clox/compiler.c

/**
  ******************************************************************************
  * @file           : compiler.c
  * @author         : simon
  * @brief          : None
  * @attention      : None
  * @date           : 2023/8/17
  ******************************************************************************
  */
#include "compiler.h"
#include "common.h"
#include "scanner.h"
#ifdef DEBUG_PRINT_CODE
#include "debug.h"
#endif

typedef struct {
  Token current;
  Token previous;
  bool hadError;
  bool panicMode;
} Parser;

typedef enum {
  PREC_NONE,
  PREC_ASSIGNMENT,// =
  PREC_OR,        // or
  PREC_AND,       // and
  PREC_EQUALITY,  // == !=
  PREC_COMPARISON,// < > <= >=
  PREC_TERM,      // + -
  PREC_FACTOR,    // * /
  PREC_UNARY,     // ! -
  PREC_CALL,      // . ()
  PREC_PRIMARY
} Precedence;

typedef void (*ParseFn)(bool canAssign);

typedef struct {
  ParseFn prefix;
  ParseFn infix;
  Precedence precedence;
} ParseRule;

typedef struct {
  Token name;
  int depth;
} Local;

typedef enum {
  TYPE_FUNCTION,
  TYPE_SCRIPT,
} FunctionType;

typedef struct Compiler {
  struct Compiler *enclosing;
  ObjFunction *function;
  FunctionType type;

  Local locals[UINT8_COUNT];
  int localCount;/// how many locals are in scope
  int scopeDepth;
} Compiler;

Parser parser;
Compiler *current = NULL;

static void parsePrecedence(Precedence);
static uint8_t parseVariable(const char *);
static void defineVariable(uint8_t);
static uint8_t identifierConstant(Token *name);
static bool identifiersEqual(Token *a, Token *b);
static ParseRule *getRule(TokenType);
static void statement();
static void declaration();
static void markInitialized();

static void initCompiler(Compiler *compiler, FunctionType type);

static void beginScope();
static void endScope();

static void block();
static ObjFunction *endCompiler();

static Chunk *currentChunk() {
  return &current->function->chunk;
}

static void errorAt(Token *token, const char *message) {
  if (parser.panicMode) return;
  parser.panicMode = true;
  fprintf(stderr, "[line %d] Error", token->line);

  if (token->type == TOKEN_EOF) {
    fprintf(stderr, " at end");
  } else if (token->type == TOKEN_ERROR) {
    ///! Nothing.
  } else {
    fprintf(stderr, " at '%.*s'", token->length, token->start);
  }

  fprintf(stderr, ": %s\n", message);
  parser.hadError = true;
}
static void error(const char *message) {
  errorAt(&parser.previous, message);
}
static void errorAtCurrent(const char *message) {
  errorAt(&parser.current, message);
}
static void advance() {
  parser.previous = parser.current;

  for (;;) {
    parser.current = scanToken();
    if (parser.current.type != TOKEN_ERROR) break;

    errorAtCurrent(parser.current.start);
  }
}
void consume(TokenType type, const char *message) {
  if (parser.current.type == type) {
    advance();
    return;
  }

  errorAtCurrent(message);
}
static bool check(TokenType type) {
  return parser.current.type == type;
}
static bool match(TokenType type) {
  if (!check(type)) return false;
  advance();
  return true;
}
static void emitByte(uint8_t byte) {
  writeChunk(currentChunk(), byte, parser.previous.line);
}
static void emitBytes(uint8_t byte1, uint8_t byte2) {
  emitByte(byte1);
  emitByte(byte2);
}
static void emitLoop(int loopStart) {
  emitByte(OP_LOOP);

  int offset = currentChunk()->count - loopStart + 2;
  if (offset > UINT16_MAX) error("Loop body too large.");

  emitByte((offset >> 8) & 0xFF);
  emitByte(offset & 0xFF);
}
static int emitJump(uint8_t instruction) {
  emitByte(instruction);
  emitByte(0xff);// 占位字节，后面填入需要跳转的 offset
  emitByte(0xff);//
  return currentChunk()->count - 2;
}
static void emitReturn() {
  emitByte(OP_NIL);
  emitByte(OP_RETURN);
}
static uint8_t makeConstant(Value value) {
  int constant = addConstant(currentChunk(), value);
  if (constant > UINT8_MAX) {
    error("Too many constants in one chunk.");
    return 0;
  }
  return (uint8_t) constant;
}
static void expression() {
  parsePrecedence(PREC_ASSIGNMENT);
}
static void function(FunctionType type) {
  Compiler compiler;
  initCompiler(&compiler, type);
  beginScope();

  consume(TOKEN_LEFT_PAREN, "Expect '(' after function name.");

