feat(copy): added cacache.get.copy api for fast copies (#107)

Author: zkat

.travis.yml

@@ -1,6 +1,6 @@
 language: node_js
 sudo: false
 node_js:
-  - "7"
+  - "8"
   - "6"
   - "4"

appveyor.yml

@@ -1,6 +1,6 @@
 environment:
   matrix:
-    - nodejs_version: "7"
+    - nodejs_version: "8"
     - nodejs_version: "6"
     - nodejs_version: "4"
 

get.js

@@ -2,6 +2,7 @@
 
 const BB = require('bluebird')
 
+const fs = require('fs')
 const index = require('./lib/entry-index')
 const memo = require('./lib/memoization')
 const pipe = require('mississippi').pipe
@@ -152,3 +153,38 @@ function info (cache, key, opts) {
 }
 
 module.exports.hasContent = read.hasContent
+
+module.exports.copy = function cp (cache, key, dest, opts) {
+  return copy(false, cache, key, dest, opts)
+}
+module.exports.copy.byDigest = function cpDigest (cache, digest, dest, opts) {
+  return copy(true, cache, digest, dest, opts)
+}
+function copy (byDigest, cache, key, dest, opts) {
+  opts = opts || {}
+  if (read.copy) {
+    return (
+      byDigest ? BB.resolve(null) : index.find(cache, key, opts)
+    ).then(entry => {
+      if (!entry && !byDigest) {
+        throw new index.NotFoundError(cache, key)
+      }
+      return read.copy(
+        cache, byDigest ? key : entry.integrity, dest, opts
+      ).then(() => byDigest ? key : {
+        metadata: entry.metadata,
+        size: entry.size,
+        integrity: entry.integrity
+      })
+    })
+  } else {
+    return getData(byDigest, cache, key, opts).then(res => {
+      return fs.writeFileAsync(dest, byDigest ? res : res.data)
+      .then(() => byDigest ? key : {
+        metadata: res.metadata,
+        size: res.size,
+        integrity: res.integrity
+      })
+    })
+  }
+}

lib/content/read.js

@@ -52,6 +52,18 @@ function readStream (cache, integrity, opts) {
   return stream
 }
 
+if (fs.copyFile) {
+  module.exports.copy = copy
+}
+function copy (cache, integrity, dest, opts) {
+  opts = opts || {}
+  return pickContentSri(cache, integrity).then(content => {
+    const sri = content.sri
+    const cpath = contentPath(cache, sri)
+    return fs.copyFileAsync(cpath, dest).then(() => content.size)
+  })
+}
+
 module.exports.hasContent = hasContent
 function hasContent (cache, integrity) {
   if (!integrity) { return BB.resolve(false) }

locales/en.js

@@ -20,6 +20,8 @@ x.get = (cache, key, opts) => get(cache, key, opts)
 x.get.byDigest = (cache, hash, opts) => get.byDigest(cache, hash, opts)
 x.get.stream = (cache, key, opts) => get.stream(cache, key, opts)
 x.get.stream.byDigest = (cache, hash, opts) => get.stream.byDigest(cache, hash, opts)
+x.get.copy = (cache, key, dest, opts) => get.copy(cache, key, dest, opts)
+x.get.copy.byDigest = (cache, hash, dest, opts) => get.copy.byDigest(cache, hash, dest, opts)
 x.get.info = (cache, key) => get.info(cache, key)
 x.get.hasContent = (cache, hash) => get.hasContent(cache, hash)
 

locales/es.js

@@ -20,6 +20,8 @@ x.saca = (cache, clave, ops) => get(cache, clave, ops)
 x.saca.porHacheo = (cache, hacheo, ops) => get.byDigest(cache, hacheo, ops)
 x.saca.flujo = (cache, clave, ops) => get.stream(cache, clave, ops)
 x.saca.flujo.porHacheo = (cache, hacheo, ops) => get.stream.byDigest(cache, hacheo, ops)
+x.sava.copia = (cache, clave, destino, opts) => get.copy(cache, clave, destino, opts)
+x.sava.copia.porHacheo = (cache, hacheo, destino, opts) => get.copy.byDigest(cache, hacheo, destino, opts)
 x.saca.info = (cache, clave) => get.info(cache, clave)
 x.saca.tieneDatos = (cache, hacheo) => get.hasContent(cache, hacheo)
 

test/benchmarks/content.read.js

@@ -1,6 +1,10 @@
 'use strict'
 
