From d1c85ae6e8a52be8c5833ba37c938e5d27d7271e Mon Sep 17 00:00:00 2001
From: Jack Parsons <jparsons@ualberta.ca>
Date: Tue, 2 Jul 2024 13:47:27 -0600
Subject: [PATCH] 64 Bit Integers (#44)

* add i64

* add extra type check tests
---
 grammar/Fusion.g4                           |  4 ++-
 include/ast/ast.h                           |  6 ++--
 include/backend/types/integer.h             |  1 +
 include/shared/context.h                    |  3 +-
 include/shared/type/integer.h               |  8 +++++
 src/ast/ast.cpp                             | 15 +++++-----
 src/ast/builder.cpp                         | 20 +++++++++----
 src/ast/passes/type_check.cpp               | 33 ++++++++++++---------
 src/backend/builtin/print.cpp               |  2 +-
 src/backend/expressions/arithmetic.cpp      |  8 ++---
 src/backend/types/boolean.cpp               |  2 +-
 src/backend/types/integer.cpp               |  7 +++++
 src/backend/visitor.cpp                     |  6 +++-
 src/shared/context.cpp                      | 11 ++++---
 src/shared/type/integer.cpp                 | 14 +++++++++
 tests/input/declaration/basic.in            |  3 ++
 tests/input/errors/syntax/int_overflow.in   |  4 +++
 tests/input/errors/type/int_sizes.in        |  4 +++
 tests/output/declaration/basic.out          |  1 +
 tests/output/errors/syntax/int_overflow.out |  1 +
 tests/output/errors/type/int_sizes.out      |  1 +
 21 files changed, 109 insertions(+), 45 deletions(-)
 create mode 100644 tests/input/errors/syntax/int_overflow.in
 create mode 100644 tests/input/errors/type/int_sizes.in
 create mode 100644 tests/output/errors/syntax/int_overflow.out
 create mode 100644 tests/output/errors/type/int_sizes.out

diff --git a/grammar/Fusion.g4 b/grammar/Fusion.g4
index 2ceb6f5..387fcfc 100644
--- a/grammar/Fusion.g4
+++ b/grammar/Fusion.g4
@@ -45,7 +45,8 @@ expr
 qualifier: CONST | LET;
 
 type
-    : I32 
+    : I32
+    | I64 
     | CHAR
     | BOOL
     ;
@@ -88,6 +89,7 @@ COMMENT: '/*' .*? '*/' -> skip;
 
 // types
 I32: 'i32';
+I64: 'i64';
 CHAR: 'ch';
 BOOL: 'bool';
 
diff --git a/include/ast/ast.h b/include/ast/ast.h
index 3ed929f..68cec38 100644
--- a/include/ast/ast.h
+++ b/include/ast/ast.h
@@ -72,11 +72,11 @@ class Expression : public Node {
 
 class IntegerLiteral : public Expression {
    private:
-    int value;
+    long long value;
 
    public:
-    explicit IntegerLiteral(int value, Token* token);
-    int get_value() const;
+    explicit IntegerLiteral(long long value, TypePtr type, Token* token);
+    long long get_value() const;
 
     void xml(int level) override;
 };
diff --git a/include/backend/types/integer.h b/include/backend/types/integer.h
index 638592f..58752c0 100644
--- a/include/backend/types/integer.h
+++ b/include/backend/types/integer.h
@@ -4,4 +4,5 @@
 
 namespace integer {
 mlir::Value create_i32(int value);
+mlir::Value create_i64(long long value);
 }  // namespace integer
diff --git a/include/shared/context.h b/include/shared/context.h
index 2866149..8414ffd 100644
--- a/include/shared/context.h
+++ b/include/shared/context.h
@@ -19,9 +19,10 @@ extern mlir::MLIRContext context;
 extern TypePtr ch;
 extern TypePtr any;
 extern TypePtr i32;
+extern TypePtr i64;
 extern TypePtr f32;
 extern TypePtr none;
-extern TypePtr t_bool;
+extern TypePtr bool_;
 
 extern std::vector<TypePtr> primitives;
 
diff --git a/include/shared/type/integer.h b/include/shared/type/integer.h
index 3a9301f..cf67e13 100644
--- a/include/shared/type/integer.h
+++ b/include/shared/type/integer.h
@@ -9,3 +9,11 @@ class I32 : public Type {
     mlir::Type get_mlir() const override;
     bool is_numeric() const override;
 };
+
+class I64 : public Type {
+   public:
+    I64();
+    std::string get_specifier() const override;
+    mlir::Type get_mlir() const override;
+    bool is_numeric() const override;
+};
diff --git a/src/ast/ast.cpp b/src/ast/ast.cpp
index b8341c0..5241d7b 100644
--- a/src/ast/ast.cpp
+++ b/src/ast/ast.cpp
@@ -111,7 +111,7 @@ TypePtr ast::Expression::get_type() const {
 }
 
 ast::BooleanLiteral::BooleanLiteral(bool value, Token* token)
-    : Expression(ctx::t_bool, token) {
+    : Expression(ctx::bool_, token) {
     this->value = value;
 }
 
@@ -135,22 +135,23 @@ char ast::CharacterLiteral::get_value() const {
 }
 
 void ast::CharacterLiteral::xml(int level) {
-    std::cout << std::string(level * 4, ' ') << "<ch value=\"" << this->value
+    std::cout << std::string(level * 4, ' ') << "<ch value=\"" << value
               << "\"/>\n";
 }
 
-ast::IntegerLiteral::IntegerLiteral(int value, Token* token)
-    : Expression(ctx::i32, token) {
+ast::IntegerLiteral::IntegerLiteral(long long value, TypePtr type, Token* token)
+    : Expression(type, token) {
+    assert(*type == *ctx::i32 || *type == *ctx::i64);
     this->value = value;
 }
 
-int ast::IntegerLiteral::get_value() const {
+long long ast::IntegerLiteral::get_value() const {
     return this->value;
 }
 
 void ast::IntegerLiteral::xml(int level) {
-    std::cout << std::string(level * 4, ' ') << "<i32 value=\"" << this->value
-              << "\"/>\n";
+    std::cout << std::string(level * 4, ' ')
+              << "<" + type->get_name() + " value=\"" << value << "\"/>\n";
 }
 
 ast::Variable::Variable(Qualifier qualifier,
diff --git a/src/ast/builder.cpp b/src/ast/builder.cpp
index 5e40eaf..2f7c9df 100644
--- a/src/ast/builder.cpp
+++ b/src/ast/builder.cpp
@@ -5,6 +5,7 @@
 #include "FusionParser.h"
 #include "ast/ast.h"
 #include "ast/builder.h"
+#include "errors/errors.h"
 #include "shared/context.h"
 #include "shared/type/type.h"
 
@@ -99,16 +100,23 @@ std::any Builder::visitQualifier(FusionParser::QualifierContext* ctx) {
 
 std::any Builder::visitLiteralInt(FusionParser::LiteralIntContext* ctx) {
     Token* token = ctx->INT()->getSymbol();
-    int value = 0;
+    std::string str = ctx->INT()->getText();
 
     try {
-        value = std::stoi(ctx->INT()->getText());
-    } catch (const std::out_of_range& oor) {
-        std::runtime_error(oor.what());
+        int num = std::stoi(str);
+        auto node = make_shared<ast::IntegerLiteral>(num, ctx::i32, token);
+        return to_node(node);
+    } catch (const std::out_of_range&) {
+        try {
+            long long num = std::stoll(str);
+            auto node = make_shared<ast::IntegerLiteral>(num, ctx::i64, token);
+            return to_node(node);
+        } catch (const std::out_of_range&) {
+            throw SyntaxError(token->getLine(), "number overflowed 64 bits");
+        }
     }
 
-    auto node = make_shared<ast::IntegerLiteral>(value, token);
-    return to_node(node);
+    throw std::runtime_error("failed to parse integer");
 }
 
 std::any Builder::visitLiteralBool(FusionParser::LiteralBoolContext* ctx) {
diff --git a/src/ast/passes/type_check.cpp b/src/ast/passes/type_check.cpp
index f6d63e3..95a5b9e 100644
--- a/src/ast/passes/type_check.cpp
+++ b/src/ast/passes/type_check.cpp
@@ -9,12 +9,17 @@ void TypeCheck::visit_declaration(shared_ptr<ast::Declaration> node) {
     visit(node->var);
     visit(node->expr);
 
