123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304 |
- package muesli
- import (
- _ "bytes"
- _ "errors"
- "fmt"
- _ "io"
- _ "reflect"
- _ "runtime"
- "strings"
- _ "unicode"
- "io"
- "os"
- "bufio"
- "unicode"
- // "gitlab.com/beoran/woe/graphviz"
- // _ "gitlab.com/beoran/woe/monolog"
- )
- type Position struct {
- FileName string
- Line int
- Column int
- }
- type Lexer struct {
- Position
- Index int
- Start int
- io.RuneScanner
- buffer []rune
- Current rune
- }
- /** Token Kind. Uses a rune to easily handle single character tokens. */
- type TokenKind rune
- const (
- TokenKindInteger = TokenKind('i')
- TokenKindFloat = TokenKind('f')
- TokenKindString = TokenKind('s')
- TokenKindBoolean = TokenKind('b')
- TokenKindWord = TokenKind('w')
- TokenKindType = TokenKind('t')
- TokenKindGet = TokenKind('$')
- TokenKindSet = TokenKind('=')
- TokenKindOpenBlock = TokenKind('{')
- TokenKindCloseBlock = TokenKind('}')
- TokenKindOpenList = TokenKind('[')
- TokenKindCloseList = TokenKind(']')
- TokenKindOpenParen = TokenKind('(')
- TokenKindCloseParen = TokenKind(')')
- TokenKindError = TokenKind('!')
- TokenKindEOX = TokenKind('\n')
- )
- func NewToken(kind TokenKind, val Value, pos Position) Token {
- return Token{kind, val, pos}
- }
- func (lexer Lexer) MakeToken(kind TokenKind) Token {
- val := StringValue(string(lexer.buffer))
- return NewToken(kind, val, lexer.Position)
- }
- func (lexer * Lexer) Next() (rune, error) {
- r, _, err := lexer.RuneScanner.ReadRune()
- if err != nil {
- return 0, err
- }
- lexer.Current = r
- lexer.buffer = append(lexer.buffer, r)
- lexer.Index++
- lexer.Position.Column++
- if r == '\n' {
- lexer.Position.Column = 1
- lexer.Position.Line++
- }
- return lexer.buffer[len(lexer.buffer) - 1], nil
- }
- func (lexer * Lexer) Previous() error {
- err := lexer.RuneScanner.UnreadRune()
- if err != nil {
- return err
- }
-
- lexer.Index--
- lexer.Position.Column--
- if (len(lexer.buffer) > 0) {
- r := lexer.buffer[len(lexer.buffer) - 1];
- lexer.buffer = lexer.buffer[0: len(lexer.buffer) - 1];
-
- if r == '\n' {
- lexer.Position.Column = 1
- lexer.Position.Line++
- }
-
- lexer.Current = r
- }
- return nil
- }
-
- func (lexer * Lexer) SkipSpace() (error) {
- var r rune
- var err error
- r = lexer.Current
-
- for unicode.IsSpace(r) {
- r, err = lexer.Next()
- if err != nil {
- return err
- }
- }
- lexer.Previous()
- return nil
- }
- func (lexer * Lexer) LexNumber() (Token, error) {
- isFloat := false
- var r rune
- var err error
-
- r = lexer.Current
-
- for unicode.IsDigit(r) || r == '.' {
- if r == '.' {
- if isFloat { // double . in floating point is an error
- tok := lexer.MakeToken(TokenKindError)
- err = fmt.Errorf("Double period . in floating point constant.")
- return tok, err
- } else {
- isFloat = true
- }
- }
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
- }
-
- lexer.Previous()
- if isFloat {
- return lexer.MakeToken(TokenKindFloat), nil
- } else {
- return lexer.MakeToken(TokenKindInteger), nil
- }
-
- }
- func (lexer * Lexer) LexString() (Token, error) {
- inEscape := false
- var r rune
- var err error
-
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
-
-
- for r != '"' || inEscape {
- if r == '\\' {
- // TODO escape parsing, now just a single character after it
- if inEscape { // double backslash
- } else {
- inEscape = true
- }
- } else {
- inEscape = false
- }
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
- }
-
- return lexer.MakeToken(TokenKindString), nil
- }
- func (lexer * Lexer) LexLongString() (Token, error) {
- var r rune
- var err error
-
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
-
- for r != '`' {
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
- }
-
- return lexer.MakeToken(TokenKindString), nil
- }
- func (lexer * Lexer) LexWord() (Token, error) {
- var r rune
- var err error
-
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
-
- for r != '`' {
- r, err = lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
- }
-
- return lexer.MakeToken(TokenKindString), nil
- }
- func (lexer * Lexer) Lex() (Token, error) {
- r, err := lexer.Next()
- if err != nil {
- return lexer.MakeToken(TokenKindError), err
- }
-
- if unicode.IsSpace(r) {
- lexer.SkipSpace()
- }
-
- if unicode.IsDigit(r) {
- return lexer.LexNumber()
- }
-
- if r == '\n' || r == '.' {
- return lexer.MakeToken(TokenKindEOX), nil
- }
-
- if r == '"' {
- return lexer.LexString()
- }
-
- if r == '`' {
- return lexer.LexLongString()
- }
-
-
- switch (TokenKind(r)) {
- case TokenKindGet : fallthrough
- case TokenKindSet : fallthrough
- case TokenKindOpenBlock : fallthrough
- case TokenKindCloseBlock: fallthrough
- case TokenKindOpenList : fallthrough
- case TokenKindCloseList : fallthrough
- case TokenKindOpenParen : fallthrough
- case TokenKindCloseParen:
- return lexer.MakeToken(TokenKind(r)), nil
- default:
- }
-
- if unicode.IsLetter(r) {
- return lexer.LexWord()
- }
- return lexer.MakeToken(TokenKindError), fmt.Errorf("Unknown character")
- }
- func NewLexer(scanner io.RuneScanner, filename string) Lexer {
- lexer := Lexer{}
- lexer.RuneScanner = scanner
- lexer.Position.FileName = filename
- lexer.Position.Column = 1
- lexer.Position.Line = 1
- return lexer
- }
- func NewLexerFromInputString(input string) Lexer {
- reader := strings.NewReader(input)
- return NewLexer(reader, "<input>")
- }
- func NewLexerFromFileName(filename string) (*Lexer, error) {
- read, err := os.Open(filename)
- if err != nil {
- bread := bufio.NewReader(read)
- lex := NewLexer(bread, filename)
- return &lex, nil
- }
- return nil , err
- }
-
|