implement page ranking

crontab/crawler.php

@@ -183,6 +183,18 @@ foreach ($db->getCrawlQueue(CRAWL_PAGE_LIMIT, time() - CRAWL_PAGE_SECONDS_OFFSET
           $hostRobots        = $host->robots;
           $hostRobotsPostfix = $host->robotsPostfix;
 
+          // Increase page rank when link does not match the current host
+          if ($hostURL->scheme . '://' .
+              $hostURL->name .
+             ($hostURL->port ? ':' . $hostURL->port : '')
+              !=
+              $queueHostPage->scheme . '://' .
+              $queueHostPage->name .
+             ($queueHostPage->port ? ':' . $queueHostPage->port : '')) {
+
+              $db->updateHostPageRank($hostId, crc32($hostPageURI->string), 1);
+          }
+
         // Register new host
         } else {
 

library/mysql.php

@@ -143,6 +143,7 @@ class MySQL {
                                          `hostPage`.`metaDescription`,
                                          `hostPage`.`data`,
                                          `hostPage`.`uri`,
+                                         `hostPage`.`rank`,
                                          `host`.`scheme`,
                                          `host`.`name`,
                                          `host`.`port`
@@ -204,6 +205,22 @@ class MySQL {
     return $query->rowCount();
   }
 
+  public function updateHostPageRank(int $hostId,
+                                     int $crc32uri,
+                                     int $increment) {
+
+    $query = $this->_db->prepare('UPDATE `hostPage` SET   `rank` = `rank` + ' . (int) $increment . '
+
+                                                    WHERE `hostId` = ?
+                                                    AND   `crc32uri` = ?
+
+                                                    LIMIT 1');
+
+    $query->execute([$hostId, $crc32uri]);
+
+    return $query->rowCount();
+  }
+
   public function deleteHostPage(int $hostPageId) {
 
     $query = $this->_db->prepare('DELETE FROM `hostPage` WHERE `hostPageId` = ? LIMIT 1');

library/sphinxql.php

@@ -13,13 +13,17 @@ class SphinxQL {
 
   public function searchHostPages(string $keyword, int $start, int $limit, int $maxMatches) {
 
-    $query = $this->_sphinx->prepare('SELECT * FROM `hostPage`
+    $query = $this->_sphinx->prepare('SELECT *, WEIGHT() AS `weight`
 
-                                               WHERE MATCH(?)
+                                      FROM `hostPage`
 
-                                               LIMIT ' . (int) ($start > $maxMatches ? ($maxMatches > 0 ? $maxMatches - 1 : 0) : $start) . ',' . (int) $limit . '
+                                      WHERE MATCH(?)
 
-                                               OPTION `max_matches`=' . (int) ($maxMatches > 1 ? $maxMatches : 1));
+                                      ORDER BY `rank` DESC, WEIGHT() DESC
+
+                                      LIMIT ' . (int) ($start > $maxMatches ? ($maxMatches > 0 ? $maxMatches - 1 : 0) : $start) . ',' . (int) $limit . '
+
+                                      OPTION `max_matches`=' . (int) ($maxMatches > 1 ? $maxMatches : 1));
 
     $query->execute([$keyword]);