From f907a7dc386e7cee729b8d96b6287dd0f1af1b72 Mon Sep 17 00:00:00 2001
From: Timothy Flynn <trflynn89@pm.me>
Date: Mon, 30 Aug 2021 15:51:08 -0400
Subject: [PATCH] LibUnicode: Canonicalize the subtag "yes" to "true"

---
 Tests/LibUnicode/TestUnicodeLocale.cpp   |  6 ++++++
 Userland/Libraries/LibUnicode/Locale.cpp | 12 ++++++++++++
 2 files changed, 18 insertions(+)

diff --git a/Tests/LibUnicode/TestUnicodeLocale.cpp b/Tests/LibUnicode/TestUnicodeLocale.cpp
index 45cd4170f6e..5a7857d6bb5 100644
--- a/Tests/LibUnicode/TestUnicodeLocale.cpp
+++ b/Tests/LibUnicode/TestUnicodeLocale.cpp
@@ -302,6 +302,10 @@ TEST_CASE(canonicalize_unicode_locale_id)
     test("EN-U-CCC-BBB-2K-AAA-1K-BBB"sv, "en-u-bbb-ccc-1k-bbb-2k-aaa"sv);
     test("en-u-1k-true"sv, "en-u-1k"sv);
     test("EN-U-1K-TRUE"sv, "en-u-1k"sv);
+    test("en-u-kb-yes"sv, "en-u-kb"sv);
+    test("EN-U-KB-YES"sv, "en-u-kb"sv);
+    test("en-u-ka-yes"sv, "en-u-ka-yes"sv);
+    test("EN-U-KA-YES"sv, "en-u-ka-yes"sv);
 
     test("en-t-en"sv, "en-t-en"sv);
     test("EN-T-EN"sv, "en-t-en"sv);
@@ -315,6 +319,8 @@ TEST_CASE(canonicalize_unicode_locale_id)
     test("EN-T-EN-K2-BBB-K1-AAA"sv, "en-t-en-k1-aaa-k2-bbb"sv);
     test("en-t-k1-true"sv, "en-t-k1-true"sv);
     test("EN-T-K1-TRUE"sv, "en-t-k1-true"sv);
+    test("en-t-k1-yes"sv, "en-t-k1-yes"sv);
+    test("EN-T-K1-YES"sv, "en-t-k1-yes"sv);
 
     test("en-0-aaa"sv, "en-0-aaa"sv);
     test("EN-0-AAA"sv, "en-0-aaa"sv);
diff --git a/Userland/Libraries/LibUnicode/Locale.cpp b/Userland/Libraries/LibUnicode/Locale.cpp
index 6cbeb241b1e..dea0d692edc 100644
--- a/Userland/Libraries/LibUnicode/Locale.cpp
+++ b/Userland/Libraries/LibUnicode/Locale.cpp
@@ -480,6 +480,17 @@ Optional<LocaleID> parse_unicode_locale_id(StringView locale)
     return locale_id;
 }
 
+static void perform_hard_coded_key_value_substitutions(String& key, String& value)
+{
+    // FIXME: In the XML export of CLDR, there are some aliases defined in the following file:
+    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/collation.xml
+    //
+    // There doesn't seem to be a counterpart in the JSON export. Since there aren't many such
+    // aliases, until an XML parser is implemented, those aliases are implemented here.
+    if (key.is_one_of("kb"sv, "kc"sv, "kh"sv, "kk"sv, "kn"sv) && (value == "yes"sv))
+        value = "true"sv;
+}
+
 static void transform_unicode_locale_id_to_canonical_syntax(LocaleID& locale_id)
 {
     auto canonicalize_language = [](LanguageID& language_id, bool force_lowercase) {
@@ -542,6 +553,7 @@ static void transform_unicode_locale_id_to_canonical_syntax(LocaleID& locale_id)
 
         for (auto& value : raw_values) {
             value = value.to_lowercase();
+            perform_hard_coded_key_value_substitutions(key, value);
 
             // Note: The spec says to remove "true" type and tfield values but that is believed to be a bug in the spec
             // because, for tvalues, that would result in invalid syntax: