nasin/internal/styles/tss/lex.go

package tss

import "io"
import "bufio"
import "unicode"
import "unicode/utf8"
import "git.tebibyte.media/sashakoshka/goparse"

const (
	Comment parse.TokenKind = iota
	LBrace
	RBrace
	LBracket
	RBracket
	Equals
	Colon
	Comma
	Semicolon
	Star
	Dot
	Dollar
	Slash
	Color
	Ident
	Number
	String
)

var tokenNames = map[parse.TokenKind] string {
	parse.EOF: "EOF",
	Comment: "Comment",
	LBrace: "LBrace",
	RBrace: "RBrace",
	LBracket: "LBracket",
	RBracket: "RBracket",
	Equals: "Equals",
	Colon: "Colon",
	Comma: "Comma",
	Semicolon: "Semicolon",
	Star: "Star",
	Dot: "Dot",
	Dollar: "Dollar",
	Slash: "Slash",
	Color: "Color",
	Ident: "Ident",
	Number: "Number",
	String: "String",
}

type lexer struct {
	filename    string
	lineScanner *bufio.Scanner
	rune        rune
	line        string
	lineFood    string

	offset   int
	row      int
	column   int

	eof bool
}

func Lex (filename string, reader io.Reader) parse.Lexer {
	lex := &lexer {
		filename:    filename,
		lineScanner: bufio.NewScanner(reader),
	}
	lex.nextRune()
	return lex
}

func (this *lexer) Next () (parse.Token, error) {
	for {
		token, err := this.next()
		if err == io.EOF { return token, this.errUnexpectedEOF() }
		if err != nil    { return token, err                     }

		if !token.Is(Comment) {
			return token, err
		}
	}
}

func (this *lexer) next () (token parse.Token, err error) {
	err = this.skipWhitespace()
	token.Position = this.pos()
	if this.eof {
		token.Kind = parse.EOF
		err = nil
		return
	}
	if err != nil { return }

	appendRune := func () {
		token.Value += string(this.rune)
		err = this.nextRune()
	}

	skipRune := func () {
		err = this.nextRune()
	}

	defer func () {
		newPos := this.pos()
		newPos.End --
		token.Position = token.Position.Union(newPos)
	} ()

	switch {
	case this.rune == '/':
		token.Kind = Comment
		skipRune()
		if err != nil { return }
		if this.rune == '/' {
			for this.rune != '\n' {
				skipRune()
				if err != nil { return }
			}
		} else {
			token.Kind = Slash
		}
		if this.eof { err = nil; return }

	case this.rune == '{':
		token.Kind = LBrace
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '}':
		token.Kind = RBrace
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '[':
		token.Kind = LBracket
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == ']':
		token.Kind = RBracket
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '=':
		token.Kind = Equals
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == ':':
		token.Kind = Colon
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == ',':
		token.Kind = Comma
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == ';':
		token.Kind = Semicolon
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '*':
		token.Kind = Star
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '.':
		token.Kind = Dot
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '$':
		token.Kind = Dollar
		appendRune()
		if this.eof { err = nil; return }

	case this.rune == '#':
		token.Kind = Color
		skipRune()
		if err != nil { return }
		for isHexDigit(this.rune) {
			appendRune()
			if this.eof { err = nil; return }
		}
		if this.eof { err = nil; return }

	case unicode.IsLetter(this.rune):
		token.Kind = Ident
		for unicode.IsLetter(this.rune) || unicode.IsNumber(this.rune) {
			appendRune()
			if this.eof { err = nil; return }
		}
		if this.eof { err = nil; return }

	case this.rune == '-':
		token.Kind = Number
		appendRune()
		for isDigit(this.rune) {
			appendRune()
			if this.eof { err = nil; return }
		}
		if this.eof { err = nil; return }
		
	case isDigit(this.rune):
		token.Kind = Number
		for isDigit(this.rune) {
			appendRune()
			if this.eof { err = nil; return }
		}
		if this.eof { err = nil; return }
		
	case this.rune == '\'', this.rune == '"':
		stringDelimiter := this.rune
		token.Kind = String
		err = this.nextRune()
		if err != nil { return }

		for this.rune != stringDelimiter {
			if this.rune == '\\' {
				var result rune
				result, err = this.escapeSequence(stringDelimiter)
				if err != nil { return }
				token.Value += string(result)
			} else {
				appendRune()
				if this.eof { err = nil; return }
				if err != nil { return }
			}
		}
		err = this.nextRune()
		if this.eof { err = nil; return }
		if err != nil { return }

	default:
		err = parse.Errorf (
			this.pos(), "unexpected rune %U",
			this.rune)
	}

	return
}

func (this *lexer) nextRune () error {
	if this.lineFood == "" {
		ok := this.lineScanner.Scan()
		if ok {
			this.line     = this.lineScanner.Text()
			this.lineFood = this.line
			this.rune     = '\n'
			this.column   = 0
			this.row      ++
		} else {
			err := this.lineScanner.Err()
			if err == nil {
				this.eof = true
				return io.EOF
			} else {
				return err
			}
		}
	} else {
		var ch   rune
		var size int
		for ch == 0 && this.lineFood != "" {
			ch, size = utf8.DecodeRuneInString(this.lineFood)
			this.lineFood = this.lineFood[size:]
		}
		this.rune = ch
		this.column ++
	}

	return nil
}

func (this *lexer) escapeSequence (stringDelimiter rune) (rune, error) {
	err := this.nextRune()
	if err != nil { return 0, err }

	if isDigit(this.rune) {
		var number rune
		for index := 0; index < 3; index ++ {
			if !isDigit(this.rune) { break }
		
			number *= 8
			number += this.rune - '0'

			err = this.nextRune()
			if err != nil { return 0, err }
		}
		return number, nil
	}

	defer this.nextRune()
	switch this.rune {
	case '\\', '\n', stringDelimiter:
		return this.rune, nil
	case 'a': return '\a', nil
	case 'b': return '\b', nil
	case 't': return '\t', nil
	case 'n': return '\n', nil
	case 'v': return '\v', nil
	case 'f': return '\f', nil
	case 'r': return '\r', nil
	default: return 0, this.errBadEscapeSequence()
	}
}

func (this *lexer) skipWhitespace () error {
	for isWhitespace(this.rune) {
		err := this.nextRune()
		if err != nil { return err }
	}
	return nil
}

func (this *lexer) pos () parse.Position {
	return parse.Position {
		File:   this.filename,
		Line:   this.lineScanner.Text(),
		Row:    this.row    - 1,
		Start:  this.column - 1,
		End:    this.column,
	}
}

func (this *lexer) errUnexpectedEOF () error {
	return parse.Errorf(this.pos(), "unexpected EOF")
}

func (this *lexer) errBadEscapeSequence () error {
	return parse.Errorf(this.pos(), "bad escape sequence")
}

func isWhitespace (char rune) bool {
	switch char {
	case ' ', '\t', '\r', '\n': return true
	default: return false
	}
}

func isSymbol (char rune) bool {
	switch char {
	case
		'~', '!', '@', '#', '$', '%', '^', '&', '-', '_', '=', '+',
		'\\', '|', ';', ',', '<', '>', '/', '?':
		return true
	default:
		return false
	}
}

func isDigit (char rune) bool {
	return char >= '0' && char <= '9'
}

func isHexDigit (char rune) bool {
	return isDigit(char) ||
		char >= 'a' && char <= 'f' ||
		char >= 'A' && char <= 'F'
}