Support parse/check/lower for char (#5901)

toolchain/check/testdata/builtins/char/basics.carbon and
toolchain/lower/testdata/builtins/char.carbon are probably the most
interesting tests here. The parse tests is required because this adds a
new node kind, and we need coverage of it; but the attached info is
minor. There's a fair amount of test churn here because I'm adding the
Core.Char and Core.CharLiteral types as new singletons.

My intent here is that `CharId` is always a unicode code point, even
when the type is a `Char` and thus must be a single UTF-8 code unit
(single byte). This mainly means the stored value of a `CharValue` can
be printed internally without knowing the type.

---------

Co-authored-by: josh11b <[email protected]>
Co-authored-by: Richard Smith <[email protected]>

Author: 3 people

core/prelude/types.carbon

@@ -5,6 +5,7 @@
 package Core library "prelude/types";
 
 export import library "prelude/types/bool";
+export import library "prelude/types/char";
 export import library "prelude/types/int";
 export import library "prelude/types/int_literal";
 export import library "prelude/types/optional";

core/prelude/types/char.carbon

@@ -0,0 +1,24 @@
+// Part of the Carbon Language project, under the Apache License v2.0 with LLVM
+// Exceptions. See /LICENSE for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+package Core library "prelude/types/char";
+
+import library "prelude/destroy";
+import library "prelude/operators";
+import library "prelude/types/uint";
+
+fn CharLiteral() -> type = "char_literal.make_type";
+
+class Char {
+  adapt u8;
+}
+
+impl CharLiteral() as ImplicitAs(Char) {
+  fn Convert[self: Self]() -> Char = "char.convert_checked";
+}
+
+// TODO: Remove these once ImplicitAs extends As.
+impl CharLiteral() as As(Char) {
+  fn Convert[self: Self]() -> Char = "char.convert_checked";
+}

toolchain/check/BUILD

@@ -131,6 +131,7 @@ cc_library(
         "//toolchain/check:scope_stack",
         "//toolchain/diagnostics:diagnostic_emitter",
         "//toolchain/diagnostics:format_providers",
+        "//toolchain/lex:token_info",
         "//toolchain/lex:token_kind",
         "//toolchain/lex:tokenized_buffer",
         "//toolchain/parse:node_kind",
@@ -296,6 +297,7 @@ cc_library(
     hdrs = ["diagnostic_emitter.h"],
     deps = [
         ":context",
+        "//common:ostream",
         "//common:raw_string_ostream",
         "//toolchain/diagnostics:diagnostic_emitter",
         "//toolchain/lex:token_index",

toolchain/check/eval.cpp

@@ -9,6 +9,7 @@
 #include <optional>
 #include <utility>
 
+#include "llvm/Support/ConvertUTF.h"
 #include "toolchain/base/canonical_value_store.h"
 #include "toolchain/base/kind_switch.h"
 #include "toolchain/check/action.h"
@@ -951,6 +952,28 @@ static auto PerformArrayIndex(EvalContext& eval_context, SemIR::ArrayIndex inst)
   return eval_context.GetConstantValue(elements[index_val.getZExtValue()]);
 }
 
+// Performs a conversion between character types, diagnosing if the value
+// doesn't fit in the destination type.
+static auto PerformCheckedCharConvert(Context& context, SemIR::LocId loc_id,
+                                      SemIR::InstId arg_id,
+                                      SemIR::TypeId dest_type_id)
+    -> SemIR::ConstantId {
+  auto arg = context.insts().GetAs<SemIR::CharLiteralValue>(arg_id);
+
+  // Values over 0x80 require multiple code units in UTF-8.
+  if (arg.value.index >= 0x80) {
+    CARBON_DIAGNOSTIC(CharTooLargeForType, Error,
+                      "character value {0} too large for type {1}",
+                      SemIR::CharId, SemIR::TypeId);
+    context.emitter().Emit(loc_id, CharTooLargeForType, arg.value,
+                           dest_type_id);
+    return SemIR::ErrorInst::ConstantId;
+  }
+
+  llvm::APInt int_val(8, arg.value.index, /*isSigned=*/false);
+  return MakeIntResult(context, dest_type_id, /*is_signed=*/false, int_val);
+}
+
 // Forms a constant int type as an evaluation result. Requires that width_id is
 // constant.
 static auto MakeIntTypeResult(Context& context, SemIR::LocId loc_id,
@@ -1590,6 +1613,10 @@ static auto MakeConstantForBuiltinCall(EvalContext& eval_context,
       return MakeFacetTypeResult(eval_context.context(), combined_info, phase);
     }
 
