@applaud/cacache

@applaud/cacache is a Node.js library for managing local key and content address caches. It's really fast, really good at concurrency, and it will never give you corrupted data, even if cache files get corrupted or manipulated.

It was originally written to be used as npm's local cache, but can just as easily be used on its own.

Install

$ npm install --save @applaud/cacache

Table of Contents

• Example
• Features
• Contributing
• API
  • Using localized APIs
  • Reading
    • ls
    • ls.stream
    • get
    • get.stream
    • get.info
    • get.hasContent
  • Writing
    • put
    • put.stream
    • put* opts
    • rm.all
    • rm.entry
    • rm.content
  • Utilities
    • setLocale
    • clearMemoized
    • tmp.mkdir
    • tmp.withTmp
  • Integrity
    • Subresource Integrity
    • verify
    • verify.lastRun

Example

1const cacache = require('@applaud/cacache/en')
2const fs = require('fs')
3
4const tarball = '/path/to/mytar.tgz'
5const cachePath = '/tmp/my-toy-cache'
6const key = 'my-unique-key-1234'
7
8// Cache it! Use `cachePath` as the root of the content cache
9cacache.put(cachePath, key, '10293801983029384').then(integrity => {
10  console.log(`Saved content to ${cachePath}.`)
11})
12
13const destination = '/tmp/mytar.tgz'
14
15// Copy the contents out of the cache and into their destination!
16// But this time, use stream instead!
17cacache.get.stream(
18  cachePath, key
19).pipe(
20  fs.createWriteStream(destination)
21).on('finish', () => {
22  console.log('done extracting!')
23})
24
25// The same thing, but skip the key index.
26cacache.get.byDigest(cachePath, integrityHash).then(data => {
27  fs.writeFile(destination, data, err => {
28    console.log('tarball data fetched based on its sha512sum and written out!')
29  })
30})

Features

• Extraction by key or by content address (shasum, etc)
• Subresource Integrity web standard support
• Multi-hash support - safely host sha1, sha512, etc, in a single cache
• Automatic content deduplication
• Fault tolerance (immune to corruption, partial writes, process races, etc)
• Consistency guarantees on read and write (full data verification)
• Lockless, high-concurrency cache access
• Streaming support
• Promise support
• Pretty darn fast -- sub-millisecond reads and writes including verification
• Arbitrary metadata storage
• Garbage collection and additional offline verification
• Thorough test coverage
• There's probably a bloom filter in there somewhere. Those are cool, right? 🤔

Contributing

The cacache team enthusiastically welcomes contributions and project participation! There's a bunch of things you can do if you want to contribute! The Contributor Guide has all the information you need for everything from reporting bugs to contributing entire new features. Please don't hesitate to jump in if you'd like to, or even ask us questions if something isn't clear.

All participants and maintainers in this project are expected to follow Code of Conduct, and just generally be excellent to each other.

Please refer to the Changelog for project history details, too.

Happy hacking!

API

Using localized APIs

cacache includes a complete API in English, with the same features as other translations. To use the English API as documented in this README, use require('@applaud/cacache/en'). This is also currently the default if you do require('@applaud/cacache'), but may change in the future.

cacache also supports other languages! You can find the list of currently supported ones by looking in ./locales in the source directory. You can use the API in that language with require('@applaud/cacache/<lang>').

Want to add support for a new language? Please go ahead! You should be able to copy ./locales/en.js and ./locales/en.json and fill them in. Translating the README.md is a bit more work, but also appreciated if you get around to it. 👍🏼

> cacache.ls(cache) -> Promise<Object>

Lists info for all entries currently in the cache as a single large object. Each entry in the object will be keyed by the unique index key, with corresponding get.info objects as the values.

Example

1cacache.ls(cachePath).then(console.log)
2// Output
3{
4  'my-thing': {
5    key: 'my-thing',
6    integrity: 'sha512-BaSe64/EnCoDED+HAsh=='
7    path: '.testcache/content/deadbeef', // joined with `cachePath`
8    time: 12345698490,
9    size: 4023948,
10    metadata: {
11      name: 'blah',
12      version: '1.2.3',
13      description: 'this was once a package but now it is my-thing'
14    }
15  },
16  'other-thing': {
17    key: 'other-thing',
18    integrity: 'sha1-ANothER+hasH=',
19    path: '.testcache/content/bada55',
20    time: 11992309289,
21    size: 111112
22  }
23}

> cacache.ls.stream(cache) -> Readable

Lists info for all entries currently in the cache as a single large object.

