wikilinks-parser/lexer.go

//go:generate stringer -type=ItemType
package wikilink

import (
	"fmt"
	"log"
	"strings"
	// "unicode"
	"unicode/utf8"
)

const (
	ItemError ItemType = iota
	ItemEOF
	ItemText
	ItemOpenLink
	ItemCloseLink
	ItemFragment
	ItemAlias
)

const (
	EOF rune = 0
)

const (
	OpenLink  = "[["
	CloseLink = "]]"
	Alias     = "|"
	BlockRef  = "#^"
)

func Lex(name, input string) *Lexer {
	l := &Lexer{
		name:  name,
		input: input,
		state: lexText,
		items: make(chan Item, 2),
	}
	go l.run()

	return l
}

func (l *Lexer) NextItem() Item {
	for {
		select {
		case item := <-l.items:
			return item
		default:
			if l.state == nil {
				return Item{
					Typ: ItemError,
					Val: "state is nil, should not be",
				}
			}

			l.state = l.state(l)
		}
	}
}

func (l *Lexer) ignore() {
	l.start = l.pos
}

func (l *Lexer) backup() {
	l.pos -= l.width
}

type Lexer struct {
	name, input       string
	start, pos, width int
	state             stateFn
	items             chan Item
}

func (l *Lexer) peek() rune {
	r := l.next()
	l.backup()

	return r
}

func (l *Lexer) accept(valid string) bool {
	if strings.ContainsRune(valid, l.next()) {
		return true
	}

	l.backup()

	return false
}

func (l *Lexer) acceptRun(valid string) {
	for strings.ContainsRune(valid, l.next()) {
	}
	l.backup()
}

func (l *Lexer) emit(t ItemType) {
	i := Item{t, l.input[l.start:l.pos]}
	log.Printf("emitting Item: %#+v\n", i)
	l.items <- i
	l.start = l.pos
}

func (l *Lexer) errorf(format string, args ...interface{}) stateFn {
	log.Printf("entering errorf: %q\n", format)
	l.items <- Item{
		ItemError,
		fmt.Sprintf(format, args...),
	}

	return nil
}

func lexFragment(l *Lexer) stateFn {
	log.Println("entering lexFragment")
	for {
		if strings.HasPrefix(l.input[l.pos:], CloseLink) {
			return lexCloseLink
		}

		if l.peek() == '^' {
			l.next()
			l.emit(ItemFragment)
			l.acceptRun("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789 -.,")

			return lexInsideLink
		}

		return lexInsideLink
	}
}

func lexAlias(l *Lexer) stateFn {
	// l.accept
	log.Println("entering lexAlias")
	if strings.HasPrefix(l.input[l.pos:], CloseLink) {
		return lexCloseLink
	}
	l.emit(ItemAlias)

	return lexInsideLink
}

func lexInsideLink(l *Lexer) stateFn {
	log.Println("entering lexInsideLink")

	for {
		if strings.HasPrefix(l.input[l.pos:], CloseLink) {
			return lexCloseLink
		}

		r := l.next()

		switch {
		case r == EOF:
		case r == '\n':
			return l.errorf("unclosed link")
		case r == '#':
			l.emit(ItemText)

			return lexFragment
		case r == '|':
			l.emit(ItemText)

			return lexAlias
		case l.peek() == ']':
			l.emit(ItemText)

			return lexCloseLink
		}
	}
}

func lexOpenLink(l *Lexer) stateFn {
	log.Println("entering lexOpenLink")
	l.pos += len(OpenLink)
	l.emit(ItemOpenLink)
	return lexInsideLink
}

func lexCloseLink(l *Lexer) stateFn {
	log.Println("entering lexCloseLink")
	l.pos += len(CloseLink)
	l.emit(ItemCloseLink)
	return lexText
}

func lexText(l *Lexer) stateFn {
	log.Println("entering lexText")

	for {
		if strings.HasPrefix(l.input[l.pos:], OpenLink) {
			if l.pos > l.start {
				l.emit(ItemText)
			}
			return lexOpenLink
		}

		if l.next() == EOF {
			break
		}
		if l.pos > l.start {
			l.emit(ItemText)
		}
		l.emit(ItemEOF)
		return nil
	}
	return nil
}

func (l *Lexer) next() rune {
	var r rune
	if l.pos >= len(l.input) {
		l.width = 0
		return EOF
	}
	r, l.width = utf8.DecodeRuneInString(l.input[l.pos:])
	l.pos += l.width
	return r
}

func (l *Lexer) run() {
	for state := lexText; state != nil; {
		state = state(l)
	}
	close(l.items)
}

type stateFn func(*Lexer) stateFn

type ItemType int

type Item struct {
	Typ ItemType
	Val string
}

func (i Item) String() string {
	switch i.Typ {

	case ItemEOF:
		return "EOF"
	case ItemError:
		return i.Val
	}

	if len(i.Val) > 10 {
		return fmt.Sprintf("%s:%.10q...", i.Typ, i.Val)
	}
	return fmt.Sprintf("%s:%q...", i.Typ, i.Val)
}