  ///! Parameters
  if (!check(TOKEN_RIGHT_PAREN)) {
    do {
      current->function->arity++;
      if (current->function->arity > 255) {
        errorAtCurrent("Can't have more than 255 parameters.");
      }
      uint8_t constant = parseVariable("Expect parameter name.");
      defineVariable(constant);
    } while (match(TOKEN_COMMA));
  }
  consume(TOKEN_RIGHT_PAREN, "Expect ')' after parameters.");
  consume(TOKEN_LEFT_BRACE, "Expect '{' before function body.");
  ///! Body
  block();

  ObjFunction *fun = endCompiler();
  emitBytes(OP_CONSTANT, makeConstant(OBJ_VAL(fun)));
}
static void funDeclaration() {
  uint8_t global = parseVariable("Expect function name.");
  markInitialized();
  function(TYPE_FUNCTION);
  defineVariable(global);
}
/// for example: var a; var b = 10;
static void varDeclaration() {
  uint8_t global = parseVariable("Expect variable name.");

  // 变量初始化
  if (match(TOKEN_EQUAL)) {
    expression();
  } else {
    emitByte(OP_NIL);
  }

  consume(TOKEN_SEMICOLON, "Expect ';' after variable declaration.");

  defineVariable(global);
}
static void expressionStatement() {
  expression();
  consume(TOKEN_SEMICOLON, "Expect ';' after expression.");
  emitByte(OP_POP);
}
static void patchJump(int offset) {
  // -2 to adjust for the bytecode for the jump offset itself.
  int jump = currentChunk()->count - offset - 2;

  if (jump > UINT16_MAX) error("Too much code to jump over.");

  currentChunk()->code[offset] = (jump >> 8) & 0xFF;
  currentChunk()->code[offset + 1] = jump & 0xFF;
}
static void forStatement() {
  /**
   * *********************************************************
   *  initializer clause
   * L4: condition expression
   *  OP_JUMP_IF_FALSE L1
   *  OP_POP
   *  OP_JUMP L3
   * L2: increment expression
   *  OP_POP
   * OP_LOOP L4
   * L3: body statement
   *  OP_LOOP L2
   *  L1: OP_POP
   *  continues...
   * *********************************************************
   */
  beginScope();
  // for(I;II;III)
  consume(TOKEN_LEFT_PAREN, "Expect '(' after 'for'.");

  ///! I.<<Initializer Part>>
  if (match(TOKEN_SEMICOLON)) {
    // No initializer.
  } else if (match(TOKEN_VAR)) {
    varDeclaration();
  } else {
    expressionStatement();// ; op_pop
  }

  ///! II.<<Condition Part>>
  int loopStart = currentChunk()->count;
  int exitJump = -1;
  if (!match(TOKEN_SEMICOLON)) {
    expression();
    consume(TOKEN_SEMICOLON, "Expect ';' after loop conditon. ");

    // Jump out of the loop if the condition is false.
    exitJump = emitJump(OP_JUMP_IF_FALSE);
    emitByte(OP_POP);
  }

  ///! III.<<Increment Clause>>
  if (!match(TOKEN_RIGHT_PAREN)) {
    int bodyJump = emitJump(OP_JUMP);
    int incrementStart = currentChunk()->count;
    expression();
    emitByte(OP_POP);
    consume(TOKEN_RIGHT_PAREN, "Expect ')' after for clauses.");

    emitLoop(loopStart);
    loopStart = incrementStart;
    patchJump(bodyJump);
  }

