@openneuro/server 4.7.0 → 4.7.2-alpha.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@openneuro/server",
-  "version": "4.7.0",
+  "version": "4.7.2-alpha.0",
   "description": "Core service for the OpenNeuro platform.",
   "license": "MIT",
   "main": "src/server.js",
@@ -18,7 +18,7 @@
   "dependencies": {
     "@apollo/client": "3.4.17",
     "@elastic/elasticsearch": "7.15.0",
-    "@openneuro/search": "^4.7.0",
+    "@openneuro/search": "^4.7.2-alpha.0",
     "@passport-next/passport-google-oauth2": "^1.0.0",
     "@sentry/node": "^4.5.3",
     "apollo-server": "2.25.3",
@@ -105,5 +105,5 @@
   "publishConfig": {
     "access": "public"
   },
-  "gitHead": "74336af513bf3128af43ad3c55e2127b3a85915b"
+  "gitHead": "1530c13f60e66a8d68129aecdaf41806db635fcf"
 }

package/src/config.js

@@ -64,6 +64,9 @@ const config = {
   elasticsearch: {
     connection: process.env.ELASTICSEARCH_CONNECTION,
   },
+  github: {
+    token: process.env.DATALAD_GITHUB_TOKEN,
+  },
 }
 
 export default config

package/src/graphql/resolvers/__tests__/derivatives.spec.ts

@@ -0,0 +1,33 @@
+import { githubDerivativeQuery, derivativeObject } from '../derivatives'
+
+describe('GraphQL derivatives', () => {
+  describe('githubDerivativeQuery()', () => {
+    it('constructs a correct URL', () => {
+      expect(githubDerivativeQuery('ds000102', 'mriqc').toString()).toEqual(
+        'https://api.github.com/repos/OpenNeuroDerivatives/ds000102-mriqc',
+      )
+    })
+  })
+  describe('derivativeObject()', () => {
+    it('returns expected values for mriqc', () => {
+      expect(derivativeObject('ds000102', 'mriqc')).toEqual({
+        dataladUrl: new URL(
+          'https://github.com/OpenNeuroDerivatives/ds000102-mriqc.git',
+        ),
+        local: false,
+        name: 'ds000102-mriqc',
+        s3Url: new URL('s3://openneuro-derivatives/mriqc/ds000102-mriqc'),
+      })
+    })
+    it('returns expected values for fmriprep', () => {
+      expect(derivativeObject('ds000102', 'fmriprep')).toEqual({
+        dataladUrl: new URL(
+          'https://github.com/OpenNeuroDerivatives/ds000102-fmriprep.git',
+        ),
+        local: false,
+        name: 'ds000102-fmriprep',
+        s3Url: new URL('s3://openneuro-derivatives/fmriprep/ds000102-fmriprep'),
+      })
+    })
+  })
+})

package/src/graphql/resolvers/dataset.js

@@ -21,6 +21,7 @@ import { getDatasetWorker } from '../../libs/datalad-service.js'
 import { getDraftHead } from '../../datalad/dataset.js'
 import { getFileName } from '../../datalad/files.js'
 import { onBrainlife } from './brainlife'
+import { derivatives } from './derivatives'
 import semver from 'semver'
 
 export const dataset = async (obj, { id }, { user, userInfo }) => {
@@ -304,6 +305,7 @@ const Dataset = {
   following,
   starred,
   onBrainlife,
+  derivatives,
   metadata,
   history,
   worker,

package/src/graphql/resolvers/derivatives.ts

@@ -0,0 +1,97 @@
+import fetch from 'node-fetch'
+import {
+  DatasetOrSnapshot,
+  getDatasetFromSnapshotId,
+} from '../../utils/datasetOrSnapshot'
+import config from '../../config'
+
+const S3_BUCKET = 'openneuro-derivatives'
+const GITHUB_ORGANIZATION = 'OpenNeuroDerivatives'
+
+// Available derivatives at this time
+type GitHubDerivative = 'mriqc' | 'fmriprep'
+
+/**
+ * Test for a derivative on GitHub via API
+ * @param datasetId
+ * @param derivative String identifying the kind of derivative
+ */
+export const githubDerivativeQuery = (
+  datasetId: string,
+  derivative: GitHubDerivative,
+): URL => {
+  const url = new URL(
+    `https://api.github.com/repos/${GITHUB_ORGANIZATION}/${datasetId}-${derivative}`,
+  )
+  return url
+}
+
+export const githubDerivative = async (
+  datasetId: string,
+  derivative: GitHubDerivative,
+): Promise<boolean> => {
+  try {
+    const url = githubDerivativeQuery(datasetId, derivative)
+    const res = await fetch(url.toString(), {
+      headers: {
+        Authorization: `token ${config.github.token}`,
+      },
+    })
+    if (res.status === 200) {
+      const body = await res.json()
+      // Verify we aren't displaying a hidden repo
+      if (!body.private) {
+        return true
+      }
+    }
+    return false
+  } catch (err) {
+    return false
+  }
+}
+
+interface DatasetDerivatives {
+  name: string
+  local: boolean
+  s3Url: URL
+  dataladUrl: URL
+}
+
+export const derivativeObject = (
+  datasetId: string,
+  derivative: GitHubDerivative,
+): DatasetDerivatives => {
+  const name = `${datasetId}-${derivative}`
+  return {
+    name,
+    // Always false to identify OpenNeuro hosted derivatives later
+    local: false,
+    s3Url: new URL(name, `s3://${S3_BUCKET}/${derivative}/`),
+    dataladUrl: new URL(
+      `${name}.git`,
+      `https://github.com/${GITHUB_ORGANIZATION}/`,
+    ),
+  }
+}
+
+/**
+ * Derivative resolver, returns an array of the available derivative dataset sources
+ */
+export const derivatives = async (
+  dataset: DatasetOrSnapshot,
+): Promise<Array<DatasetDerivatives>> => {
+  let datasetId
+  if ('tag' in dataset) {
+    datasetId = getDatasetFromSnapshotId(dataset.id)
+  } else {
+    datasetId = dataset.id
+  }
+  const available: Array<DatasetDerivatives> = []
+  if (await githubDerivative(datasetId, 'mriqc')) {
+    available.push(derivativeObject(datasetId, 'mriqc'))
+  }
+  if (await githubDerivative(datasetId, 'fmriprep')) {
+    available.push(derivativeObject(datasetId, 'fmriprep'))
+  }
+  return available
+}

package/src/graphql/schema.js

@@ -384,6 +384,8 @@ export const typeDefs = `
     publishDate: DateTime
     # Is the dataset available for analysis on Brainlife?
     onBrainlife: Boolean @cacheControl(maxAge: 10080, scope: PUBLIC)
+    # Available derivatives of this dataset
+    derivatives: [DatasetDerivatives] @cacheControl(maxAge: 3600, scope: PUBLIC)
     # Dataset Metadata
     metadata: Metadata
     # Return the version history for a dataset (git log)
@@ -394,6 +396,17 @@ export const typeDefs = `
     reviewers: [DatasetReviewer]
   }
 
+  type DatasetDerivatives {
+    # Remote reference to display
+    name: String
+    # Is this an OpenNeuro derivative?
+    local: Boolean
+    # S3-like URL if available
+    s3Url: String
+    # DataLad GitHub URL if available
+    dataladUrl: String
+  }
+
   type DatasetCommit {
     # Git commit hash
     id: ID!