clean up, clean up

everybody, everywhere

lex.go

@@ -9,6 +9,8 @@ import (
 	"unicode"
 )
 
+const eof = -1
+
 type tokenType int
 
 func (t tokenType) String() string {
@@ -39,6 +41,8 @@ func (t tokenType) String() string {
 		return "t_object_separator"
 	case t_object_end:
 		return "t_object_end"
+	case t_int:
+		return "t_int"
 	default:
 		panic(fmt.Sprintf("unknown token type: %v", t))
 	}
@@ -58,9 +62,10 @@ const (
 	t_object_start                      // {
 	t_object_end                        // }
 	t_object_separator                  // :
+	t_int                               // an integer
 )
 
-type stateFn func(*lexer) (stateFn, error)
+type stateFn func(*lexer) stateFn
 
 type token struct {
 	t tokenType
@@ -72,34 +77,36 @@ type lexer struct {
 	out    chan token
 	buf    []rune // running buffer for current lexeme
 	backup []rune
+	err    error
 }
 
 func (l *lexer) lex() {
 	defer close(l.out)
-	var err error
-	fn := lexRoot
-	for {
-		fn, err = fn(l)
-		switch err {
-		case nil:
-		case io.EOF:
-			l.out <- token{t_eof, ""}
-			return
-		default:
-			l.out <- token{t_error, err.Error()}
-			return
+	for fn := lexRoot; fn != nil; {
+		fn = fn(l)
+		if l.err != nil {
+			fn = lexErrorf("read error: %s", l.err)
 		}
 	}
 }
 
-func (l *lexer) next() (rune, error) {
+func (l *lexer) next() rune {
 	if len(l.backup) > 0 {
 		r := l.backup[len(l.backup)-1]
 		l.backup = l.backup[:len(l.backup)-1]
-		return r, nil
+		return r
 	}
 	r, _, err := l.in.ReadRune()
-	return r, err
+	switch err {
+	case io.EOF:
+		return eof
+	case nil:
+		return r
+	default:
+		l.err = err
+		return eof
+	}
+	return r
 }
 
 func (l *lexer) keep(r rune) {
@@ -144,139 +151,205 @@ func fullTokens(c chan token) ([]token, error) {
 	return tokens, nil
 }
 
-func lexRoot(l *lexer) (stateFn, error) {
-	r, err := l.next()
-	if err != nil {
-		return nil, err
+func lexErrorf(t string, args ...interface{}) stateFn {
+	return func(l *lexer) stateFn {
+		l.out <- token{t_error, fmt.Sprintf(t, args...)}
+		return nil
 	}
+}
+
+func lexRoot(l *lexer) stateFn {
+	r := l.next()
 	switch {
+	case r == eof:
+		return nil
 	case r == '=':
 		l.keep(r)
 		l.emit(t_equals)
-		return lexRoot, nil
+		return lexRoot
 	case r == '"', r == '`':
-		return lexStringLiteral(r), nil
+		return lexStringLiteral(r)
 	case r == '#':
-		return lexComment, nil
+		return lexComment
 	case r == '[':
 		l.keep(r)
 		l.emit(t_list_start)
-		return lexRoot, nil
+		return lexRoot
 	case r == ']':
 		l.keep(r)
 		l.emit(t_list_end)
-		return lexRoot, nil
+		return lexRoot
 	case r == ',':
 		l.keep(r)
 		l.emit(t_list_separator)
-		return lexRoot, nil
+		return lexRoot
 	case r == '{':
 		l.keep(r)
 		l.emit(t_object_start)
-		return lexRoot, nil
+		return lexRoot
 	case r == '}':
 		l.keep(r)
 		l.emit(t_object_end)
-		return lexRoot, nil
+		return lexRoot
 	case r == ':':
 		l.keep(r)
 		l.emit(t_object_separator)
-		return lexRoot, nil
+		return lexRoot
+	// case strings.IndexRune("-0123456789", r) >= 0:
+	// 	l.unread(r)
+	// 	return lexNumber, nil
 	case unicode.IsSpace(r):
-		return lexRoot, nil
+		return lexRoot
 	case unicode.IsLower(r):
 		l.keep(r)
-		return lexName, nil
+		return lexName
 	case unicode.IsUpper(r):
 		l.keep(r)
-		return lexType, nil
+		return lexType
 	default:
-		return nil, fmt.Errorf("unexpected rune in lexRoot: %c", r)
+		return lexErrorf("unexpected rune in lexRoot: %c", r)
 	}
 }
 
-func lexComment(l *lexer) (stateFn, error) {
-	r, err := l.next()
-	switch err {
-	case io.EOF:
+func lexComment(l *lexer) stateFn {
+	switch r := l.next(); r {
+	case '\n':
 		l.emit(t_comment)
-		return nil, io.EOF
-	case nil:
-	default:
-		return nil, err
-	}
-	switch {
-	case r == '\n':
+		return lexRoot
+	case eof:
 		l.emit(t_comment)
-		return lexRoot, nil
+		return nil
 	default:
 		l.keep(r)
-		return lexComment, nil
+		return lexComment
 	}
 }
 
 func lexStringLiteral(delim rune) stateFn {
-	return func(l *lexer) (stateFn, error) {
-		r, err := l.next()
-		if err != nil {
-			return nil, err
-		}
-		switch r {
+	return func(l *lexer) stateFn {
+		switch r := l.next(); r {
 		case delim:
 			l.emit(t_string)
-			return lexRoot, nil
+			return lexRoot
 		case '\\':
-			r, err := l.next()
-			if err != nil {
-				return nil, err
+			switch r := l.next(); r {
+			case eof:
+				return lexErrorf("unexpected eof in string literal")
+			default:
+				l.keep(r)
+				return lexStringLiteral(delim)
 			}
-			l.keep(r)
-			return lexStringLiteral(delim), nil
+		case eof:
+			return lexErrorf("unexpected eof in string literal")
 		default:
 			l.keep(r)
-			return lexStringLiteral(delim), nil
+			return lexStringLiteral(delim)
 		}
 	}
 }
 
