summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--inc/Sitemapper.php201
-rw-r--r--inc/actions.php55
-rw-r--r--inc/common.php15
-rw-r--r--inc/load.php1
-rw-r--r--lib/exe/indexer.php97
5 files changed, 272 insertions, 97 deletions
diff --git a/inc/Sitemapper.php b/inc/Sitemapper.php
new file mode 100644
index 000000000..52c71c545
--- /dev/null
+++ b/inc/Sitemapper.php
@@ -0,0 +1,201 @@
+<?php
+/**
+ * Sitemap handling functions
+ *
+ * @license GPL 2 (http://www.gnu.org/licenses/gpl.html)
+ * @author Michael Hamann <michael@content-space.de>
+ */
+
+if(!defined('DOKU_INC')) die('meh.');
+
+/**
+ * A class for building sitemaps and pinging search engines with the sitemap URL.
+ *
+ * @author Michael Hamann
+ */
+class Sitemapper {
+ /**
+ * Builds a Google Sitemap of all public pages known to the indexer
+ *
+ * The map is placed in the cache directory named sitemap.xml.gz - This
+ * file needs to be writable!
+ *
+ * @author Michael Hamann
+ * @author Andreas Gohr
+ * @link https://www.google.com/webmasters/sitemaps/docs/en/about.html
+ * @link http://www.sitemaps.org/
+ */
+ public function generate(){
+ global $conf;
+ if($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) return false;
+
+ $sitemap = Sitemapper::getFilePath();
+
+ if(@file_exists($sitemap)){
+ if(!is_writable($sitemap)) return false;
+ }else{
+ if(!is_writable(dirname($sitemap))) return false;
+ }
+
+ if(@filesize($sitemap) &&
+ @filemtime($sitemap) > (time()-($conf['sitemap']*86400))){ // 60*60*24=86400
+ dbglog('Sitemapper::generate(): Sitemap up to date'); // FIXME: only in debug mode
+ return false;
+ }
+
+ dbglog("Sitemapper::generate(): using $sitemap"); // FIXME: Only in debug mode
+
+ $pages = idx_getIndex('page', '');
+ dbglog('Sitemapper::generate(): creating sitemap using '.count($pages).' pages');
+ $items = array();
+
+ // build the sitemap items
+ foreach($pages as $id){
+ //skip hidden, non existing and restricted files
+ if(isHiddenPage($id)) continue;
+ if(auth_aclcheck($id,'','') < AUTH_READ) continue;
+ $items[] = SitemapItem::createFromID($id);
+ }
+
+ $eventData = array('items' => &$items, 'sitemap' => &$sitemap);
+ $event = new Doku_Event('SITEMAP_GENERATE', $eventData);
+ if ($event->advise_before(true)) {
+ //save the new sitemap
+ $result = io_saveFile($sitemap, Sitemapper::getXML($items));
+ }
+ $event->advise_after();
+
+ return $result;
+ }
+
+ /**
+ * Builds the sitemap XML string from the given array auf SitemapItems.
+ *
+ * @param $items array The SitemapItems that shall be included in the sitemap.
+ * @return string The sitemap XML.
+ * @author Michael Hamann
+ */
+ private function getXML($items) {
+ ob_start();
+ echo '<?xml version="1.0" encoding="UTF-8"?>'.NL;
+ echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
+ foreach ($items as $item) {
+ echo $item->toXML();
+ }
+ echo '</urlset>'.NL;
+ $result = ob_get_contents();
+ ob_end_clean();
+ return $result;
+ }
+
+ /**
+ * Helper function for getting the path to the sitemap file.
+ *
+ * @return The path to the sitemap file.
+ * @author Michael Hamann
+ */
+ public function getFilePath() {
+ global $conf;
+
+ $sitemap = $conf['cachedir'].'/sitemap.xml';
+ if($conf['compression'] === 'bz2' || $conf['compression'] === 'gz'){
+ $sitemap .= '.gz';
+ }
+
+ return $sitemap;
+ }
+
+ /**
+ * Pings search engines with the sitemap url. Plugins can add or remove
+ * urls to ping using the SITEMAP_PING event.
+ *
+ * @author Michael Hamann
+ */
+ public function pingSearchEngines() {
+ //ping search engines...
