allow all kinds of chars in selector, fixes #4

Author: Rct567

src/Rct567/DomQuery/CssToXpath.php

@@ -164,12 +164,12 @@ function ($matches) {
             $segment->attribute_filters[] = $matches[2];
         }
 
-        while (preg_match('/(.*)\.([a-z]+[a-z0-9\-\_]*)$/i', $segment->selector, $matches)) { // class selector
+        while (preg_match('/(.*)\.([^\.\#]+)$/i', $segment->selector, $matches)) { // class selector
             $segment->selector = $matches[1];
             $segment->attribute_filters[] = 'class~="'.$matches[2].'"';
         }
 
-        while (preg_match('/(.*)\#([a-z]+[a-z0-9\-\_]*)$/i', $segment->selector, $matches)) { // id selector
+        while (preg_match('/(.*)\#([^\.\#]+)$/i', $segment->selector, $matches)) { // id selector
             $segment->selector = $matches[1];
             $segment->attribute_filters[] = 'id="'.$matches[2].'"';
         }

tests/Rct567/DomQuery/Tests/DomQuerySelectorsTest.php

@@ -40,6 +40,7 @@ public function testCssToXpath()
             'div :header' => '//div//*[self::h1 or self::h2 or self::h3 or self::h5 or self::h5 or self::h6]',
             ':odd' => '//*[position() mod 2 = 0]',
             '.h' => '//*[contains(concat(\' \', normalize-space(@class), \' \'), \' h \')]',
+            '.😾-_😾' => '//*[contains(concat(\' \', normalize-space(@class), \' \'), \' 😾-_😾 \')]',
             '.hidden' => '//*[contains(concat(\' \', normalize-space(@class), \' \'), \' hidden \')]',
             '.hidden-something' => '//*[contains(concat(\' \', normalize-space(@class), \' \'), \' hidden-something \')]',
             'a.hidden[href]' => '//a[contains(concat(\' \', normalize-space(@class), \' \'), \' hidden \')][@href]',
@@ -127,6 +128,16 @@ public function testIdSelector()
         $this->assertEquals(1, $dom->find('#here')->length);
     }
 
+    /*
+     * Test emoji as id selector
+     */
+    public function testEmojiIdSelector()
+    {
+        $dom = new DomQuery('<div><a>1</a><b id="🐝">2</b></div><a >3</a>');
+        $this->assertEquals('2', $dom->find('#🐝')->text());
+        $this->assertEquals(1, $dom->find('#🐝')->length);
+    }
+
     /*
      * Test id selector with meta character
      */
@@ -228,6 +239,16 @@ public function testClassSelectorWithUnderscore()
         $this->assertEquals('3', $dom->find('.monkey-moon')->text());
     }
 
+    /*
+     * Test emoji as class selector
+     */
+    public function testEmojiClassSelector()
+    {
+        $dom = new DomQuery('<div><a>1</a><b class="🐝">2</b></div><a>3</a>');
+        $this->assertEquals('2', $dom->find('.🐝')->text());
+        $this->assertEquals(1, $dom->find('b.🐝')->length);
+    }
+
     /*
      * Test not filter selector
      */