LibWeb: Implement text-transform: capitalize

Author: https://github.com/Cr4xy
Commit: https://github.com/SerenityOS/serenity/commit/bbfe0d3a82
Pull-request: https://github.com/SerenityOS/serenity/pull/21266
Reviewed-by: https://github.com/trflynn89 ✅

Userland/Libraries/LibUnicode/CharacterTypes.cpp

@@ -53,10 +53,10 @@ ErrorOr<DeprecatedString> to_unicode_uppercase_full(StringView string, Optional<
     return builder.to_deprecated_string();
 }
 
-ErrorOr<String> to_unicode_titlecase_full(StringView string, Optional<StringView> const& locale)
+ErrorOr<String> to_unicode_titlecase_full(StringView string, Optional<StringView> const& locale, TrailingCodePointTransformation trailing_code_point_transformation)
 {
     StringBuilder builder;
-    TRY(Detail::build_titlecase_string(Utf8View { string }, builder, locale));
+    TRY(Detail::build_titlecase_string(Utf8View { string }, builder, locale, trailing_code_point_transformation));
     return builder.to_string();
 }
 

Userland/Libraries/LibUnicode/CharacterTypes.h

@@ -34,6 +34,13 @@ struct BlockName {
     StringView display_name;
 };
 
+enum class TrailingCodePointTransformation : u8 {
+    // Default behaviour; Puts the first typographic letter unit of each word, if lowercase, in titlecase; the other characters in lowercase.
+    Lowercase,
+    // Puts the first typographic letter unit of each word, if lowercase, in titlecase; other characters are unaffected. (https://drafts.csswg.org/css-text/#valdef-text-transform-capitalize)
+    PreserveExisting,
+};
+
 Optional<DeprecatedString> code_point_display_name(u32 code_point);
 Optional<StringView> code_point_block_display_name(u32 code_point);
 Optional<StringView> code_point_abbreviation(u32 code_point);
@@ -50,7 +57,7 @@ u32 to_unicode_titlecase(u32 code_point);
 
 ErrorOr<DeprecatedString> to_unicode_lowercase_full(StringView, Optional<StringView> const& locale = {});
 ErrorOr<DeprecatedString> to_unicode_uppercase_full(StringView, Optional<StringView> const& locale = {});
-ErrorOr<String> to_unicode_titlecase_full(StringView, Optional<StringView> const& locale = {});
+ErrorOr<String> to_unicode_titlecase_full(StringView, Optional<StringView> const& locale = {}, TrailingCodePointTransformation trailing_code_point_transformation = TrailingCodePointTransformation::Lowercase);
 ErrorOr<String> to_unicode_casefold_full(StringView);
 
 Optional<GeneralCategory> general_category_from_string(StringView);

Userland/Libraries/LibUnicode/Forward.h

@@ -19,6 +19,7 @@ enum class Property : u8;
 enum class Script : u8;
 enum class SentenceBreakProperty : u8;
 enum class WordBreakProperty : u8;
+enum class TrailingCodePointTransformation : u8;
 
 struct CodePointDecomposition;
 struct CurrencyCode;

Userland/Libraries/LibUnicode/String.cpp

@@ -8,6 +8,7 @@
 #include <AK/StringBuilder.h>
 #include <AK/Utf32View.h>
 #include <AK/Utf8View.h>
+#include <LibUnicode/CharacterTypes.h>
 #include <LibUnicode/UnicodeUtils.h>
 
 // This file contains definitions of AK::String methods which require UCD data.
@@ -31,7 +32,7 @@ ErrorOr<String> String::to_uppercase(Optional<StringView> const& locale) const
 ErrorOr<String> String::to_titlecase(Optional<StringView> const& locale) const
 {
     StringBuilder builder;
-    TRY(Unicode::Detail::build_titlecase_string(code_points(), builder, locale));
+    TRY(Unicode::Detail::build_titlecase_string(code_points(), builder, locale, Unicode::TrailingCodePointTransformation::Lowercase));
     return builder.to_string();
 }
 

Userland/Libraries/LibUnicode/UnicodeUtils.cpp

@@ -266,7 +266,7 @@ ErrorOr<void> build_uppercase_string([[maybe_unused]] Utf8View code_points, [[ma
 }
 
 // https://www.unicode.org/versions/Unicode15.0.0/ch03.pdf#G34078
-ErrorOr<void> build_titlecase_string([[maybe_unused]] Utf8View code_points, [[maybe_unused]] StringBuilder& builder, [[maybe_unused]] Optional<StringView> const& locale)
+ErrorOr<void> build_titlecase_string([[maybe_unused]] Utf8View code_points, [[maybe_unused]] StringBuilder& builder, [[maybe_unused]] Optional<StringView> const& locale, TrailingCodePointTransformation trailing_code_point_transformation)
 {
 #if ENABLE_UNICODE_DATA
     // toTitlecase(X): Find the word boundaries in X according to Unicode Standard Annex #29,
@@ -317,8 +317,15 @@ ErrorOr<void> build_titlecase_string([[maybe_unused]] Utf8View code_points, [[ma
             boundary = code_point_offset + code_point_length;
         }
 
-        auto substring_to_lowercase = code_points.substring_view(boundary, *next_boundary - boundary);
-        TRY(build_lowercase_string(substring_to_lowercase, builder, locale));
+        auto remaining_code_points = code_points.substring_view(boundary, *next_boundary - boundary);
+        switch (trailing_code_point_transformation) {
+        case TrailingCodePointTransformation::Lowercase:
+            TRY(build_lowercase_string(remaining_code_points, builder, locale));
+            break;
+        case TrailingCodePointTransformation::PreserveExisting:
+            TRY(builder.try_append(remaining_code_points.as_string()));
+            break;
+        }
 
         boundary = *next_boundary;
     }

Userland/Libraries/LibUnicode/UnicodeUtils.h

@@ -17,7 +17,7 @@ namespace Unicode::Detail {
 
 ErrorOr<void> build_lowercase_string(Utf8View code_points, StringBuilder& builder, Optional<StringView> const& locale);
 ErrorOr<void> build_uppercase_string(Utf8View code_points, StringBuilder& builder, Optional<StringView> const& locale);
-ErrorOr<void> build_titlecase_string(Utf8View code_points, StringBuilder& builder, Optional<StringView> const& locale);
+ErrorOr<void> build_titlecase_string(Utf8View code_points, StringBuilder& builder, Optional<StringView> const& locale, TrailingCodePointTransformation trailing_code_point_transformation);
 ErrorOr<void> build_casefold_string(Utf8View code_points, StringBuilder& builder);
 Utf32View casefold_code_point(u32 const& code_point);
 

Userland/Libraries/LibWeb/Layout/TextNode.cpp

@@ -287,7 +287,9 @@ static ErrorOr<DeprecatedString> apply_text_transform(DeprecatedString const& st
         return string;
     case CSS::TextTransform::MathAuto:
         return apply_math_auto_text_transform(string);
-    case CSS::TextTransform::Capitalize:
+    case CSS::TextTransform::Capitalize: {
+        return TRY(Unicode::to_unicode_titlecase_full(string, {}, Unicode::TrailingCodePointTransformation::PreserveExisting)).to_deprecated_string();
+    }
     case CSS::TextTransform::FullSizeKana:
     case CSS::TextTransform::FullWidth:
         // FIXME: Implement these!