  ///! <<Body Part>>
  statement();
  emitLoop(loopStart);

  if (exitJump != -1) {
    patchJump(exitJump);
    emitByte(OP_POP);// Condition.
  }
  endScope();
}
static void returnStatement() {
  if (current->type == TYPE_SCRIPT) {
    error("Can't return from top-level code.");
  }

  if (match(TOKEN_SEMICOLON)) {
    emitReturn();
  } else {
    expression();
    consume(TOKEN_SEMICOLON, "Expect ';' after return value.");
    emitByte(OP_RETURN);
  }
}
static void ifStatement() {
  consume(TOKEN_LEFT_PAREN, "Expect '(' after 'if'.");
  expression();
  consume(TOKEN_RIGHT_PAREN, "Expect ')' after condition.");

  int thenJump = emitJump(OP_JUMP_IF_FALSE);
  emitByte(OP_POP);
  statement();// then branch statement

  int elseJump = emitJump(OP_JUMP);

  patchJump(thenJump);
  emitByte(OP_POP);

  if (match(TOKEN_ELSE)) statement();// else branch statement
  patchJump(elseJump);
}
static void whileStatement() {
  int loopStart = currentChunk()->count;
  consume(TOKEN_LEFT_PAREN, "Expect '(' after 'while'.");
  expression();// while condition.
  consume(TOKEN_RIGHT_PAREN, "Expect ')' after condition.");

  int exitJump = emitJump(OP_JUMP_IF_FALSE);
  emitByte(OP_POP);

  statement();// while body.
  emitLoop(loopStart);

  patchJump(exitJump);
  emitByte(OP_POP);
}
static void printStatement() {
  expression();
  consume(TOKEN_SEMICOLON, "Expect ';' after value.");
  emitByte(OP_PRINT);
}
static void synchronize() {
  parser.panicMode = false;
  while (parser.current.type != TOKEN_EOF) {
    if (parser.previous.type == TOKEN_SEMICOLON) return;
    switch (parser.current.type) {
      case TOKEN_CLASS:
      case TOKEN_FUN:
      case TOKEN_VAR:
      case TOKEN_FOR:
      case TOKEN_IF:
      case TOKEN_WHILE:
      case TOKEN_PRINT:
      case TOKEN_RETURN:
        return;

      default:;// Do nothing.
    }
    advance();
  }
}
///declaration    → classDecl
///               | funDecl
///               | varDecl
///               | statement ;
static void declaration() {
  if (match(TOKEN_FUN)) {
    funDeclaration();
  } else if (match(TOKEN_VAR)) {
    varDeclaration();
  } else {
    statement();
  }

  if (parser.panicMode) synchronize();
}
static void block() {
  while (!check(TOKEN_RIGHT_BRACE) && !check(TOKEN_EOF)) {
    declaration();
  }

  consume(TOKEN_RIGHT_BRACE, "Expect '}' after block.");
}
///statement      → exprStmt
///               | forStmt
///               | ifStmt
///               | printStmt
///               | returnStmt
///               | whileStmt
///               | block ;
static void statement() {
  if (match(TOKEN_PRINT)) {
    printStatement();
  } else if (match(TOKEN_FOR)) {
    forStatement();
  } else if (match(TOKEN_IF)) {
    ifStatement();
  } else if (match(TOKEN_RETURN)) {
    returnStatement();
  } else if (match(TOKEN_WHILE)) {
    whileStatement();
  } else if (match(TOKEN_LEFT_BRACE)) {
    ///! block → "{" declaration* "}" ;
    beginScope();
    block();
    endScope();
  } else {
    expressionStatement();
  }
}
static void beginScope() {
  current->scopeDepth++;
}
static void endScope() {
  current->scopeDepth--;

  while (current->localCount > 0
         && current->locals[current->localCount - 1].depth > current->scopeDepth) {
    emitByte(OP_POP);
    current->localCount--;
  }
}
static void emitConstant(Value value) {
  emitBytes(OP_CONSTANT, makeConstant(value));
}
static void initCompiler(Compiler *compiler, FunctionType type) {
  compiler->enclosing = current;// 上一极的 compiler
  compiler->function = NULL;
  compiler->type = type;
  compiler->localCount = 0;
  compiler->scopeDepth = 0;
  compiler->function = newFunction();
  current = compiler;
  if (type != TYPE_SCRIPT) {
    ///! Function name
    current->function->name = copyString(parser.previous.start, parser.previous.length);
  }

