Merge pull request minkphp#720 from klausi/xpath-manipulator-union

Allow pipes (|) in attributes by implementing better XPath parsing
jhedstrom · Oct 14, 2016 · 39e3309 · 39e3309
2 parents 64e3a4b + 4846634
commit 39e3309
Show file tree

Hide file tree

Showing 3 changed files with 50 additions and 5 deletions.
diff --git a/src/Selector/Xpath/Manipulator.php b/src/Selector/Xpath/Manipulator.php
@@ -44,7 +44,7 @@ public function prepend($xpath, $prefix)
         }
 
         // Split any unions into individual expressions.
-        foreach (preg_split(self::UNION_PATTERN, $xpath) as $expression) {
+        foreach ($this->splitUnionParts($xpath) as $expression) {
             $expression = trim($expression);
             $parenthesis = '';
 
@@ -66,4 +66,44 @@ public function prepend($xpath, $prefix)
 
         return implode(' | ', $expressions);
     }
+
+    /**
+     * Splits the XPath into parts that are separated by the union operator.
+     *
+     * @param string $xpath
+     *
+     * @return string[]
+     */
+    private function splitUnionParts($xpath)
+    {
+        // Split any unions into individual expressions. We need to iterate
+        // through the string to correctly parse opening/closing quotes and
+        // braces which is not possible with regular expressions.
+        $unionParts = array();
+        $inSingleQuotedString = false;
+        $inDoubleQuotedString = false;
+        $openedBrackets = 0;
+        $lastUnion = 0;
+        for ($i = 0; $i < strlen($xpath); $i++) {
+            $char = $xpath[$i];
+
+            if ($char === "'" && !$inDoubleQuotedString) {
+                $inSingleQuotedString = !$inSingleQuotedString;
+            } elseif ($char === '"' && !$inSingleQuotedString) {
+                $inDoubleQuotedString = !$inDoubleQuotedString;
+            } elseif (!$inSingleQuotedString && !$inDoubleQuotedString) {
+                if ($char === '[') {
+                    $openedBrackets++;
+                } elseif ($char === ']') {
+                    $openedBrackets--;
+                } elseif ($char === '|' && $openedBrackets === 0) {
+                    $unionParts[] = substr($xpath, $lastUnion, $i - $lastUnion);
+                    $lastUnion = $i + 1;
+                }
+            }
+        }
+        $unionParts[] = substr($xpath, $lastUnion);
+        return $unionParts;
+    }
+
 }
diff --git a/tests/Element/NodeElementTest.php b/tests/Element/NodeElementTest.php
@@ -553,8 +553,8 @@ public function testSubmitForm()
     public function testFindAllUnion()
     {
         $node = new NodeElement('some_xpath', $this->session);
-        $xpath = "some_tag1 | some_tag2[@foo =\n 'bar|'']\n | some_tag3[foo | bar]";
-        $expected = "some_xpath/some_tag1 | some_xpath/some_tag2[@foo =\n 'bar|''] | some_xpath/some_tag3[foo | bar]";
+        $xpath = "some_tag1 | some_tag2[@foo =\n 'bar|']\n | some_tag3[foo | bar]";
+        $expected = "some_xpath/some_tag1 | some_xpath/some_tag2[@foo =\n 'bar|'] | some_xpath/some_tag3[foo | bar]";
 
         $this->driver
             ->expects($this->exactly(1))

diff --git a/tests/Selector/Xpath/ManipulatorTest.php b/tests/Selector/Xpath/ManipulatorTest.php
@@ -56,8 +56,13 @@ public function getPrependedXpath()
             ),
             'multiline' => array(
                 'some_xpath',
-                "some_tag1 | some_tag2[@foo =\n 'bar|'']\n | some_tag3[foo | bar]",
-                "some_xpath/some_tag1 | some_xpath/some_tag2[@foo =\n 'bar|''] | some_xpath/some_tag3[foo | bar]",
+                "some_tag1 | some_tag2[@foo =\n 'bar|']\n | some_tag3[foo | bar]",
+                "some_xpath/some_tag1 | some_xpath/some_tag2[@foo =\n 'bar|'] | some_xpath/some_tag3[foo | bar]",
+            ),
+            'containing pipe' => array(
+                'some_xpath',
+                "some_tag[(contains(normalize-space(string(.)), 'foo|bar') | other_tag[contains(./@some_attribute, 'foo|bar')])]",
+                "some_xpath/some_tag[(contains(normalize-space(string(.)), 'foo|bar') | other_tag[contains(./@some_attribute, 'foo|bar')])]",
             ),
         );
     }