[BUGFIX] Parse attribute selectors

Don't treat spaces or characters like `~` within them as a stop character
representing a combinator.

The bug was introduced by #1496 and related changes that split selector
representation and parsing into compound selectors and combinators.

Addresses some of the issues in #1533.

Author: JakeQZ

CHANGELOG.md

@@ -27,7 +27,7 @@ Please also have a look at our
   `getPropertyName()` (#1506)
 - `Selector` is now represented as a sequence of `Selector\Component` objects
   which can be accessed via `getComponents()`, manipulated individually, or set
-  via `setComponents()` (#1478, #1486, #1487, #1488, #1494, #1496)
+  via `setComponents()` (#1478, #1486, #1487, #1488, #1494, #1496, #1536)
 - `Selector::setSelector()` and `Selector` constructor will now throw exception
   upon provision of an invalid selectior (#1498, #1502)
 - Clean up extra whitespace in CSS selector (#1398)

src/Property/Selector/CompoundSelector.php

@@ -27,6 +27,8 @@ class CompoundSelector implements Component
         '"',
         '(',
         ')',
+        '[',
+        ']',
         ',',
         ' ',
         "\t",
@@ -46,7 +48,7 @@ class CompoundSelector implements Component
             (?:
                 (?:
                     # any sequence of valid unescaped characters, except quotes
-                    [a-zA-Z0-9\\x{00A0}-\\x{FFFF}_^$|*=\\[\\]()\\-\\.:#,\\s]++
+                    [a-zA-Z0-9\\x{00A0}-\\x{FFFF}_^$|*~=\\[\\]()\\-\\.:#,\\s]++
                     |
                     # one or more escaped characters
                     (?:\\\\.)++
@@ -100,6 +102,7 @@ public static function parse(ParserState $parserState, array &$comments = []): s
         $selectorParts = [];
         $stringWrapperCharacter = null;
         $functionNestingLevel = 0;
+        $isWithinAttribute = false;
 
         while (true) {
             $selectorParts[] = $parserState->consumeUntil(self::PARSER_STOP_CHARACTERS, false, false, $comments);
@@ -140,6 +143,32 @@ public static function parse(ParserState $parserState, array &$comments = []): s
                         --$functionNestingLevel;
                     }
                     break;
+                case '[':
+                    if (!\is_string($stringWrapperCharacter)) {
+                        if ($isWithinAttribute) {
+                            throw new UnexpectedTokenException(
+                                'anything but',
+                                '[',
+                                'literal',
+                                $parserState->currentLine()
+                            );
+                        }
+                        $isWithinAttribute = true;
+                    }
+                    break;
+                case ']':
+                    if (!\is_string($stringWrapperCharacter)) {
+                        if (!$isWithinAttribute) {
+                            throw new UnexpectedTokenException(
+                                'anything but',
+                                ']',
+                                'literal',
+                                $parserState->currentLine()
+                            );
+                        }
+                        $isWithinAttribute = false;
+                    }
+                    break;
                 case '{':
                     // The fallthrough is intentional.
                 case '}':
@@ -162,7 +191,7 @@ public static function parse(ParserState $parserState, array &$comments = []): s
                 case '+':
                     // The fallthrough is intentional.
                 case '~':
-                    if (!\is_string($stringWrapperCharacter) && $functionNestingLevel === 0) {
+                    if (!\is_string($stringWrapperCharacter) && $functionNestingLevel === 0 && !$isWithinAttribute) {
                         break 2;
                     }
                     break;

tests/Unit/Property/Selector/CompoundSelectorTest.php

@@ -57,6 +57,17 @@ public static function provideCompoundSelectorAndSpecificity(): array
             '`not` with multiple arguments' => [':not(#your-mug, .their-mug)', 110],
             'attribute with `"`' => ['[alt="{}()[]\\"\',"]', 10],
             'attribute with `\'`' => ['[alt=\'{}()[]"\\\',\']', 10],
+            // TODO, broken: specificity should be 11, but the calculator doesn't realize the `#` is in a string.
+            'attribute with `^=`' => ['a[href^="#"]', 111],
+            'attribute with `*=`' => ['a[href*="example"]', 11],
+            // TODO, broken: specificity should be 11, but the calculator doesn't realize the `.` is in a string.
+            'attribute with `$=`' => ['a[href$=".org"]', 21],
+            'attribute with `~=`' => ['span[title~="bonjour"]', 11],
+            'attribute with `|=`' => ['[lang|="en"]', 10],
+            // TODO, broken: specificity should be 11, but the calculator doesn't realize the `i` is in an attribute.
+            'attribute with case insensitive modifier' => ['a[href*="insensitive" i]', 12],
+            // TODO, broken: specificity should be 21, but the calculator doesn't realize the `.` is in a string.
+            'multiple attributes' => ['a[href^="https://"][href$=".org"]', 31],
         ];
     }
 
@@ -173,6 +184,8 @@ public static function provideInvalidCompoundSelectorForParse(): array
             'attribute value missing closing double quote' => ['a[href="#top]'],
             'attribute value with mismatched quotes, single quote opening' => ['a[href=\'#top"]'],
             'attribute value with mismatched quotes, double quote opening' => ['a[href="#top\']'],
+            'attribute value with extra `[`' => ['a[[href="#top"]'],
+            'attribute value with extra `]`' => ['a[href="#top"]]'],
         ];
     }