+    case SemIR::BuiltinFunctionKind::CharLiteralMakeType: {
+      return context.constant_values().Get(SemIR::CharLiteralType::TypeInstId);
+    }
+
     case SemIR::BuiltinFunctionKind::IntLiteralMakeType: {
       return context.constant_values().Get(SemIR::IntLiteralType::TypeInstId);
     }
@@ -1619,6 +1646,15 @@ static auto MakeConstantForBuiltinCall(EvalContext& eval_context,
       return context.constant_values().Get(SemIR::BoolType::TypeInstId);
     }
 
+    // Character conversions.
+    case SemIR::BuiltinFunctionKind::CharConvertChecked: {
+      if (phase != Phase::Concrete) {
+        return MakeConstantResult(context, call, phase);
+      }
+      return PerformCheckedCharConvert(context, loc_id, arg_ids[0],
+                                       call.type_id);
+    }
+
     // Integer conversions.
     case SemIR::BuiltinFunctionKind::IntConvert: {
       if (phase != Phase::Concrete) {

toolchain/check/handle_literal.cpp

@@ -35,6 +35,17 @@ auto HandleParseNode(Context& context, Parse::BoolLiteralTrueId node_id)
   return true;
 }
 
+auto HandleParseNode(Context& context, Parse::CharLiteralId node_id) -> bool {
+  auto value = context.tokens().GetCharLiteralValue(
+      context.parse_tree().node_token(node_id));
+  auto inst_id = AddInst<SemIR::CharLiteralValue>(
+      context, node_id,
+      {.type_id = GetSingletonType(context, SemIR::CharLiteralType::TypeInstId),
+       .value = SemIR::CharId(value.value)});
+  context.node_stack().Push(node_id, inst_id);
+  return true;
+}
+
 auto HandleParseNode(Context& context, Parse::IntLiteralId node_id) -> bool {
   auto int_literal_id = MakeIntLiteral(
       context, node_id,

toolchain/check/literal.cpp

@@ -9,6 +9,7 @@
 #include "toolchain/check/convert.h"
 #include "toolchain/check/name_lookup.h"
 #include "toolchain/check/type.h"
+#include "toolchain/lex/token_info.h"
 #include "toolchain/sem_ir/ids.h"
 
 namespace Carbon::Check {

toolchain/check/literal.h

@@ -6,6 +6,7 @@
 #define CARBON_TOOLCHAIN_CHECK_LITERAL_H_
 
 #include "toolchain/check/context.h"
+#include "toolchain/lex/token_info.h"
 #include "toolchain/sem_ir/ids.h"
 
