1 files changed, 33 insertions, 24 deletions

diff --git a/crawler/Database.php b/crawler/Database.php
index 6d04009..0d500ad 100644
--- a/crawler/Database.php
+++ b/crawler/Database.php
@@ -13,46 +13,47 @@ class Database
         $conn = self::initDbConnection();
         $checkStmt = $conn->query('SELECT url FROM queue ORDER BY id ' . $sort . ' LIMIT 1');
 
-        return $checkStmt->fetchAll(PDO::FETCH_ASSOC)[0]['url'];
+        return $checkStmt->fetchAll(PDO::FETCH_ASSOC)[0]['url'] ?? '';
     }
 
-    private static function initDbConnection(): PDO
-    {
-        global $servername, $dbname, $username, $password;
-        $conn = new PDO("mysql:host=$servername;dbname=$dbname;charset=utf8", $username, $password);
-        $conn->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
-        return $conn;
-    }
-
-    public static function insertIntoQueue($url): bool
+    public static function insertIntoQueue($urls)
     {
-        if (!self::alreadyCrawled($url)) {
-            $conn = self::initDbConnection();
-            $hash = md5($url);
-            $stmt = $conn->prepare('INSERT IGNORE INTO queue (url, hash) VALUES (:url, :hash)');
-            $stmt->execute([':url' => $url, 'hash' => $hash]);
-            return $stmt->rowCount() > 0;
+        foreach ($urls as $url) {
+            if (self::alreadyCrawled($url)) {
+                print "\t\e[91mUrl already queued " . $url . "\n";
+            } else {
+                print "\t\e[92mQueueing url " . $url . "\n";
+                $conn = self::initDbConnection();
+                $hash = md5($url);
+                $stmt = $conn->prepare('INSERT IGNORE INTO queue (url, hash) VALUES (:url, :hash)');
+                $stmt->execute([':url' => $url, 'hash' => $hash]);
+            }
         }
     }
 
-    public static function alreadyCrawled($url): bool
+    public static function removeFromQueue($url)
     {
         $hash = md5($url);
         $conn = self::initDbConnection();
-        $checkStmt = $conn->prepare('(SELECT null FROM url_data WHERE hash = :hash) UNION (SELECT null FROM error_url WHERE hash = :hash)');
+        $checkStmt = $conn->prepare('DELETE FROM queue WHERE hash = :hash');
         $checkStmt->execute([':hash' => $hash]);
-        return $checkStmt->rowCount() !== 0; // return true if already crawled
     }
 
-    public static function removeFromQueue($url): void
+    public static function alreadyCrawled($url): bool
     {
+        print "\t\e[96mChecking if url already has been crawled " . $url . "\n";
         $hash = md5($url);
         $conn = self::initDbConnection();
-        $checkStmt = $conn->prepare('DELETE FROM queue WHERE hash = :hash');
+        $checkStmt = $conn->prepare('(SELECT null FROM url_data WHERE hash = :hash) UNION (SELECT null FROM error_url WHERE hash = :hash)');
         $checkStmt->execute([':hash' => $hash]);
+        $alreadyCrawled = $checkStmt->rowCount() !== 0;
+        if ($alreadyCrawled) {
+            print "\t\e[91mUrl already crawled " . $url . "\n";
+        }
+        return $alreadyCrawled; // return true if already crawled
     }
 
-    public static function urlHasError($url): void
+    public static function urlHasError($url)
     {
         $hash = md5($url);
         $conn = self::initDbConnection();
@@ -60,10 +61,18 @@ class Database
         $checkStmt->execute([':url' => $url, 'hash' => $hash]);
     }
 
-    public static function saveUrlData($data): void
+    public static function saveUrlData($url, $data)
     {
         $conn = self::initDbConnection();
         $stmt = $conn->prepare('INSERT IGNORE INTO url_data (url, title, description, lang, hash) VALUES (:url, :title, :description, :lang, :hash)');
-        $stmt->execute([':url' => $data[0], ':title' => $data[1], ':description' => $data[2], ':lang' => $data[3], ':hash' => $data[4]]);
+        $stmt->execute([':url' => $url, ':title' => $data['title'] ?? '', ':description' => $data['description'] ?? '', ':lang' => $data['lang'] ?? 'en', ':hash' => md5($url)]);
+    }
+
+    private static function initDbConnection(): PDO
+    {
+        global $servername, $dbname, $username, $password;
+        $conn = new PDO("mysql:host=$servername;dbname=$dbname;charset=utf8", $username, $password);
+        $conn->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
+        return $conn;
     }
 }
\ No newline at end of file