-    Type var = *node->var->get_type();
-    Type expr = *node->expr->get_type();
-    if (var != expr) {
+    TypePtr var = node->var->get_type();
+    TypePtr expr = node->expr->get_type();
+    if (*var == *ctx::i64 && *expr == *ctx::i32) {
+        node->expr->set_type(ctx::i64);
+        return;
+    }
+
+    if (*var != *expr) {
         throw TypeError(node->token->getLine(),
-                        "mismatched lhs(" + var.get_name() + ") and rhs(" +
-                            expr.get_name() + ") types on declaration");
+                        "mismatched lhs(" + var->get_name() + ") and rhs(" +
+                            expr->get_name() + ") types on declaration");
     }
 }
 
@@ -96,7 +101,7 @@ void TypeCheck::visit_binary_operator(shared_ptr<ast::BinaryOperator> node) {
         case ast::BinaryOpType::GTE:
             check_numeric(lhs, line);
             check_numeric(rhs, line);
-            node->set_type(ctx::t_bool);
+            node->set_type(ctx::bool_);
             break;
         case ast::BinaryOpType::ADD:
         case ast::BinaryOpType::SUB:
@@ -112,11 +117,11 @@ void TypeCheck::visit_binary_operator(shared_ptr<ast::BinaryOperator> node) {
         case ast::BinaryOpType::OR:
             check_bool(lhs, line);
             check_bool(rhs, line);
-            node->set_type(ctx::t_bool);
+            node->set_type(ctx::bool_);
             break;
         case ast::BinaryOpType::EQ:
         case ast::BinaryOpType::NE:
-            node->set_type(ctx::t_bool);
+            node->set_type(ctx::bool_);
             break;
     }
 }
@@ -125,17 +130,17 @@ void TypeCheck::visit_unary_operator(shared_ptr<ast::UnaryOperator> node) {
     visit(node->rhs);
 
     size_t line = node->token->getLine();
-    Type rhs = *node->get_type();
-    if (node->type == ast::UnaryOpType::MINUS && rhs != *ctx::i32) {
+    TypePtr rhs = node->get_type();
+    if (node->type == ast::UnaryOpType::MINUS && !rhs->is_numeric()) {
         throw TypeError(
             line, "unary minus only works on numeric types, found type: " +
-                      rhs.get_name());
+                      rhs->get_name());
     }
 
-    if (node->type == ast::UnaryOpType::NOT && rhs != *ctx::t_bool) {
+    if (node->type == ast::UnaryOpType::NOT && *rhs != *ctx::bool_) {
         throw TypeError(line,
                         "unary minus only works on booleans, found type: " +
-                            rhs.get_name());
+                            rhs->get_name());
     }
 }
 
@@ -146,7 +151,7 @@ void TypeCheck::check_numeric(TypePtr type, size_t line) {
 }
 
 void TypeCheck::check_bool(TypePtr type, size_t line) {
-    if (*type != *ctx::t_bool) {
+    if (*type != *ctx::bool_) {
         throw TypeError(line, "type(" + type->get_name() + ") is not boolean");
     }
 }
diff --git a/src/backend/builtin/print.cpp b/src/backend/builtin/print.cpp
index 65a8644..3bd7e60 100644
--- a/src/backend/builtin/print.cpp
+++ b/src/backend/builtin/print.cpp
@@ -29,7 +29,7 @@ void setupPrintf() {
 
 void newline_type_str(TypePtr type) {
     std::string ty = type->get_specifier() + "\n";
-    auto gvalue = mlir::StringRef(ty.c_str(), 4);
+    auto gvalue = mlir::StringRef(ty.c_str(), ty.size() + 1);
 
