Sitemap rewrite

4 files changed, 174 insertions, 97 deletions

diff --git a/inc/actions.php b/inc/actions.php
index 3e0cb1207..2d70ac8ed 100644
--- a/inc/actions.php
+++ b/inc/actions.php
@@ -56,6 +56,10 @@ function act_dispatch(){
         //check permissions
         $ACT = act_permcheck($ACT);
 
+        //sitemap
+        if ($ACT == 'sitemap')
+            $ACT = act_sitemap($ACT);
+
         //register
         $nil = array();
         if($ACT == 'register' && $_POST['save'] && register()){
@@ -205,7 +209,7 @@ function act_clean($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($act),-1);
         return 'show';
     }
@@ -233,7 +237,8 @@ function act_permcheck($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'){
         $permneed = AUTH_ADMIN;
@@ -587,6 +592,54 @@ function act_export($act){
 }
 
 /**
+ * Handle sitemap delivery
+ *
+ * @author Michael Hamann <michael@content-space.de>
+ */
+function act_sitemap($act) {
+    global $conf;
+
+    if (!$conf['sitemap']) {
+        header("HTTP/1.0 404 Not Found");
+        print "Sitemap generation is disabled.";
+        exit;
+    }
+    
+    $sitemap = $conf['cachedir'].'/sitemap.xml';
+    if($conf['compression'] == 'bz2' || $conf['compression'] == 'gz'){
+        $mime = 'application/x-gzip';
+        $sitemap .= '.gz';
+    } else {
+        $mime = 'application/xml; charset=utf-8';
+    }
+
+    // Check if sitemap file exists, otherwise create it
+    if (!is_readable($sitemap)) {
+        require_once DOKU_INC.'inc/sitemap.php';
+        sitemapGenerate();
+    }
+
+    if (is_readable($sitemap)) {
+        // Send headers
+        header('Content-Type: '.$mime);
+
+        // Send file
+        //use x-sendfile header to pass the delivery to compatible webservers
+        if (http_sendfile($sitemap)) exit;
+
+        $fp = @fopen($sitemap,"rb");
+        if($fp){
+            http_rangeRequest($fp,filesize($sitemap),$mime);
+            exit;
+        }
+    }
+
+    header("HTTP/1.0 500 Internal Server Error");
+    print "Could not read $sitemap - bad permissions?";
+    exit;
+}
+
+/**
  * Handle page 'subscribe'
  *
  * Throws exception on error.
diff --git a/inc/common.php b/inc/common.php
index bf5987c28..0816d9fbb 100644
--- a/inc/common.php
+++ b/inc/common.php
@@ -1267,6 +1267,21 @@ function dformat($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/sitemap.php b/inc/sitemap.php
new file mode 100644
index 000000000..bbed7d269
--- /dev/null
+++ b/inc/sitemap.php
@@ -0,0 +1,101 @@
+<?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.');
+
+/**
+ * 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 Andreas Gohr
+ * @link   https://www.google.com/webmasters/sitemaps/docs/en/about.html
+ */
+function sitemapGenerate(){
+    global $conf;
+    dbglog('sitemapGenerate(): started');
+    if(!$conf['sitemap']) return false;
+
+    $sitemap = sitemapGetFilePath();
+    dbglog("runSitemapper(): using $sitemap");
+
+    if(@file_exists($sitemap)){
+        if(!is_writable($sitemap)) return false;
+    }else{
+        if(!is_writable(dirname($sitemap))) return false;
+    }
+
+    if(@filesize($sitemap) &&
+       @filemtime($sitemap) > (time()-($conf['sitemap']*60*60*24))){
+       dbglog('runSitemapper(): Sitemap up to date');
+       return false;
+    }
+
+    $pages = idx_getIndex('page', '');
+    dbglog('runSitemapper(): creating sitemap using '.count($pages).' pages');
+
+    // 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;
+    $data = ob_get_contents();
+    ob_end_clean();
+
+    //save the new sitemap
+    return io_saveFile($sitemap,$data);
+}
+
+function sitemapGetFilePath() {
+    global $conf;
+
+    $sitemap = $conf['cachedir'].'/sitemap.xml';
+    if($conf['compression'] == 'bz2' || $conf['compression'] == 'gz'){
+        $sitemap .= '.gz';
+    }
+
+    return $sitemap;
+}
+
+function sitemapPingSearchEngines() {
+    //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,
+    );
+
+    foreach ($ping_urls as $name => $url) {
+        dbglog("sitemapPingSearchEngines(): pinging $name");
+        $resp = $http->get($url);
+        if($http->error) dbglog("runSitemapper(): $http->error");
+        dbglog('runSitemapper(): '.preg_replace('/[\n\r]/',' ',strip_tags($resp)));
+    }
+
+    return true;
+}
diff --git a/lib/exe/indexer.php b/lib/exe/indexer.php
index f8e2f7981..63ad5931f 100644
--- a/lib/exe/indexer.php
+++ b/lib/exe/indexer.php
@@ -232,88 +232,11 @@ 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;
-
+    require_once DOKU_INC.'inc/sitemap.php';
+    $result = sitemapGenerate() && sitemapPingSearchEngines();
     print 'runSitemapper(): finished'.NL;
-    return true;
+    return $result;
 }
 
 /**
@@ -409,21 +332,6 @@ 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>