wikilinks-parser/lexer.go

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

import (
	"fmt"
	// "os"
	"strings"
	"sync"

	// "unicode"
	"unicode/utf8"

	"go.uber.org/zap"
	"go.uber.org/zap/zapcore"
)

const (
	LexError LexemeType = iota
	LexEOF
	LexIdent
	LexOpenLink
	LexCloseLink
	LexHeading
	LexBlockRef
	LexAlias
	LexText
)

const (
	EOF rune = 0
)

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

func Lex(name, input string) *Lexer {
	encoderCfg := zap.NewProductionEncoderConfig()
	encoderCfg.EncodeTime = zapcore.ISO8601TimeEncoder

	config := zap.Config{
		Level:         zap.NewAtomicLevelAt(zap.InfoLevel),
		EncoderConfig: encoderCfg,
		OutputPaths: []string{
			"./lexer.log",
			"stdout",
		},
		Encoding: "console",
		ErrorOutputPaths: []string{
			"stderr",
		},
		InitialFields: map[string]interface{}{
			"lexer": name,
			// "pid": os.Getpid(),
		},
	}

	l := &Lexer{
		L:     zap.Must(config.Build()).Named("lexer"),
		name:  name,
		input: input,
		state: lexText,
		Items: make([]Lexeme, 0),
	}
	// go l.run()
	l.run()

	return l
}

/* shouldn't need this in non-concurrent implementation
func (l *Lexer) NextItem() Item {
	for {
		select {
		case item := <-l.items:
			return item
		default:
			if l.state == nil {
				l.L.Named("NextItem").Errorw("state should not be nil")
				return Item{
					Typ: ItemError,
					Val: "state is nil, should not be",
				}
			}

			l.state = l.state(l)
		}
	}
}
*/

func (l *Lexer) ignore() {
	l.SetStart(l.pos)
}

func (l *Lexer) backup() {
	l.SetPos(l.GetPos() - l.GetWidth())
}

type Lexer struct {
	L                 *zap.Logger
	name, input       string
	start, pos, width int
	state             stateFn
	// Items                                       chan Item
	Items                                       []Lexeme
	widthMutex, startMutex, posMutex, chanMutex sync.Mutex
}

func (l *Lexer) peek() rune {
	L := l.L.Named("peek")
	L.Debug("peeking")
	r := l.next()
	l.backup()

	return r
}

func (l *Lexer) accept(valid string) bool {
	L := l.L.Named("accept").With(
		zap.String("input", valid),
	)
	if strings.ContainsRune(valid, l.next()) {
		L.Debug("matched input")
		return true
	}

	L.Debug("rejected input")
	l.backup()

	return false
}

func (l *Lexer) acceptRun(valid string) {
	L := l.L.Named("acceptRun").With(
		zap.String("input", valid),
	)
	L.Debug("scanning")
	for strings.ContainsRune(valid, l.next()) {
	}
	l.backup()
}

func (l *Lexer) emit(t LexemeType) {
	i := Lexeme{t, l.input[l.GetStart():l.GetPos()]}
	L := l.L.Named("emit").With(
		zap.String("item", i.String()),
	)
	L.Info("emitting lexeme")
	l.Items = append(l.Items, i)
	l.SetStart(l.GetPos())
	/* original concurrent implementation
	defer l.chanMutex.Unlock()
	l.chanMutex.Lock()
	i := Item{t, l.input[l.GetStart():l.GetPos()]}
	L := l.L.With(
		zap.Int("pos", l.GetPos()),
		zap.Int("width", l.GetWidth()),
	).Named("emit")

	L.Debug("emitting item",
		zap.String("item", i.String()),
	)
	l.items <- i
	*/
}

func (l *Lexer) errorf(format string, args ...interface{}) stateFn {
	L := l.L.Named("errorf")
	errorItem := Lexeme{
		LexError,
		fmt.Sprintf(format, args...),
	}
	L.Debug("emitting errorItem",
		zap.String("error", errorItem.String()),
	)

	l.Items = append(l.Items, errorItem)
	return nil
}

func (l *Lexer) next() rune {
	var r rune
	L := l.L.Named("next")
	if l.GetPos() >= len(l.input) {
		L.Debug("end of input reached")
		l.SetWidth(0)
		return EOF
	}
	r, width := utf8.DecodeRuneInString(l.input[l.GetPos():])
	L.Debug("found rune",
		zap.String("rune", string(r)),
		zap.Int("width", width),
	)
	l.SetWidth(width)
	l.SetPos(l.GetPos() + l.GetWidth())
	return r
}

func (l *Lexer) run() {
	for state := lexText; state != nil; {
		state = state(l)
	}
	/* original concurrent implementation
	defer l.chanMutex.Unlock()
	for state := lexText; state != nil; {
		state = state(l)
	}
	l.chanMutex.Lock()
	close(l.items)
	*/
}

func (l *Lexer) GetPos() int {
	defer l.posMutex.Unlock()
	l.posMutex.Lock()
	l.L.Named("GetPos").Debug("getting current position",
		zap.Int("old", l.pos),
	)
	return l.pos
}

func (l *Lexer) SetPos(pos int) {
	defer l.posMutex.Unlock()
	l.posMutex.Lock()
	l.L.Named("SetPos").Debug("setting new position",
		zap.Int("new", pos),
		zap.Int("old", l.pos),
	)
	l.pos = pos
}

func (l *Lexer) GetWidth() int {
	defer l.widthMutex.Unlock()
	l.widthMutex.Lock()
	l.L.Named("GetWidth").Debug("setting new width",
		zap.Int("old", l.width),
	)
	return l.width
}

func (l *Lexer) SetWidth(width int) {
	defer l.widthMutex.Unlock()
	l.widthMutex.Lock()
	l.L.Named("SetWidth").Debug("setting new width",
		zap.Int("new", width),
		zap.Int("old", l.width),
	)
	l.width = width
}

func (l *Lexer) GetStart() int {
	defer l.startMutex.Unlock()
	l.startMutex.Lock()
	l.L.Named("GetStart").Debug("getting old start",
		zap.Int("old", l.start),
	)
	return l.start
}

func (l *Lexer) SetStart(start int) {
	defer l.startMutex.Unlock()
	l.startMutex.Lock()
	l.L.Named("SetStart").Debug("setting new start",
		zap.Int("new", start),
		zap.Int("old", l.start),
	)
	l.start = start
}

type stateFn func(*Lexer) stateFn

type LexemeType int

type Lexeme struct {
	Typ LexemeType
	Val string
}

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

	case LexEOF:
		return "EOF"
	case LexError:
		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)
	}
	return fmt.Sprintf("%s:%q", i.Typ, i.Val)
}