waiig15/parser/parser.go

package parser

import (
	"fmt"

	"code.osinet.fr/fgm/waiig15/ast"
	"code.osinet.fr/fgm/waiig15/lexer"
	"code.osinet.fr/fgm/waiig15/token"
)

// Precedence constants.
const (
	_ int = iota
	LOWEST
	EQUALS      // ==
	LESSGREATER // > or <
	SUM         // +
	PRODUCT     // *
	PREFIX      // -X or !X
	CALL        // myFunction(X)
)

// Parser implements the parsing mechanism top-level layer.
type Parser struct {
	errors []string
	l      *lexer.Lexer

	curToken  token.Token
	peekToken token.Token

	prefixParseFns map[token.TokenType]prefixParseFn
	infixParseFns  map[token.TokenType]infixParseFn
}

type (
	prefixParseFn func() ast.Expression
	infixParseFn  func(ast.Expression) ast.Expression
)

var precedences = map[token.TokenType]int{
	token.EQ:       EQUALS,
	token.NOT_EQ:   EQUALS,
	token.LT:       LESSGREATER,
	token.GT:       LESSGREATER,
	token.PLUS:     SUM,
	token.MINUS:    SUM,
	token.SLASH:    PRODUCT,
	token.ASTERISK: PRODUCT,
}

// New returns a new Parser instance with the first two parser tokens already
// loaded.
func New(l *lexer.Lexer) *Parser {
	p := &Parser{
		l:      l,
		errors: []string{},
	}

	p.infixParseFns = make(map[token.TokenType]infixParseFn)
	for _, tok := range []token.TokenType{
		token.ASTERISK,
		token.EQ,
		token.GT,
		token.LT,
		token.MINUS,
		token.NOT_EQ,
		token.PLUS,
		token.SLASH,
	} {
		p.registerInfix(tok, p.parseInfixExpression)
	}

	p.prefixParseFns = make(map[token.TokenType]prefixParseFn)
	p.registerPrefix(token.BANG, p.parsePrefixExpression)
	p.registerPrefix(token.IDENT, p.parseIdentifier)
	p.registerPrefix(token.INT, p.parseIntegerLiteral)
	p.registerPrefix(token.MINUS, p.parsePrefixExpression)

	// Read two tokens, so curToken and peeToken are both set.
	p.nextToken()
	p.nextToken()

	return p
}

// Errors is a getter for Parser.errors.
func (p *Parser) Errors() []string {
	return p.errors
}

// ParseProgram is the outermost parsing logic, accumulating statements in a
// Program instance and returning that instance once parsing is done.
func (p *Parser) ParseProgram() *ast.Program {
	defer untrace(trace("ParseProgram"))
	program := &ast.Program{
		Statements: []ast.Statement{},
	}
	for !p.curTokenIs(token.EOF) {
		stmt := p.parseStatement()
		if stmt != nil {
			program.Statements = append(program.Statements, stmt)
		}
		p.nextToken()
	}

	return program
}

// Return the precedence for the current token without advancing.
func (p *Parser) curPrecedence() int {
	if precedence, ok := precedences[p.curToken.Type]; ok {
		return precedence
	}

	return LOWEST

}

// Is the current token in the parser of the given type ?
func (p *Parser) curTokenIs(t token.TokenType) bool {
	return p.curToken.Type == t
}

// Is the next token in the parser of the given type ? If it is, consume it,
// else don't.
func (p *Parser) expectPeek(t token.TokenType) bool {
	if p.peekTokenIs(t) {
		p.nextToken()
		return true
	}

	p.peekError(t)
	return false
}

func (p *Parser) nextToken() {
	p.curToken = p.peekToken
	p.peekToken = p.l.NextToken()
}

func (p *Parser) parseStatement() ast.Statement {
	defer untrace(trace("parseStatement"))
	switch p.curToken.Type {
	case token.LET:
		return p.parseLetStatement()
	case token.RETURN:
		return p.parseReturnStatement()
	default:
		return p.parseExpressionStatement()
	}
}

// Log a mismatch error on the peek token type in the parser instance.
//
//   - t is the type of token that was expected
func (p *Parser) peekError(t token.TokenType) {
	msg := fmt.Sprintf("expected next token to be %s, got %s instead",
		t, p.peekToken.Type)
	p.errors = append(p.errors, msg)
}

// Look forward for the precedence of the next token without advancing.
func (p *Parser) peekPrecedence() int {
	if precedence, ok := precedences[p.peekToken.Type]; ok {
		return precedence
	}

	return LOWEST
}

// Is the next token in the parser of the given type ? Don't consume it.
func (p *Parser) peekTokenIs(t token.TokenType) bool {
	return p.peekToken.Type == t
}