This works just like ls, except get.info entries are returned as 'data' events on the returned stream.

Example

1cacache.ls.stream(cachePath).on('data', console.log)
2// Output
3{
4  key: 'my-thing',
5  integrity: 'sha512-BaSe64HaSh',
6  path: '.testcache/content/deadbeef', // joined with `cachePath`
7  time: 12345698490,
8  size: 13423,
9  metadata: {
10    name: 'blah',
11    version: '1.2.3',
12    description: 'this was once a package but now it is my-thing'
13  }
14}
15
16{
17  key: 'other-thing',
18  integrity: 'whirlpool-WoWSoMuchSupport',
19  path: '.testcache/content/bada55',
20  time: 11992309289,
21  size: 498023984029
22}
23
24{
25  ...
26}

> cacache.get(cache, key, [opts]) -> Promise({data, metadata, integrity})

Returns an object with the cached data, digest, and metadata identified by key. The data property of this object will be a Buffer instance that presumably holds some data that means something to you. I'm sure you know what to do with it! cacache just won't care.

integrity is a Subresource Integrity string. That is, a string that can be used to verify data, which looks like <hash-algorithm>-<base64-integrity-hash>.

If there is no content identified by key, or if the locally-stored data does not pass the validity checksum, the promise will be rejected.

A sub-function, get.byDigest may be used for identical behavior, except lookup will happen by integrity hash, bypassing the index entirely. This version of the function only returns data itself, without any wrapper.

Note

This function loads the entire cache entry into memory before returning it. If you're dealing with Very Large data, consider using get.stream instead.

Example

1// Look up by key
2cache.get(cachePath, 'my-thing').then(console.log)
3// Output:
4{
5  metadata: {
6    thingName: 'my'
7  },
8  integrity: 'sha512-BaSe64HaSh',
9  data: Buffer#<deadbeef>,
10  size: 9320
11}
12
13// Look up by digest
14cache.get.byDigest(cachePath, 'sha512-BaSe64HaSh').then(console.log)
15// Output:
16Buffer#<deadbeef>

> cacache.get.stream(cache, key, [opts]) -> Readable

Returns a Readable Stream of the cached data identified by key.

If there is no content identified by key, or if the locally-stored data does not pass the validity checksum, an error will be emitted.

metadata and integrity events will be emitted before the stream closes, if you need to collect that extra data about the cached entry.

A sub-function, get.stream.byDigest may be used for identical behavior, except lookup will happen by integrity hash, bypassing the index entirely. This version does not emit the metadata and integrity events at all.

Example

1// Look up by key
2cache.get.stream(
3  cachePath, 'my-thing'
4).on('metadata', metadata => {
5  console.log('metadata:', metadata)
6}).on('integrity', integrity => {
7  console.log('integrity:', integrity)
8}).pipe(
9  fs.createWriteStream('./x.tgz')
10)
11// Outputs:
12metadata: { ... }
13integrity: 'sha512-SoMeDIGest+64=='
14
15// Look up by digest
16cache.get.stream.byDigest(
17  cachePath, 'sha512-SoMeDIGest+64=='
18).pipe(
19  fs.createWriteStream('./x.tgz')
20)

> cacache.get.info(cache, key) -> Promise

Looks up key in the cache index, returning information about the entry if one exists.

Fields

• key - Key the entry was looked up under. Matches the key argument.
• integrity - Subresource Integrity hash for the content this entry refers to.
• path - Filesystem path relative to cache argument where content is stored.
• time - Timestamp the entry was first added on.
• metadata - User-assigned metadata associated with the entry/content.

Example

1cacache.get.info(cachePath, 'my-thing').then(console.log)
2
3// Output
4{
5  key: 'my-thing',
6  integrity: 'sha256-MUSTVERIFY+ALL/THINGS=='
7  path: '.testcache/content/deadbeef',
8  time: 12345698490,
9  size: 849234,
10  metadata: {
11    name: 'blah',
12    version: '1.2.3',
13    description: 'this was once a package but now it is my-thing'
14  }
15}

> cacache.get.hasContent(cache, integrity) -> Promise

Looks up a Subresource Integrity hash in the cache. If content exists for this integrity, it will return an object, with the specific single integrity hash that was found in sri key, and the size of the found content as size. If no content exists for this integrity, it will return false.

