arcthird

A modern and more abstract version of the Arcsecond library made with Coffeescript.

Usage no npm install needed!

<script type="module">
  import arcthird from 'https://cdn.skypack.dev/arcthird';
</script>

README

Arcthird

A modern and more abstract version of the Arcsecond library made with Coffeescript.

Arcsecond is a simple parsing library inspired by Haskell's Parsec, made from scratch by francisrstrokes who runs the LLJS YouTube channel.

Arcthird makes it more abstract by introducing a new class called PStream, from which your parsers are gonna parse from. This allows for example to make a parser of tokens, represented as JS objects.

The documentation is still very empty. It will get more full once I have more free time. If you'd like to contribute, don't hesitate to create pull requests!

Minor changes in functions

I've decided to change some of the functions from Arcsecond. Overall:

  • The ParserState is stored in its own class. Therefore:
    • createParserState is replaced by the ParserState constructor;
    • updateError is replaced by the errorify method;
    • updateResult is replaced by the resultify method;
    • updateData is replaced by the dataify method;
    • updateParserState is replaced by the update method.
  • The Parser's run method is called parse, and has to take a PStream. Thus the curried parse function also takes a PSteam as an argument.
  • For string / buffer / typed array specific parsers from Arcsecond, I created a StringPStream class instance of PSteam. So you would write for example
    char('h').parse(new StringPStream("hello world"))
parse(str("hello"))(new StringPStream("hello world"))
  • Because writing new StringPStream everytime would be kinda annoying, I created the strparse function to directly take a string instead:
    strparse(str("hello"))("hello world")
  • There is no many1 function. Instead, there is atLeast and atLeast1, which is more general.

Classes

PStream

PStream is an abstract class which is gonna be the input stream of every parser you're gonna use and create.

-- This is just for type annotation
class PStream
  constructor :: t -> PStream t
  length :: PStream t ~> () -> Int
  elementAt :: PStream t ~> Int -> a
  next :: PStream t ~> () -> a
  nexts :: PStream t ~> Int -> [a]
  • next: gives the next element.
  • nexts: gives an array of the next n elements.

The next and nexts methods are automatically defined in terms of elementAt. Although they aren't used anywhere as I didn't realize that I had used the index property of the ParserState class all this time. Replacing that with the index property of the PStream is an improvement to make in the future.

Because it is an abstract class, you cannot use it as is, you have to create your own extension and implement the length and elementAt functions yourself. For example:

class NumberStream extends PStream {
  constructor(target) {
    if (!(target instanceof Array))
      throw new TypeError("Target must be an Array");
    super(target);
  }

  length() {
    return this.structure.length;
  }

  elementAt(i) {
    try {
      return this.structure[i];
    } catch (err) {
      return null;
    }
  }
}

const nstream = new NumberStream([1, 2, 3, 42, 69]);
console.log(nstream.next());       // 1
console.log(nstream.nexts(2));     // [2, 3]
console.log(nstream.elementAt(0)); // 1
console.log(nstream.nexts(3));     // [42, 69, null]
console.log(nstream.next());       // null
console.log(nstream.length());     // 5

Parser

The type of a parser as used in the documentation is different from Arcsecond. Because of the new abstraction, there is now a new generic type to take into account: t as a PStream.

-- This is just for type annotation
class Parser
  constructor :: PStream t => (ParserState t * d -> ParserState t a d) -> Parser t a d
  parse :: PStream t => Parser t a d ~> t -> ParserState t a d
  fork :: PStream t => Parser t a d ~> (t, (ParserState t a d -> x), (ParserState t a d -> y)) -> (Either x y)
  map :: PStream t => Parser t a d ~> (a -> b) -> Parser t b d
  chain :: PStream t => Parser t a d ~> (a -> Parser t b d) -> Parser t b d
  ap :: PStream t => Parser t a d ~> Parser t (a -> b) d -> Parser t b d
  errorMap :: PStream t => Parser t a d ~> ({d, String, Int} -> String) -> Parser t a d
  errorChain :: PStream t => Parser t a d ~> ({d, String, Int} -> Parser t a d) -> Parser t a d
  mapFromData :: PStream t => Parser t a d ~> ({a, d} -> b) -> Parser t b d
  chainFromData :: PStream t => Parser t a d ~> ({a, d} -> Parser t b e) -> Parser t b e
  mapData :: PStream t => Parser t a d ~> (d -> e) -> Parser t a e
  of :: PStream t => x -> Parser t x d

All the errors returned in parser states are strings. Now this doesn't change anything from Arcsecond, although Arcsecond explicitly allowed for different error types. After verification, I realized that my code does also allow for different error types; however, this isn't reflected in the documentation where errors are described as strings even for parser combinators.

