Need help with parjs?
Click the “chat” button below for chat support from the developer who created it, or find similar developers for support.

About the developer

GregRos
134 Stars 12 Forks MIT License 191 Commits 15 Opened issues

Description

JavaScript parser-combinator library

Services available

!
?

Need anything else?

Contributors list

Parjs - Parser Combinator Library

build codecov npm

API Documentation

Parjs is a JavaScript library of parser combinators, similar in principle and in design to the likes of Parsec and in particular its F# adaptation FParsec.

Parjs was originally meant to be pronounced "Paris", because the j kind'a looks like an i, but I just kept calling it par-js in my head anyway, so make of that what you will.

It's also similar to the parsimmon library, but intends to be superior to it. Some of its features:

  1. Many more combinators and basic parsers.
  2. Support for parsing Unicode characters.
  3. Written in TypeScript with ES6 features.
  4. Systematically documented.
  5. Advanced debugging features and ability to parse very complex languages.

Parjs is written in TypeScript, using features of ES6+ such as classes, getter/setters, and other things. It's designed to be used from TypeScript too, but that's not necessary.

Starting from version 0.12.0, Parjs is written to leverage tree-shaking. If you don't use something in the library, it won't be inserted into your bundle. This includes the Unicode parsers. They're pretty heavy.

Parjs parsers are called "Parjsers". I'm not sure how to pronounce that either.

Parjs 0.12.0

I decided to make lots of changes to the library, including refactoring lots of code, renaming things, and redesigning the API to support things like tree shaking. I started this library when I had less experience, so after a few years I think I can make some cool improvements.

I've been seeing more and more stars and views recently, so I was thinking this might be the last time I can do something like this.

Some major things that have been changed:

  • No quiet parsers that don't return values. Now all parsers return values. This seemed like a nice feature in my head, but turned out to cause difficulties down the road.
  • Combinators use a
    pipe
    method and function operators instead of instance methods. See more on this below.
  • Names for some types and objects.

If you want to use the previous version with the old API, you can install

[email protected]
.

Example Parsers

You can see implementations of example parsers in the

examples
folder:
  1. Tuple Parser
  2. JSON parser
  3. Math Expression Parser

What's a parser-combinator library?

It's a library for building complex parsers out of smaller, simpler ones. It also provides a set of those simpler building block parsers.

For example, if you have a parser

digit
for parsing decimal digits, you can parse a number by applying
digit
multiple times until it fails, and then producing the consumed text as a result. Then you can use another combinator to convert the result to a number.

By combining different parsers in different ways, you can construct parsers for arbitrary expressions and languages.

Here is how you might construct a parser for text in the form

(a, b, c, ...)
where
a, b, c
are floating point numbers. One feature of the expression is that arbitrary amounts of whitespace are allowed in between the numbers.
import {float, string, whitespace} from "parjs";
import {between, manySepBy} from "parjs/combinators"

// Built-in building block parser for floating point numbers. let tupleElement = float();

// Allow whitespace around elements: let paddedElement = tupleElement.pipe( between(whitespace()) );

// Multiple instances of {paddedElement}, separated by a comma: let separated = paddedElement.pipe( manySepBy(",") );

// Surround everything with parentheses: let surrounded = separated.pipe( between("(", ")") );

// Prints [1, 2, 3]: console.log(surrounded.parse("(1, 2 , 3 )"));

In the above example,

float
,
string
, and
whitespace
are building-block parsers that parse certain kinds of text, and
between
and
manySepBy
are combinators, which take those parsers and apply them in different ways.

What can you use it for?

Parsing, generally. You can parse all sorts of things:

  1. A custom DSL specifying an algorithm for chicken counting.
  2. Your own flavor of markdown, just to make things even more confusing.
  3. A custom data-interchange format inspired by chess notation.

The possibilities are limitless.

Since it's written in JavaScript, it can be used in web environments.

Installing

# yarn:
yarn add parjs

npm:

npm install parjs --save

What's this weird
pipe
method?