+const BB = require('bluebird')
+
 const CacheContent = require('../util/cache-content')
+const fs = BB.promisifyAll(require('fs'))
+const path = require('path')
 const Tacks = require('tacks')
 const ssri = require('ssri')
 
@@ -58,6 +62,32 @@ module.exports = (suite, CACHE) => {
     }
   })
 
+  suite.add('content.read.copy()', {
+    defer: true,
+    setup () {
+      const fixture = new Tacks(CacheContent({
+        [BIGINTEGRITY]: BIGCONTENT
+      }))
+      fixture.create(CACHE)
+    },
+    fn (deferred) {
+      if (read.copy) {
+        read.copy(CACHE, BIGINTEGRITY, path.join(CACHE, 'bigdata'))
+        .then(
+          () => deferred.resolve(),
+          err => deferred.reject(err)
+        )
+      } else {
+        read(CACHE, BIGINTEGRITY)
+        .then(data => fs.writeFileAsync(path.join(CACHE, 'bigdata'), data))
+        .then(
+          () => deferred.resolve(),
+          err => deferred.reject(err)
+        )
+      }
+    }
+  })
+
   suite.add('content.read.stream() small data', {
     defer: true,
     setup () {

test/content.read.js

@@ -4,12 +4,15 @@ const Buffer = require('safe-buffer').Buffer
 const BB = require('bluebird')
 
 const finished = BB.promisify(require('mississippi').finished)
+const fs = require('fs')
 const path = require('path')
 const ssri = require('ssri')
 const Tacks = require('tacks')
 const test = require('tap').test
 const testDir = require('./util/test-dir')(__filename)
 
+BB.promisifyAll(fs)
+
 const CACHE = path.join(testDir, 'cache')
 const CacheContent = require('./util/cache-content')
 
@@ -147,3 +150,20 @@ test('hasContent: returns { sri, size } when a cache file exists', function (t)
     })
   )
 })
+
+test('copy: copies content to a destination path', {
+  skip: !fs.copyFile && 'Not supported on node versions without fs.copyFile'
+}, t => {
+  const CONTENT = Buffer.from('foobarbaz')
+  const INTEGRITY = ssri.fromData(CONTENT)
+  const DEST = path.join(CACHE, 'foobar-file')
+  const fixture = new Tacks(CacheContent({
+    [INTEGRITY]: CONTENT
+  }))
+  fixture.create(CACHE)
+  return read.copy(CACHE, INTEGRITY, DEST).then(() => {
+    return fs.readFileAsync(DEST)
+  }).then(data => {
+    t.deepEqual(data, CONTENT, 'file successfully copied')
+  })
+})

test/get.js

@@ -4,6 +4,7 @@ const Buffer = require('safe-buffer').Buffer
 const BB = require('bluebird')
 
 const finished = BB.promisify(require('mississippi').finished)
+const fs = require('fs')
 const index = require('../lib/entry-index')
 const memo = require('../lib/memoization')
 const path = require('path')
@@ -13,6 +14,8 @@ const test = require('tap').test
 const testDir = require('./util/test-dir')(__filename)
 const ssri = require('ssri')
 
+BB.promisifyAll(fs)
+
 const CacheContent = require('./util/cache-content')
 
 const CACHE = path.join(testDir, 'cache')
@@ -104,6 +107,34 @@ test('basic stream get', t => {
   })
 })
 
+test('get.copy', t => {
+  const DEST = path.join(CACHE, 'copymehere')
+  const fixture = new Tacks(CacheContent({
+    [INTEGRITY]: CONTENT
+  }))
+  fixture.create(CACHE)
+  return index.insert(CACHE, KEY, INTEGRITY, opts())
+  .then(() => get.copy(CACHE, KEY, DEST))
+  .then(res => {
+    t.deepEqual(res, {
+      metadata: METADATA,
+      integrity: INTEGRITY,
+      size: SIZE
+    }, 'copy operation returns basic metadata')
+    return fs.readFileAsync(DEST)
+  })
+  .then(data => {
+    t.deepEqual(data, CONTENT, 'data copied by key matches')
+    return rimraf(DEST)
+  })
+  .then(() => get.copy.byDigest(CACHE, INTEGRITY, DEST))
+  .then(() => fs.readFileAsync(DEST))
+  .then(data => {
+    t.deepEqual(data, CONTENT, 'data copied by digest matches')
+    return rimraf(DEST)
+  })
+})
+
 test('ENOENT if not found', t => {
   return get(CACHE, KEY).then(() => {
     throw new Error('lookup should fail')