feat: Add support for specifying hash algorithms in files.add

Author: vasco-santos

src/cli/commands/files/add.js

@@ -10,6 +10,7 @@ const zip = require('pull-zip')
 const getFolderSize = require('get-folder-size')
 const byteman = require('byteman')
 const waterfall = require('async/waterfall')
+const mh = require('multihashes')
 const utils = require('../../utils')
 const print = require('../../utils').print
 const createProgressBar = require('../../utils').createProgressBar
@@ -162,6 +163,11 @@ module.exports = {
       type: 'integer',
       describe: 'Cid version. Non-zero value will change default of \'raw-leaves\' to true. (experimental)'
     },
+    hash: {
+      type: 'string',
+      choices: Object.keys(mh.names),
+      describe: 'Hash function to use. Will set Cid version to 1 if used. (experimental)'
+    },
     quiet: {
       alias: 'q',
       type: 'boolean',
@@ -191,7 +197,8 @@ module.exports = {
         : Infinity,
       cidVersion: argv.cidVersion,
       rawLeaves: argv.rawLeaves,
-      onlyHash: argv.onlyHash
+      onlyHash: argv.onlyHash,
+      hashAlg: argv.hash
     }
 
     // Temporary restriction on raw-leaves:
@@ -206,6 +213,18 @@ module.exports = {
       throw new Error('Implied argument raw-leaves must be passed and set to false when cid-version is > 0')
     }
 
+    // Temporary restriction on raw-leaves:
+    // When hash != undefined then raw-leaves MUST be present and false.
+    //
+    // This is because raw-leaves is not yet implemented in js-ipfs,
+    // and go-ipfs changes the value of raw-leaves to true when
+    // hash != undefined unless explicitly set to false.
+    //
+    // This retains feature parity without having to implement raw-leaves.
+    if (options.hash && options.rawLeaves !== false) {
+      throw new Error('Implied argument raw-leaves must be passed and set to false when hash argument is specified')
+    }
+
     if (options.rawLeaves) {
       throw new Error('Not implemented: raw-leaves')
     }

src/core/components/files.js

@@ -22,15 +22,15 @@ function noop () {}
 function prepareFile (self, opts, file, callback) {
   opts = opts || {}
 
-  waterfall([
-    (cb) => opts.onlyHash ? cb(null, file) : self.object.get(file.multihash, cb),
-    (node, cb) => {
-      let cid = new CID(node.multihash)
+  let cid = new CID(file.multihash)
 
-      if (opts['cid-version'] === 1) {
-        cid = cid.toV1()
-      }
+  if (opts.cidVersion === 1) {
+    cid = cid.toV1()
+  }
 
+  waterfall([
+    (cb) => opts.onlyHash ? cb(null, file) : self.object.get(file.multihash, opts, cb),
+    (node, cb) => {
       const b58Hash = cid.toBaseEncodedString()
 
       cb(null, {
@@ -110,6 +110,10 @@ module.exports = function files (self) {
         : Infinity
     }, options)
 
+    if (opts.hashAlg && opts.cidVersion !== 1) {
+      opts.cidVersion = 1
+    }
+
     let total = 0
     let prog = opts.progress || (() => {})
     const progress = (bytes) => {
@@ -182,7 +186,7 @@ module.exports = function files (self) {
   }
 
   return {
-    add: promisify((data, options, callback) => {
+    add: promisify((data, options = {}, callback) => {
       if (typeof options === 'function') {
         callback = options
         options = {}
@@ -200,6 +204,11 @@ module.exports = function files (self) {
         return callback(new Error('first arg must be a buffer, readable stream, an object or array of objects'))
       }
 
+      // CID v0 is for multihashes encoded with sha2-256
+      if (options.hashAlg && options.cidVersion !== 1) {
+        options.cidVersion = 1
+      }
+
       pull(
         pull.values([data]),
         _addPullStream(options),

src/core/components/object.js

@@ -191,7 +191,11 @@ module.exports = function object (self) {
       } catch (err) {
         return callback(err)
       }
-      const cid = new CID(mh)
+      let cid = new CID(mh)
+
+      if (options.cidVersion === 1) {
+        cid = cid.toV1()
+      }
 
       self._ipld.get(cid, (err, result) => {
         if (err) {
@@ -214,6 +218,7 @@ module.exports = function object (self) {
         if (err) {
           return callback(err)
         }
+
         callback(null, node.data)
       })
     }),

src/http/api/resources/files.js

@@ -207,7 +207,8 @@ exports.add = {
       cidVersion: request.query['cid-version'],
       rawLeaves: request.query['raw-leaves'],
       progress: request.query.progress ? progressHandler : null,
-      onlyHash: request.query['only-hash']
+      onlyHash: request.query['only-hash'],
+      hashAlg: request.query['hash']
     }
 
     const aborter = abortable()

test/cli/files.js

@@ -7,8 +7,22 @@ const expect = require('chai').expect
 const path = require('path')
 const compareDir = require('dir-compare').compareSync
 const rimraf = require('rimraf').sync
+const CID = require('cids')
+const mh = require('multihashes')
 const runOnAndOff = require('../utils/on-and-off')
 
+// TODO: Test against all algorithms Object.keys(mh.names)
+// This subset is known to work with both go-ipfs and js-ipfs as of 2017-09-05
+const HASH_ALGS = [
+  'sha1',
+  'sha2-256',
+  'sha2-512',
+  'keccak-224',
+  'keccak-256',
+  'keccak-384',
+  'keccak-512'
+]
+
 describe('files', () => runOnAndOff((thing) => {
   let ipfs
   const readme = fs.readFileSync(path.join(process.cwd(), '/src/init-files/init-docs/readme'))
@@ -300,6 +314,19 @@ describe('files', () => runOnAndOff((thing) => {
       })
   })
 
+  HASH_ALGS.forEach((name) => {
+    it(`add with hash=${name} and raw-leaves=false`, function () {
+      this.timeout(30 * 1000)
+
+      return ipfs(`add src/init-files/init-docs/readme --hash=${name} --raw-leaves=false`)
+        .then((out) => {
+          const hash = out.split(' ')[1]
+          const cid = new CID(hash)
+          expect(mh.decode(cid.multihash).name).to.equal(name)
+        })
+    })
+  })
+
   it('cat', function () {
     this.timeout(30 * 1000)