Prior to version 0.12.0,

parjs
implemented combinators as instance methods. This is nice, but has downsides:
  1. It's less convenient to add your own combinators, since that required modifying the prototype.
  2. More importantly, it makes tree-shaking impossible. So even if you don't use one of the combinators, you still have to put it into your bundle. This is a big deal for web environments.

Parjs makes the same move as rxjs, another much more popular library of combinators (though a different kind) and introduces an API based on higher-order function operators, what rxjs calls lettable operators.

Each parser supports one key prototype method:

pipe
, and this method accepts a chain of functions that feed into each other, transforming the original, source parser.

These transformations are created by calling other functions and giving them various arguments. For example, the factory function for the

map
combinator looks like this:
function map(projection: (x: A) => B): ParjsCombinator

Where:

type ParjsCombinator = (source: Parjser) => Parjser;

So it's a function that returns another function. Here is how it's used in practice:

import {map} from "parjs/combinators";
import {string} from "parjs";

const parser = string("test").pipe( map(result => result.length) );

In most ways, this API is identical to the prototype-based API. Instead of writing

parser.map(f)
, you write
parser.pipe(map(f))
. This is more long-winded, but the benefits outweigh the drawbacks.

Using combinators like regular functions

You don't really have to use the

.pipe
method. Combinators are just functions that return functions, and you can call the combinator and the function it creates in a single expression. For example:
let hiOrHello = or("hi")("hello");

Import paths

  • "parjs"
    - here you will find the building-block parsers and no combinators. Also, commonly used types if you're working from TypeScript.
  • "parjs/combinators"
    - here you will find combinators, the stuff you give the
    pipe
    method.
  • "parjs/errors"
    - here are some error types
    parjs
    can throw.
  • "parjs/trace"
    - Visualizing parser failures.
  • "parjs/internal"
    - here be dragons. See "creating custom parsers" from more details.

Parsing Unicode

Parjs can parse Unicode characters in the BMP (Basic Multilingual Plane), which includes all but the most exotic of characters. This is done using the [

char-info
][char-info] package of character recognizers.

Parsers such as

upper()
only parse the ASCII subset of Unicode. Only parsers with names beginning with
uni
, such as
uniUpper()
, parse all Unicode characters. These parsers are inherently slower, because each character needs to be looked up in a tree-like data structure.

Parjs supports tree-shaking, so if you have this feature enabled, it will only embed Unicode data into your bundle if you use the Unicode features.

If you want to parse characters from specific scripts or with other special properties, you should import the

char-info
package yourself.

Implicit parsers

Combinators that accept parsers as parameters can be given a literal instead. Two types of literals are supported:

  1. A string.
  2. A regular expression.

These literals are automatically converted to parsers that parse them, using the

string
and
regexp
parsers, respectively. So while it is clearer to write:
let p = string("a").pipe(
    then(string("b"))
);

You can simply write:

let p = string("a").pipe(
    then("b")
);

And your code will behave the same way.

Immutability

Parjs parsers are functionally immutable. That is, once a Parjser is created, it will always do the same thing, whether it's applied for the first or tenth time. Some things, like non-functional metadata, may be subject to change though.

This allows you to write such idiomatic code as:

let myString = string("my personal string");
let variant1 = myString.pipe(
    then(" is okay.")
);
let variant2 = myString.pipe(
    then(" is the best.")
);

If

myString
changed when it parsed something, it would influence both
variant1
and
variant2
, which is obviously undesirable.

Success and failure

When the

.parse
method of a Parjser you get a result which can indicate either success or failure. You can tell these apart using the
.kind
property, or using the shorthand
.isOkay
:
let result = parser.parse("hello");

// true if result.kind === "OK" if (result.isOkay) { // it succeeded } else { // result.kind === "Soft", "Hard", "Fatal" }

You can get the parser's return value by using the

.value
property of the result object, but it will throw an error if the object is a failure.
let finalResult = result.value;

Failures have the extra property

trace
which gives you a trace of where the failure happened, which parser caused it, and other information you can use to diagnose it.
let {
    trace,
    reason,
    kind,
} = result;

The

trace
property contains a detailed object with a trace of parsers that led to the failure, the location of the failure, and more.

