style: reorganize scarlet's parser and codegen (#58)

* reorganize parser * nit * done with parser reorg * reorg codegen * update docs
Sh0g0-1758 · Feb 20, 2025 · a90adc3 · a90adc3
1 parent 9087598
commit a90adc3
Show file tree

Hide file tree

Showing 34 changed files with 2,869 additions and 2,649 deletions.
diff --git a/README.md b/README.md
@@ -6,10 +6,10 @@ Scarlet is a custom compiler developed in C++, designed with the primary goal of
 
 [![ubuntu](https://github.com/Sh0g0-1758/scarlet/actions/workflows/UBUNTU_test_gcc.yml/badge.svg)](https://github.com/Sh0g0-1758/scarlet/actions/workflows/UBUNTU_test_gcc.yml) [![macos](https://github.com/Sh0g0-1758/scarlet/actions/workflows/MACOS_test_clang.yml/badge.svg)](https://github.com/Sh0g0-1758/scarlet/actions/workflows/MACOS_test_clang.yml)
 
-To build Scarlet, invoke `cmake` with the `DCMAKE_CXX_COMPILER` flag within the build directory and then run `make`.
+To build Scarlet, invoke `cmake` with the `DCMAKE_CXX_COMPILER` flag within the build directory and then run `make`. By default, scarlet builds in `Release` mode (-O3).
 
 ```sh
-cmake -DCMAKE_CXX_COMPILER=<compiler> ..
+cmake -DCMAKE_CXX_COMPILER=<compiler> -DCMAKE_BUILD_TYPE=<Release/Debug> ..
 make
 ```
 

diff --git a/codegen/CMakeLists.txt b/codegen/CMakeLists.txt
@@ -1,6 +1,19 @@
 add_library(scarCodegen STATIC
     codegen.cc
     codegen.hh
+    common.hh
+
+    scargen/scar.cc
+    scargen/scar_statement.cc
+    scargen/scar_factor.cc
+    scargen/scar_expression.cc
+    scargen/scar_declaration.cc
+    scargen/scar_block.cc
+    scargen/pretty_print.cc
+
+    scasmgen/scasm_generation.cc
+    scasmgen/fix_pseudo_registers.cc
+    scasmgen/fix_instructions.cc
 )
 
 set_basic_compile_options(scarCodegen)

diff --git a/codegen/codegen.cc b/codegen/codegen.cc
diff --git a/codegen/codegen.hh b/codegen/codegen.hh
@@ -36,6 +36,11 @@ namespace scarlet {
 namespace codegen {
 
 #define NOTNULL(x) x != nullptr
+#define UNREACHABLE()                                                          \
+  std::cout << "Unreachable code reached in " << __FILE__ << " at line "       \
+            << __LINE__ << std::endl;                                          \
+  __builtin_unreachable();
+#define MAKE_SHARED(a, b) std::shared_ptr<a> b = std::make_shared<a>()
 
 class Codegen {
 private:

