readdirp

What is readdirp?

The readdirp npm package is a Node.js module that allows for reading the contents of directories recursively with a stream API. It provides a flexible and powerful way to filter, map, and reduce directory contents in an efficient manner. This package is particularly useful for tasks that involve file system operations, such as building file trees, searching for files with specific patterns, or processing files in batches.

What are readdirp's main functionalities?

Streaming directory contents

This feature allows you to stream the contents of a directory, filtering for specific file types (in this case, JavaScript files). It's useful for processing files as they are found.

const readdirp = require('readdirp');

readdirp('.', { fileFilter: '*.js' })
  .on('data', (entry) => {
    console.log(entry.path);
  })
  .on('end', () => console.log('Done'));

Promise API for directory reading

This feature provides a Promise-based API for reading directories, allowing for asynchronous file processing with better error handling and integration with async/await syntax.

const readdirp = require('readdirp');

readdirp('.', { depth: 1, fileFilter: '*.js' })
  .then(files => {
    files.forEach(file => console.log(file.path));
  })
  .catch(error => console.error('Error:', error));

Custom filter and entry formatting

This feature demonstrates how to use custom filters for file selection and how to format the entries returned by readdirp. It's useful for more complex file selection criteria and custom output formatting.

const readdirp = require('readdirp');

const options = {
  fileFilter: (entry) => entry.basename.startsWith('test'),
  entryType: 'files',
  alwaysStat: true,
  depth: 2
};

readdirp('.', options)
  .on('data', (entry) => {
    console.log(`${entry.path} - ${entry.stats.size} bytes`);
  });

Other packages similar to readdirp

glob

The 'glob' package provides pattern matching functionality to select files in directories. It's similar to readdirp in that it can be used to search for files, but it uses glob patterns instead of providing a stream API or Promise-based interface.

fs-extra

The 'fs-extra' package extends the built-in Node.js 'fs' module with additional file system methods, including recursive directory reading. While it offers similar functionality to readdirp, fs-extra provides a broader set of file system operations, making it more of a general-purpose library.

walk

The 'walk' package is another Node.js module for recursively reading directory contents. It is similar to readdirp but focuses more on event-based directory walking. Compared to readdirp, 'walk' might offer a simpler API for some use cases but lacks the advanced filtering and mapping capabilities.

README

readdirp

Recursive version of fs.readdir. Exposes a stream API and a promise API.

npm install readdirp

const readdirp = require('readdirp');

// Use streams to achieve small RAM & CPU footprint.
// 1) Streams example with for-await.
for await (const entry of readdirp('.')) {
  const {path} = entry;
  console.log(`${JSON.stringify({path})}`);
}

// 2) Streams example, non for-await.
// Print out all JS files along with their size within the current folder & subfolders.
readdirp('.', {fileFilter: '*.js', alwaysStat: true})
  .on('data', (entry) => {
    const {path, stats: {size}} = entry;
    console.log(`${JSON.stringify({path, size})}`);
  })
  // Optionally call stream.destroy() in `warn()` in order to abort and cause 'close' to be emitted
  .on('warn', error => console.error('non-fatal error', error))
  .on('error', error => console.error('fatal error', error))
  .on('end', () => console.log('done'));

// 3) Promise example. More RAM and CPU than streams / for-await.
const files = await readdirp.promise('.');
console.log(files.map(file => file.path));

// Other options.
readdirp('test', {
  fileFilter: '*.js',
  directoryFilter: ['!.git', '!*modules']
  // directoryFilter: (di) => di.basename.length === 9
  type: 'files_directories',
  depth: 1
});

For more examples, check out examples directory.

API

const stream = readdirp(root[, options]) — Stream API

• Reads given root recursively and returns a stream of entry infos
• Optionally can be used like for await (const entry of stream) with node.js 10+ (asyncIterator).
• on('data', (entry) => {}) entry info for every file / dir.
• on('warn', (error) => {}) non-fatal Error that prevents a file / dir from being processed. Example: inaccessible to the user.
• on('error', (error) => {}) fatal Error which also ends the stream. Example: illegal options where passed.
• on('end') — we are done. Called when all entries were found and no more will be emitted.
• on('close') — stream is destroyed via stream.destroy(). Could be useful if you want to manually abort even on a non fatal error. At that point the stream is no longer readable and no more entries, warning or errors are emitted
• To learn more about streams, consult the very detailed nodejs streams documentation or the stream-handbook

const entries = await readdirp.promise(root[, options]) — Promise API. Returns a list of entry infos.

First argument is awalys root, path in which to start reading and recursing into subdirectories.

options

• fileFilter: ["*.js"]: filter to include or exclude files. A Function, Glob string or Array of glob strings.
  • Function: a function that takes an entry info as a parameter and returns true to include or false to exclude the entry
  • Glob string: a string (e.g., *.js) which is matched using picomatch, so go there for more information. Globstars (**) are not supported since specifying a recursive pattern for an already recursive function doesn't make sense. Negated globs (as explained in the minimatch documentation) are allowed, e.g., !*.txt matches everything but text files.
  • Array of glob strings: either need to be all inclusive or all exclusive (negated) patterns otherwise an error is thrown. ['*.json', '*.js'] includes all JavaScript and Json files. ['!.git', '!node_modules'] includes all directories except the '.git' and 'node_modules'.
  • Directories that do not pass a filter will not be recursed into.
• directoryFilter: ['!.git']: filter to include/exclude directories found and to recurse into. Directories that do not pass a filter will not be recursed into.
• depth: 5: depth at which to stop recursing even if more subdirectories are found
• type: 'files': determines if data events on the stream should be emitted for 'files' (default), 'directories', 'files_directories', or 'all'. Setting to 'all' will also include entries for other types of file descriptors like character devices, unix sockets and named pipes.
• alwaysStat: false: always return stats property for every file. Default is false, readdirp will return Dirent entries. Setting it to true can double readdir execution time - use it only when you need file size, mtime etc. Cannot be enabled on node <10.10.0.
• lstat: false: include symlink entries in the stream along with files. When true, fs.lstat would be used instead of fs.stat

EntryInfo

Has the following properties:

• path: 'assets/javascripts/react.js': path to the file/directory (relative to given root)
• fullPath: '/Users/dev/projects/app/assets/javascripts/react.js': full path to the file/directory found
• basename: 'react.js': name of the file/directory
• dirent: fs.Dirent: built-in dir entry object - only with alwaysStat: false
• stats: fs.Stats: built in stat object - only with alwaysStat: true

Changelog

• 3.5 (Oct 13, 2020) disallows recursive directory-based symlinks. Before, it could have entered infinite loop.
• 3.4 (Mar 19, 2020) adds support for directory-based symlinks.
• 3.3 (Dec 6, 2019) stabilizes RAM consumption and enables perf management with highWaterMark option. Fixes race conditions related to for-await looping.
• 3.2 (Oct 14, 2019) improves performance by 250% and makes streams implementation more idiomatic.
• 3.1 (Jul 7, 2019) brings bigint support to stat output on Windows. This is backwards-incompatible for some cases. Be careful. It you use it incorrectly, you'll see "TypeError: Cannot mix BigInt and other types, use explicit conversions".
• 3.0 brings huge performance improvements and stream backpressure support.
• Upgrading 2.x to 3.x:
  • Signature changed from readdirp(options) to readdirp(root, options)
  • Replaced callback API with promise API.
  • Renamed entryType option to type
  • Renamed entryType: 'both' to 'files_directories'
  • EntryInfo
    • Renamed stat to stats
      • Emitted only when alwaysStat: true
      • dirent is emitted instead of stats by default with alwaysStat: false
    • Renamed name to basename
    • Removed parentDir and fullParentDir properties
• Supported node.js versions:
  • 3.x: node 8+
  • 2.x: node 0.6+

License

Copyright (c) 2012-2019 Thorsten Lorenz, Paul Miller (https://paulmillr.com)

MIT License, see LICENSE file.