LibJS+LibUnicode: Make the collation punctation default locale-aware

Author: https://github.com/trflynn89
Commit: 78625c746d
Pull-request: https://github.com/LadybirdBrowser/ladybird/pull/1078
Reviewed-by: https://github.com/awesomekling

Userland/Libraries/LibJS/Runtime/Intl/CollatorConstructor.cpp

@@ -171,14 +171,16 @@ ThrowCompletionOr<NonnullGCPtr<Object>> CollatorConstructor::construct(FunctionO
     // 34. Set collator.[[Sensitivity]] to sensitivity.
     collator->set_sensitivity(sensitivity.as_string().utf8_string_view());
 
-    // FIXME: 35. Let defaultIgnorePunctuation be resolvedLocaleData.[[ignorePunctuation]].
+    // 35. Let defaultIgnorePunctuation be resolvedLocaleData.[[ignorePunctuation]].
-    auto default_ignore_punctuation = false;
+    // NOTE: We do not acquire the default [[ignorePunctuation]] here. Instead, we default the option to null,
+    //       and let LibUnicode fill in the default value if an override was not provided here.
 
     // 36. Let ignorePunctuation be ? GetOption(options, "ignorePunctuation", boolean, empty, defaultIgnorePunctuation).
-    auto ignore_punctuation = TRY(get_option(vm, *options, vm.names.ignorePunctuation, OptionType::Boolean, {}, default_ignore_punctuation));
+    auto ignore_punctuation_value = TRY(get_option(vm, *options, vm.names.ignorePunctuation, OptionType::Boolean, {}, Empty {}));
 
-    // 37. Set collator.[[IgnorePunctuation]] to ignorePunctuation.
+    Optional<bool> ignore_punctuation;
-    collator->set_ignore_punctuation(ignore_punctuation.as_bool());
+    if (!ignore_punctuation_value.is_undefined())
+        ignore_punctuation = ignore_punctuation_value.as_bool();
 
     // Non-standard, create an ICU collator for this Intl object.
     auto icu_collator = Unicode::Collator::create(
@@ -188,9 +190,12 @@ ThrowCompletionOr<NonnullGCPtr<Object>> CollatorConstructor::construct(FunctionO
         collator->sensitivity(),
         collator->case_first(),
         collator->numeric(),
-        collator->ignore_punctuation());
+        ignore_punctuation);
     collator->set_collator(move(icu_collator));
 
+    // 37. Set collator.[[IgnorePunctuation]] to ignorePunctuation.
+    collator->set_ignore_punctuation(collator->collator().ignore_punctuation());
+
     // 38. Return collator.
     return collator;
 }

Userland/Libraries/LibJS/Tests/builtins/Intl/Collator/Collator.prototype.resolvedOptions.js

@@ -38,9 +38,15 @@ describe("correct behavior", () => {
         const en1 = new Intl.Collator("en");
         expect(en1.resolvedOptions().ignorePunctuation).toBeFalse();
 
+        const th1 = new Intl.Collator("th");
+        expect(th1.resolvedOptions().ignorePunctuation).toBeTrue();
+
         [true, false].forEach(ignorePunctuation => {
             const en2 = new Intl.Collator("en", { ignorePunctuation: ignorePunctuation });
             expect(en2.resolvedOptions().ignorePunctuation).toBe(ignorePunctuation);
+
+            const th2 = new Intl.Collator("th", { ignorePunctuation: ignorePunctuation });
+            expect(th2.resolvedOptions().ignorePunctuation).toBe(ignorePunctuation);
         });
     });
 

Userland/Libraries/LibUnicode/Collator.cpp

@@ -129,6 +129,16 @@ static constexpr UColAttributeValue icu_case_first(CaseFirst case_first)
     VERIFY_NOT_REACHED();
 }
 
+static bool ignore_punctuation_for_collator(icu::Collator const& collator)
+{
+    UErrorCode status = U_ZERO_ERROR;
+
+    auto attribute = collator.getAttribute(UCOL_ALTERNATE_HANDLING, status);
+    VERIFY(icu_success(status));
+
+    return attribute == UCOL_SHIFTED;
+}
+
 class CollatorImpl : public Collator {
 public:
     explicit CollatorImpl(NonnullOwnPtr<icu::Collator> collator)
@@ -155,6 +165,11 @@ public:
         VERIFY_NOT_REACHED();
     }
 
+    virtual bool ignore_punctuation() const override
+    {
+        return ignore_punctuation_for_collator(*m_collator);
+    }
+
 private:
     NonnullOwnPtr<icu::Collator> m_collator;
 };
@@ -166,7 +181,7 @@ NonnullOwnPtr<Collator> Collator::create(
     Sensitivity sensitivity,
     CaseFirst case_first,
     bool numeric,
-    bool ignore_punctuation)
+    Optional<bool> ignore_punctuation)
 {
     UErrorCode status = U_ZERO_ERROR;
 
@@ -183,11 +198,14 @@ NonnullOwnPtr<Collator> Collator::create(
         VERIFY(icu_success(status));
     };
 
+    if (!ignore_punctuation.has_value())
+        ignore_punctuation = ignore_punctuation_for_collator(*collator);
+
     set_attribute(UCOL_STRENGTH, icu_sensitivity(sensitivity));
     set_attribute(UCOL_CASE_LEVEL, sensitivity == Sensitivity::Case ? UCOL_ON : UCOL_OFF);
     set_attribute(UCOL_CASE_FIRST, icu_case_first(case_first));
     set_attribute(UCOL_NUMERIC_COLLATION, numeric ? UCOL_ON : UCOL_OFF);
-    set_attribute(UCOL_ALTERNATE_HANDLING, ignore_punctuation ? UCOL_SHIFTED : UCOL_NON_IGNORABLE);
+    set_attribute(UCOL_ALTERNATE_HANDLING, *ignore_punctuation ? UCOL_SHIFTED : UCOL_NON_IGNORABLE);
     set_attribute(UCOL_NORMALIZATION_MODE, UCOL_ON);
 
     return adopt_own(*new CollatorImpl(move(collator)));

Userland/Libraries/LibUnicode/Collator.h

@@ -44,7 +44,7 @@ public:
         Sensitivity,
         CaseFirst,
         bool numeric,
-        bool ignore_punctuation);
+        Optional<bool> ignore_punctuation);
 
     virtual ~Collator() = default;
 
@@ -55,6 +55,8 @@ public:
     };
     virtual Order compare(StringView, StringView) const = 0;
 
+    virtual bool ignore_punctuation() const = 0;
+
 protected:
     Collator() = default;
 };