Example

1cacache.get.hasContent(cachePath, 'sha256-MUSTVERIFY+ALL/THINGS==').then(console.log)
2
3// Output
4{
5  sri: {
6    source: 'sha256-MUSTVERIFY+ALL/THINGS==',
7    algorithm: 'sha256',
8    digest: 'MUSTVERIFY+ALL/THINGS==',
9    options: []
10  },
11  size: 9001
12}
13
14cacache.get.hasContent(cachePath, 'sha521-NOT+IN/CACHE==').then(console.log)
15
16// Output
17false

> cacache.put(cache, key, data, [opts]) -> Promise

Inserts data passed to it into the cache. The returned Promise resolves with a digest (generated according to opts.algorithms) after the cache entry has been successfully written.

Example

1fetch(
2  'https://registry.npmjs.org/@applaud/cacache/-/@applaud/cacache-1.0.0.tgz'
3).then(data => {
4  return cacache.put(cachePath, 'registry.npmjs.org|@applaud/cacache@1.0.0', data)
5}).then(integrity => {
6  console.log('integrity hash is', integrity)
7})

> cacache.put.stream(cache, key, [opts]) -> Writable

Returns a Writable Stream that inserts data written to it into the cache. Emits an integrity event with the digest of written contents when it succeeds.

Example

1request.get(
2  'https://registry.npmjs.org/@applaud/cacache/-/@applaud/cacache-1.0.0.tgz'
3).pipe(
4  cacache.put.stream(
5    cachePath, 'registry.npmjs.org|@applaud/cacache@1.0.0'
6  ).on('integrity', d => console.log(`integrity digest is ${d}`))
7)

> cacache.put options

cacache.put functions have a number of options in common.

opts.metadata

Arbitrary metadata to be attached to the inserted key.

opts.size

If provided, the data stream will be verified to check that enough data was passed through. If there's more or less data than expected, insertion will fail with an EBADSIZE error.

opts.integrity

If present, the pre-calculated digest for the inserted content. If this option if provided and does not match the post-insertion digest, insertion will fail with an EINTEGRITY error.

algorithms has no effect if this option is present.

opts.algorithms

Default: ['sha512']

Hashing algorithms to use when calculating the subresource integrity digest for inserted data. Can use any algorithm listed in crypto.getHashes() or 'omakase'/'お任せします' to pick a random hash algorithm on each insertion. You may also use any anagram of 'modnar' to use this feature.

Currently only supports one algorithm at a time (i.e., an array length of exactly 1). Has no effect if opts.integrity is present.

opts.uid/opts.gid

If provided, cacache will do its best to make sure any new files added to the cache use this particular uid/gid combination. This can be used, for example, to drop permissions when someone uses sudo, but cacache makes no assumptions about your needs here.

opts.memoize

Default: null

If provided, cacache will memoize the given cache insertion in memory, bypassing any filesystem checks for that key or digest in future cache fetches. Nothing will be written to the in-memory cache unless this option is explicitly truthy.

If opts.memoize is an object or a Map-like (that is, an object with get and set methods), it will be written to instead of the global memoization cache.

Reading from disk data can be forced by explicitly passing memoize: false to the reader functions, but their default will be to read from memory.

> cacache.rm.all(cache) -> Promise

Clears the entire cache. Mainly by blowing away the cache directory itself.

Example

1cacache.rm.all(cachePath).then(() => {
2  console.log('THE APOCALYPSE IS UPON US 😱')
3})

> cacache.rm.entry(cache, key) -> Promise

Alias: cacache.rm

Removes the index entry for key. Content will still be accessible if requested directly by content address (get.stream.byDigest).

To remove the content itself (which might still be used by other entries), use rm.content. Or, to safely vacuum any unused content, use verify.

Example

1cacache.rm.entry(cachePath, 'my-thing').then(() => {
2  console.log('I did not like it anyway')
3})

> cacache.rm.content(cache, integrity) -> Promise

Removes the content identified by integrity. Any index entries referring to it will not be usable again until the content is re-added to the cache with an identical digest.

