_pos+1);
2599 + if ($parts['port'] < 1) $parts['port'] = null;
2600 + } else {
2601 + $parts['host'] = $parts['auth'];
2602 + }
2603 + }
2604 + unset($parts['auth']);
2605 + $parts['path'] = $match[5];
2606 + if (isset($match[6]) && ($match[6] != '')) $parts['query'] = $match[7];
2607 + if (isset($match[8]) && ($match[8] != '')) $parts['fragment'] = $match[9];
2608 + return $parts;
2609 + }
2610 + // shouldn't reach here
2611 + return array('path'=>'');
2612 + }
2613 +
2614 + function _encode($string)
2615 + {
2616 + static $replace = array();
2617 + if (!count($replace)) {
2618 + $find = array(32, 34, 60, 62, 123, 124, 125, 91, 92, 93, 94, 96, 127);
2619 + $find = array_merge(range(0, 31), $find);
2620 + $find = array_map('chr', $find);
2621 + foreach ($find as $char) {
2622 + $replace[$char] = '%'.bin2hex($char);
2623 + }
2624 + }
2625 + // escape control characters and a few other characters
2626 + $encoded = strtr($string, $replace);
2627 + // remove any character outside the hex range: 21 - 7E (see www.asciitable.com)
2628 + return preg_replace('/[^\x21-\x7e]/', '', $encoded);
2629 + }
2630 + } // class Relative_URI
1182 2631 }
1183 2632
1184 2633 // take your best guess at the realname and e-mail, given a string
1185 2634 define('FWP_REGEX_EMAIL_ADDY', '([^@"(<\s]+@[^"@(<\s]+\.[^"@(<\s]+)');