Add `intrinsics.constant_utf16_cstring`

1 files changed, 71 insertions, 0 deletions

diff --git a/src/llvm_backend_proc.cpp b/src/llvm_backend_proc.cpp
index 261e2819c..7bc7fb61f 100644
--- a/src/llvm_backend_proc.cpp
+++ b/src/llvm_backend_proc.cpp
@@ -2122,6 +2122,77 @@ lbValue lb_build_builtin_proc(lbProcedure *p, Ast *expr, TypeAndValue const &tv,
 	case BuiltinProc_objc_find_class:        return lb_handle_objc_find_class(p, expr);
 	case BuiltinProc_objc_register_selector: return lb_handle_objc_register_selector(p, expr);
 	case BuiltinProc_objc_register_class:    return lb_handle_objc_register_class(p, expr);
+
+
+	case BuiltinProc_constant_utf16_cstring:
+		{
+			auto const encode_surrogate_pair = [](Rune r, u16 *r1, u16 *r2) {
+				if (r < 0x10000 || r > 0x10ffff) {
+					*r1 = 0xfffd;
+					*r2 = 0xfffd;
+				} else {
+					r -= 0x10000;
+					*r1 = 0xd800 + ((r>>10)&0x3ff);
+					*r2 = 0xdc00 + (r&0x3ff);
+				}
+			};
+
+			lbModule *m = p->module;
+
+			auto tav = type_and_value_of_expr(ce->args[0]);
+			GB_ASSERT(tav.value.kind == ExactValue_String);
+			String value = tav.value.value_string;
+
+			LLVMTypeRef llvm_u16 = lb_type(m, t_u16);
+
+			isize max_len = value.len*2 + 1;
+			LLVMValueRef *buffer = gb_alloc_array(temporary_allocator(), LLVMValueRef, max_len);
+			isize n = 0;
+			while (value.len > 0) {
+				Rune r = 0;
+				isize w = gb_utf8_decode(value.text, value.len, &r);
+				value.text += w;
+				value.len  -= w;
+				if ((0 <= r && r < 0xd800) || (0xe000 <= r && r < 0x10000)) {
+					buffer[n++] = LLVMConstInt(llvm_u16, cast(u16)r, false);
+				} else if (0x10000 <= r && r <= 0x10ffff) {
+					u16 r1, r2;
+					encode_surrogate_pair(r, &r1, &r2);
+					buffer[n++] = LLVMConstInt(llvm_u16, r1, false);
+					buffer[n++] = LLVMConstInt(llvm_u16, r2, false);
+				} else {
+					buffer[n++] = LLVMConstInt(llvm_u16, 0xfffd, false);
+				}
+			}
+
+			buffer[n++] = LLVMConstInt(llvm_u16, 0, false);
+
+			LLVMValueRef array = LLVMConstArray(llvm_u16, buffer, cast(unsigned int)n);
+
+			char *name = nullptr;
+			{
+				isize max_len = 7+8+1;
+				name = gb_alloc_array(permanent_allocator(), char, max_len);
+				u32 id = m->gen->global_array_index.fetch_add(1);
+				isize len = gb_snprintf(name, max_len, "csbs$%x", id);
+				len -= 1;
+			}
+			LLVMValueRef global_data = LLVMAddGlobal(m->mod, LLVMTypeOf(array), name);
+			LLVMSetInitializer(global_data, array);
+			LLVMSetLinkage(global_data, LLVMInternalLinkage);
+
+
+
+			LLVMValueRef indices[] = {
+				LLVMConstInt(lb_type(m, t_u32), 0, false),
+				LLVMConstInt(lb_type(m, t_u32), 0, false),
+			};
+			lbValue res = {};
+			res.type = tv.type;
+			res.value = LLVMBuildInBoundsGEP(p->builder, global_data, indices, gb_count_of(indices), "");
+			return res;
+
+		}
 	}
 
 	GB_PANIC("Unhandled built-in procedure %.*s", LIT(builtin_procs[id].name));