+ $http = new DokuHTTPClient();
+ $http->timeout = 8;
+
+ $encoded_sitemap_url = urlencode(wl('', array('do' => 'sitemap'), true, '&'));
+ $ping_urls = array(
+ 'google' => 'http://www.google.com/webmasters/sitemaps/ping?sitemap='.$encoded_sitemap_url,
+ 'yahoo' => 'http://search.yahooapis.com/SiteExplorerService/V1/updateNotification?appid=dokuwiki&url='.$encoded_sitemap_url,
+ 'microsoft' => 'http://www.bing.com/webmaster/ping.aspx?siteMap='.$encoded_sitemap_url,
+ );
+
+ $data = array('ping_urls' => $ping_urls,
+ 'encoded_sitemap_url' => $encoded_sitemap_url
+ );
+ $event = new Doku_Event('SITEMAP_PING', $data);
+ if ($event->advise_before(true)) {
+ foreach ($data['ping_urls'] as $name => $url) {
+ dbglog("Sitemapper::PingSearchEngines(): pinging $name");
+ $resp = $http->get($url);
+ if($http->error) dbglog("Sitemapper:pingSearchengines(): $http->error");
+ dbglog('Sitemapper:pingSearchengines(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp)));
+ }
+ }
+ $event->advise_after();
+
+ return true;
+ }
+}
+
+/**
+ * An item of a sitemap.
+ *
+ * @author Michael Hamann
+ */
+class SitemapItem {
+ public $url;
+ public $lastmod;
+ public $changefreq;
+ public $priority;
+
+ /**
+ * Create a new item.
+ *
+ * @param $url string The url of the item
+ * @param $lastmod int Timestamp of the last modification
+ * @param $changefreq string How frequently the item is likely to change. Valid values: always, hourly, daily, weekly, monthly, yearly, never.
+ * @param $priority float|string The priority of the item relative to other URLs on your site. Valid values range from 0.0 to 1.0.
+ */
+ public function __construct($url, $lastmod, $changefreq = null, $priority = null) {
+ $this->url = $url;
+ $this->lastmod = $lastmod;
+ $this->changefreq = $changefreq;
+ $this->priority = $priority;
+ }
+
+ /**
+ * Helper function for creating an item for a wikipage id.
+ *
+ * @param $id string A wikipage id.
+ * @param $changefreq string How frequently the item is likely to change. Valid values: always, hourly, daily, weekly, monthly, yearly, never.
+ * @param $priority float|string The priority of the item relative to other URLs on your site. Valid values range from 0.0 to 1.0.
+ * @return The sitemap item.
+ */
+ public static function createFromID($id, $changefreq = null, $priority = null) {
+ $id = trim($id);
+ $date = @filemtime(wikiFN($id));
+ if(!$date) return NULL;
+ return new SitemapItem(wl($id, '', true), $date, $changefreq, $priority);
+ }
+
+ /**
+ * Get the XML representation of the sitemap item.
+ *
+ * @return The XML representation.
+ */
+ public function toXML() {
+ $result = ' <url>'.NL
+ .' <loc>'.hsc($this->url).'</loc>'.NL
+ .' <lastmod>'.date_iso8601($this->lastmod).'</lastmod>'.NL;
+ if ($this->changefreq !== NULL)
+ $result .= ' <changefreq>'.hsc($this->changefreq).'</changefreq>'.NL;
+ if ($this->priority !== NULL)
+ $result .= ' <priority>'.hsc($this->priority).'</priority>'.NL;
+ $result .= ' </url>'.NL;
+ return $result;
+ }
+}
diff --git a/inc/actions.php b/inc/actions.php
index 0a6e6d8c7..9db7d5f24 100644
--- a/inc/actions.php
+++ b/inc/actions.php
@@ -53,6 +53,11 @@ function act_dispatch(){
//check permissions
$ACT = act_permcheck($ACT);
+ //sitemap
+ if ($ACT == 'sitemap'){
+ $ACT = act_sitemap($ACT);
+ }
+
//register
if($ACT == 'register' && $_POST['save'] && register()){
$ACT = 'login';
@@ -205,7 +210,7 @@ function act_clean($act){
