LeetSheet · Designer · Scroll · Subreddit

What is Parsers?

Parsers is a language for making languages.

Parsers is built on Particles (Particle Syntax).

Can you show me an example?

Below is some code written in Parsers to create a microlanguage called Panda.

pandaParser
 catchAllParser expressionParser
operatorAtom
 enum + -
integerAtom
expressionParser
 atoms operatorAtom
 catchAllAtomType integerAtom

The Parsers program above can generate a parser that parses statements like + 1 2 5 10 and - 10 2.

What is a longer description?

Parsers is a language built on Particles for building other languages built on Particles. The parsers not only allow parsing of those languages but also can include code written in a host language (such as Javascript) for compiling and/or runtime interpreting of these languages.

What is built on Parsers?

Many things are built on Parsers including Scroll and PLDB.

Can I try it online?

Yes! The Parser Designer is an online tool for writing Parsers code to build languages right in your browser.

Parsers Roadmap

The following major features are likely to be added to a future version of Parsers:

Import statements

Turning Parsers into a 2 pass language where the first pass resolves any imports
Imports will leverage the existing import code used by Scroll for posix style paths
Other protocols may be supported such as https:// imports and perhaps something like node_modules://

Dependent Types

The inScope parsers could change based upon current values of the particle. For example, you might have parsers where if you set type website certain parsers would come into/out of scope.
Scope Refactor. As part of the dependent type work the inScope implementation may undergo a refactor to allow for further future possibilities. For example, we may want to consider alternate approaches to inScope than the current inheritance system.

Union Atom Types

Sometimes a language may want to allow a number in a hole, or potentially something like an "N/A" value. Union Atom Types would be one way to do this.

Pattern Matching

You might want to have parsers with the same cue but different forms (for example, different atoms). Currently you might be able to do this using Regex parsers, but perhaps we could add some pattern matching code to make this easier.

Mixins

You might want to share code between parsers without creating an inheritance link. Mixins where code is copied at compile time would be a way to do this.

Positional Parsers

A common pattern in languages is to have different scopes for the first line of the file. For example, shebang lines in executable scripts or header lines in CSV files. We might want to create a way to have certain parsers inScope for line 0 and out of scope the rest of the lines.

LSP Support

We should move away from ParticleComponentFramework and the Designer App code base and move efforts toward having languages written in Parsers get an LSP implementation for free.

BNF Parsing

It's worth exploring using Parsers to also generate parsers for BNF languages.

Upgrading to CodeMirror Version 6

It may be worth it to upgrade the CodeMirror Mode Code to Code Mirror Version 6.

Parsers Release Notes

Here's a list of the notable changes in Parsers:

6.0.0 2023-4-2

⚠️ BREAKING:

instead of NodeTypeDefinitions ending in Node it is now ParserDefinitions ending in Parser
baseNodeType > baseParser
blobNode > blobParser
errorNode > errorParser
catchAllNodeType > catchAllParser

5.0.0 2023-3-30

🎉 You can now add scoped line parsers inside other line parser definitions.
⚠️ BREAKING: A lot of methods that were formerly getX() have been changed to getters like get x().

4.0.0 2023-3-28

🎉 blank lines are now OK
🎉 added // comments
⚠️ BREAKING:

Removed todo comments. Use // todo instead
Removed tooling keyword. Use // tooling instead

Parsers: A language for making languages