kuebiko/src/parser.js

import { Stream } from './iterator.js'
import { Result } from './result.js'
import { curry, range } from './utils.js'
import * as string from './string.js'

export class ParseError extends Error {
  name
  index
  description

  constructor(index, description) {
    super()
    this.name = 'ParseError'
    this.index = index
    this.description = description
  }
}

export class EofError extends ParseError {
  name = 'EofError'
  description = 'End of stream reached'
}

// convenience factory functions and cached callbacks
const eofErr = index => () => new EofError(index)
const parseErr = (index, desc) => () => new ParseError(index, desc)
const eq = a => b => a === b
const toString = value => value.toString()

export const pure = curry((value, input) => Result.ok([value, input]))
export const fail = curry((error, input) => Result.err(new ParseError(input.index, error)))

export const anyItem = () => input => {
  return input.peek().okOrElse(eofErr(input.index))
    .map(value => [value, input.drop()])
}

export const satisfy = curry((predicate, input) => {
  return input.peek()
    .okOrElse(eofErr(input.index))
    .filterOrElse(
      predicate,
      value => parseErr(input.index, `Value did not match predicate: ${value}`)
    )
    .map(value => [value, input.drop()])
})

export const literal = curry((value, input) => (
  satisfy(eq(value), input)
))

export const bind = curry((parser, transform, input) =>
  parser(input).andThen(([value, rest]) => transform(value)(rest))
)

export const map = curry((parser, morphism, input) => (
  parser(input).map(([value, rest]) => [morphism(value), rest])
))

export const seq = curry((a, b, input) => (
  bind(a, x => map(b, y => [x, y]), input)
))

export const alt = (...parsers) => input => {
  for (const p of parsers) {
    const result = p(input.clone())

    if (result.isOk()) {
      return result
    }
  }

  return Result.err(new ParseError(input.index, "No parsers matched alt"))
}

export const many = curry((parser, input) => {
  const results = []
  let stream = input

  while (true) {
    const result = parser(stream.clone())
    if (result.isOk()) {
      const [value, rest] = result.unwrap()
      results.push(value)
      stream = rest
    } else {
      break
    }
  }

  return Result.ok([results, stream])
})

export const many1 = curry((parser, input) => (
  parser(input).andThen(([first, rest]) => (
    many(parser, rest).map(([others, rest]) => (
      [[first, ...others], rest]
    ))
  ))
))

export const optional = curry((parser, input) =>
  parser(input.clone()).orElse(() => Result.ok([undefined, input]))
)

export const eof = () => input => (
  anyItem()(input)
    .andThen(([value, rest]) =>
      Result.err(
        new ParseError(rest.index, `Expected EOF, found ${value}`)
      )
    )
    .orElse(err => {
      if (err instanceof EofError) {
        return Result.ok([null, input])
      } else {
        return Result.err(err)
      }
    })
)

export const take = curry((limit, input) => {
  const result = []
  const next = anyItem()
  let stream = input

  for (const _index of range(limit)) {
    const nextResult = next(stream)
    if (nextResult.isOk()) {
      const [value, rest] = nextResult.unwrap()
      result.push(value)
      stream = rest
    } else {
      return nextResult
    }
  }

  return Result.ok([result, stream])
})

export const drop = curry((limit, input) =>
  skip(take(limit), input)
)

export const skip = curry((parser, input) =>
  parser(input).map(([_, rest]) => [undefined, rest])
)

const streamInfo = (stream, n = 3) => {
  const clone = stream.clone()
  const values = clone.take(n).map(toString).join(', ')
  const hasMore = clone.peek().isSome()
  const ellip = hasMore ? '...' : ''

  return `Stream { index = ${stream.index}, values = [${values}${ellip}]`
}

export const trace = curry((parser, label, input) => {
  const name = label || parser.name || 'anonymous parser'
  console.log(`trace(parser = ${name}, input = ${streamInfo(input)})`)
  const result = parser(input)
  if (result.isOk()) {
    const [value, rest] = result.unwrap()
    console.log(`    success: value = ${toString(value)}, remaining = ${streamInfo(rest)}`)
  } else {
    const err = result.unwrapErr()
    console.log(`    fail: error = ${err}`)
  }
})

// cached parsers
const LetterParser = alt(...Array.from(string.AsciiLetters).map(x => literal(x)))
const DigitParser = alt(...Array.from(string.Digits).map(x => literal(x)))
const WhitespaceParser = alt(...Array.from(string.Whitespace).map(x => literal(x)))

export const letter = () => LetterParser
export const digit = () => DigitParser
export const whitespace = () => WhitespaceParser

export const parseSome = curry((parser, value) =>
  parser(new Stream(value))
)

export const parse = curry((parser, value) => {
  const result = parseSome(seq(parser, eof()), value)

  if (result.isOk()) {
    const [[value, _rest], _stream] = result.unwrap()
    return Result.ok(value)
  } else {
    return result
  }
})