From 5b71560c65f6d7a8b4fd05eeabb814b379c854c3 Mon Sep 17 00:00:00 2001
From: Sokwhan Huh <sokwhan@google.com>
Date: Fri, 3 Mar 2023 21:03:40 +0000
Subject: [PATCH] Add test cases for String standard functions with Unicode in
 Supplemental Multilingual Plane

---
 tests/simple/testdata/string.textproto | 29 ++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)

diff --git a/tests/simple/testdata/string.textproto b/tests/simple/testdata/string.textproto
index 10d4142d..5458080e 100644
--- a/tests/simple/testdata/string.textproto
+++ b/tests/simple/testdata/string.textproto
@@ -72,6 +72,11 @@ section {
     expr: "'завтра'.startsWith('за')"
     value: { bool_value: true }
   }
+  test {
+    name: "unicode_smp"
+    expr: "'\U0001F431\U0001F600\U0001F61B'.startsWith('\U0001F431')"
+    value: { bool_value: true }
+  }
 }
 section {
   name: "ends_with"
@@ -106,6 +111,11 @@ section {
     expr: "'forté'.endsWith('té')"
     value: { bool_value: true }
   }
+  test {
+    name: "unicode_smp"
+    expr: "'\U0001F431\U0001F600\U0001F61B'.endsWith('\U0001F61B')"
+    value: { bool_value: true }
+  }
 }
 section {
   name: "matches"
@@ -150,6 +160,11 @@ section {
     expr: "'mañana'.matches('a+ñ+a+')"
     value: { bool_value: true }
   }
+  test {
+    name: "unicode_smp"
+    expr: "'\U0001F431\U0001F600\U0001F600'.matches('(a|\U0001F600){2}')"
+    value: { bool_value: true }
+  }
 }
 section {
   name: "concatentation"
@@ -157,12 +172,12 @@ section {
   test {
     name: "concat_true"
     expr: "'he' + 'llo'"
-    value: { string_value: "hello"}
+    value: { string_value: "hello" }
   }
   test {
     name: "concat_with_spaces"
     expr: "'hello' + ' ' == 'hello'"
-    value: { bool_value: false}
+    value: { bool_value: false }
   }
   test {
     name: "concat_empty_string_beginning"
@@ -189,6 +204,11 @@ section {
     expr: "'r' + 'ô' + 'le'"
     value: { string_value: "rôle" }
   }
+  test {
+    name: "ascii_unicode_unicode_smp"
+    expr: "'a' + 'ÿ' + '\U0001F431'"
+    value: { string_value: "aÿ\xf0\x9f\x90\xb1" }
+  }
   test {
     name: "empty_unicode"
     expr: "'' + 'Ω' + ''"
@@ -223,6 +243,11 @@ section {
     expr: "'Straße'.contains('aß')"
     value: { bool_value: true }
   }
+  test {
+    name: "contains_unicode_smp"
+    expr: "'\U0001F431\U0001F600\U0001F601'.contains('\U0001F600')"
+    value: { bool_value: true }
+  }
   test {
     name: "empty_contains"
     expr: "''.contains('something')"