From 6615d13f6e8332a789bf71b24ba2b6b94c512359 Mon Sep 17 00:00:00 2001
From: Daniel Winzen <daniel@danwin1210.de>
Date: Sat, 11 Mar 2023 12:18:40 +0100
Subject: [PATCH] Handle alternative languages correctly in sitemap

---
 www/sitemap.php | 115 +++++++++++++++++++++++++++++++-----------------
 1 file changed, 74 insertions(+), 41 deletions(-)

diff --git a/www/sitemap.php b/www/sitemap.php
index 4e94bcc..1c7c3fa 100644
--- a/www/sitemap.php
+++ b/www/sitemap.php
@@ -8,68 +8,101 @@ try{
 	die(_('No database connection!'));
 }
 $links = [];
-$links []= ['loc' => CANONICAL_URL . '/test.php', 'changefreq' => 'weekly', 'priority' => '0.8'];
-$links []= ['loc' => CANONICAL_URL . '/', 'changefreq' => 'daily', 'priority' => '1'];
+$alts = [];
+foreach (LANGUAGES as $lang_code => $data) {
+	$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . '/test.php?lang='.$lang_code];
+}
+$links []= ['loc' => CANONICAL_URL . '/test.php', 'changefreq' => 'weekly', 'priority' => '0.8', 'alt' => $alts];
 $links []= ['loc' => CANONICAL_URL . '/?format=json', 'changefreq' => 'daily', 'priority' => '0.2'];
 $links []= ['loc' => CANONICAL_URL . '/?format=text', 'changefreq' => 'daily', 'priority' => '0.2'];
 $admin_approval = '';
 if(REQUIRE_APPROVAL){
 	$admin_approval = PREFIX . 'onions.approved = 1 AND';
 }
-foreach (LANGUAGES as $lang_code => $data){
-	$links []= ['loc' => CANONICAL_URL . "/test.php?lang=$lang_code", 'changefreq' => 'weekly', 'priority' => '0.4'];
-	$links []= ['loc' => CANONICAL_URL . "/?lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.5'];
-	$stmt=$db->prepare('SELECT COUNT(*) FROM ' . PREFIX . "onions WHERE $admin_approval category=? AND address!='' AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing) AND timediff<604800;');
-	foreach($categories as $cat => $name){
-		$links []= ['loc' => CANONICAL_URL . "/?cat=$cat&lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3'];
-		$stmt->execute([$cat]);
-		$num=$stmt->fetch(PDO::FETCH_NUM);
-		$pages=ceil($num[0]/PER_PAGE);
-		if($pages > 1) {
-			while ( $pages > 1 ) {
-				$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=$pages&lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3' ];
-				--$pages;
-			}
-			$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=0&lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3' ];
-		}
+$stmt=$db->prepare('SELECT COUNT(*) FROM ' . PREFIX . "onions WHERE $admin_approval category=? AND address!='' AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing) AND timediff<604800;');
+foreach($categories as $cat => $name){
+	$alts = [];
+	foreach (LANGUAGES as $lang_code => $data) {
+		$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?cat=$cat&lang=$lang_code"];
 	}
-	$special=[
-		'all'=>"address!='' AND category!=15 AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing) AND timediff<604800',
-		'lastadded'=>"address!='' AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing)',
-		'offline'=>"address!='' AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing) AND timediff>604800'
-	];
-	$cat=count($categories);
-	foreach($special as $query){
-		$links []= ['loc' => CANONICAL_URL . "/?".($cat===count($categories) ? '' : "cat=$cat&")."lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3'];
-		if($cat===count($categories)+1){
-			$num[0]=PER_PAGE;
-		}else{
-			$num=$db->query('SELECT COUNT(*) FROM ' . PREFIX . "onions WHERE $admin_approval $query;")->fetch(PDO::FETCH_NUM);
-		}
-		$pages=ceil($num[0]/PER_PAGE);
-		if($pages > 1) {
-			while ( $pages > 1 ) {
-				$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=$pages&lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3' ];
-				--$pages;
+	$links []= ['loc' => CANONICAL_URL . "/?cat=$cat", 'changefreq' => 'daily', 'priority' => '0.3', 'alt' => $alts];
+	$stmt->execute([$cat]);
+	$num=$stmt->fetch(PDO::FETCH_NUM);
+	$pages=ceil($num[0]/PER_PAGE);
+	if($pages > 1) {
+		while ( $pages > 1 ) {
+			$alts = [];
+			foreach (LANGUAGES as $lang_code => $data) {
+				$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?cat=$cat&pg=$pages&lang=$lang_code"];
 			}
-			$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=0&lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3' ];
+			$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=$pages", 'changefreq' => 'daily', 'priority' => '0.3', 'alt' => $alts ];
+			--$pages;
 		}
-		++$cat;
+		$alts = [];
+		foreach (LANGUAGES as $lang_code => $data) {
+			$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?cat=$cat&pg=0&lang=$lang_code"];
+		}
+		$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=0", 'changefreq' => 'daily', 'priority' => '0.3', 'alt' => $alts ];
 	}
-	$links []= ['loc' => CANONICAL_URL . "/?cat=$cat&lang=$lang_code", 'changefreq' => 'daily', 'priority' => '0.3'];
 }
+$special=[
+	'all'=>"address!='' AND category!=15 AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing) AND timediff<604800',
+	'lastadded'=>"address!='' AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing)',
+	'offline'=>"address!='' AND id NOT IN (SELECT onion_id FROM " . PREFIX . 'phishing) AND timediff>604800'
+];
+$cat=count($categories);
+foreach($special as $query){
+	$alts = [];
+	foreach (LANGUAGES as $lang_code => $data) {
+		$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?".($cat===count($categories) ? '' : "cat=$cat&")."lang=$lang_code"];
+	}
+	$links []= ['loc' => CANONICAL_URL . "/".($cat===count($categories) ? '' : "?cat=$cat"), 'changefreq' => 'daily', 'priority' => $cat===count($categories) ? '1.0' : '0.3', 'alt' => $alts];
+	if($cat===count($categories)+1){
+		$num[0]=PER_PAGE;
+	}else{
+		$num=$db->query('SELECT COUNT(*) FROM ' . PREFIX . "onions WHERE $admin_approval $query;")->fetch(PDO::FETCH_NUM);
+	}
+	$pages=ceil($num[0]/PER_PAGE);
+	if($pages > 1) {
+		while ( $pages > 1 ) {
+			$alts = [];
+			foreach (LANGUAGES as $lang_code => $data) {
+				$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?cat=$cat&pg=$pages&lang=$lang_code"];
+			}
+			$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=$pages", 'changefreq' => 'daily', 'priority' => '0.3', 'alt' => $alts ];
+			--$pages;
+		}
+		$alts = [];
+		foreach (LANGUAGES as $lang_code => $data) {
+			$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?cat=$cat&pg=0&lang=$lang_code"];
+		}
+		$links [] = [ 'loc' => CANONICAL_URL . "/?cat=$cat&pg=0", 'changefreq' => 'daily', 'priority' => '0.3', 'alt' => $alts ];
+	}
+	++$cat;
+}
+$alts = [];
+foreach (LANGUAGES as $lang_code => $data) {
+	$alts []= ['hreflang' => $lang_code, 'href' => CANONICAL_URL . "/?cat=$cat&lang=$lang_code"];
+}
+$links []= ['loc' => CANONICAL_URL . "/?cat=$cat", 'changefreq' => 'daily', 'priority' => '0.3', 'alt' => $alts];
 $dom = new DOMDocument('1.0', 'UTF-8');
 try {
 	$urlset = $dom->createElement( 'urlset' );
 	$urlset->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
-	$urlset->setAttribute('xmlns:xsi', 'http://www.w3.org/2001/XMLSchema-instance');
-	$urlset->setAttribute('xsi:schemaLocation', 'http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd');
+	$urlset->setAttribute('xmlns:xhtml', 'http://www.w3.org/1999/xhtml');
 	$dom->appendChild($urlset);
 	foreach ($links as $link) {
 		$url = $dom->createElement('url');
 		$urlset->appendChild($url);
 		$loc = $dom->createElement('loc', htmlspecialchars($link['loc']));
 		$url->appendChild($loc);
+		foreach ($link['alt'] as $alt){
+			$link_alt = $dom->createElement('xhtml:link');
+			$link_alt->setAttribute('rel', 'alternate');
+			$link_alt->setAttribute('hreflang', $alt['hreflang']);
+			$link_alt->setAttribute('href', $alt['href']);
+			$url->appendChild($link_alt);
+		}
 		$changefreq = $dom->createElement('changefreq', $link['changefreq']);
 		$url->appendChild($changefreq);
 		$priority = $dom->createElement('priority', $link['priority']);