muesli/lexer.go

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"
)




/* A Lexer splits scanned input into tokens.
 */
type Lexer struct {
    Position
    Index       int
    Start       int
    io.RuneScanner
    buffer      []rune
    Current     rune
}


func (lexer * Lexer) ClearBuffer() {
  lexer.buffer = make([]rune, 0)
}

func (lexer * Lexer) MakeToken(kind TokenKind) Token {
    val := StringValue(string(lexer.buffer))
    lexer.ClearBuffer()
    return NewToken(kind, val, lexer.Position)
}

func (lexer Lexer) MakeErrorToken(err error) Token {
    return NewToken(TokenKindError, err.Error(), lexer.Position)
}

func (lexer Lexer) MakeErrorfToken(format string, va ... interface{}) Token {
    err := fmt.Errorf(format, va...)
    return lexer.MakeErrorToken(err)
}


func (lexer Lexer) MakeEOFToken() Token {
    return NewToken(TokenKindEOF, "", lexer.Position)
}


func (lexer * Lexer) Peek() (rune, error) {
    r, _, err := lexer.RuneScanner.ReadRune()
    err2 := lexer.RuneScanner.UnreadRune()
    
    if err == nil {
        err = err2
    }
    return r, err
}

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) oldPrevious() error {
    fmt.Printf("Previous: now %c \n", lexer.Current)
    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 // XXX wrong
            lexer.Position.Line--
        }
        
        lexer.Current = r
    }
    return nil
}


func (lexer * Lexer) NextIf(predicate func(rune) bool) (bool, error) {
  r, err := lexer.Peek()
  if err != nil {
    return false, err
  }
  if (predicate(r)) {
    r, err = lexer.Next()
    if err != nil {
      return true, err
    }
    return true, nil
  }
  return false, nil
}

func (lexer * Lexer) NextWhile(predicate func(rune) bool) (bool, error) {
    result := true
    ok, err := lexer.NextIf(predicate)
    result = result || ok
    for ; ok && (err == nil) ; ok, err = lexer.NextIf(predicate) {
        result = result || ok
    }
    return result, err
}


func isSpace(r rune) bool {
  return r == ' ' || r == '\t'
}


func (lexer * Lexer) SkipSpace() (error) {
    _, err := lexer.NextWhile(isSpace)
    if err == nil { 
      lexer.ClearBuffer()
    }
    return err
 }

/* Handles errors including EOF by either returning an error token or an 
 * EOF token.
 */
func (lexer * Lexer) handleError(err error) Token {
    if err == io.EOF {
        return lexer.MakeEOFToken()
    } else {
        return lexer.MakeErrorToken(err)
    }
}


func (lexer * Lexer) LexNumber() Token { 
  isFloat := false
  
  _, err := lexer.NextWhile(func (r rune) bool {
    if unicode.IsDigit(r) {
        return true
    } else if r == '.' {
      if isFloat { 
          return false // double point in floating point
      } else {
          isFloat = true
          return true
      }
    } else {
        return false
    }
  })
  
  
  if err != nil {
      return lexer.MakeErrorfToken("when parsing number: %s", err)
  }
  if isFloat {
      return lexer.MakeToken(TokenKindFloat)
  } else {
      return lexer.MakeToken(TokenKindInteger)  
  }  
}

func isDoubleQuote(r rune) bool {
  return r == '"'
}

func (lexer * Lexer) LexString() Token { 
  inEscape := false
  var err error
  
  _, err = lexer.Next() 
  if err != nil {
      return lexer.handleError(err)
  }
    
  _, err = lexer.NextWhile(func (r rune) bool {
      if r == '"' && !inEscape {
        return false
      }
      if r == '\\' {
        // TODO escape parsing, now just a single character after it      
          if inEscape { // double backslash
            inEscape = false
          } else {
            inEscape = true
          }
      } else {
            inEscape = false
      }
      return true // still inside the string
  })
  if err != nil {
        return lexer.MakeErrorfToken("when parsing string: %s", err)
  }
  
  _, err = lexer.Next()
  if err != nil {
      return lexer.handleError(err)
  }
  
  return lexer.MakeToken(TokenKindString)
}

func (lexer * Lexer) LexLongString() Token { 
  var err error
  
  _, err = lexer.Next()
  if err != nil {
      return lexer.handleError(err)
  }
  
  _, err = lexer.NextWhile(func (r rune) bool {
      return r != '`'
  })

  if err != nil {
        return lexer.MakeErrorfToken("when parsing long string: %s", err)
  }
  
  _, err = lexer.Next()
  if err != nil {
      return lexer.handleError(err)
  }
  
  return lexer.MakeToken(TokenKindString)
}

func (lexer * Lexer) LexWord() Token { 
  var err error
  
  _, err = lexer.Next()
  if err != nil {
    return lexer.handleError(err)
  }
  
  lexer.NextWhile(func(r rune) bool {
      return unicode.IsLetter(r)
  })
  
  return lexer.MakeToken(TokenKindWord)
}

func (lexer * Lexer) lex() Token {
    r, err := lexer.Peek()
 
    if err != nil {
      return lexer.handleError(err)
    }
    
    if isSpace(r) { 
        err = lexer.SkipSpace()
        if err != nil {
          return lexer.handleError(err)
        } 
        r, err = lexer.Peek()
        if err != nil {
          return lexer.handleError(err)
        }
    }
    

    if unicode.IsDigit(r) {
        return lexer.LexNumber()
    }
    
    if r == '\n' || r == '.' {
       lexer.Next()
       return lexer.MakeToken(TokenKindEOX)
    }
        
    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: 
            lexer.Next()
            return lexer.MakeToken(TokenKind(r))
        default:
    }
    
    if unicode.IsLetter(r) {
        return lexer.LexWord()
    }

    return lexer.MakeErrorfToken("Unknown character: %c", r)
}

func (lexer * Lexer) Lex() Token {
  res := lexer.lex()
  lexer.ClearBuffer() // ensure buffer is cleared after lexing, always.
  return res
}

func (lexer * Lexer) LexAll() []Token {
    var token Token

    res := make([]Token, 0)

    for token = lexer.Lex() ; ! token.IsLast() ; token = lexer.Lex() {
          fmt.Printf("token: %s %v\n", token.String(), token.IsLast())
          res = append(res, token)
    } 
    
    fmt.Printf("Last token: %s %v\n", token.String(), token.IsLast())
    res = append(res, token)
 
    return res
}


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
}