  Local *local = &current->locals[current->localCount++];
  local->depth = 0;
  local->name.start = "";
  local->name.length = 0;
}
static ObjFunction *endCompiler() {
  emitReturn();
  ObjFunction *function = current->function;
#ifdef DEBUG_PRINT_CODE
  if (!parser.hadError) {
    disassembleChunk(currentChunk(), function->name != NULL ? function->name->chars : "<script>");
  }
#endif

  current = current->enclosing;// 当前的 compiler 变成上级 compiler
  return function;
}
static void grouping(__attribute__((unused)) bool canAssign) {
  expression();
  consume(TOKEN_RIGHT_PAREN, "Expect ')' after expression.");
}
/// \brief parse number token
/// Number literals: 123
static void number(__attribute__((unused)) bool canAssign) {
  double value = strtod(parser.previous.start, NULL);
  emitConstant(NUMBER_VAL(value));
}
static void string(__attribute__((unused)) bool canAssign) {
  emitConstant(OBJ_VAL(copyString(parser.previous.start + 1,
                                  parser.previous.length - 2)));
}
static void and_(__attribute__((unused)) bool canAssign) {
  /// left operand expression
  /// OP_JUMP_IF_FALSE b
  /// OP_POP
  /// right operand expression
  /// b: continues...
  int endJump = emitJump(OP_JUMP_IF_FALSE);

  emitByte(OP_POP);

  parsePrecedence(PREC_AND);// right operand expression

  patchJump(endJump);
}
static void or_(__attribute__((unused)) bool canAssign) {
  /// left operand expression
  /// OP_JUMP_IF_FALSE b1
  /// OP_JUMP b2
  /// b1: OP_POP
  /// right operand expression
  /// b2: continue
  int elseJump = emitJump(OP_JUMP_IF_FALSE);
  int endJump = emitJump(OP_JUMP);

  patchJump(elseJump);
  emitByte(OP_POP);

  parsePrecedence(PREC_OR);// right operand expression
  patchJump(endJump);
}
static int resolveLocal(Compiler *compile, Token *name) {
  for (int i = compile->localCount - 1; i >= 0; i--) {
    Local *local = &compile->locals[i];
    if (identifiersEqual(name, &local->name)) {
      if (local->depth == -1) {
        error("Can't read local variable in ints own initializer.");
      }
      return i;
    }
  }
  return -1;
}
static void namedVariable(Token name, bool canAssign) {
  uint8_t getOp, setOp;
  int arg = resolveLocal(current, &name);
  if (arg != -1) {
    getOp = OP_GET_LOCAL;
    setOp = OP_SET_LOCAL;
  } else {
    arg = identifierConstant(&name);
    getOp = OP_GET_GLOBAL;
    setOp = OP_SET_GLOBAL;
  }

  if (canAssign && match(TOKEN_EQUAL)) {
    // 如 menu.brunch(sunday).beverage = "mimosa";
    expression();
    emitBytes(setOp, (uint8_t) arg);
  } else {
    emitBytes(getOp, (uint8_t) arg);
  }
}
static void variable(bool canAssign) {
  namedVariable(parser.previous, canAssign);
}
/// Unary negation: -123
static void unary(__attribute__((unused)) bool canAssign) {
  TokenType operatorType = parser.previous.type;

  // Compile the operand.
  parsePrecedence(PREC_UNARY);

  // Emit the operator instruction.
  switch (operatorType) {
    case TOKEN_BANG:
      emitByte(OP_NOT);
      break;
    case TOKEN_MINUS:
      emitByte(OP_NEGATE);
      break;
    default: return;// Unreachable.
  }
}
/// \brief infix parser
static void binary(__attribute__((unused)) bool canAssign) {
  TokenType operatorType = parser.previous.type;
  ParseRule *rule = getRule(operatorType);
  parsePrecedence((Precedence) rule->precedence + 1);

