provide StringView::char_code_at

builtin/pkg.generated.mbti

@@ -855,10 +855,12 @@ fn Bytes::of_string(String) -> Bytes
 fn Bytes::to_unchecked_string(Bytes, offset? : Int, length? : Int) -> String
 
 #alias("_[_]")
+#deprecated
 fn StringView::at(Self, Int) -> Int
 fn StringView::char_length(Self) -> Int
 fn StringView::char_length_eq(Self, Int) -> Bool
 fn StringView::char_length_ge(Self, Int) -> Bool
+fn StringView::code_unit_at(Self, Int) -> UInt16
 fn StringView::contains(Self, Self) -> Bool
 fn StringView::contains_any(Self, chars~ : Self) -> Bool
 fn StringView::contains_char(Self, Char) -> Bool

builtin/stringview.mbt

@@ -31,23 +31,26 @@ fn StringView::make_view(str : String, start : Int, end : Int) -> StringView = "
 /// Returns the charcode(UTF-16 code unit) at the given index.
 /// 
 /// This method has O(1) complexity.
-/// 
-/// # Example
-/// 
-/// ```mbt
-///   let str = "Hello🤣🤣🤣"
-///   let view = str.view(start_offset = str.offset_of_nth_char(1).unwrap(), end_offset = str.offset_of_nth_char(6).unwrap())
-///   inspect(view[0].to_char(), content="Some('e')")
-///   inspect(view[4], content="55358")
-/// ```
 #alias("_[_]")
+#deprecated("Use `code_unit_at` instead which returns UInt16")
 pub fn StringView::at(self : StringView, index : Int) -> Int {
   guard index >= 0 && index < self.length() else {
     abort("Index out of bounds")
   }
   self.str().unsafe_charcode_at(self.start() + index)
 }
 
+///|
+/// Returns the UTF-16 code unit at the given index.
+/// 
+/// This method has O(1) complexity.
+pub fn StringView::code_unit_at(self : StringView, index : Int) -> UInt16 {
+  guard index >= 0 && index < self.length() else {
+    abort("Index out of bounds")
+  }
+  self.str().code_unit_at(self.start() + index)
+}
+
 ///|
 /// Returns the length of the view.
 /// 

builtin/stringview_test.mbt

@@ -30,3 +30,22 @@ test "StringView::suffixes" {
     .collect()
   @json.inspect(suffixes_no_empty, content=["a🤣b", "🤣b", "b"])
 }
+
+///|
+test "code_unit_at" {
+  let s = "Hello, 世界!😀"
+  assert_true(s[:].code_unit_at(0) is 'H')
+  assert_true(s[:].code_unit_at(7) is '世')
+  assert_true(s[:].code_unit_at(8) is '界')
+  assert_true(s[:].code_unit_at(0) is ('A'..='Z'))
+  inspect(s[:].code_unit_at(0), content="72") // 'H'
+  inspect(s[:].code_unit_at(7), content="19990") // '世'
+  inspect(s[:].code_unit_at(8), content="30028") // '界'
+  inspect(s[:].code_unit_at(10), content="55357") // High surrogate of '😀'
+  inspect(s[:].code_unit_at(11), content="56832") // Low surrogate of '😀'
+  let view = s[7:] // "世界!😀"
+  inspect(view.code_unit_at(0), content="19990") // '世'
+  inspect(view.code_unit_at(1), content="30028") // '界'
+  inspect(view.code_unit_at(3), content="55357") // High surrogate of '😀'
+  inspect(view.code_unit_at(4), content="56832") // Low surrogate of '😀'
+}

encoding/utf16/encode.mbt

@@ -38,15 +38,15 @@ pub fn encode(
     arr[0] = 0xFE
     arr[1] = 0xFF
     for i in 0..<str.length() {
-      let code_unit = str[i]
+      let code_unit = str.code_unit_at(i)
       arr[2 + i * 2] = (code_unit >> 8).to_byte()
       arr[2 + i * 2 + 1] = (code_unit & 0xFF).to_byte()
     }
     arr.unsafe_reinterpret_as_bytes()
   } else {
     let arr = FixedArray::make(str.length() * 2, b'\x00')
     for i in 0..<str.length() {
-      let code_unit = str[i]
+      let code_unit = str.code_unit_at(i)
       arr[i * 2] = (code_unit >> 8).to_byte()
       arr[i * 2 + 1] = (code_unit & 0xFF).to_byte()
     }

string/regex/internal/regexp/internal/vm/impl.mbt

@@ -117,8 +117,8 @@ fn add_thread(
       let assertion = match pred {
         BeginText => sp == 0
         EndText => sp == content.length()
-        BeginLine => sp == 0 || content[sp - 1] == '\n'
-        EndLine => sp == content.length() || content[sp] == '\n'
+        BeginLine => sp == 0 || content.code_unit_at(sp - 1) == '\n'
+        EndLine => sp == content.length() || content.code_unit_at(sp) == '\n'
         WordBoundary =>
           is_word_char_at(content, sp - 1) != is_word_char_at(content, sp)
         NoWordBoundary =>

string/view_test.mbt

@@ -1044,7 +1044,7 @@ test "StringView add" {
 test "panic view op_get out of bounds" {
   let str = "Hello"
   let view = str.view(start_offset=0, end_offset=5)
-  view[5] |> ignore
+  view.code_unit_at(5) |> ignore
 }
 
 // Added test to cover start index pointing to trailing surrogate