`for in string16`; Support `string16` across core

5 files changed, 166 insertions, 6 deletions

diff --git a/src/check_builtin.cpp b/src/check_builtin.cpp
index 4abace637..66ea0cfbd 100644
--- a/src/check_builtin.cpp
+++ b/src/check_builtin.cpp
@@ -7179,7 +7179,11 @@ gb_internal bool check_builtin_procedure(CheckerContext *c, Operand *operand, As
 				return false;
 			}
 			operand->mode = Addressing_Value;
-			operand->type = alloc_type_multi_pointer(t_u16);
+			if (type_hint != nullptr && is_type_cstring16(type_hint)) {
+				operand->type = type_hint;
+			} else {
+				operand->type = alloc_type_multi_pointer(t_u16);
+			}
 			operand->value = {};
 			break;
 		}
diff --git a/src/check_expr.cpp b/src/check_expr.cpp
index 8d2e4d637..34149f92b 100644
--- a/src/check_expr.cpp
+++ b/src/check_expr.cpp
@@ -3426,7 +3426,7 @@ gb_internal bool check_is_castable_to(CheckerContext *c, Operand *operand, Type
 	if (are_types_identical(src, t_cstring16) && is_type_u16_multi_ptr(dst)) {
 		return !is_constant;
 	}
-	// cstring -> rawptr
+	// cstring16 -> rawptr
 	if (are_types_identical(src, t_cstring16) && is_type_rawptr(dst)) {
 		return !is_constant;
 	}
diff --git a/src/check_stmt.cpp b/src/check_stmt.cpp
index bc9b6c5dd..ae88ff333 100644
--- a/src/check_stmt.cpp
+++ b/src/check_stmt.cpp
@@ -974,7 +974,14 @@ gb_internal void check_unroll_range_stmt(CheckerContext *ctx, Ast *node, u32 mod
 			Type *t = base_type(operand.type);
 			switch (t->kind) {
 			case Type_Basic:
-				if (is_type_string(t) && t->Basic.kind != Basic_cstring) {
+				if (is_type_string16(t) && t->Basic.kind != Basic_cstring) {
+					val0 = t_rune;
+					val1 = t_int;
+					inline_for_depth = exact_value_i64(operand.value.value_string.len);
+					if (unroll_count > 0) {
+						error(node, "#unroll(%lld) does not support strings", cast(long long)unroll_count);
+					}
+				} else if (is_type_string(t) && t->Basic.kind != Basic_cstring) {
 					val0 = t_rune;
 					val1 = t_int;
 					inline_for_depth = exact_value_i64(operand.value.value_string.len);
@@ -1236,7 +1243,11 @@ gb_internal void check_switch_stmt(CheckerContext *ctx, Ast *node, u32 mod_flags
 
 				add_to_seen_map(ctx, &seen, upper_op, x, lhs, rhs);
 
-				if (is_type_string(x.type)) {
+				if (is_type_string16(x.type)) {
+					// NOTE(bill): Force dependency for strings here
+					add_package_dependency(ctx, "runtime", "string16_le");
+					add_package_dependency(ctx, "runtime", "string16_lt");
+				} else if (is_type_string(x.type)) {
 					// NOTE(bill): Force dependency for strings here
 					add_package_dependency(ctx, "runtime", "string_le");
 					add_package_dependency(ctx, "runtime", "string_lt");
@@ -1770,7 +1781,16 @@ gb_internal void check_range_stmt(CheckerContext *ctx, Ast *node, u32 mod_flags)
 
 			switch (t->kind) {
 			case Type_Basic:
-				if (t->Basic.kind == Basic_string || t->Basic.kind == Basic_UntypedString) {
+				if (t->Basic.kind == Basic_string16) {
+					is_possibly_addressable = false;
+					array_add(&vals, t_rune);
+					array_add(&vals, t_int);
+					if (is_reverse) {
+						add_package_dependency(ctx, "runtime", "string16_decode_last_rune");
+					} else {
+						add_package_dependency(ctx, "runtime", "string16_decode_rune");
+					}
+				} else if (t->Basic.kind == Basic_string || t->Basic.kind == Basic_UntypedString) {
 					is_possibly_addressable = false;
 					array_add(&vals, t_rune);
 					array_add(&vals, t_int);
diff --git a/src/llvm_backend_stmt.cpp b/src/llvm_backend_stmt.cpp
index 027837f3f..5481ca447 100644
--- a/src/llvm_backend_stmt.cpp
+++ b/src/llvm_backend_stmt.cpp
@@ -622,6 +622,121 @@ gb_internal void lb_build_range_string(lbProcedure *p, lbValue expr, Type *val_t
 	if (done_) *done_ = done;
 }
 