-func lexName(l *lexer) (stateFn, error) {
-	r, err := l.next()
-	switch err {
-	case io.EOF:
-		l.emit(t_name)
-		return nil, io.EOF
-	case nil:
-	default:
-		return nil, err
-	}
+func lexName(l *lexer) stateFn {
+	r := l.next()
 	switch {
 	case unicode.IsLetter(r), unicode.IsDigit(r), r == '_':
 		l.keep(r)
-		return lexName, nil
+		return lexName
+	case r == eof:
+		l.emit(t_name)
+		return nil
 	default:
 		l.emit(t_name)
 		l.unread(r)
-		return lexRoot, nil
+		return lexRoot
 	}
 }
 
-func lexType(l *lexer) (stateFn, error) {
-	r, err := l.next()
-	switch err {
-	case io.EOF:
-		l.emit(t_type)
-		return nil, io.EOF
-	case nil:
-	default:
-		return nil, err
-	}
+func lexType(l *lexer) stateFn {
+	r := l.next()
 	switch {
 	case unicode.IsLetter(r), unicode.IsDigit(r), r == '_':
 		l.keep(r)
-		return lexType, nil
+		return lexType
+	case r == eof:
+		l.emit(t_type)
+		return nil
 	default:
 		l.emit(t_type)
 		l.unread(r)
-		return lexRoot, nil
+		return lexRoot
 	}
 }
+
+// func lexNumber(l *lexer) (stateFn, error) {
+// 	r, err := l.next()
+// 	if err != nil {
+// 		return nil, err
+// 	}
+//
+// 	switch {
+// 	case r == '-', r == '+':
+// 		l.keep(r)
+// 		return lexNumber, nil
+// 	case r == '0':
+// 		l.keep(r)
+// 		return lexHexOct, nil
+// 	case strings.IndexRune("123456789", r) >= 0:
+// 		l.keep(r)
+// 		return lexDecimal, nil
+// 	default:
+// 		return nil, fmt.Errorf("unexpected rune in lexNumber: %c", r)
+// 	}
+// }
+//
+// func lexHexOct(l *lexer) (stateFn, error) {
+// 	r, err := l.next()
+// 	if err != nil {
+// 		return nil, err
+// 	}
+//
+// 	switch {
+// 	case r == 'x', r == 'X':
+// 		l.keep(r)
+// 		return lexHex, nil
+//     case r == 'e', r == 'E':
+//         l.keep(r)
+//         return lexExponent, nil
+// 	case r == '8', r == '9':
+// 		return nil, fmt.Errorf("unexpected 8 or 9 in lexHexOct.  there's no 8 or 9 in octal!")
+// 	case strings.IndexRune("01234567", r) >= 0:
+// 		l.keep(r)
+// 		return lexOct, nil
+// 	default:
+//         // we get here for the literals -0, +0 and 0
+// 		l.unread(r)
+//         l.emit(t_int)
+//         return lexRoot, nil
+// 	}
+// }
+//
+// func lexHex(l *lexer) (stateFn, error) {
+// 	r, err := l.next()
+// 	if err != nil {
+// 		return nil, err
+// 	}
+//
+//     switch {
+//
+//     }
+// }
+//
+// func lexDecimal(l *lexer) (stateFn, error) {
+// 	r, err := l.next()
+// 	if err != nil {
+// 		return nil, err
+// 	}
+//
+// 	switch {
+//     case strings.IndexRune("0123456789", r) >= 0:
+//         l.keep(r)
+//         return lexDecimal, nil
+//     case r == '.':
+//         l.keep(r)
+//         return lexFloat, nil
+// 	}
+// }

lex_test.go

@@ -58,7 +58,7 @@ func TestLexPrimities(t *testing.T) {
 			t.Error(err)
 			continue
 		}
-		tokens = tokens[:len(tokens)-1]
+		// tokens = tokens[:len(tokens)-1]
 		if len(tokens) != len(test.out) {
 			t.Errorf("expected %d token, saw %d: %v", len(test.out), len(tokens), tokens)
 			continue

node.go

@@ -36,7 +36,7 @@ func (n *rootNode) parse(p *parser) error {
 		t := p.next()
 		switch t.t {
 		case t_error:
-			return fmt.Errorf("parse error: saw lex error while parsing root node: %v", t.s)
+			return fmt.Errorf("parse error: saw lex error while parsing root node: %v", t)
 		case t_eof:
 			return nil
 		case t_comment:

parse.go

@@ -39,7 +39,11 @@ func (p *parser) next() token {
 		p.backup = p.backup[:len(p.backup)-1]
 		return t
 	}
-	return <-p.input
+	t, ok := <-p.input
+	if !ok {
+		return token{t_eof, "eof"}
+	}
+	return t
 }
 
 func (p *parser) peek() token {