From ce5c224211a94bfd4c898b51d15febdf2ed9d6f2 Mon Sep 17 00:00:00 2001
From: Charlie Stanton <charlie@shtanton.xyz>
Date: Fri, 26 Aug 2022 11:51:46 +0100
Subject: Refactors some stuff and adds lexing and parsing

---
 main/lex.go | 224 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 224 insertions(+)
 create mode 100644 main/lex.go

(limited to 'main/lex.go')

diff --git a/main/lex.go b/main/lex.go
new file mode 100644
index 0000000..6977f8a
--- /dev/null
+++ b/main/lex.go
@@ -0,0 +1,224 @@
+package main
+
+import (
+	"fmt"
+	"strings"
+	"unicode/utf8"
+)
+
+type stateFunc func(*lexer) stateFunc
+
+type lexer struct {
+	input string
+	start int
+	pos int
+	width int
+	tokenStream chan Token
+}
+
+func (l *lexer) run() {
+	for state := lexCommand; state != nil; {
+		state = state(l)
+	}
+	close(l.tokenStream)
+}
+
+func (l *lexer) emit(t TokenType) {
+	l.tokenStream <- Token{
+		typ: t,
+		val: l.input[l.start:l.pos],
+	}
+	l.start = l.pos
+}
+
+func (l *lexer) errorf(format string, args ...interface{}) stateFunc {
+	l.tokenStream <- Token{
+		typ: TokenErr,
+		val: fmt.Sprintf(format, args...),
+	}
+	return nil
+}
+
+const eof rune = -1
+
+func (l *lexer) next() rune {
+	if l.pos >= len(l.input) {
+		l.width = 0
+		return eof
+	}
+	var r rune
+	r, l.width = utf8.DecodeRuneInString(l.input[l.pos:])
+	l.pos += l.width
+	return r
+}
+
+func (l *lexer) backup() {
+	l.pos -= l.width
+}
+
+func (l *lexer) ignore() {
+	l.start = l.pos
+}
+
+func (l *lexer) reset() {
+	l.pos = l.start
+}
+
+func (l *lexer) peek() rune {
+	w := l.width
+	r := l.next()
+	l.backup()
+	l.width = w
+	return r
+}
+
+func (l *lexer) accept(valid string) bool {
+	if strings.IndexRune(valid, l.next()) >= 0 {
+		return true
+	}
+	l.backup()
+	return false
+}
+
+func (l *lexer) acceptAll(valid string) {
+	for strings.IndexRune(valid, l.next()) >= 0 {}
+	l.backup()
+}
+
+func (l *lexer) acceptPassing(valid func(rune) bool) bool {
+	if valid(l.next()) {
+		return true
+	}
+	l.backup()
+	return false
+}
+
+func (l *lexer) acceptAllPassing(valid func(rune) bool) {
+	for valid(l.next()) {}
+	l.backup()
+}
+
+type TokenType int
+
+const (
+	TokenErr TokenType = iota // Lexing error
+	TokenEOF // end of file
+	TokenSemicolon // ;
+	TokenLParen // (
+	TokenRParen // )
+	TokenLBrace // {
+	TokenRBrace // }
+	TokenLBrack // [
+	TokenRBrack // ]
+	TokenCommand // A command character
+	TokenHash // #
+	TokenAt // @
+	TokenDot // .
+	TokenAst // *
+	TokenPatternStringIndex // A string index in a pattern
+	TokenPatternIntegerIndex // An integer index in a pattern
+)
+
+type Token struct {
+	typ TokenType
+	val string
+}
+
+func (t Token) String() string {
+	switch t.typ {
+	case TokenEOF:
+		return "EOF"
+	case TokenErr:
+		return t.val
+	}
+	if len(t.val) > 10 {
+		return fmt.Sprintf("%.10q...", t.val)
+	}
+	return fmt.Sprintf("%q", t.val)
+}
+
+func Lex(input string) chan Token {
+	l := &lexer{
+		input: input,
+		tokenStream: make(chan Token),
+	}
+	go l.run()
+	return l.tokenStream
+}
+
+const (
+	whitespace string = " \t"
+	whitespaceNewlines string = " \t\r\n"
+)
+
+func isAlpha(r rune) bool {
+	return ('a' <= r && r < 'z') || ('A' <= r && r <= 'Z')
+}
+func isDigit(r rune) bool {
+	return '0' <= r && r <= '9'
+}
+func isAlphaNumeric(r rune) bool {
+	return isAlpha(r) || isDigit(r)
+}
+func isStringIndexChar(r rune) bool {
+	return isAlphaNumeric(r) || r == '_' || r == '-'
+}
+
+func lexCommand(l *lexer) stateFunc {
+	l.acceptAll(whitespace)
+	l.ignore()
+	if l.peek() == eof {
+		l.emit(TokenEOF)
+		return nil
+	}
+	r := l.next()
+	switch r {
+		case '#':
+			l.emit(TokenHash)
+			return lexPatternStringIndex
+		case '@':
+			l.emit(TokenAt)
+			return lexPatternIntegerIndex
+		case '.':
+			l.emit(TokenDot)
+			return lexCommand
+		case '*':
+			l.emit(TokenAst)
+			return lexCommand
+		case '{':
+			l.emit(TokenLBrace)
+			return lexCommand
+		case '}':
+			l.emit(TokenRBrace)
+			return lexCommandEnd
+	}
+	if isAlpha(r) {
+		l.emit(TokenCommand)
+		return lexCommandEnd
+	}
+	return l.errorf("Expected command found something else")
+}
+
+func lexPatternStringIndex(l *lexer) stateFunc {
+	l.acceptAllPassing(isStringIndexChar)
+	l.emit(TokenPatternStringIndex)
+	return lexCommand
+}
+
+func lexPatternIntegerIndex(l *lexer) stateFunc {
+	l.acceptAllPassing(isDigit)
+	l.emit(TokenPatternIntegerIndex)
+	return lexCommand
+}
+
+func lexCommandEnd(l *lexer) stateFunc {
+	if l.peek() == eof {
+		l.emit(TokenEOF)
+		return nil
+	}
+	if l.accept(";") {
+		l.emit(TokenSemicolon)
+		return lexCommand
+	}
+	return l.errorf("Expected ; found something else")
+}
-- 
cgit v1.2.3