find-entry-points

find-entry-points

Find the entry points in a set of JavaScript files.

Install

$ npm i find-entry-points

Huh?

Example 1

Given a sync or async iterable of paths to the JavaScript files with the following dependencies between them (a directed arrow from a.js to b.js means a.js imports b.js):

The findEntryPoints function would return [['a.js']] because a.js is the entry point of the above JavaScript library or application (it is not imported by any JavaScript file). The findSingleEntryPoints function would return ['a.js'].

Example 2

Given a sync or async iterable of paths to the JavaScript files with the following dependencies between them:

The findEntryPoints function would return [['a.js'], ['f.js'], ['g.js']] (order not guaranteed) because a.js, f.js, and g.js are all not imported by any JavaScript file. Notice that it is possible to have a disconnected dependency graph. The findSingleEntryPoints function would return ['a.js', 'f.js', 'g.js'] (order not guaranteed).

Example 3

Given a sync or async iterable of paths to the JavaScript files with the following dependencies between them:

The findEntryPoints function would return [['a.js'], ['g.js']] (order not guaranteed). Notice that cycles are possible because imports are cached. The findSingleEntryPoints function would return ['a.js', 'g.js'] (order not guaranteed).

Example 4

Given a sync or async iterable of paths to the JavaScript files with the following dependencies between them:

The findEntryPoints function would return [['a.js'], ['g.js', 'h.js', 'i.js']] (order not guaranteed). Notice that this is the first example where an inner array contains more than one element. This is because g.js, h.js, and i.js are all valid entry points for their graph component. The findSingleEntryPoints function would return ['a.js'] (order not guaranteed).

Usage

Suppose the JavaScript files with the following dependencies between them are all in a src directory and that h.js imports i.js via a dynamic import:

import { findEntryPoints, findSingleEntryPoints } from 'find-entry-points'
import globby from 'globby'
import * as swc from '@swc/core'

// `globby.stream` returns an async iterable of file paths
console.log(await findEntryPoints(globby.stream('src/**/*.js')))
//=> [['src/a.js'], ['src/g.js', 'src/h.js', 'src/i.js']]

// `findSingleEntryPoints` ignores cyclic entry points
console.log(await findSingleEntryPoints(globby.stream('src/**/*.js')))
// => ['src/a.js']

console.log(
  await findEntryPoints(globby.stream('src/**/*.js'), {
    followDynamicImports: false
  })
)
//=> [['src/a.js'], ['src/i.js']]

console.log(
  await findSingleEntryPoints(globby.stream('src/**/*.js'), {
    followDynamicImports: false
  })
)
// => ['src/a.js', 'src/i.js']

// Use the `transform` option to transform non-standard syntax
// like JSX to standard ECMAScript so that imports can be parsed
console.log(
  await findEntryPoints(globby.stream('src/**/*.js'), {
    transform: async ({ path, code }) =>
      (
        await transform(code, {
          filename: path,
          jsc: { parser: { jsx: true } }
        })
      ).code
  })
)
//=> [['src/a.js'], ['src/g.js', 'src/h.js', 'src/i.js']]

console.log(
  await findSingleEntryPoints(globby.stream('src/**/*.js'), {
    transform: async ({ path, code }) =>
      (
        await transform(code, {
          filename: path,
          jsc: { parser: { jsx: true } }
        })
      ).code
  })
)
//=> ['src/a.js']

// Use the `parseImports` option to parse imports yourself
console.log(
  await findEntryPoints(globby.stream('src/**/*.js'), {
    parseImports: async ({ followDynamicImports, file }) => {
      const { path, read } = file

      // Read the file if you want
      const code = await read()
      const importedFilenames = yourFancyImportParser(code)

      return importedFilenames
    }
  })
)

See the commented type definitions for clarification.

How?

The package uses parse-imports (another package of mine) to construct a dependency graph, which is a directed graph, from the given set of JavaScript files. Then the package finds the strongly connected components of the dependency graph using Tarjan's strongly connected components algorithm, and constructs a directed acyclic graph from the strongly connected components. Finally, the package returns the strongly connected components corresponding to the vertices with in-degree 1 in the new directed acyclic graph.

Contributing

Stars are always welcome!

For bugs and feature requests, please create an issue.

For pull requests, please read the contributing guidelines.

License

Apache 2.0

This is not an official Google product.