'preview','search','show','check','index','revisions',
'diff','recent','backlink','admin','subscribe','revert',
'unsubscribe','profile','resendpwd','recover',
- 'draftdel','subscribens','unsubscribens',)) && substr($act,0,7) != 'export_' ) {
+ 'draftdel','subscribens','unsubscribens','sitemap')) && substr($act,0,7) != 'export_' ) {
msg('Command unknown: '.htmlspecialchars($act),-1);
return 'show';
}
@@ -233,7 +238,8 @@ function act_permcheck($act){
}else{
$permneed = AUTH_CREATE;
}
- }elseif(in_array($act,array('login','search','recent','profile','index'))){
+ }elseif(in_array($act,array('login','search','recent','profile','index', 'sitemap'))){
+ }elseif(in_array($act,array('login','search','recent','profile','sitemap'))){
$permneed = AUTH_NONE;
}elseif($act == 'revert'){
$permneed = AUTH_ADMIN;
@@ -587,6 +593,51 @@ function act_export($act){
}
/**
+ * Handle sitemap delivery
+ *
+ * @author Michael Hamann <michael@content-space.de>
+ */
+function act_sitemap($act) {
+ global $conf;
+
+ if ($conf['sitemap'] < 1 || !is_numeric($conf['sitemap'])) {
+ header("HTTP/1.0 404 Not Found");
+ print "Sitemap generation is disabled.";
+ exit;
+ }
+
+ $sitemap = Sitemapper::getFilePath();
+ if(strrchr($sitemap, '.') === '.gz'){
+ $mime = 'application/x-gzip';
+ }else{
+ $mime = 'application/xml; charset=utf-8';
+ }
+
+ // Check if sitemap file exists, otherwise create it
+ if (!is_readable($sitemap)) {
+ Sitemapper::generate();
+ }
+
+ if (is_readable($sitemap)) {
+ // Send headers
+ header('Content-Type: '.$mime);
+
+ http_conditionalRequest(filemtime($sitemap));
+
+ // Send file
+ //use x-sendfile header to pass the delivery to compatible webservers
+ if (http_sendfile($sitemap)) exit;
+
+ readfile($sitemap);
+ exit;
+ }
+
+ header("HTTP/1.0 500 Internal Server Error");
+ print "Could not read the sitemap file - bad permissions?";
+ exit;
+}
+
+/**
* Handle page 'subscribe'
*
* Throws exception on error.
diff --git a/inc/common.php b/inc/common.php
index 3e760419f..894b8a020 100644
--- a/inc/common.php
+++ b/inc/common.php
@@ -1271,6 +1271,21 @@ function dformat($dt=null,$format=''){
}
/**
+ * Formats a timestamp as ISO 8601 date
+ *
+ * @author <ungu at terong dot com>
+ * @link http://www.php.net/manual/en/function.date.php#54072
+ */
+function date_iso8601($int_date) {
+ //$int_date: current date in UNIX timestamp
+ $date_mod = date('Y-m-d\TH:i:s', $int_date);
+ $pre_timezone = date('O', $int_date);
+ $time_zone = substr($pre_timezone, 0, 3).":".substr($pre_timezone, 3, 2);
+ $date_mod .= $time_zone;
+ return $date_mod;
+}
+
+/**
* return an obfuscated email address in line with $conf['mailguard'] setting
*
* @author Harry Fuecks <hfuecks@gmail.com>
diff --git a/inc/load.php b/inc/load.php
index 2f5be6d63..478ee7c76 100644
--- a/inc/load.php
+++ b/inc/load.php
@@ -74,6 +74,7 @@ function load_autoload($name){
'DokuWikiFeedCreator' => DOKU_INC.'inc/feedcreator.class.php',
'Doku_Parser_Mode' => DOKU_INC.'inc/parser/parser.php',
'SafeFN' => DOKU_INC.'inc/SafeFN.class.php',
+ 'Sitemapper' => DOKU_INC.'inc/Sitemapper.php',
'DokuWiki_Action_Plugin' => DOKU_PLUGIN.'action.php',
'DokuWiki_Admin_Plugin' => DOKU_PLUGIN.'admin.php',
diff --git a/lib/exe/indexer.php b/lib/exe/indexer.php
index 3a9673ed6..f35f9ed72 100644
--- a/lib/exe/indexer.php
+++ b/lib/exe/indexer.php
@@ -229,88 +229,10 @@ function metaUpdate(){
* @link https://www.google.com/webmasters/sitemaps/docs/en/about.html
*/