This information is used when you stringify a failed result object with

.toString()
:
console.log(result.toString());
// Soft failure at Ln 1 Col 1
// 1 | hello!
//     ^expecting 'hi'
// Stack: string

Failure types

There are several failure types recognized by the library. They're used for different purposes that can drastically change how a parser behaves. From least to most severe, they are:

  1. Soft failure
  2. Hard failure
  3. Fatal failure

Different failure types let you accept alternative inputs while not swallowing important syntax errors and not backtracking too much.

Failure bubbles up the parser tree until a parser can handle it, like an exception does. Parsers that deal with Soft failures will usually not handle Hard ones.

Soft failure

A parser fails softly if it receives input which it immediately sees as inappropriate.

This kind of failure allows alternative parser combinators like the

or
combinators to work. The
or
combinator looks like this:
let option1 = string("a");
let option2 = string("b");
let either = option1.pipe(
    or(option2)
);

The combinator will try the

option2
parser if
option1
does not work. This not-working is signalled by a soft failure. If
option2
reports a soft failure too then
either
will bubble that failure up, possibly with some more information.

One way of describing a soft failure is that it doesn't consume much of the input and doesn't require much backtracking.

Hard failure

A hard failure means the parser started parsing the input but encountered something unexpected.

Hard failures are common and are usually caused by syntax errors. A common hard failure appears when you have a parser that uses the

then
sequential combinator:
let p = string("a").pipe(
    then("b")
);

And you give it input that makes the 1st parser succeed and the 2nd parser to fail.

p.parse("ac");
// Hard failure at Ln 1 Col 2
// 1 | ac
//      ^expecting 'b'
// Stack: string < then

In this case, combinators like

or
will not work. Even if we added the
or
combinator to the above:
// doesn't work, still fails

let p2 = p.pipe( or("ac") );

p2.parse("ac");

This still results in the same failure as before. The idea behind this is that the input made the parser break an expectation. When the first parser for

"a"
succeeds, it convinces the
then
combinator that this really is the parser it's supposed to be using. When the parser for
"b"
fails, it sees it as a syntax error and not just an alternative input type.

The recommended way to solve this problem is to write parsers that quickly determine if the input is right for them, without having to apply multiple parsers and backtrack a non-constant amount to recover from a failure. For example, we could write the above parser like this:

let example = string("a").pipe(
    then(
        or("b")("c")
    )
);

This failure can also appear in other parsers. For example:

let floatParser = float();

floatParser.parse("5.0e+hello");

Here we're using the

float()
parser to parse a floating-point number. We give it an input with a number in scientific notation, except the exponent is gibberish. By the time the parser reached the exponent, it had already chosen to interpret it as a number in scientific notation, so the lack of a valid exponent breaks this expectation.

See the section below to learn about how to recover from a Hard failure if you really need to.

Fatal failure

This is an extra failure type which isn't emitted by Parjs by default, but you can build your own parsers to emit it. It won't be handled by any combinator and will cause parsing to fail.

Recovering from (most) failures

You can use the

recover
combinator to recover from non-Fatal failures.

You can give it a handler that will be called if the parser it's used on fails, together with all the failure information. You can then alter the parser's result to something else or return nothing to signal nothing should change. The function will not be called if the parser succeeds.

Here is an example of it being used:

let hardFailingParser = fail({kind: "Hard", reason: "who knows"});

let recovered = hardFailingParser.pipe( recover(failure => { if (failure.reason === "who knows") { return { kind: "OK", value: "some value to return" } } }) );

Signalling failure

You can signal a failure in several different ways.