+gb_internal void lb_build_range_string16(lbProcedure *p, lbValue expr, Type *val_type,
+                                       lbValue *val_, lbValue *idx_, lbBlock **loop_, lbBlock **done_,
+                                       bool is_reverse) {
+
+	lbModule *m = p->module;
+	lbValue count = lb_const_int(m, t_int, 0);
+	Type *expr_type = base_type(expr.type);
+	switch (expr_type->kind) {
+	case Type_Basic:
+		count = lb_string_len(p, expr);
+		break;
+	default:
+		GB_PANIC("Cannot do range_string of %s", type_to_string(expr_type));
+		break;
+	}
+
+	lbValue val = {};
+	lbValue idx = {};
+	lbBlock *loop = nullptr;
+	lbBlock *done = nullptr;
+	lbBlock *body = nullptr;
+
+	loop = lb_create_block(p, "for.string16.loop");
+	body = lb_create_block(p, "for.string16.body");
+	done = lb_create_block(p, "for.string16.done");
+
+	lbAddr offset_ = lb_add_local_generated(p, t_int, false);
+	lbValue offset = {};
+	lbValue cond = {};
+
+	if (!is_reverse) {
+		/*
+			for c, offset in str {
+				...
+			}
+
+			offset := 0
+			for offset < len(str) {
+				c, _w := string16_decode_rune(str[offset:])
+				...
+				offset += _w
+			}
+		*/
+		lb_addr_store(p, offset_, lb_const_int(m, t_int, 0));
+
+		lb_emit_jump(p, loop);
+		lb_start_block(p, loop);
+
+
+		offset = lb_addr_load(p, offset_);
+		cond = lb_emit_comp(p, Token_Lt, offset, count);
+	} else {
+		// NOTE(bill): REVERSED LOGIC
+		/*
+			#reverse for c, offset in str {
+				...
+			}
+
+			offset := len(str)
+			for offset > 0 {
+				c, _w := string16_decode_last_rune(str[:offset])
+				offset -= _w
+				...
+			}
+		*/
+		lb_addr_store(p, offset_, count);
+
+		lb_emit_jump(p, loop);
+		lb_start_block(p, loop);
+
+		offset = lb_addr_load(p, offset_);
+		cond = lb_emit_comp(p, Token_Gt, offset, lb_const_int(m, t_int, 0));
+	}
+	lb_emit_if(p, cond, body, done);
+	lb_start_block(p, body);
+
+
+	lbValue rune_and_len = {};
+	if (!is_reverse) {
+		lbValue str_elem = lb_emit_ptr_offset(p, lb_string_elem(p, expr), offset);
+		lbValue str_len  = lb_emit_arith(p, Token_Sub, count, offset, t_int);
+		auto args = array_make<lbValue>(permanent_allocator(), 1);
+		args[0] = lb_emit_string16(p, str_elem, str_len);
+
+		rune_and_len = lb_emit_runtime_call(p, "string16_decode_rune", args);
+		lbValue len  = lb_emit_struct_ev(p, rune_and_len, 1);
+		lb_addr_store(p, offset_, lb_emit_arith(p, Token_Add, offset, len, t_int));
+
+		idx = offset;
+	} else {
+		// NOTE(bill): REVERSED LOGIC
+		lbValue str_elem = lb_string_elem(p, expr);
+		lbValue str_len  = offset;
+		auto args = array_make<lbValue>(permanent_allocator(), 1);
+		args[0] = lb_emit_string16(p, str_elem, str_len);
+
+		rune_and_len = lb_emit_runtime_call(p, "string16_decode_last_rune", args);
+		lbValue len  = lb_emit_struct_ev(p, rune_and_len, 1);
+		lb_addr_store(p, offset_, lb_emit_arith(p, Token_Sub, offset, len, t_int));
+
+		idx = lb_addr_load(p, offset_);
+	}
+
+
+	if (val_type != nullptr) {
+		val = lb_emit_struct_ev(p, rune_and_len, 0);
+	}
+
+	if (val_)  *val_  = val;
+	if (idx_)  *idx_  = idx;
+	if (loop_) *loop_ = loop;
+	if (done_) *done_ = done;
+}
+
+
 
 gb_internal Ast *lb_strip_and_prefix(Ast *ident) {
 	if (ident != nullptr) {
@@ -1138,7 +1253,11 @@ gb_internal void lb_build_range_stmt(lbProcedure *p, AstRangeStmt *rs, Scope *sc
 			}
 			Type *t = base_type(string.type);
 			GB_ASSERT(!is_type_cstring(t));
-			lb_build_range_string(p, string, val0_type, &val, &key, &loop, &done, rs->reverse);
+			if (is_type_string16(t)) {
+				lb_build_range_string16(p, string, val0_type, &val, &key, &loop, &done, rs->reverse);
+			} else {
+				lb_build_range_string(p, string, val0_type, &val, &key, &loop, &done, rs->reverse);
+			}
 			break;
 		}
 		case Type_Tuple:
diff --git a/src/llvm_backend_utility.cpp b/src/llvm_backend_utility.cpp
index ea1bae4e9..dcb95a9a2 100644
--- a/src/llvm_backend_utility.cpp
+++ b/src/llvm_backend_utility.cpp
@@ -191,6 +191,23 @@ gb_internal lbValue lb_emit_clamp(lbProcedure *p, Type *t, lbValue x, lbValue mi
 	return z;
 }
 
+gb_internal lbValue lb_emit_string16(lbProcedure *p, lbValue str_elem, lbValue str_len) {
+	if (false && lb_is_const(str_elem) && lb_is_const(str_len)) {
+		LLVMValueRef values[2] = {
+			str_elem.value,
+			str_len.value,
+		};
+		lbValue res = {};
+		res.type = t_string16;
+		res.value = llvm_const_named_struct(p->module, t_string16, values, gb_count_of(values));
+		return res;
+	} else {
+		lbAddr res = lb_add_local_generated(p, t_string16, false);
+		lb_emit_store(p, lb_emit_struct_ep(p, res.addr, 0), str_elem);
+		lb_emit_store(p, lb_emit_struct_ep(p, res.addr, 1), str_len);
+		return lb_addr_load(p, res);
+	}
+}
 
 
 gb_internal lbValue lb_emit_string(lbProcedure *p, lbValue str_elem, lbValue str_len) {