  • constructor: constructs a new parser by taking a ParserState transformer as an argument, which is a function that takes a ParserState and returns a new ParserState.
    const consume = new Parser(state => new ParserState({
  ...state.props(),
  index: state.target.length()
}));
  • parse: directly parses a PStream into a ParserState.
    char('a').parse(new StringPStream("abc"));

Parser Generators

char

char :: StringPStream t => Char -> Parser t Char d

Parses a single character.

strparse(char('h'))("hello world")
// -> ParserState result='h' index=1

anyChar

anyChar :: StringPStream t => Parser t Char d

Parses any character.

strparse(anyChar)("bruh")
// -> ParserState result='b' index=1

peek

peek :: PStream t => Parser t a d

Returns the element at the current index as a result without moving forward.

strparse(peek)("something")
// -> ParserState result='s' index=0

str

str :: StringPStream t => String -> Parser t String d

regex

regex :: StringPStream t => RegExp -> Parser t String d

digit

digit :: StringPStream t => Parser t String d

digits

digits :: StringPStream t => Parser t String d

letter

letter :: StringPStream t => Parser t String d

letters

letters :: StringPStream t => Parser t String d

anyOfString

anyOfString :: StringPStream t => String -> Parser t Char d

endOfInput

endOfInput :: PStream t => Parser t Null d

whitespace

whitespace :: StringPStream t => Parser t String d

optionalWhitespace

optionalWhitespace :: StringPStream t => Parser t String d

Parser Combinators

getData

getData :: PStream t => Parser t a d

setData

setData :: PStream t => d -> Parser t a d

mapData

mapData :: PStream t => (d -> e) -> Parser t a e

withData

withData :: PStream t => Parser t a d -> e -> Parser t a e

pipe

pipe :: PStream t => [Parser t * *] -> Parser t * *

compose

compose :: PStream t => [Parser t * *] -> Parser t * *

tap

tap :: PStream t => (ParserState t a d -> IO ()) -> Parser t a d

parse

parse :: PStream t => Parser t a d -> t -> Either String a

strparse

strparse :: StringPStream t => Parser t a d -> String -> Either String a

decide

decide :: PStream t => (a -> Parser t b d) -> Parser t b d

fail

fail :: PStream t => String -> Parser t a d

succeedWith

succeedWith :: PStream t => a -> Parser t a d

either

either :: PStream t => Parser t a d -> Parser t (Either String a) d

coroutine

coroutine :: PStream t => (() -> Iterator (Parser t a d)) -> Parser t a d

exactly

exactly :: PStream t => Int -> Parser t a d -> Parser t [a] d

many

many :: PStream t => Parser t a d -> Parser t [a] d

atLeast

atLeast :: PStream t => Int -> Parser t a d -> Parser t [a] d

atLeast1

atLeast1 :: PStream t => Parser t a d -> Parser t [a] d

mapTo

mapTo :: PStream t => (a -> b) -> Parser t b d

errorMapTo

errorMapTo :: PStream t => ({d, String, Int} -> String) -> Parser t a d

namedSequenceOf

namedSequenceOf :: PStream t => [(String, Parser t * *)] -> Parser t (StrMap *) d

sequenceOf

sequenceOf :: PStream t => [Parser t * *] -> Parser t [*] *

sepBy

sepBy :: PStream t => Parser t a d -> Parser t b d -> Parser t [b] d

sepBy1

sepBy1 :: PStream t => Parser t a d -> Parser t b d -> Parser t [b] d

choice

choice :: PStream t => [Parser t * *] -> Parser t * *

between

between :: PStream t => Parser t a d -> Parser t b d -> Parser t c d -> Parser t b d

everythingUntil

everythingUntil :: StringPStream t => Parser t a d -> Parser t String d

everyCharUntil

everyCharUntil :: StringPStream t => Parser t a d -> Parser t String d

anythingExcept

anythingExcept :: StringPStream t => Parser t a d -> Parser t Char d

anyCharExcept

anyCharExcept :: StringPStream t => Parser t a d -> Parser t Char d

lookAhead

lookAhead :: PStream t => Parser t a d -> Parser t a d

possibly

possibly :: PStream t => Parser t a d -> Parser t (a | Null) d

skip

skip :: PStream t => Parser t a d -> Parser t a d

recursiveParser

recursiveParser :: PStream t => (() -> Parser t a d) -> Parser t a d

takeRight

takeRight :: PStream t => Parser t a d -> Parser t b d -> Parser t b d

takeLeft

takeLeft :: PStream t => Parser t a d -> Parser t b d -> Parser t a d

toPromise

toPromise :: PStream t => ParserState t a d -> Promise (String, Integer, d) a

toValue

toValue :: PStream t => ParserState t a d -> a