Add 'string::utf8ByteLength' (#1412)

`string::utf8ByteLength` is `char::utf8ByteCount` but for whole strings:
efficient on NodeJS and LLVM, but I also included a version written in
Effekt because I don't know what Chez does...

Author: jiribenes

examples/stdlib/string/utf8bytelength.check

@@ -0,0 +1,20 @@
+: 0 (internal: 0, bytearray: 0)
+a: 1 (internal: 1, bytearray: 1)
+ab: 2 (internal: 2, bytearray: 2)
+abc: 3 (internal: 3, bytearray: 3)
+é: 2 (internal: 2, bytearray: 2)
+λ: 2 (internal: 2, bytearray: 2)
+💩: 4 (internal: 4, bytearray: 4)
+aé💩b: 8 (internal: 8, bytearray: 8)
+§é💩: 8 (internal: 8, bytearray: 8)
+©: 2 (internal: 2, bytearray: 2)
+¢: 2 (internal: 2, bytearray: 2)
+€: 3 (internal: 3, bytearray: 3)
+漢: 3 (internal: 3, bytearray: 3)
+अ: 3 (internal: 3, bytearray: 3)
+😀: 4 (internal: 4, bytearray: 4)
+𝄞: 4 (internal: 4, bytearray: 4)
+𐍈: 4 (internal: 4, bytearray: 4)
+مرحبا: 10 (internal: 10, bytearray: 10)
+👨‍👩‍👧‍👦: 25 (internal: 25, bytearray: 25)
+é: 3 (internal: 3, bytearray: 3)

examples/stdlib/string/utf8bytelength.effekt

@@ -0,0 +1,11 @@
+def one(str: String) = {
+  val bytes = str.utf8ByteLength
+  val bytes2 = str.internal::utf8ByteLength
+  val bytes3 = str.bytearray::fromString.size
+  println(s"${str}: ${bytes.show} (internal: ${bytes2.show}, bytearray: ${bytes3.show})")
+}
+
+def main() = {
+  val cases = ["", "a", "ab", "abc", "é", "λ", "💩", "aé💩b", "§é💩", "©", "¢", "€", "漢", "अ", "😀", "𝄞", "𐍈", "مرحبا", "👨‍👩‍👧‍👦", "e\u0301"]
+  for { cases.each } { (s: String) => one(s) }
+}

libraries/common/effekt.effekt

@@ -152,6 +152,9 @@ extern def infixPlusPlus(s1: String, s2: String) at {}: String =
     ret %Pos %spz
   """
 
+/// Beware that this function is backend-dependent.
+/// If you want to get the number of bytes that a string would take up in a UTF-8 encoding,
+/// use `string::utf8ByteLength` instead.
 extern def length(str: String) at {}: Int =
   js "${str}.length"
   chez "(string-length ${str})"

libraries/common/string.effekt

@@ -204,6 +204,23 @@ def lastIndexOf(str: String, sub: String, from: Int): Option[Int] = {
   go(from)
 }
 
+namespace internal {
+  def utf8ByteLength(str: String): Int = stream::sum {
+    for {str.each} { (c: Char) =>
+      do emit(c.utf8ByteCount)
+    }
+  }
+}
+
+/// Returns the number of bytes that the given `str` would take up in a UTF-8 encoding.
+extern def utf8ByteLength(str: String) at {io, global}: Int =
+  jsNode "Buffer.byteLength(${str}, 'utf8')"
+  llvm """
+    %x = call %Int @c_bytearray_size(%Pos ${str})
+    ret %Int %x
+  """ // via bytearray::fromString(str).length
+  default { internal::utf8ByteLength(str) }
+
 // TODO more efficient (native) implementations could be integrated as follows
 
 // extern def indexOf(str: String, sub: String) at {}: Option[Int] =