function runSitemapper(){
- global $conf;
print "runSitemapper(): started".NL;
- if(!$conf['sitemap']) return false;
-
- if($conf['compression'] == 'bz2' || $conf['compression'] == 'gz'){
- $sitemap = 'sitemap.xml.gz';
- }else{
- $sitemap = 'sitemap.xml';
- }
- print "runSitemapper(): using $sitemap".NL;
-
- if(@file_exists(DOKU_INC.$sitemap)){
- if(!is_writable(DOKU_INC.$sitemap)) return false;
- }else{
- if(!is_writable(DOKU_INC)) return false;
- }
-
- if(@filesize(DOKU_INC.$sitemap) &&
- @filemtime(DOKU_INC.$sitemap) > (time()-($conf['sitemap']*60*60*24))){
- print 'runSitemapper(): Sitemap up to date'.NL;
- return false;
- }
-
- $pages = idx_getIndex('page', '');
- print 'runSitemapper(): creating sitemap using '.count($pages).' pages'.NL;
-
- // build the sitemap
- ob_start();
- print '<?xml version="1.0" encoding="UTF-8"?>'.NL;
- print '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.NL;
- foreach($pages as $id){
- $id = trim($id);
- $file = wikiFN($id);
-
- //skip hidden, non existing and restricted files
- if(isHiddenPage($id)) continue;
- $date = @filemtime($file);
- if(!$date) continue;
- if(auth_aclcheck($id,'','') < AUTH_READ) continue;
-
- print ' <url>'.NL;
- print ' <loc>'.wl($id,'',true).'</loc>'.NL;
- print ' <lastmod>'.date_iso8601($date).'</lastmod>'.NL;
- print ' </url>'.NL;
- }
- print '</urlset>'.NL;
- $data = ob_get_contents();
- ob_end_clean();
-
- //save the new sitemap
- io_saveFile(DOKU_INC.$sitemap,$data);
-
- //ping search engines...
- $http = new DokuHTTPClient();
- $http->timeout = 8;
-
- //ping google
- print 'runSitemapper(): pinging google'.NL;
- $url = 'http://www.google.com/webmasters/sitemaps/ping?sitemap=';
- $url .= urlencode(DOKU_URL.$sitemap);
- $resp = $http->get($url);
- if($http->error) print 'runSitemapper(): '.$http->error.NL;
- print 'runSitemapper(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp)).NL;
-
- //ping yahoo
- print 'runSitemapper(): pinging yahoo'.NL;
- $url = 'http://search.yahooapis.com/SiteExplorerService/V1/updateNotification?appid=dokuwiki&url=';
- $url .= urlencode(DOKU_URL.$sitemap);
- $resp = $http->get($url);
- if($http->error) print 'runSitemapper(): '.$http->error.NL;
- print 'runSitemapper(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp)).NL;
-
- //ping microsoft
- print 'runSitemapper(): pinging microsoft'.NL;
- $url = 'http://www.bing.com/webmaster/ping.aspx?siteMap=';
- $url .= urlencode(DOKU_URL.$sitemap);
- $resp = $http->get($url);
- if($http->error) print 'runSitemapper(): '.$http->error.NL;
- print 'runSitemapper(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp)).NL;
-
+ $result = Sitemapper::generate() && Sitemapper::pingSearchEngines();
print 'runSitemapper(): finished'.NL;
- return true;
+ return $result;
}
/**
@@ -406,21 +328,6 @@ function sendDigest() {
}
/**
- * Formats a timestamp as ISO 8601 date
- *
- * @author <ungu at terong dot com>
- * @link http://www.php.net/manual/en/function.date.php#54072
- */
-function date_iso8601($int_date) {
- //$int_date: current date in UNIX timestamp
- $date_mod = date('Y-m-d\TH:i:s', $int_date);
- $pre_timezone = date('O', $int_date);
- $time_zone = substr($pre_timezone, 0, 3).":".substr($pre_timezone, 3, 2);
- $date_mod .= $time_zone;
- return $date_mod;
-}
-
-/**
* Just send a 1x1 pixel blank gif to the browser
*
* @author Andreas Gohr <andi@splitbrain.org>
generated by cgit v1.2.3 (git 2.25.1) at 2024-04-28 21:42:45 +0000