support non-latin characters in matches-path

browser/brave_shields/ad_block_service_browsertest.cc

@@ -2827,10 +2827,14 @@ IN_PROC_BROWSER_TEST_F(AdBlockServiceTest, ProceduralFilterMatchesPath) {
       "p.odd:matches-path(cosmetic_filtering.html)\n"
       "a.com##:matches-path(/c[aeiou]smetic\\_[a-z]{9}/) section "
       ".positive-regex-case p.odd\n"
-      "a.com##section .negative-case:matches-path(/some-other-page.html)");
+      "a.com##section .negative-case:matches-path(/some-other-page.html)\n"
+      "a.com##:matches-path(😎) section .positive-unicode-case p.odd");
 
-  GURL tab_url =
-      embedded_test_server()->GetURL("a.com", "/cosmetic_filtering.html");
+  GURL::Replacements replacements;
+  replacements.SetQueryStr("😎");
+  GURL tab_url = embedded_test_server()
+                     ->GetURL("a.com", "/cosmetic_filtering.html")
+                     .ReplaceComponents(replacements);
   NavigateToURL(tab_url);
 
   content::WebContents* contents = web_contents();

components/cosmetic_filters/resources/data/procedural_filters.ts

@@ -123,6 +123,7 @@ const _extractKeyMatchRuleFromStr = (text: string): [TextMatchRule, number] => {
 }
 
 const _extractValueMatchRuleFromStr = (text: string,
+                                       uriEncode = false,
                                        needlePosition = 0): TextMatchRule => {
   const isQuotedCase = text[needlePosition] === '"'
   let endIndex
@@ -140,7 +141,10 @@ const _extractValueMatchRuleFromStr = (text: string,
     endIndex = text.length
   }
 
-  const testCaseStr = text.slice(needlePosition, endIndex)
+  let testCaseStr = text.slice(needlePosition, endIndex)
+  if (uriEncode) {
+    testCaseStr = testCaseStr.replace(/\P{ASCII}/gu, c => encodeURIComponent(c))
+  }
   const testCaseFunc = _testMatches.bind(undefined, testCaseStr)
   return testCaseFunc
 }
@@ -159,7 +163,7 @@ const _extractValueMatchRuleFromStr = (text: string,
 // }
 const _parseKeyValueMatchRules = (arg: string): KeyValueMatchRules => {
   const [keyMatchRule, needlePos] = _extractKeyMatchRuleFromStr(arg)
-  const valueMatchRule = _extractValueMatchRuleFromStr(arg, needlePos)
+  const valueMatchRule = _extractValueMatchRuleFromStr(arg, false, needlePos)
   return [keyMatchRule, valueMatchRule]
 }
 
@@ -387,7 +391,7 @@ const operatorMatchesMedia = (instruction: string,
 const operatorMatchesPath = (instruction: string,
                              element: HTMLElement): OperatorResult => {
   const pathAndQuery = W.location.pathname + W.location.search
-  const matchRule = _extractValueMatchRuleFromStr(instruction)
+  const matchRule = _extractValueMatchRuleFromStr(instruction, true)
   return matchRule(pathAndQuery) ? [element] : []
 }
 

test/data/cosmetic_filtering.html

@@ -288,6 +288,13 @@ const classes = (expectedClasses) =>
         <p data-expect="visible">Should not be hidden</p>
       </div>
 
+      <div class="positive-unicode-case">
+        <h2><code>:matches-path</code>: positive unicode case</h2>
+        <p class="odd" data-expect="hidden">Should be hidden</p>
+        <p class="even" data-expect="visible">Should not be hidden</p>
+        <p data-expect="visible">Should not be hidden</p>
+      </div>
+
       <div class="negative-case">
         <h2><code>:matches-path</code>: negative case</h2>
         <p class="odd" data-expect="visible">Should not be hidden</p>