codegen

This crate contains local commands used to auto-generate source code.

`cargo codegen grammar`

This command transforms the js.ungram file into the rome_js_syntax crate.

The project uses ungrammar to define the syntax of the language.

ungrammar uses a DSL to define and parse the grammar of a language.

Once the library parses the DSL files, some custom logic generates the AST APIs.

Conventions to create syntax

Here's a list of internal conventions that we follow to write grammar:

use the prefix manual__ to mark methods that are implemented manually;
text
```
MyDeclaration = manual__decl:Body
```
If you define like this, it means that somewhere in your code you will create something like:
rust
```
impl MyDeclaration {
	fn decl(self) -> Option<Body> {
		// custom logic goes here
	}
}
```

unions of tokens have to have a label. In this way, the code generation can handle it properly and generate a better AST.

text

BinExpr = left: Expr op: ('+' | '-' | '*') right: Expr

This will generate the correct implantation:

rust

impl BinExpr {
	fn op(self) -> Option<SyntaxToken> {
		// custom logic goes here
		support::find_token(
		&self.syntax,
		&[
			T![+],
			T![-],
			T![*],
		],
	)
	}
}

Nodes used to track broken code should contain the Bogus word in its name (case sensitive). This is needed because it will generate a different type of code, useful in case of errors inside the source code Rome will parse.

`cargo codegen test`

This command extracts inline comment tests inside rome_js_parser into the directory rome_js_parser/test_data/.

A usual workflow would be:

bash

# (modify inline comment tests inside the parser)
cargo codegen test
cargo test parser # for checking failed tests
UPDATE_EXPECT=1 cargo test parser # for committing the changes

`cargo codegen unicode`

This command downloads unicode data from unicode.org and writes it crates/rome_js_lexer/src/tables.rs. Use this command when unicode support has changed.