pxasme/lexer.go

package main

import (
	"bufio"
	"fmt"
	"io"
	"strings"
)

type lexer struct {
	r      *bufio.Reader
	lineno int

	peek Token
}

func NewLexer(src io.Reader) *lexer {
	return &lexer{
		r:      bufio.NewReader(src),
		lineno: 0,
	}
}

func (l *lexer) Peek() (Token, error) {
	tok, err := l.Next()
	if err != nil {
		return nil, err
	}

	l.peek = tok
	return tok, nil
}

func (l *lexer) Fields(line string) ([]string, error) {
	const (
		STATE_START            = 0
		STATE_IN_WORD          = 1
		STATE_IN_QUOTED_STRING = 3
		STATE_BACKSLASH        = 4
		STATE_LINE_COMMENT     = 5
	)

	var (
		ret   = []string{}
		state = STATE_START
		buff  = ""
	)

	for _, c := range line {

		switch state {
		case STATE_START:
			if c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == ',' {
				continue
			} else if c == ';' {
				state = STATE_LINE_COMMENT
			} else if c == '"' {
				state = STATE_IN_QUOTED_STRING
			} else {
				buff = string(c)
				state = STATE_IN_WORD
			}

		case STATE_IN_WORD:
			if c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == ',' {
				ret = append(ret, buff)
				buff = ""
				state = STATE_START
			} else if c == ';' {
				ret = append(ret, buff)
				buff = ""
				state = STATE_LINE_COMMENT
			} else {
				buff += string(c)
			}

		case STATE_IN_QUOTED_STRING:
			if c == '"' {
				ret = append(ret, buff)
				buff = ""
				state = STATE_START
			} else if c == '\\' {
				state = STATE_BACKSLASH
			} else {
				buff += string(c)
			}

		case STATE_BACKSLASH:
			if c == 'n' {
				buff += "\n"
			} else if c == 'r' {
				buff += "\r"
			} else if c == 't' {
				buff += "\t"
			} else if c == 's' {
				buff += " "
			} else if c == '\\' {
				buff += `\`
			} else {
				return nil, fmt.Errorf("Unknown backslash escape sequence")
			}
			state = STATE_IN_QUOTED_STRING

		case STATE_LINE_COMMENT:
			if c == '\n' {
				state = STATE_START
			} // ignore all else

		default:
			panic("impossible state")
		}
	}

	if state == STATE_IN_QUOTED_STRING {
		return nil, fmt.Errorf("unterminated string literal")
	}

	if state == STATE_IN_WORD {
		ret = append(ret, buff)
		buff = ""
		state = STATE_START
	}

	return ret, nil
}

func (l *lexer) Next() (Token, error) {
	if l.peek != nil {
		ret := l.peek
		l.peek = nil
		return ret, nil
	}

	//

	l.lineno++
	line, err := l.r.ReadString('\n')
	if err != nil {
		return nil, err
	}

	//

	fields, err := l.Fields(line)
	if err != nil {
		return nil, err
	}

	if len(fields) == 0 {
		// This line only contained comments
		// Continue to the next line
		return l.Next()
	}

	switch strings.ToLower(fields[0]) {
	case "section":
		return SectionToken{fields[1]}, nil

	case "global":
		return LabelToken{strings.TrimRight(fields[1], `:`), true}, nil

	case "mov":
		for i, _ := range fields {
			fields[i] = strings.TrimRight(fields[i], `,`)
		}
		return MovInstrToken{fields[1:]}, nil

	case "syscall":
		return SyscallInstrToken{}, nil

	case "ret":
		return RetInstrToken{}, nil

	case "nop":
		return NopInstrToken{}, nil

	default:
		// If the field ends with `:`, it's a (local) label
		if strings.HasSuffix(fields[0], `:`) {
			return LabelToken{strings.TrimRight(fields[0], `:`), false}, nil
		}

		// If the field starts with `$`, it's a "variable"
		if strings.HasPrefix(fields[0], `$`) {
			// 1: =
			if fields[1] != `=` {
				return nil, fmt.Errorf("Invalid syntax at line %d (expected = in declaration)", l.lineno)
			}

			// 2: sizeclass
			// 3+++: literal initializer
			return DataVariableInstrToken{
				VarName:   fields[0][1:],
				Sizeclass: fields[2],
				Value:     strings.Join(fields[3:], " "), // FIXME consecutive spaces are ruined
			}, nil
		}
	}

	return nil, fmt.Errorf("Unknown token '%s' at line %d", fields[0], l.lineno)
}