diff --git a/lexer.go b/lexer.go
new file mode 100644
index 0000000..999fd78
--- /dev/null
+++ b/lexer.go
@@ -0,0 +1,227 @@
+package glob
+
+import (
+	"fmt"
+	"strings"
+	"unicode/utf8"
+)
+
+var eof int = 0
+
+type stateFn func(*lexer) stateFn
+
+type itemType int
+
+const (
+	item_eof itemType = iota
+	item_error
+	item_text
+	item_any
+	item_single
+	item_range_open
+	item_range_not
+	item_range_minus
+	item_range_close
+)
+
+type item struct {
+	t itemType
+	s string
+}
+
+type lexer struct {
+	input string
+	start int
+	pos   int
+	width int
+	runes int
+	items chan item
+}
+
+func (l *lexer) run() {
+	for state := lexText; state != nil; {
+		state = state(l)
+	}
+	close(l.items)
+}
+
+func (l *lexer) read() (rune int) {
+	if l.pos >= len(l.input) {
+		return eof
+	}
+
+	rune, l.width = utf8.DecodeRuneInString(l.input[l.pos:])
+	l.pos += l.width
+	l.runes++
+
+	return
+}
+
+func (l *lexer) unread() {
+	l.pos -= l.width
+	l.runes--
+}
+
+func (l *lexer) ignore() {
+	l.start = l.pos
+	l.runes = 0
+}
+
+func (l *lexer) lookahead() int {
+	r := l.read()
+	l.unread()
+	return r
+}
+
+func (l *lexer) accept(valid string) bool {
+	if strings.IndexRune(valid, l.read()) != -1 {
+		return true
+	}
+	l.unread()
+	return false
+}
+
+func (l *lexer) acceptAll(valid string) {
+	for strings.IndexRune(valid, l.read()) != -1 {
+	}
+	l.unread()
+}
+
+func (l *lexer) emit(t itemType) {
+	l.items <- item{t, l.input[l.start:l.pos]}
+	l.start = l.pos
+	l.runes = 0
+	l.width = 0
+}
+
+func (l *lexer) flush(t itemType) {
+	if l.pos > l.start {
+		l.emit(t)
+	}
+}
+
+func (l *lexer) errorf(format string, args ...interface{}) {
+	l.emit(item{item_error, fmt.Sprintf(format, args...)})
+}
+
+func lex(source string) *lexer {
+	l := &lexer{
+		input: strings.NewReader(source),
+		items: make(chan item),
+	}
+
+	go l.run()
+
+	return l
+}
+
+func lexText(l *lexer) stateFn {
+	for {
+		switch l.input[l.pos] {
+		case escape:
+			if l.read() == eof {
+				l.errorf("unclosed '%s' character", string(escape))
+				return nil
+			}
+		case single:
+			l.flush(item_text)
+			return lexSingle
+		case any:
+			l.flush(item_text)
+			return lexAny
+		case range_open:
+			l.flush(item_text)
+			return lexRangeOpen
+		}
+
+		if l.read() == eof {
+			break
+		}
+	}
+
+	if l.pos > l.start {
+		l.emit(item_text)
+	}
+
+	l.emit(item_eof)
+
+	return nil
+}
+
+func lexRangeOpen(l *lexer) stateFn {
+	l.pos += 1
+	l.emit(item_range_open)
+	return lexInsideRange
+}
+
+func lexInsideRange(l *lexer) stateFn {
+	for {
+		switch l.input[l.pos] {
+
+		case inside_range_not:
+			// only first char makes sense
+			if l.pos == l.start {
+				l.emit(item_range_not)
+			}
+
+		case inside_range_minus:
+			if len(l.runes) != 1 {
+				l.errorf("unexpected character '%s'", string(inside_range_minus))
+				return nil
+			}
+
+			l.emit(item_text)
+
+			l.pos += 1
+			l.emit(item_range_minus)
+
+			switch l.input[l.pos] {
+			case eof, range_close:
+				l.errorf("unexpected end of range: character is expected")
+				return nil
+			default:
+				l.read()
+				l.emit(item_text)
+			}
+
+			return lexText
+
+		case range_close:
+			l.flush(item_text)
+			return lexRangeClose
+		}
+
+		if l.read() == eof {
+			l.errorf("unclosed range construction")
+			return nil
+		}
+	}
+}
+
+func lexAny(l *lexer) stateFn {
+	l.pos += 1
+	l.emit(item_any)
+	return lexText
+}
+
+func lexRangeHiLo(l *lexer) stateFn {
+
+	l.emit(item_text)
+	return lexRangeMinus
+
+	l.pos += 1
+	l.emit(item_range_minus)
+	return lexInsideRange
+}
+
+func lexSingle(l *lexer) stateFn {
+	l.pos += 1
+	l.emit(item_single)
+	return lexText
+}
+
+func lexRangeClose(l *lexer) stateFn {
+	l.pos += 1
+	l.emit(item_range_close)
+	return lexText
+}