Fix in HTML_Index::detect_charset() for too short html #91

duzun · duzun · commit 37b416052d8a · 2024-07-03T10:26:56.000+03:00
diff --git a/examples/playground.php b/examples/playground.php
@@ -18,7 +18,7 @@
         hQuery::$cache_path = sys_get_temp_dir() . '/hQuery/';
         isset($ch) && $ch !== '' and hQuery::$cache_expires = (int)$ch;
 
-        // Results acumulator
+        // Results accumulator
         $return = array();
 
         // If we have $url to parse and $sel (selector) to fetch, we a good to go
diff --git a/src/hQuery/HTML_Index.php b/src/hQuery/HTML_Index.php
@@ -402,9 +402,9 @@ public static function abs_url($url, $base)
      */
     public static function detect_charset($str)
     {
-        $l    = 1024;
-        $str  = substr($str, 0, $l);
+        $str  = substr($str, 0, 1024);
         $str_ = strtolower($str);
+        $l    = strlen($str_);
         $p    = 0;
         while ($p < $l) {
             $p = strpos($str_, '<meta', $p);
diff --git a/tests/hQueryCore.Test.php b/tests/hQueryCore.Test.php
@@ -122,6 +122,8 @@ class hQueryCore extends PHPUnit_BaseClass
     </body></html>
 EOS;
 
+    public static $badHTML1 = '<iframe><meta http-equiv="refresh" content="1;/>';
+
     // Before any test
     public static function mySetUpBeforeClass()
     {
@@ -216,6 +218,10 @@ public function test_fromHTML()
         $emptyDoc = $doc::fromHTML('');
         $this->assertInstanceOf(get_class($doc), $emptyDoc);
         $this->assertEquals('', $emptyDoc->html);
+
+        // Bad HTML
+        $badDoc = hQueryTestSurrogate::fromHTML(self::$badHTML1);
+        $this->assertInstanceOf(get_class($doc), $badDoc);
     }
 
     // -----------------------------------------------------
@@ -347,6 +353,14 @@ public function test_find()
         $b = $edoc->find('body');
         $this->assertEquals(1, count($b));
 
+        // 6)
+        $bdoc = hQueryTestSurrogate::fromHTML(self::$badHTML1, self::$baseUrl . 'index.html');
+        $a = $bdoc->find('iframe');
+        $this->assertEquals(1, count($a));
+
+        $b = $bdoc->find('meta');
+        $this->assertEquals(1, count($b));
+
         return $ff;
     }
 