Example

1cacache.rm.content(cachePath, 'sha512-SoMeDIGest/IN+BaSE64==').then(() => {
2  console.log('data for my-thing is gone!')
3})

> cacache.setLocale(locale)

Configure the language/locale used for messages and errors coming from cacache. The list of available locales is in the ./locales directory in the project root.

Interested in contributing more languages! Submit a PR!

> cacache.clearMemoized()

Completely resets the in-memory entry cache.

> tmp.mkdir(cache, opts) -> Promise<Path>

Returns a unique temporary directory inside the cache's tmp dir. This directory will use the same safe user assignment that all the other stuff use.

Once the directory is made, it's the user's responsibility that all files within are made according to the same opts.gid/opts.uid settings that would be passed in. If not, you can ask cacache to do it for you by calling tmp.fix(), which will fix all tmp directory permissions.

If you want automatic cleanup of this directory, use tmp.withTmp()

Example

1cacache.tmp.mkdir(cache).then(dir => {
2  fs.writeFile(path.join(dir, 'blablabla'), Buffer#<1234>, ...)
3})

> tmp.withTmp(cache, opts, cb) -> Promise

Creates a temporary directory with tmp.mkdir() and calls cb with it. The created temporary directory will be removed when the return value of cb() resolves -- that is, if you return a Promise from cb(), the tmp directory will be automatically deleted once that promise completes.

The same caveats apply when it comes to managing permissions for the tmp dir's contents.

Example

1cacache.tmp.withTmp(cache, dir => {
2  return fs.writeFileAsync(path.join(dir, 'blablabla'), Buffer#<1234>, ...)
3}).then(() => {
4  // `dir` no longer exists
5})

Subresource Integrity Digests

For content verification and addressing, cacache uses strings following the Subresource Integrity spec. That is, any time cacache expects an integrity argument or option, it should be in the format <hashAlgorithm>-<base64-hash>.

One deviation from the current spec is that cacache will support any hash algorithms supported by the underlying Node.js process. You can use crypto.getHashes() to see which ones you can use.

Generating Digests Yourself

If you have an existing content shasum, they are generally formatted as a hexadecimal string (that is, a sha1 would look like: 5f5513f8822fdbe5145af33b64d8d970dcf95c6e). In order to be compatible with cacache, you'll need to convert this to an equivalent subresource integrity string. For this example, the corresponding hash would be: sha1-X1UT+IIv2+UUWvM7ZNjZcNz5XG4=.

If you want to generate an integrity string yourself for existing data, you can use something like this:

1const crypto = require('crypto')
2const hashAlgorithm = 'sha512'
3const data = 'foobarbaz'
4
5const integrity = (
6  hashAlgorithm +
7  '-' +
8  crypto.createHash(hashAlgorithm).update(data).digest('base64')
9)

You can also use ssri to have a richer set of functionality around SRI strings, including generation, parsing, and translating from existing hex-formatted strings.

> cacache.verify(cache, opts) -> Promise

Checks out and fixes up your cache:

• Cleans up corrupted or invalid index entries.
• Custom entry filtering options.
• Garbage collects any content entries not referenced by the index.
• Checks integrity for all content entries and removes invalid content.
• Fixes cache ownership.
• Removes the tmp directory in the cache and all its contents.

When it's done, it'll return an object with various stats about the verification process, including amount of storage reclaimed, number of valid entries, number of entries removed, etc.

Options

• opts.uid - uid to assign to cache and its contents
• opts.gid - gid to assign to cache and its contents
• opts.filter - receives a formatted entry. Return false to remove it. Note: might be called more than once on the same entry.

Example

1echo somegarbage >> $CACHEPATH/content/deadbeef

1cacache.verify(cachePath).then(stats => {
2  // deadbeef collected, because of invalid checksum.
3  console.log('cache is much nicer now! stats:', stats)
4})

> cacache.verify.lastRun(cache) -> Promise

Returns a Date representing the last time cacache.verify was run on cache.

Example

1cacache.verify(cachePath).then(() => {
2  cacache.verify.lastRun(cachePath).then(lastTime => {
3    console.log('cacache.verify was last called on' + lastTime)
4  })
5})