     auto gtype =
         mlir::LLVM::LLVMArrayType::get(ctx::ch->get_mlir(), gvalue.size());
diff --git a/src/backend/expressions/arithmetic.cpp b/src/backend/expressions/arithmetic.cpp
index d73ff61..6410da7 100644
--- a/src/backend/expressions/arithmetic.cpp
+++ b/src/backend/expressions/arithmetic.cpp
@@ -36,11 +36,11 @@ mlir::Value arithmetic::mod(mlir::Value lhs, mlir::Value rhs, TypePtr type) {
 }
 
 mlir::Value arithmetic::and_(mlir::Value lhs, mlir::Value rhs, TypePtr type) {
-    return binop<mlir::LLVM::AndOp>(lhs, rhs, ctx::t_bool);
+    return binop<mlir::LLVM::AndOp>(lhs, rhs, ctx::bool_);
 }
 
 mlir::Value arithmetic::or_(mlir::Value lhs, mlir::Value rhs, TypePtr type) {
-    return binop<mlir::LLVM::OrOp>(lhs, rhs, ctx::t_bool);
+    return binop<mlir::LLVM::OrOp>(lhs, rhs, ctx::bool_);
 }
 
 mlir::Value arithmetic::pow(mlir::Value lhs, mlir::Value rhs, TypePtr type) {
@@ -79,7 +79,7 @@ mlir::Value arithmetic::lte(mlir::Value lhs, mlir::Value rhs, TypePtr type) {
 
 mlir::Value arithmetic::not_(mlir::Value value) {
     mlir::Value f = boolean::create_bool(false);
-    return eq(f, value, ctx::t_bool);
+    return eq(f, value, ctx::bool_);
 }
 
 mlir::Value arithmetic::negate(mlir::Value value, TypePtr type) {
@@ -92,7 +92,7 @@ mlir::Value arithmetic::binary_equality(mlir::Value lhs,
                                         TypePtr type,
                                         mlir::LLVM::ICmpPredicate predicate) {
     mlir::Value value = ctx::builder->create<mlir::LLVM::ICmpOp>(
-        *ctx::loc, ctx::t_bool->get_mlir(), predicate, lhs, rhs);
+        *ctx::loc, ctx::bool_->get_mlir(), predicate, lhs, rhs);
 
     return value;
 }
diff --git a/src/backend/types/boolean.cpp b/src/backend/types/boolean.cpp
index 41f8521..2ef0720 100644
--- a/src/backend/types/boolean.cpp
+++ b/src/backend/types/boolean.cpp
@@ -3,7 +3,7 @@
 
 mlir::Value boolean::create_bool(bool value) {
     mlir::Value val = ctx::builder->create<mlir::LLVM::ConstantOp>(
-        *ctx::loc, ctx::t_bool->get_mlir(), value);
+        *ctx::loc, ctx::bool_->get_mlir(), value);
 
     return val;
 }
diff --git a/src/backend/types/integer.cpp b/src/backend/types/integer.cpp
index c4ee90d..d858d7d 100644
--- a/src/backend/types/integer.cpp
+++ b/src/backend/types/integer.cpp
@@ -7,3 +7,10 @@ mlir::Value integer::create_i32(int value) {
 
     return val;
 }
+
+mlir::Value integer::create_i64(long long value) {
+    mlir::Value val = ctx::builder->create<mlir::LLVM::ConstantOp>(
+        *ctx::loc, ctx::i64->get_mlir(), value);
+
+    return val;
+}
diff --git a/src/backend/visitor.cpp b/src/backend/visitor.cpp
index 2f88764..89c0218 100644
--- a/src/backend/visitor.cpp
+++ b/src/backend/visitor.cpp
@@ -46,7 +46,11 @@ mlir::Value Backend::visit_block(shared_ptr<ast::Block> node) {
 
 mlir::Value Backend::visit_integer_literal(
     shared_ptr<ast::IntegerLiteral> node) {
-    return integer::create_i32(node->get_value());
+    if (*node->get_type() == *ctx::i32) {
+        return integer::create_i32(node->get_value());
+    }
+
+    return integer::create_i64(node->get_value());
 }
 
 mlir::Value Backend::visit_character_literal(
diff --git a/src/shared/context.cpp b/src/shared/context.cpp
index 61962c7..fa8e7bd 100644
--- a/src/shared/context.cpp
+++ b/src/shared/context.cpp
@@ -14,9 +14,10 @@ std::unique_ptr<mlir::ModuleOp> ctx::module;
 TypePtr ctx::ch;
 TypePtr ctx::any;
 TypePtr ctx::i32;
+TypePtr ctx::i64;
 TypePtr ctx::f32;
 TypePtr ctx::none;
-TypePtr ctx::t_bool;
+TypePtr ctx::bool_;
 
 std::vector<TypePtr> ctx::primitives;
 
