Add `string16` and `cstring16` (UTF-16 based strings)

1 files changed, 72 insertions, 0 deletions

diff --git a/src/llvm_backend_expr.cpp b/src/llvm_backend_expr.cpp
index 74aea82f1..fbf0dea11 100644
--- a/src/llvm_backend_expr.cpp
+++ b/src/llvm_backend_expr.cpp
@@ -1656,6 +1656,8 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) {
 				res.type = t;
 				res.value = llvm_cstring(m, str);
 				return res;
+			} else if (src->kind == Type_Basic && src->Basic.kind == Basic_string16 && dst->Basic.kind == Basic_cstring16) {
+				GB_PANIC("TODO(bill): UTF-16 string");
 			}
 			// if (is_type_float(dst)) {
 			// 	return value;
@@ -1795,6 +1797,38 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) {
 	}
 
 
+
+	if (is_type_cstring16(src) && is_type_u16_ptr(dst)) {
+		return lb_emit_transmute(p, value, dst);
+	}
+	if (is_type_u16_ptr(src) && is_type_cstring16(dst)) {
+		return lb_emit_transmute(p, value, dst);
+	}
+	if (is_type_cstring16(src) && is_type_u16_multi_ptr(dst)) {
+		return lb_emit_transmute(p, value, dst);
+	}
+	if (is_type_u8_multi_ptr(src) && is_type_cstring16(dst)) {
+		return lb_emit_transmute(p, value, dst);
+	}
+	if (is_type_cstring16(src) && is_type_rawptr(dst)) {
+		return lb_emit_transmute(p, value, dst);
+	}
+	if (is_type_rawptr(src) && is_type_cstring16(dst)) {
+		return lb_emit_transmute(p, value, dst);
+	}
+
+	if (are_types_identical(src, t_cstring16) && are_types_identical(dst, t_string16)) {
+		TEMPORARY_ALLOCATOR_GUARD();
+
+		lbValue c = lb_emit_conv(p, value, t_cstring16);
+		auto args = array_make<lbValue>(temporary_allocator(), 1);
+		args[0] = c;
+		lbValue s = lb_emit_runtime_call(p, "cstring16_to_string16", args);
+		return lb_emit_conv(p, s, dst);
+	}
+
+
+
 	// integer -> boolean
 	if (is_type_integer(src) && is_type_boolean(dst)) {
 		lbValue res = {};
@@ -2296,6 +2330,14 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) {
 		return res;
 	}
 
+	// []u16 <-> string16
+	if (is_type_u16_slice(src) && is_type_string16(dst)) {
+		return lb_emit_transmute(p, value, t);
+	}
+	if (is_type_string16(src) && is_type_u16_slice(dst)) {
+		return lb_emit_transmute(p, value, t);
+	}
+
 	// []byte/[]u8 <-> string
 	if (is_type_u8_slice(src) && is_type_string(dst)) {
 		return lb_emit_transmute(p, value, t);
@@ -2304,6 +2346,7 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) {
 		return lb_emit_transmute(p, value, t);
 	}
 
+
 	if (is_type_array_like(dst)) {
 		Type *elem = base_array_type(dst);
 		isize index_count = cast(isize)get_array_type_count(dst);
@@ -2483,6 +2526,12 @@ gb_internal lbValue lb_emit_conv(lbProcedure *p, lbValue value, Type *t) {
 
 
 	if (is_type_untyped(src)) {
+		if (is_type_string(src) && is_type_string16(dst)) {
+			GB_PANIC("TODO(bill): UTF-16 string");
+			lbAddr result = lb_add_local_generated(p, t, false);
+			lb_addr_store(p, result, value);
+			return lb_addr_load(p, result);
+		}
 		if (is_type_string(src) && is_type_string(dst)) {
 			lbAddr result = lb_add_local_generated(p, t, false);
 			lb_addr_store(p, result, value);
@@ -3056,6 +3105,13 @@ gb_internal lbValue lb_emit_comp_against_nil(lbProcedure *p, TokenKind op_kind,
 				res.value = LLVMBuildIsNotNull(p->builder, x.value, "");
 			}
 			return res;
+		case Basic_cstring16:
+			if (op_kind == Token_CmpEq) {
+				res.value = LLVMBuildIsNull(p->builder, x.value, "");
+			} else if (op_kind == Token_NotEq) {
+				res.value = LLVMBuildIsNotNull(p->builder, x.value, "");
+			}
+			return res;
 		case Basic_any:
 			{
 				// TODO(bill): is this correct behaviour for nil comparison for any?
@@ -4432,6 +4488,22 @@ gb_internal lbAddr lb_build_addr_slice_expr(lbProcedure *p, Ast *expr) {
 	}
 
 	case Type_Basic: {
+		if (is_type_string16(type)) {
+			GB_ASSERT_MSG(are_types_identical(type, t_string16), "got %s", type_to_string(type));
+			lbValue len = lb_string_len(p, base);
+			if (high.value == nullptr) high = len;
+
+			if (!no_indices) {
+				lb_emit_slice_bounds_check(p, se->open, low, high, len, se->low != nullptr);
+			}
+
+			lbValue elem    = lb_emit_ptr_offset(p, lb_string_elem(p, base), low);
+			lbValue new_len = lb_emit_arith(p, Token_Sub, high, low, t_int);
+
+			lbAddr str = lb_add_local_generated(p, t_string16, false);
+			lb_fill_string(p, str, elem, new_len);
+			return str;
+		}
 		GB_ASSERT_MSG(are_types_identical(type, t_string), "got %s", type_to_string(type));
 		lbValue len = lb_string_len(p, base);
 		if (high.value == nullptr) high = len;