gigaparsec/gigaparsec.go

package gigaparsec

import (
	"errors"
	"fmt"
	"io"
	"slices"

	"git.codemonkeysoftware.net/b/gigaparsec/cursor"
)

type Result[In, Out any] struct {
	Value Out
	State[In]
	Message
}

type Message struct {
	Pos      uint64
	Got      string
	Expected []string
}

func (m Message) expect(s string) Message {
	m.Expected = []string{s}
	return m
}

func MessageOK(pos uint64) Message { return Message{Pos: pos} }

func MessageEnd(pos uint64) Message { return Message{Pos: pos, Got: "end of input"} }

type ParseError Message

func (pe ParseError) Error() string {
	return fmt.Sprintf("parse error: %d: %s", pe.Pos, pe.Got)
}

func MakeState[In any](c cursor.Cursor[In]) State[In] {
	return State[In]{cursor: c}
}

type State[In any] struct {
	cursor cursor.Cursor[In]
}

func (s State[In]) Cursor() cursor.Cursor[In] {
	return s.cursor
}

func (s State[In]) Read(dst []In) (n uint64, next State[In], err error) {
	n, c, err := s.cursor.Read(dst)
	return n, State[In]{cursor: c}, err
}

func (s State[In]) Pos() uint64 {
	return s.cursor.Pos()
}

type Parser[In, Out any] func(State[In]) (consumed bool, reply Result[In, Out], err error)

func Return[In, Out any](value Out) Parser[In, Out] {
	return func(state State[In]) (bool, Result[In, Out], error) {
		return false, Result[In, Out]{
			Value:   value,
			State:   state,
			Message: MessageOK(state.Pos()),
		}, nil
	}
}

func Satisfy[T any](pred func(T) bool) Parser[T, T] {
	return func(state State[T]) (bool, Result[T, T], error) {
		token := make([]T, 1)
		n, next, err := state.Read(token)
		if errors.Is(err, io.EOF) {
			return false, Result[T, T]{}, ParseError(MessageEnd(state.Pos()))
		}
		if err != nil {
			return false, Result[T, T]{}, err
		}
		if n != 1 {
			panic(fmt.Sprintf("expected 1 element from Read, but got %d", n))
		}
		if pred(token[0]) {
			return true, Result[T, T]{
				Value:   token[0],
				State:   next,
				Message: MessageOK(state.Pos()),
			}, nil
		}
		return false, Result[T, T]{}, ParseError{
			Pos: state.Pos(),
			Got: fmt.Sprint(token),
		}
	}
}

func Slice[T comparable](s []T) Parser[T, []T] {
	expected := fmt.Sprint(s)
	return func(state State[T]) (consumed bool, reply Result[T, []T], err error) {
		token := make([]T, len(s))
		_, next, err := state.Read(token)
		if errors.Is(err, io.EOF) {
			return false, Result[T, []T]{}, ParseError(MessageEnd(state.Pos()))
		}
		if err != nil {
			return false, Result[T, []T]{}, err
		}
		if !slices.Equal(s, token) {
			return false, Result[T, []T]{}, ParseError{
				Pos:      state.Pos(),
				Got:      fmt.Sprint(token),
				Expected: []string{expected},
			}
		}
		return true, Result[T, []T]{
			Value:   token,
			State:   next,
			Message: MessageOK(state.Pos()),
		}, nil
	}
}

func Bind[In, A, B any](p Parser[In, A], f func(A) Parser[In, B]) Parser[In, B] {
	return func(input State[In]) (bool, Result[In, B], error) {
		consumed, resultA, err := p(input)
		if err != nil {
			return false, Result[In, B]{}, err
		}
		consumed2, replyB, err := f(resultA.Value)(resultA.State)
		return consumed || consumed2, replyB, err
	}
}

func Choose[In, Out any](p Parser[In, Out], ps ...Parser[In, Out]) Parser[In, Out] {
	all := append([]Parser[In, Out]{p}, ps...)
	return func(input State[In]) (bool, Result[In, Out], error) {
		expecteds := make([][]string, 0, len(all))
		var value Out
		var got string
		var gotGot bool
		var failed bool
		for _, q := range all {
			consumed, result, err := q(input)
			if consumed {
				return consumed, result, err
			}
			var qMsg Message
			if err != nil {
				var parseErr ParseError
				if !errors.As(err, &parseErr) {
					// It broke. Give up.
					return consumed, result, err
				}
				failed = failed && true
				qMsg = Message(parseErr)
			} else {
				if failed {
					value = result.Value
					failed = false
				}
				qMsg = result.Message
			}
			if !gotGot {
				got = qMsg.Got
				gotGot = true
			}
		}
		msg := Message{Pos: input.Pos(), Got: got, Expected: slices.Concat(expecteds...)}
		if failed {
			return false, Result[In, Out]{}, ParseError(msg)
		}
		return false, Result[In, Out]{
			Value:   value,
			State:   input,
			Message: msg,
		}, nil
	}
}

// Try behaves identically to p, except that if p returns an error,
// Try will pretend that no input was consumed. This allows infinite
// lookahead: Since Choose only calls another parser when the previous
// parser consumed nothing, Try will allow backing out of a complex
// parser that did partially succeeded.
func Try[In, Out any](p Parser[In, Out]) Parser[In, Out] {
	return func(input State[In]) (bool, Result[In, Out], error) {
		consumed, reply, err := p(input)
		if err != nil {
			return false, Result[In, Out]{}, err
		}
		return consumed, reply, err
	}
}

func Label[In, Out any](p Parser[In, Out], l string) Parser[In, Out] {
	return func(input State[In]) (consumed bool, reply Result[In, Out], err error) {
		consumed, reply, err = p(input)
		if consumed {
			return
		}
		if err == nil {
			reply.Message = reply.Message.expect(l)
			return
		}
		var parseErr ParseError
		if errors.As(err, &parseErr) {
			err = ParseError(Message(parseErr).expect(l))
			return
		}
		return
	}
}

func Map[In, Out1, Out2 any](p Parser[In, Out1], f func(Out1) Out2) Parser[In, Out2] {
	return Bind(p, func(out Out1) Parser[In, Out2] {
		return Return[In](f(out))
	})
}