Prints Babel AST to readable JavaScript. Use 🐊Putout to parse your code.
You may also use Babel 8 with estree-to-babel
for ESTree and Babel AST to put .extra.raw
to .raw
(which is simpler for transforms, no need to use Optional Chaining and add extra values every time).
- ☝️ Similar to Recast, but twice faster, also simpler and easier in maintenance, since it supports only Babel.
- ☝️ As opinionated as Prettier, but has more user-friendly output and works directly with AST.
- ☝️ Like ESLint but works directly with Babel AST.
- ☝️ Easily extendable with help of Overrides.
Supports:
- ✅ ES2023;
- ✅ JSX;
- ✅ TypeScript;
- ✅ JSON;
npm i @putout/printer
Printer has first class support from 🐊Putout with help of @putout/plugin-printer
. So install:
npm i @putout/plugin-printer -aD
And update .putout.json
:
{
"printer": "putout",
"plugins": ["printer"]
}
To benefit from it.
const {print} = require('@putout/printer');
const {parse} = require('putout');
const ast = parse('const a = (b, c) => {const d = 5; return a;}');
print(ast);
// returns
`
const a = (b, c) => {
const d = 5;
return a;
};
`;
When you need to extend syntax of @putout/printer
just pass a function which receives:
-
path
, Babel Path -
print
, a function to output result of printing into token array;
When path
contains to dashes __
and name, it is the same as: write(path.get('right'))
, and this is
actually traverse(path.get('right'))
shortened to simplify read and process.
Here is how you can override AssignmentPattern
:
const ast = parse('const {a = 5} = b');
print(ast, {
format: {
indent: ' ',
newline: '\n',
space: ' ',
splitter: '\n',
quote: `'`,
endOfFile: '\n',
},
semantics: {
comments: true,
maxSpecifiersInOneLine: 2,
maxElementsInOneLine: 3,
maxVariablesInOneLine: 4,
maxPropertiesInOneLine: 2,
maxPropertiesLengthInOneLine: 15,
trailingComma: true,
encodeSingleQuote: true,
encodeDoubleQuote: false,
roundBraces: {
arrow: true,
sequence: true,
assign: false,
},
},
visitors: {
AssignmentPattern(path, {print}) {
print('/* [hello world] */= ');
print('__right');
},
},
});
// returns
'const {a/* [hello world] */= 5} = b;\n';
Options related to visuals and not related to logic of output can be changed with help of format
,
you can override next options:
const overrides = {
format: {
indent: ' ',
newline: '\n',
space: ' ',
splitter: '\n',
endOfFile: '\n',
},
};
-
indent
- use two spaces, tabs, or anything you want; -
newline
- symbol used for line separation; -
space
- default symbol used for space character; -
splitter
- mandatory symbol that used inside of statements like this:
Default options produce:
if (a > 3)
console.log('ok');
else
console.log('not ok');
But you can override them with:
const overrides = {
format: {
indent: '',
newline: '',
space: '',
splitter: ' ',
},
};
And have minified code:
if(a>3)console.log('ok');else console.log('not ok');
Options used to configure logic of output, similar to ESLint rules:
- ✅
maxElementsInOneLine
- count ofArrayExpression
andArrayPattern
elements placed in one line. - ✅
maxVariablesInOneLine
- count ofVariableDeclarators
in one line. - ✅
maxPropertiesInOneLine
- count ofObjectProperties
in one line. - ✅
maxPropertiesLengthInOneLine
- maximum length ofObject Property
, when violated splits event ifmaxPropertiesInOneLine
satisfies; - ✅
roundBraces
to output braces or not-
arrow
: In a single argument arrow function expressions enabled:(a) => {}
, disabled:a => {}
; -
sequence
: In sequence expressions: enabled:for(let e of l) (a(), b())
, disabled:for(let e of l) a(), b()
; -
assign
: In assignment expressions: enabled:(e.o=w(e.o)
, disabled:e.o=w(e.o)
;
-
When you want to improve support of existing visitor or extend Printer with a new ones, you need next base operations:
When you need to override behavior of existing visitor use:
import {
print,
visitors as v,
} from '@putout/printer';
print(ast, {
visitors: {
CallExpression(path, printer, semantics) {
const {print} = printer;
if (!path.node.goldstein)
return v.CallExpression(path, printer, semantics);
print('__goldstein');
},
},
});
Used in previous example print
can be used for a couple purposes:
- to write
string
; - to write
node
whenobject
passed; - to write
node
whenstring
started with__
;
print(ast, {
visitors: {
AssignmentPattern(path, {print, maybe}) {
maybe.write.newline(path.parentPath.isCallExpression());
print('/* [hello world] */= ');
print('__right');
},
},
});
When you need some condition use maybe
. For example, to add newline only when parent node is CallExpression
you
can use maybe.write.newline(condition)
:
print(ast, {
visitors: {
AssignmentPattern(path, {write, maybe}) {
maybe.write.newline(path.parentPath.isCallExpression());
write(' /* [hello world] */= ');
write('__right');
},
},
});
When you going to output string you can use low-level function write
:
print(ast, {
visitors: {
BlockStatement(path, {write}) {
write('hello');
},
},
});
When you need to add indentation use indent
, for example when you output body,
you need to increment indentation, and then decrement it back:
print(ast, {
visitors: {
BlockStatement(path, {write, indent}) {
write('{');
indent.inc();
indent();
write('some;');
indent.dec();
write('{');
},
},
});
When you need to traverse node path, you can use traverse
:
print(ast, {
visitors: {
AssignmentExpression(path, {traverse}) {
traverse(path.get('left'));
},
},
});
This is the same as print('__left')
but more low-level, and supports only objects.
About speed, for file speed.js
:
const {readFileSync} = require('node:fs');
const putout = require('putout');
const parser = require('@babel/parser');
const code = readFileSync('./lib/tokenize/tokenize.js', 'utf8');
const ast = parser.parse(code);
speed('recast');
speed('putout');
function speed(printer) {
console.time(printer);
for (let i = 0; i < 1000; i++) {
putout(code, {
printer,
plugins: ['remove-unused-variables'],
});
}
console.timeEnd(printer);
}
With contents of tokenize.js
, we have:
MIT