parser: implement initial tokeniser

only logging position, errors and non-whitespace elements

Signed-off-by: Alejandro Mery <amery@jpi.io>

parser/lexer.go

@@ -0,0 +1,66 @@
+package parser
+
+import (
+	"log"
+
+	"asciigoat.org/core/lexer"
+)
+
+// Run parses the source
+func (p *Parser) Run() error {
+	p.pos.Reset()
+	return lexer.Run(p.lexStart)
+}
+
+func (p *Parser) lexStart() (lexer.StateFn, error) {
+	for {
+		r, _, err := p.src.ReadRune()
+		switch {
+		case err != nil:
+			// read error
+			log.Printf("%s: %s: %s", p.pos, "error", err)
+			return nil, err
+		case IsNewLine(r):
+			// new line
+			p.lexNewLine(r)
+			p.src.Discard()
+			p.pos.StepLine()
+		case IsSpace(r):
+			// whitespace
+			p.src.Discard()
+			p.pos.Step()
+		default:
+			// token
+			p.src.UnreadRune()
+			return p.lexToken, nil
+		}
+	}
+}
+
+func (p *Parser) lexToken() (lexer.StateFn, error) {
+	p.src.AcceptAll(IsNotSpace)
+
+	s := p.src.Emit()
+	log.Printf("%s: %s: %q", p.pos, "token", s)
+	p.pos.StepN(len(s))
+
+	return p.lexStart, nil
+}
+
+func (p *Parser) lexNewLine(r1 rune) {
+	// r1 is warrantied to be either \n or \r
+	r2, _, err := p.src.ReadRune()
+
+	switch {
+	case r1 == '\r' && r2 == '\n':
+		// CR LN
+	case r1 == '\r' && err == nil:
+		// CR
+		p.src.UnreadRune()
+	case r2 == '\r':
+		// LN CR
+	case err == nil:
+		// LN
+		p.src.UnreadRune()
+	}
+}

parser/lexer_runes.go

@@ -0,0 +1,14 @@
+package parser
+
+import "asciigoat.org/core/lexer"
+
+var (
+	// IsNewLine tells if a rune represents a line break or the start of one
+	IsNewLine = lexer.NewIsIn("\n\r")
+	// IsSpace tells if a rune is considered whitespace by unicode
+	IsSpace = lexer.IsSpace
+	// IsNotNewLine tells if a rune is anything other than line breaks
+	IsNotNewLine = lexer.NewIsNot(IsNewLine)
+	// IsNotSpace tells if a rune is anything other than whitespace
+	IsNotSpace = lexer.NewIsNot(IsSpace)
+)

parser/parser.go

@@ -10,6 +10,8 @@ import (
 // Parser parses a dosini-style document
 type Parser struct {
 	src *lexer.Reader
+
+	pos lexer.Position
 }
 
 // NewParser creates a dosini-style parser using