The

fail
basic parser is a parser that fails immediately with the information you give it when it's created. It's mainly provided for completeness.

The

must
combinator makes sure that a parser has the correct user state and return value. Otherwise it makes a parser fail with the info and severity you give it. There also exist several more parsers accepting predicates like this.

The

recover
combinator can be used to re-emit failures too.

User State

User state is a feature that can help you to parse complex languages, like mathematical expressions with operator precedence and languages like XML where you need to match up an end tag to a start tag.

Every time you invoke the

.parse
method Parjs creates a unique, mutable user state object. The object is propagated throughout the parsing process and some combinators and building block parsers can modify it or inspect it. The only information in it will be what you put inside it and it won't change how the rest of the library behaves.

The

.parse
method accepts an additional parameter
initialState
that contains properties and methods that are merged with the user state:
// p is called with a parser state initialized with properties and methods.
let example = p.parse("hello", {token: "hi", method() {return 1;});

The combinator

map
is a projection combinator. You can give it a function taking two parameters: the parser result and the parser state.
let example = string("a").pipe(
    map((result, state) => state.flag)
);

each
is a combinator that doesn't change the parser result, so you can use it to only modify the user state.

User state is a less idiomatic and elegant feature meant to be used together with, rather than instead of, parser results.

Replacing user state

The combinator

replaceState
lets you replace the user state object, but only in the scope of the parser it applied to.

It creates a brand new user state object, merged with properties from the object you specify, and gives it to the parser. Once the parser is finished, the old user state object is restored. This means you will need to use that parser's result value to communicate out of it, and it serves the isolate other parsers from what happens inside.

Replacing user state is powerful, and can allow you to write recursive parsers that need a hierarchy of nested user states to work.

Writing a parser with custom low-level logic

In most cases, it should be easy to use existing combinators and building block parsers to create what you want. You shouldn't automatically write a custom parser.

Writing a parser with totally custom logic lets you read the input and manage the position directly. This can allow you to implement new kinds of building-block parsers. While Parjs is meant to be easily extensible, this API will probably change more than more outward facing APIs, so be warned.

Parser flow

When parsing, a unique mutable

ParsingState
object is created. This object has the following shape:
interface ParsingState {
    readonly input : string;
    position : number;
    value : any;
    userState : UserState;
    reason : string;
    kind : ReplyKind;
    //...
}

Each parser gets handed this object and needs to mutate its properties to return information and change the position.

The

kind
,
value
, and
reason
properties are used to send data out of the parser.
  1. The
    kind
    gives the result type: success, failure, and which type of failure.
  2. value
    is used to output the parser result. It must be assigned if the
    kind
    is
    OK
    , and must not be assigned otherwise.
  3. reason
    is used to communicate the reason for an error, if any. You should only set it in case of an error. If you don't set it and signal an error, the
    expecting
    field in your parser object will be used as the error.

You can modify the other properties too, except for

input
which you almost certainly should not modify.

Creating the parser

To create a custom

Parjs
parser you need to extend the class
ParjserBase
, which you import from
parjs/internal
.
  • Override the
    _apply
    method to set the logic of the parser (this method is the one that takes the parsing state above).
  • You also need to set
    expecting
    which is a default error string the parser will use in case of error.
  • Finally, set the
    type
    string of the parser. This string is used to identify the parser and isn't only for informational purposes. It could be used in things like optimizations for example.

Here is a simple implementation of the

eof
parser, which detects the end of the input. Add type annotations as desired.
import {ParjserBase, ParsingState} from "parjs/internal";

export class Eof extends ParjserBase { type = "eof"; expecting = "expecting end of input"; constResult = undefined;

constructor(constResult: any) {
    this.constResult = constResult;
}

_apply(state: ParsingState) {
    if (state.position === state.input.length) {
        state.kind = "OK";
        state.value = this.constResult;
        return;
    }
    // we don't set the `reason` so it will be taken out of the `expecting
    // property
    state.kind = "Soft";
    return;
}

}

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.