Ensure that quoted strings are matched correctly

We need to match the initial quote, and then all characters up to but
not including a matching final quote, taking into account escaped quotes
as well.

This patch uses a technique learned from http://blog.stevenlevithan.com/archives/match-quoted-string
to do so, and updates existing regular expressions, including the one
modified for #21, to properly match quoted strings, introducing several
new test cases to verify the changes.

src/Document/Query.php

@@ -88,7 +88,7 @@ public static function cssToXpath($path)
 
         // Arbitrary attribute value contains whitespace
         $path = preg_replace_callback(
-            '/\[\S+?(["\'])(.+?)\1\]/',
+            '/\[\S+?([\'"])((?:(?!\1)[^\\\]|\\.)*)\1\]/',
             function ($matches) {
                 return str_replace($matches[2], preg_replace('/\s+/', '\s', $matches[2]), $matches[0]);
             },
@@ -147,29 +147,29 @@ protected static function _tokenize($expression)
 
         // arbitrary attribute strict equality
         $expression = preg_replace_callback(
-            '|\[@?([a-z0-9_-]+)=[\'"]([^\'"]+)[\'"]\]|i',
+            '/\[@?([a-z0-9_-]+)=([\'"])((?:(?!\2)[^\\\]|\\.)*)\2\]/i',
             function ($matches) {
-                return '[@' . strtolower($matches[1]) . "='" . $matches[2] . "']";
+                return sprintf("[@%s='%s']", strtolower($matches[1]), str_replace("'", "\\'", $matches[3]));
             },
             $expression
         );
 
         // arbitrary attribute contains full word
         $expression = preg_replace_callback(
-            '|\[([a-z0-9_-]+)~=[\'"]([^\'"]+)[\'"]\]|i',
+            '/\[([a-z0-9_-]+)~=([\'"])((?:(?!\2)[^\\\]|\\.)*)\2\]/i',
             function ($matches) {
                 return "[contains(concat(' ', normalize-space(@" . strtolower($matches[1]) . "), ' '), ' "
-                     . $matches[2] . " ')]";
+                     . $matches[3] . " ')]";
             },
             $expression
         );
 
         // arbitrary attribute contains specified content
         $expression = preg_replace_callback(
-            '|\[([a-z0-9_-]+)\*=[\'"]([^\'"]+)[\'"]\]|i',
+            '/\[([a-z0-9_-]+)\*=([\'"])((?:(?!\2)[^\\\]|\\.)*)\2\]/i',
             function ($matches) {
                 return "[contains(@" . strtolower($matches[1]) . ", '"
-                     . $matches[2] . "')]";
+                     . $matches[3] . "')]";
             },
             $expression
         );

test/Document/QueryTest.php

@@ -173,9 +173,33 @@ public function testCanTransformWithAttributeAndDot()
         $this->assertEquals("//a[@href='http://example.com']", $test);
     }
 
-    public function testTransformNestedAttributeSelectors()
+    public function nestedAttributeSelectors()
+    {
+        return [
+            'with-double-quotes' => [
+                'select[name="foo"] option[selected="selected"]',
+                "//select[@name='foo']//option[@selected='selected']",
+            ],
+            'with-single-quotes' => [
+                "select[name='foo'] option[selected='selected']",
+                "//select[@name='foo']//option[@selected='selected']",
+            ],
+            'double-quotes-containing-single-quotes' => [
+                "select[name=\"f'oo\"] option[selected=\"sel'ected\"]",
+                "//select[@name='f\'oo']//option[@selected='sel\'ected']",
+            ],
+            'single-quotes-containing-double-quotes' => [
+                "select[name='f\"oo'] option[selected='sel\"ected']",
+                "//select[@name='f\"oo']//option[@selected='sel\"ected']",
+            ],
+        ];
+    }
+
+    /**
+     * @dataProvider nestedAttributeSelectors
+     */
+    public function testTransformNestedAttributeSelectors($selector, $expectedXpath)
     {
-        $test = Query::cssToXpath('select[name="foo"] option[selected="selected"]');
-        $this->assertEquals("//select[@name='foo']//option[@selected='selected']", $test);
+        $this->assertEquals($expectedXpath, Query::cssToXpath($selector));
     }
 }