Add support of multiple pseudoclasses

Imangazaliev · Jul 26, 2021 · 8f5200d · 8f5200d
1 parent dda9476
commit 8f5200d
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 16 deletions.
diff --git a/src/DiDom/Query.php b/src/DiDom/Query.php
@@ -292,13 +292,15 @@ public static function buildXpath(array $segments, $prefix = '//')
         }
 
         // if the pseudo class specified
-        if (isset($segments['pseudo'])) {
-            $expression = isset($segments['pseudo-expression']) ? trim($segments['pseudo-expression']) : '';
+        if (array_key_exists('pseudo', $segments)) {
+            foreach ($segments['pseudo'] as $pseudo) {
+                $expression = $pseudo['expression'] !== null ? $pseudo['expression'] : '';
 
-            $parameters = explode(',', $expression);
-            $parameters = array_map('trim', $parameters);
+                $parameters = explode(',', $expression);
+                $parameters = array_map('trim', $parameters);
 
-            $attributes[] = self::convertPseudo($segments['pseudo'], $tagName, $parameters);
+                $attributes[] = self::convertPseudo($pseudo['type'], $tagName, $parameters);
+            }
         }
 
         if (count($attributes) === 0 && ! isset($segments['tag'])) {
@@ -511,10 +513,15 @@ public static function getSegments($selector)
 
         // if the pseudo class specified
         if (isset($segments['pseudo']) && $segments['pseudo'] !== '') {
-            $result['pseudo'] = $segments['pseudo'];
+            preg_match_all('/:(?P<type>[\w\-]+)(?:\((?P<expr>[^\)]+)\))?/', $segments['pseudo'], $pseudoClasses);
+
+            $result['pseudo'] = [];
 
-            if (isset($segments['pseudoExpr']) && $segments['pseudoExpr'] !== '') {
-                $result['pseudo-expression'] = $segments['pseudoExpr'];
+            foreach ($pseudoClasses['type'] as $index => $pseudoType) {
+                $result['pseudo'][] = [
+                    'type' => $pseudoType,
+                    'expression' => $pseudoClasses['expr'][$index] !== '' ? $pseudoClasses['expr'][$index] : null,
+                ];
             }
         }
 
@@ -532,9 +539,9 @@ private static function getSelectorRegex()
         $id = '(?:#(?P<id>[\w|\-]+))?';
         $classes = '(?P<classes>\.[\w|\-|\.]+)*';
         $attrs = '(?P<attrs>(?:\[.+?\])*)?';
-        $pseudoType = '(?P<pseudo>[\w\-]+)';
-        $pseudoExpr = '(?:\((?P<pseudoExpr>[^\)]+)\))';
-        $pseudo = '(?::' . $pseudoType . $pseudoExpr . '?)?';
+        $pseudoType = '[\w\-]+';
+        $pseudoExpr = '(?:\([^\)]+\))?';
+        $pseudo = '(?P<pseudo>(?::' . $pseudoType  . $pseudoExpr . ')+)?';
         $rel = '\s*(?P<rel>>)?';
 
         return '/' . $tag . $id . $classes . $attrs . $pseudo . $rel . '/is';

diff --git a/tests/QueryTest.php b/tests/QueryTest.php
@@ -258,6 +258,7 @@ public function compileCssTests()
             ['li:first-child', '//li[position() = 1]'],
             ['li:last-child', '//li[position() = last()]'],
             ['*:not(a[href*="example.com"])', '//*[not(self::a[contains(@href, "example.com")])]'],
+            ['*:not(a[href*="example.com"]):not(.foo)', '//*[(not(self::a[contains(@href, "example.com")])) and (not(self::*[contains(concat(" ", normalize-space(@class), " "), " foo ")]))]'],
             ['ul:empty', '//ul[count(descendant::*) = 0]'],
             ['ul:not-empty', '//ul[count(descendant::*) > 0]'],
             ['li:nth-child(odd)', '//*[(name()="li") and (position() mod 2 = 1 and position() >= 1)]'],
@@ -388,8 +389,8 @@ public function buildXpathTests()
             ['tag' => 'a', 'attributes' => ['href*' => 'example']],
             ['tag' => 'a', 'attributes' => ['href!' => 'http://foo.com/']],
             ['tag' => 'script', 'attributes' => ['!src' => null]],
-            ['tag' => 'li', 'pseudo' => 'first-child'],
-            ['tag' => '*', 'id' => 'id', 'classes' => ['foo'], 'attributes' => ['name' => 'value'], 'pseudo' => 'first-child', 'rel' => '>'],
+            ['tag' => 'li', 'pseudo' => [['type' => 'first-child', 'expression' => null]]],
+            ['tag' => '*', 'id' => 'id', 'classes' => ['foo'], 'attributes' => ['name' => 'value'], 'pseudo' => [['type' => 'first-child', 'expression' => null]], 'rel' => '>'],
         ];
 
         $parameters = [];
@@ -416,10 +417,10 @@ public function getSegmentsTests()
             ['selector' => 'a[href=http://example.com/][title=Example Domain]', 'tag' => 'a', 'attributes' => ['href' => 'http://example.com/', 'title' => 'Example Domain']],
             ['selector' => 'a[href=http://example.com/][href=http://example.com/404]', 'tag' => 'a', 'attributes' => ['href' => 'http://example.com/404']],
             ['selector' => 'a[href^=https]', 'tag' => 'a', 'attributes' => ['href^' => 'https']],
-            ['selector' => 'li:first-child', 'tag' => 'li', 'pseudo' => 'first-child'],
+            ['selector' => 'li:first-child', 'tag' => 'li', 'pseudo' => [['type' => 'first-child', 'expression' => null]]],
             ['selector' => 'ul >', 'tag' => 'ul', 'rel' => '>'],
-            ['selector' => '#id.foo[name=value]:first-child >', 'id' => 'id', 'classes' => ['foo'], 'attributes' => ['name' => 'value'], 'pseudo' => 'first-child', 'rel' => '>'],
-            ['selector' => 'li.bar:nth-child(2n)', 'tag' => 'li', 'classes' => ['bar'], 'pseudo' => 'nth-child', 'pseudo-expression' => '2n'],
+            ['selector' => '#id.foo[name=value]:first-child >', 'id' => 'id', 'classes' => ['foo'], 'attributes' => ['name' => 'value'], 'pseudo' => [['type' => 'first-child', 'expression' => null]], 'rel' => '>'],
+            ['selector' => 'li.bar:nth-child(2n)', 'tag' => 'li', 'classes' => ['bar'], 'pseudo' => [['type' => 'nth-child', 'expression' => '2n']]],
         ];
 
         $parameters = [];