From afac87e1a9d1238ba15c8d3380d5c5fecec550e0 Mon Sep 17 00:00:00 2001
From: mappu <mappu04@gmail.com>
Date: Mon, 11 Dec 2023 19:32:37 +1300
Subject: [PATCH] lexer: parse fields with a state machine

---
 lexer.go | 107 ++++++++++++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 99 insertions(+), 8 deletions(-)

diff --git a/lexer.go b/lexer.go
index 56063cd..d00dd4d 100644
--- a/lexer.go
+++ b/lexer.go
@@ -31,6 +31,99 @@ func (l *lexer) Peek() (Token, error) {
 	return tok, nil
 }
 
+func (l *lexer) Fields(line string) ([]string, error) {
+	const (
+		STATE_START            = 0
+		STATE_IN_WORD          = 1
+		STATE_IN_QUOTED_STRING = 3
+		STATE_BACKSLASH        = 4
+		STATE_LINE_COMMENT     = 5
+	)
+
+	var (
+		ret   = []string{}
+		state = STATE_START
+		buff  = ""
+	)
+
+	for _, c := range line {
+
+		switch state {
+		case STATE_START:
+			if c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == ',' {
+				continue
+			} else if c == ';' {
+				state = STATE_LINE_COMMENT
+			} else if c == '"' {
+				state = STATE_IN_QUOTED_STRING
+			} else {
+				buff = string(c)
+				state = STATE_IN_WORD
+			}
+
+		case STATE_IN_WORD:
+			if c == ' ' || c == '\t' || c == '\r' || c == '\n' || c == ',' {
+				ret = append(ret, buff)
+				buff = ""
+				state = STATE_START
+			} else if c == ';' {
+				ret = append(ret, buff)
+				buff = ""
+				state = STATE_LINE_COMMENT
+			} else {
+				buff += string(c)
+			}
+
+		case STATE_IN_QUOTED_STRING:
+			if c == '"' {
+				ret = append(ret, buff)
+				buff = ""
+				state = STATE_START
+			} else if c == '\\' {
+				state = STATE_BACKSLASH
+			} else {
+				buff += string(c)
+			}
+
+		case STATE_BACKSLASH:
+			if c == 'n' {
+				buff += "\n"
+			} else if c == 'r' {
+				buff += "\r"
+			} else if c == 't' {
+				buff += "\t"
+			} else if c == 's' {
+				buff += " "
+			} else if c == '\\' {
+				buff += `\`
+			} else {
+				return nil, fmt.Errorf("Unknown backslash escape sequence")
+			}
+			state = STATE_IN_QUOTED_STRING
+
+		case STATE_LINE_COMMENT:
+			if c == '\n' {
+				state = STATE_START
+			} // ignore all else
+
+		default:
+			panic("impossible state")
+		}
+	}
+
+	if state == STATE_IN_QUOTED_STRING {
+		return nil, fmt.Errorf("unterminated string literal")
+	}
+
+	if state == STATE_IN_WORD {
+		ret = append(ret, buff)
+		buff = ""
+		state = STATE_START
+	}
+
+	return ret, nil
+}
+
 func (l *lexer) Next() (Token, error) {
 	if l.peek != nil {
 		ret := l.peek
@@ -46,21 +139,19 @@ func (l *lexer) Next() (Token, error) {
 		return nil, err
 	}
 
-	// Strip leading spaces
-	line = strings.TrimLeft(line, " \t\r\n")
+	//
 
-	// Strip trailing line-comments (;)
-	line, _, _ = strings.Cut(line, `;`)
+	fields, err := l.Fields(line)
+	if err != nil {
+		return nil, err
+	}
 
-	if len(line) == 0 {
+	if len(fields) == 0 {
 		// This line only contained comments
 		// Continue to the next line
 		return l.Next()
 	}
 
-	fields := strings.Fields(line)
-	// FIXME commas!?
-
 	switch strings.ToLower(fields[0]) {
 	case "section":
 		return SectionToken{fields[1]}, nil