@@ -26,16 +27,14 @@ void ctx::initialize_context() {
 
     ctx::any = make_shared<Any>();
     ctx::i32 = make_shared<I32>();
+    ctx::i64 = make_shared<I64>();
     ctx::f32 = make_shared<F32>();
     ctx::none = make_shared<None>();
     ctx::ch = make_shared<Character>();
-    ctx::t_bool = make_shared<Boolean>();
+    ctx::bool_ = make_shared<Boolean>();
 
     ctx::primitives = {
-        ctx::ch,
-        ctx::i32,
-        ctx::f32,
-        ctx::t_bool,
+        ctx::ch, ctx::i32, ctx::i64, ctx::f32, ctx::bool_,
     };
 
     module = std::make_unique<mlir::ModuleOp>(
diff --git a/src/shared/type/integer.cpp b/src/shared/type/integer.cpp
index d38da35..aecceb7 100644
--- a/src/shared/type/integer.cpp
+++ b/src/shared/type/integer.cpp
@@ -14,3 +14,17 @@ mlir::Type I32::get_mlir() const {
 bool I32::is_numeric() const {
     return true;
 }
+
+I64::I64() : Type("i64") {}
+
+std::string I64::get_specifier() const {
+    return "%ld";
+}
+
+mlir::Type I64::get_mlir() const {
+    return ctx::builder->getI64Type();
+}
+
+bool I64::is_numeric() const {
+    return true;
+}
diff --git a/tests/input/declaration/basic.in b/tests/input/declaration/basic.in
index d843647..fc4f015 100644
--- a/tests/input/declaration/basic.in
+++ b/tests/input/declaration/basic.in
@@ -14,5 +14,8 @@ fn main(): i32 {
     println(t);
     println(f);
 
+    let r: i64 = 10000000000000;
+    println(r);
+
     return 0;
 }
diff --git a/tests/input/errors/syntax/int_overflow.in b/tests/input/errors/syntax/int_overflow.in
new file mode 100644
index 0000000..b26134b
--- /dev/null
+++ b/tests/input/errors/syntax/int_overflow.in
@@ -0,0 +1,4 @@
+fn main(): i32 {
+    let x: i64 = 100000000000000000000000000000000000;
+    return 0;
+}
diff --git a/tests/input/errors/type/int_sizes.in b/tests/input/errors/type/int_sizes.in
new file mode 100644
index 0000000..af8869f
--- /dev/null
+++ b/tests/input/errors/type/int_sizes.in
@@ -0,0 +1,4 @@
+fn main(): i32 {
+    let x: i32 = 10000000000000;
+    return 0;
+}
\ No newline at end of file
diff --git a/tests/output/declaration/basic.out b/tests/output/declaration/basic.out
index a833f4c..b2cfae9 100644
--- a/tests/output/declaration/basic.out
+++ b/tests/output/declaration/basic.out
@@ -1,3 +1,4 @@
 55a
 1
 0
+10000000000000
diff --git a/tests/output/errors/syntax/int_overflow.out b/tests/output/errors/syntax/int_overflow.out
new file mode 100644
index 0000000..8142565
--- /dev/null
+++ b/tests/output/errors/syntax/int_overflow.out
@@ -0,0 +1 @@
+SyntaxError on Line 2
\ No newline at end of file
diff --git a/tests/output/errors/type/int_sizes.out b/tests/output/errors/type/int_sizes.out
new file mode 100644
index 0000000..cae1341
--- /dev/null
+++ b/tests/output/errors/type/int_sizes.out
@@ -0,0 +1 @@
+TypeError on Line 2
\ No newline at end of file