diff --git a/codegen/common.hh b/codegen/common.hh
@@ -0,0 +1,23 @@
+#pragma once
+
+#include "codegen.hh"
+
+namespace scarlet {
+namespace codegen {
+
+#define SETVARCONSTANTREG(src)                                                 \
+  if (!variable_buffer.empty()) {                                              \
+    src->set_type(scar::val_type::VAR);                                        \
+    src->set_reg_name(variable_buffer);                                        \
+    variable_buffer.clear();                                                   \
+  } else if (!constant_buffer.empty()) {                                       \
+    src->set_type(scar::val_type::CONSTANT);                                   \
+    src->set_value(constant_buffer);                                           \
+    constant_buffer.clear();                                                   \
+  } else {                                                                     \
+    src->set_type(scar::val_type::VAR);                                        \
+    src->set_reg_name(get_prev_reg_name());                                    \
+  }
+
+} // namespace codegen
+} // namespace scarlet
diff --git a/codegen/scargen/pretty_print.cc b/codegen/scargen/pretty_print.cc
@@ -0,0 +1,96 @@
+#include <codegen/codegen.hh>
+
+namespace scarlet {
+namespace codegen {
+
+void Codegen::pretty_print() {
+  std::cout << "Program(" << std::endl;
+  for (auto function : scar.get_functions()) {
+    std::cout << "\tFunction(" << std::endl;
+    std::cout << "\t\tname=\"" << function->get_identifier()->get_value()
+              << "\"," << std::endl;
+    std::cout << "\t\tbody=[" << std::endl;
+    for (auto statement : function->get_instructions()) {
+      std::cout << "\t\t\t" << to_string(statement->get_type()) << "(";
+      if (statement->get_type() == scar::instruction_type::RETURN) {
+        if (statement->get_src1()->get_type() == scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_src1()->get_value() << ")";
+        } else if (statement->get_src1()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_src1()->get_reg() << ")";
+        }
+        std::cout << ")" << std::endl;
+      } else if (statement->get_type() == scar::instruction_type::UNARY) {
+        std::cout << unop::to_string(statement->get_unop()) << ", ";
+        if (statement->get_src1()->get_type() == scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_src1()->get_value() << ")";
+        } else if (statement->get_src1()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_src1()->get_reg() << ")";
+        }
+        std::cout << ", ";
+        if (statement->get_dst()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_dst()->get_reg() << ")";
+        } else if (statement->get_dst()->get_type() ==
+                   scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_dst()->get_value() << ")";
+        }
+        std::cout << ")" << std::endl;
+      } else if (statement->get_type() == scar::instruction_type::BINARY) {
+        std::cout << binop::to_string(statement->get_binop()) << ", ";
+        if (statement->get_src1()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_src1()->get_reg() << ")";
+        } else if (statement->get_src1()->get_type() ==
+                   scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_src1()->get_value() << ")";
+        }
+        std::cout << ", ";
+        if (statement->get_src2()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_src2()->get_reg() << ")";
+        } else if (statement->get_src2()->get_type() ==
+                   scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_src2()->get_value() << ")";
+        }
+        std::cout << ", ";
+        if (statement->get_dst()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_dst()->get_reg() << ")";
+        } else if (statement->get_dst()->get_type() ==
+                   scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_dst()->get_value() << ")";
+        }
+        std::cout << ")" << std::endl;
+      } else if (statement->get_type() == scar::instruction_type::COPY) {
+        if (statement->get_src1()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_src1()->get_reg() << ")";
+        } else if (statement->get_src1()->get_type() ==
+                   scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_src1()->get_value() << ")";
+        }
+        std::cout << " ,";
+        if (statement->get_dst()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_dst()->get_reg() << ")";
+        }
+        std::cout << ")" << std::endl;
+      } else if (statement->get_type() == scar::instruction_type::JUMP or
+                 statement->get_type() == scar::instruction_type::LABEL) {
+        std::cout << statement->get_src1()->get_value() << ")" << std::endl;
+      } else if (statement->get_type() ==
+                     scar::instruction_type::JUMP_IF_ZERO or
+                 statement->get_type() ==
+                     scar::instruction_type::JUMP_IF_NOT_ZERO) {
+        if (statement->get_src1()->get_type() == scar::val_type::VAR) {
+          std::cout << "Var(" << statement->get_src1()->get_reg() << ")";
+        } else if (statement->get_src1()->get_type() ==
+                   scar::val_type::CONSTANT) {
+          std::cout << "Constant(" << statement->get_src1()->get_value() << ")";
+        }
+        std::cout << ", ";
+        std::cout << statement->get_dst()->get_value() << ")" << std::endl;
+      }
+    }
+    std::cout << "\t\t]" << std::endl;
+    std::cout << "\t)," << std::endl;
+  }
+  std::cout << ")" << std::endl;
+}
+
+} // namespace codegen
+} // namespace scarlet
diff --git a/codegen/scargen/scar.cc b/codegen/scargen/scar.cc
@@ -0,0 +1,33 @@
+#include <codegen/codegen.hh>
+
+namespace scarlet {
+namespace codegen {
+
+void Codegen::gen_scar() {
+  scar::scar_Program_Node scar_program;
+  for (auto it : program.get_functions()) {
+    MAKE_SHARED(scar::scar_Function_Node, scar_function);
+    MAKE_SHARED(scar::scar_Identifier_Node, identifier);
+    identifier->set_value(it->get_identifier()->get_value());
+    scar_function->set_identifier(identifier);
+
+    gen_scar_block(it->get_block(), scar_function);
+
+    // Add a complementary return 0 at the end of the function
+    // in case there is no return statement
+    MAKE_SHARED(scar::scar_Instruction_Node, scar_instruction);
+    scar_instruction->set_type(scar::instruction_type::RETURN);
+    MAKE_SHARED(scar::scar_Val_Node, scar_val_src);
+    scar_val_src->set_type(scar::val_type::CONSTANT);
+    scar_val_src->set_value("0");
+    scar_instruction->set_src1(scar_val_src);
+    scar_function->add_instruction(scar_instruction);
+
+    scar_program.add_function(scar_function);
+  }
+
+  this->scar = scar_program;
+}
+
+} // namespace codegen
+} // namespace scarlet
diff --git a/codegen/scargen/scar_block.cc b/codegen/scargen/scar_block.cc
@@ -0,0 +1,24 @@
+#include <codegen/codegen.hh>
+
+namespace scarlet {
+namespace codegen {
+
+void Codegen::gen_scar_block(
+    std::shared_ptr<ast::AST_Block_Node> block,
+    std::shared_ptr<scar::scar_Function_Node> scar_function) {
+  for (auto inst : block->get_blockItems()) {
+    switch (inst->get_type()) {
+    case ast::BlockItemType::STATEMENT:
+      gen_scar_statement(inst->get_statement(), scar_function);
+      break;
+    case ast::BlockItemType::DECLARATION:
+      gen_scar_declaration(inst->get_declaration(), scar_function);
+      break;
+    case ast::BlockItemType::UNKNOWN:
+      UNREACHABLE()
+    }
+  }
+}
+
+} // namespace codegen
+} // namespace scarlet
diff --git a/codegen/scargen/scar_declaration.cc b/codegen/scargen/scar_declaration.cc
@@ -0,0 +1,26 @@
+#include <codegen/common.hh>
+
+namespace scarlet {
+namespace codegen {
+
+void Codegen::gen_scar_declaration(
+    std::shared_ptr<ast::AST_Declaration_Node> declaration,
+    std::shared_ptr<scar::scar_Function_Node> scar_function) {
+  // if there is no definition, we ignore it
+  if (declaration->get_exp() != nullptr) {
+    gen_scar_exp(declaration->get_exp(), scar_function);
+    MAKE_SHARED(scar::scar_Instruction_Node, scar_instruction);
+    scar_instruction->set_type(scar::instruction_type::COPY);
+    MAKE_SHARED(scar::scar_Val_Node, scar_val_src);
+    SETVARCONSTANTREG(scar_val_src);
+    scar_instruction->set_src1(scar_val_src);
+    MAKE_SHARED(scar::scar_Val_Node, scar_val_dst);
+    scar_val_dst->set_type(scar::val_type::VAR);
+    scar_val_dst->set_reg_name(declaration->get_identifier()->get_value());
+    scar_instruction->set_dst(scar_val_dst);
+    scar_function->add_instruction(scar_instruction);
+  }
+}
+
+} // namespace codegen
+} // namespace scarlet