php-css-parser-commit-10a2501.patch

From fa139f65c5b098ae652c970b25e6eb03fc495eb4 Mon Sep 17 00:00:00 2001
From: Weston Ruter <weston@xwp.co>
Date: Wed, 25 Jul 2018 10:38:59 -0700
Subject: [PATCH 1/2] Fix parsing CSS selectors which contain commas

---
 .../CSS/RuleSet/DeclarationBlock.php          | 60 ++++++++++++++++++-
 tests/Sabberworm/CSS/ParserTest.php           |  6 ++
 tests/files/specificity.css                   |  4 +-
 3 files changed, 68 insertions(+), 2 deletions(-)

diff --git a/lib/Sabberworm/CSS/RuleSet/DeclarationBlock.php b/lib/Sabberworm/CSS/RuleSet/DeclarationBlock.php
index e18f5d8..26c2e12 100644
--- a/lib/Sabberworm/CSS/RuleSet/DeclarationBlock.php
+++ b/lib/Sabberworm/CSS/RuleSet/DeclarationBlock.php
@@ -28,7 +28,19 @@ public function setSelectors($mSelector) {
 		if (is_array($mSelector)) {
 			$this->aSelectors = $mSelector;
 		} else {
-			$this->aSelectors = explode(',', $mSelector);
+			list( $sSelectors, $aPlaceholders ) = $this->addSelectorExpressionPlaceholders( $mSelector );
+			if ( empty( $aPlaceholders ) ) {
+				$this->aSelectors = explode(',', $sSelectors);
+			} else {
+				$aSearches = array_keys( $aPlaceholders );
+				$aReplaces = array_values( $aPlaceholders );
+				$this->aSelectors = array_map(
+					function( $sSelector ) use ( $aSearches, $aReplaces ) {
+						return str_replace( $aSearches, $aReplaces, $sSelector );
+					},
+					explode(',', $sSelectors)
+				);
+			}
 		}
 		foreach ($this->aSelectors as $iKey => $mSelector) {
 			if (!($mSelector instanceof Selector)) {
@@ -37,6 +49,52 @@ public function setSelectors($mSelector) {
 		}
 	}
 
+	/**
+	 * Add placeholders for parenthetical/bracketed expressions in selectors which may contain commas that break exploding.
+	 *
+	 * This prevents a single selector like `.widget:not(.foo, .bar)` from erroneously getting parsed in setSelectors as
+	 * two selectors `.widget:not(.foo` and `.bar)`.
+	 *
+	 * @param string $sSelectors Selectors.
+	 * @return array First array value is the selectors with placeholders, and second value is the array of placeholders mapped to the original expressions.
+	 */
+	private function addSelectorExpressionPlaceholders( $sSelectors ) {
+		$iOffset = 0;
+		$aPlaceholders = array();
+
+		while ( preg_match( '/\(|\[/', $sSelectors, $aMatches, PREG_OFFSET_CAPTURE, $iOffset ) ) {
+			$sMatchString = $aMatches[0][0];
+			$iMatchOffset = $aMatches[0][1];
+			$iStyleLength = strlen( $sSelectors );
+			$iOpenParens  = 1;
+			$iStartOffset = $iMatchOffset + strlen( $sMatchString );
+			$iFinalOffset = $iStartOffset;
+			for ( ; $iFinalOffset < $iStyleLength; $iFinalOffset++ ) {
+				if ( '(' === $sSelectors[ $iFinalOffset ] || '[' === $sSelectors[ $iFinalOffset ] ) {
+					$iOpenParens++;
+				} elseif ( ')' === $sSelectors[ $iFinalOffset ] || ']' === $sSelectors[ $iFinalOffset ] ) {
+					$iOpenParens--;
+				}
+
+				// Found the end of the expression, so replace it with a placeholder.
+				if ( 0 === $iOpenParens ) {
+					$sMatchedExpr = substr( $sSelectors, $iMatchOffset, $iFinalOffset - $iMatchOffset + 1 );
+					$sPlaceholder = sprintf( '{placeholder:%d}', count( $aPlaceholders ) + 1 );
+					$aPlaceholders[ $sPlaceholder ] = $sMatchedExpr;
+
+					// Update the CSS to replace the matched calc() with the placeholder function.
+					$sSelectors = substr( $sSelectors, 0, $iMatchOffset ) . $sPlaceholder . substr( $sSelectors, $iFinalOffset + 1 );
+					// Update offset based on difference of length of placeholder vs original matched calc().
+					$iFinalOffset += strlen( $sPlaceholder ) - strlen( $sMatchedExpr );
+					break;
+				}
+			}
+			// Start matching at the next byte after the match.
+			$iOffset = $iFinalOffset + 1;
+		}
+		return array( $sSelectors, $aPlaceholders );
+	}
+
 	// remove one of the selector of the block
 	public function removeSelector($mSelector) {
 		if($mSelector instanceof Selector) {
diff --git a/tests/Sabberworm/CSS/ParserTest.php b/tests/Sabberworm/CSS/ParserTest.php
index 43c22e2..a5073be 100644
--- a/tests/Sabberworm/CSS/ParserTest.php
+++ b/tests/Sabberworm/CSS/ParserTest.php
@@ -148,6 +148,12 @@ function testSpecificity() {
 				case "li.green":
 					$this->assertSame(11, $oSelector->getSpecificity());
 					break;
+				case "div:not(.foo[title=\"a,b\"], .bar)":
+					$this->assertSame(31, $oSelector->getSpecificity());
+					break;
+				case "div[title=\"a,b\"]":
+					$this->assertSame(11, $oSelector->getSpecificity());
+					break;
 				default:
 					$this->fail("specificity: untested selector " . $oSelector->getSelector());
 			}
diff --git a/tests/files/specificity.css b/tests/files/specificity.css
index 82a2939..df03ff0 100644
--- a/tests/files/specificity.css
+++ b/tests/files/specificity.css
@@ -2,6 +2,8 @@
 #file,
 .help:hover,
 li.green,
-ol li::before {
+ol li::before,
+div:not(.foo[title="a,b"], .bar),
+div[title="a,b"] {
 	font-family: Helvetica;
 }

From 10a2501c119abafced3e4014aa3c0a3453a86f67 Mon Sep 17 00:00:00 2001
From: Weston Ruter <westonruter@google.com>
Date: Mon, 20 Apr 2020 13:53:44 -0700
Subject: [PATCH 2/2] Update SELECTOR_VALIDATION_RX to account for
 parenthetical groups

---
 lib/Sabberworm/CSS/Property/Selector.php | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/Sabberworm/CSS/Property/Selector.php b/lib/Sabberworm/CSS/Property/Selector.php
index bd04b88..0f1c643 100644
--- a/lib/Sabberworm/CSS/Property/Selector.php
+++ b/lib/Sabberworm/CSS/Property/Selector.php
@@ -43,6 +43,7 @@ class Selector {
 			[a-zA-Z0-9\x{00A0}-\x{FFFF}_^$|*="\'~\[\]()\-\s\.:#+>]* # any sequence of valid unescaped characters
 			(?:\\\\.)?                                              # a single escaped character
 			(?:([\'"]).*?(?<!\\\\)\2)?                              # a quoted text like [id="example"]
+			(?:\(.*?\))?                                            # an argument for pseudo selector like :not(a,b) or :lang(en,es).
 		)*
 	)$
 	/ux';