  switch (operatorType) {
    case TOKEN_BANG_EQUAL:
      emitBytes(OP_EQUAL, OP_NOT);
      break;
    case TOKEN_EQUAL_EQUAL:
      emitByte(OP_EQUAL);
      break;
    case TOKEN_GREATER:
      emitByte(OP_GREATER);
      break;
    case TOKEN_GREATER_EQUAL:
      emitBytes(OP_LESS, OP_NOT);
      break;
    case TOKEN_LESS:
      emitByte(OP_LESS);
      break;
    case TOKEN_LESS_EQUAL:
      emitBytes(OP_GREATER, OP_NOT);
      break;
    case TOKEN_PLUS:
      emitByte(OP_ADD);
      break;
    case TOKEN_MINUS:
      emitByte(OP_SUBTRACT);
      break;
    case TOKEN_STAR:
      emitByte(OP_MULTIPLY);
      break;
    case TOKEN_SLASH:
      emitByte(OP_DIVIDE);
      break;
    default: return;// Unreachable.
  }
}
static uint8_t argumentList() {
  uint8_t argCount = 0;
  if (!check(TOKEN_RIGHT_PAREN)) {
    do {
      expression();
      if (argCount == 255) error("Can't have more than 255 arguments.");
      argCount++;
    } while (match(TOKEN_COMMA));
  }
  consume(TOKEN_RIGHT_PAREN, "Expect ')' after arguments.");
  return argCount;
}
/// Function call expression is kind of an infix ( operator.
/// You have a high-precedence expression on the left for the thing
/// being called - usually just a single identifier.
/// Then the ( in the middle, followed by the argument expressions
/// separated by commas, and a final ) to wrap it up at the end.
static void call(__attribute__((unused)) bool canAssign) {
  uint8_t argCount = argumentList();
  emitBytes(OP_CALL, argCount);
}
static void literal(__attribute__((unused)) bool canAssign) {
  switch (parser.previous.type) {
    case TOKEN_FALSE:
      emitByte(OP_FALSE);
      break;
    case TOKEN_NIL:
      emitByte(OP_NIL);
      break;
    case TOKEN_TRUE:
      emitByte(OP_TRUE);
      break;
    default: return;// Unreachable
  }
}
ParseRule rules[] = {
    [TOKEN_LEFT_PAREN] = {grouping, call, PREC_CALL},
    [TOKEN_RIGHT_PAREN] = {NULL, NULL, PREC_NONE},
    [TOKEN_LEFT_BRACE] = {NULL, NULL, PREC_NONE},
    [TOKEN_RIGHT_BRACE] = {NULL, NULL, PREC_NONE},
    [TOKEN_COMMA] = {NULL, NULL, PREC_NONE},
    [TOKEN_DOT] = {NULL, NULL, PREC_NONE},
    [TOKEN_MINUS] = {unary, binary, PREC_TERM},
    [TOKEN_PLUS] = {NULL, binary, PREC_TERM},
    [TOKEN_SEMICOLON] = {NULL, NULL, PREC_NONE},
    [TOKEN_SLASH] = {NULL, binary, PREC_FACTOR},
    [TOKEN_STAR] = {NULL, binary, PREC_FACTOR},
    [TOKEN_BANG] = {unary, NULL, PREC_NONE},
    [TOKEN_BANG_EQUAL] = {NULL, binary, PREC_EQUALITY},
    [TOKEN_EQUAL] = {NULL, NULL, PREC_NONE},
    [TOKEN_EQUAL_EQUAL] = {NULL, binary, PREC_EQUALITY},
    [TOKEN_GREATER] = {NULL, binary, PREC_COMPARISON},
    [TOKEN_GREATER_EQUAL] = {NULL, binary, PREC_COMPARISON},
    [TOKEN_LESS] = {NULL, binary, PREC_COMPARISON},
    [TOKEN_LESS_EQUAL] = {NULL, binary, PREC_COMPARISON},
    [TOKEN_IDENTIFIER] = {variable, NULL, PREC_NONE},
    [TOKEN_STRING] = {string, NULL, PREC_NONE},
    [TOKEN_NUMBER] = {number, NULL, PREC_NONE},
    [TOKEN_AND] = {NULL, and_, PREC_AND},
    [TOKEN_CLASS] = {NULL, NULL, PREC_NONE},
    [TOKEN_ELSE] = {NULL, NULL, PREC_NONE},
    [TOKEN_FALSE] = {literal, NULL, PREC_NONE},
    [TOKEN_FOR] = {NULL, NULL, PREC_NONE},
    [TOKEN_FUN] = {NULL, NULL, PREC_NONE},
    [TOKEN_IF] = {NULL, NULL, PREC_NONE},
    [TOKEN_NIL] = {literal, NULL, PREC_NONE},
    [TOKEN_OR] = {NULL, or_, PREC_OR},
    [TOKEN_PRINT] = {NULL, NULL, PREC_NONE},
    [TOKEN_RETURN] = {NULL, NULL, PREC_NONE},
    [TOKEN_SUPER] = {NULL, NULL, PREC_NONE},
    [TOKEN_THIS] = {NULL, NULL, PREC_NONE},
    [TOKEN_TRUE] = {literal, NULL, PREC_NONE},
    [TOKEN_VAR] = {NULL, NULL, PREC_NONE},
    [TOKEN_WHILE] = {NULL, NULL, PREC_NONE},
    [TOKEN_ERROR] = {NULL, NULL, PREC_NONE},
    [TOKEN_EOF] = {NULL, NULL, PREC_NONE},
};