 namespace Carbon::Check {

toolchain/check/testdata/alias/export_name.carbon

@@ -111,7 +111,7 @@ var d: D* = &c;
 // CHECK:STDOUT:   %Main.C = import_ref Main//base, C, unloaded
 // CHECK:STDOUT:   %Main.D: type = import_ref Main//base, D, loaded [concrete = constants.%C]
 // CHECK:STDOUT:   %Main.import_ref.8f2: <witness> = import_ref Main//base, loc4_10, loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Main.import_ref.2c4 = import_ref Main//base, inst16 [no loc], unloaded
+// CHECK:STDOUT:   %Main.import_ref.2c4 = import_ref Main//base, inst17 [no loc], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {
@@ -142,7 +142,7 @@ var d: D* = &c;
 // CHECK:STDOUT:   %Main.C: type = import_ref Main//base, C, loaded [concrete = constants.%C]
 // CHECK:STDOUT:   %Main.D = import_ref Main//base, D, unloaded
 // CHECK:STDOUT:   %Main.import_ref.8f2: <witness> = import_ref Main//base, loc4_10, loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Main.import_ref.2c4 = import_ref Main//base, inst16 [no loc], unloaded
+// CHECK:STDOUT:   %Main.import_ref.2c4 = import_ref Main//base, inst17 [no loc], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {
@@ -173,8 +173,8 @@ var d: D* = &c;
 // CHECK:STDOUT:
 // CHECK:STDOUT: imports {
 // CHECK:STDOUT:   %Main.D: type = import_ref Main//export, D, loaded [concrete = constants.%C]
-// CHECK:STDOUT:   %Main.import_ref.8db: <witness> = import_ref Main//export, inst22 [indirect], loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Main.import_ref.6a9 = import_ref Main//export, inst23 [indirect], unloaded
+// CHECK:STDOUT:   %Main.import_ref.8db: <witness> = import_ref Main//export, inst23 [indirect], loaded [concrete = constants.%complete_type]
+// CHECK:STDOUT:   %Main.import_ref.6a9 = import_ref Main//export, inst24 [indirect], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {
@@ -257,8 +257,8 @@ var d: D* = &c;
 // CHECK:STDOUT: imports {
 // CHECK:STDOUT:   %Main.D: type = import_ref Main//export, D, loaded [concrete = constants.%C]
 // CHECK:STDOUT:   %Main.C: type = import_ref Main//export_orig, C, loaded [concrete = constants.%C]
-// CHECK:STDOUT:   %Main.import_ref.8db: <witness> = import_ref Main//export_orig, inst22 [indirect], loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Main.import_ref.6a9 = import_ref Main//export_orig, inst23 [indirect], unloaded
+// CHECK:STDOUT:   %Main.import_ref.8db: <witness> = import_ref Main//export_orig, inst23 [indirect], loaded [concrete = constants.%complete_type]
+// CHECK:STDOUT:   %Main.import_ref.6a9 = import_ref Main//export_orig, inst24 [indirect], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {

toolchain/check/testdata/alias/import.carbon

@@ -120,7 +120,7 @@ var c: () = a_alias_alias;
 // CHECK:STDOUT:   %Main.c_alias: type = import_ref Main//class1, c_alias, loaded [concrete = constants.%C]
 // CHECK:STDOUT:   %Main.a = import_ref Main//class1, a, unloaded
 // CHECK:STDOUT:   %Main.import_ref.8f2: <witness> = import_ref Main//class1, loc4_10, loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Main.import_ref.2c4 = import_ref Main//class1, inst16 [no loc], unloaded
+// CHECK:STDOUT:   %Main.import_ref.2c4 = import_ref Main//class1, inst17 [no loc], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {
@@ -166,8 +166,8 @@ var c: () = a_alias_alias;
 // CHECK:STDOUT: imports {
 // CHECK:STDOUT:   %Main.c_alias_alias: type = import_ref Main//class2, c_alias_alias, loaded [concrete = constants.%C]
 // CHECK:STDOUT:   %Main.b = import_ref Main//class2, b, unloaded
-// CHECK:STDOUT:   %Main.import_ref.8db: <witness> = import_ref Main//class2, inst23 [indirect], loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Main.import_ref.6a9 = import_ref Main//class2, inst24 [indirect], unloaded
+// CHECK:STDOUT:   %Main.import_ref.8db: <witness> = import_ref Main//class2, inst24 [indirect], loaded [concrete = constants.%complete_type]
+// CHECK:STDOUT:   %Main.import_ref.6a9 = import_ref Main//class2, inst25 [indirect], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {

toolchain/check/testdata/alias/import_access.carbon

@@ -97,7 +97,7 @@ var inst: Test.A = {};
 // CHECK:STDOUT:   %Test.C = import_ref Test//def, C, unloaded
 // CHECK:STDOUT:   %Test.A: type = import_ref Test//def, A, loaded [concrete = constants.%C]
 // CHECK:STDOUT:   %Test.import_ref.8f2: <witness> = import_ref Test//def, loc4_10, loaded [concrete = constants.%complete_type]
-// CHECK:STDOUT:   %Test.import_ref.2c4 = import_ref Test//def, inst16 [no loc], unloaded
+// CHECK:STDOUT:   %Test.import_ref.2c4 = import_ref Test//def, inst17 [no loc], unloaded
 // CHECK:STDOUT: }
 // CHECK:STDOUT:
 // CHECK:STDOUT: file {