From ed55f2c4ad0ab96ac6332c5929295a60847ac4d1 Mon Sep 17 00:00:00 2001
From: Jordan Orelli <jordanorelli@gmail.com>
Date: Sat, 14 Mar 2015 16:35:45 -0400
Subject: [PATCH] lex

---
 lex.go      | 124 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 lex_test.go |  31 +++++++++++++
 2 files changed, 155 insertions(+)
 create mode 100644 lex.go
 create mode 100644 lex_test.go

diff --git a/lex.go b/lex.go
new file mode 100644
index 0000000..a6b26f0
--- /dev/null
+++ b/lex.go
@@ -0,0 +1,124 @@
+package main
+
+import (
+	"errors"
+	"fmt"
+	"io"
+	"strings"
+)
+
+type tokenType int
+
+const (
+	t_error  tokenType = iota // a stored lex error
+	t_string                  // a string literal
+)
+
+type stateFn func(*lexer) (stateFn, error)
+
+type token struct {
+	t tokenType
+	s string
+}
+
+type lexer struct {
+	in  io.RuneReader
+	out chan token
+	buf []rune // running buffer for current lexeme
+}
+
+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:
+			return
+		default:
+			l.out <- token{t_error, err.Error()}
+			return
+		}
+	}
+}
+
+func (l *lexer) next() (rune, error) {
+	r, _, err := l.in.ReadRune()
+	return r, err
+}
+
+func (l *lexer) keep(r rune) {
+	if l.buf == nil {
+		l.buf = make([]rune, 0, 18)
+	}
+	l.buf = append(l.buf, r)
+}
+
+func (l *lexer) emit(t tokenType) {
+	l.out <- token{t, string(l.buf)}
+	l.buf = l.buf[0:0]
+}
+
+func lexString(in string) chan token {
+	r := strings.NewReader(in)
+	return lex(r)
+}
+
+func lex(r io.RuneReader) chan token {
+	l := lexer{
+		in:  r,
+		out: make(chan token),
+	}
+	go l.lex()
+	return l.out
+}
+
+func fullTokens(c chan token) ([]token, error) {
+	tokens := make([]token, 0, 32)
+	for t := range c {
+		if t.t == t_error {
+			return nil, errors.New(t.s)
+		}
+		tokens = append(tokens, t)
+	}
+	return tokens, nil
+}
+
+func lexRoot(l *lexer) (stateFn, error) {
+	r, err := l.next()
+	if err != nil {
+		return nil, err
+	}
+	switch r {
+	case '"', '`':
+		return lexStringLiteral(r), nil
+	default:
+		return nil, fmt.Errorf("unexpected rune in lexRoot: %c", r)
+	}
+}
+
+func lexStringLiteral(delim rune) stateFn {
+	return func(l *lexer) (stateFn, error) {
+		r, err := l.next()
+		if err != nil {
+			return nil, err
+		}
+		switch r {
+		case delim:
+			l.emit(t_string)
+			return lexRoot, nil
+		case '\\':
+			r, err := l.next()
+			if err != nil {
+				return nil, err
+			}
+			l.keep(r)
+			return lexStringLiteral(delim), nil
+		default:
+			l.keep(r)
+			return lexStringLiteral(delim), nil
+		}
+	}
+}
diff --git a/lex_test.go b/lex_test.go
new file mode 100644
index 0000000..d358083
--- /dev/null
+++ b/lex_test.go
@@ -0,0 +1,31 @@
+package main
+
+import (
+	"testing"
+)
+
+var primitivesTests = []struct {
+	in  string
+	out token
+}{
+	{`"x"`, token{t_string, "x"}},
+	{`"yes"`, token{t_string, "yes"}},
+	{`"this one has spaces"`, token{t_string, "this one has spaces"}},
+	{`"this one has \"quotes\" in it"`, token{t_string, `this one has "quotes" in it`}},
+	{"`this one is delimited by backticks`", token{t_string, "this one is delimited by backticks"}},
+}
+
+func TestLexPrimities(t *testing.T) {
+	for _, test := range primitivesTests {
+		tokens, err := fullTokens(lexString(test.in))
+		if err != nil {
+			t.Error(err)
+			continue
+		}
+		if len(tokens) > 1 {
+			t.Errorf("expected 1 token, saw %d: %v", len(tokens), tokens)
+			continue
+		}
+		t.Logf("OK: %s", test.in)
+	}
+}