/// \brief 优先级处理
/// \param precedence
static void parsePrecedence(Precedence precedence) {
  advance();
  ParseFn prefixRule = getRule(parser.previous.type)->prefix;
  if (prefixRule == NULL) {
    error("Expect expression.");
    return;
  }

  bool canAssign = precedence <= PREC_ASSIGNMENT;
  prefixRule(canAssign);///! 执行具体函数

  while (precedence <= getRule(parser.current.type)->precedence) {
    advance();
    ParseFn infixRule = getRule(parser.previous.type)->infix;
    infixRule(canAssign);///! 执行具体函数
  }

  if (canAssign && match(TOKEN_EQUAL)) {
    error("Invalid assignment target.");
  }
}
static uint8_t identifierConstant(Token *name) {
  return makeConstant(OBJ_VAL(copyString(name->start, name->length)));
}
static bool identifiersEqual(Token *a, Token *b) {
  if (a->length != b->length) return false;
  return memcmp(a->start, b->start, a->length) == 0;
}
static void addLocal(Token name) {
  if (current->localCount == UINT8_COUNT) {
    error("Too many local variables in function.");
    return;
  }

  Local *local = &current->locals[current->localCount++];
  local->name = name;
  local->depth = -1;// 未初始化
}
static void declareVariable() {
  if (current->scopeDepth == 0) return;

  Token *name = &parser.previous;

  // we detect the error like:
  // {
  //   var a = "first";
  //   var a = "second";
  // }
  for (int i = current->localCount - 1; i >= 0; i--) {
    Local *local = &current->locals[i];
    if (local->depth != -1 && local->depth < current->scopeDepth) {
      break;
    }

    if (identifiersEqual(name, &local->name)) {
      error("Already a variable with the name in this scope.");
    }
  }

  addLocal(*name);
}
static uint8_t parseVariable(const char *errorMessage) {
  consume(TOKEN_IDENTIFIER, errorMessage);

  declareVariable();// 处理本地变量
  if (current->scopeDepth > 0) return 0;

  return identifierConstant(&parser.previous);
}
static void markInitialized() {
  if (current->scopeDepth == 0) return;
  current->locals[current->localCount - 1].depth = current->scopeDepth;
}
static void defineVariable(uint8_t global) {

  if (current->scopeDepth > 0) {
    // markInitialized 未初始化时值 为 -1
    markInitialized();
    // 本地变量，直接退出
    return;
  }
  emitBytes(OP_DEFINE_GLOBAL, global);
}
static ParseRule *getRule(TokenType type) {
  return &rules[type];
}

/**
  ******************************************************************************
  * statement      → exprStmt
                      | forStmt
                      | ifStmt
                      | printStmt
                      | returnStmt
                      | whileStmt
                      | block ;
    block          → "{" declaration* "}" ;

    declaration    → classDecl
                   | funDecl
                   | varDecl
                   | statement ;
  ******************************************************************************
  */
ObjFunction *compile(const char *source) {
  initScanner(source);
  Compiler compiler;
  initCompiler(&compiler, TYPE_SCRIPT);

  parser.hadError = false;
  parser.panicMode = false;

  advance();

  while (!match(TOKEN_EOF)) {
    declaration();
  }

  ObjFunction *function = endCompiler();
  return parser.hadError ? NULL : function;
}