Merge pull request #3315 from OpenNeuroOrg/feat/brain-initiative-resolver

feat(server): Add resolver to flag Brain Initiative datasets

Author: nellh

.codespellrc

@@ -1,6 +1,6 @@
 # Ref: https://github.com/codespell-project/codespell#using-a-config-file
 [codespell]
-skip = .git,*.pdf,*.svg,go.sum,*.lock,.codespellrc,.yarn,node_modules,schema-validator.js,.pnp.cjs
+skip = .git,*.pdf,*.svg,go.sum,*.lock,.codespellrc,.yarn,node_modules,schema-validator.js,.pnp.cjs,funded_awards.json
 check-hidden = true
 ignore-regex = ^.{300,}$|/((.*\|){4,}.*\))|\b(afterAll)\b
 ignore-words-list = chack

packages/openneuro-search/src/mappings/datasets-mapping.json

@@ -3,6 +3,7 @@
     "id": { "type": "keyword" },
     "created": { "type": "date" },
     "public": { "type": "boolean" },
+    "brainInitiative": { "type": "boolean" },
     "metadata": {
       "properties": {
         "datasetName": { "type": "keyword" },

packages/openneuro-search/src/query.ts

@@ -9,6 +9,7 @@ export const INDEX_DATASET_FRAGMENT = gql`
     id
     created
     public
+    brainInitiative
     metadata {
       datasetName
       datasetUrl

packages/openneuro-server/src/cache/types.ts

@@ -13,4 +13,5 @@ export enum CacheType {
   participantCount = "participantCount",
   snapshotDownload = "download",
   draftRevision = "revision",
+  brainInitiative = "brainInitiative"
 }

packages/openneuro-server/src/data/README.md

@@ -0,0 +1 @@
+funded_awards.json retrieved originally from https://braininitiative.nih.gov/sites/default/files/views_export/funded_awards/funded_awards.json

packages/openneuro-server/src/data/funded_awards.json

@@ -0,0 +1,65 @@
+import { redis } from "../../libs/redis"
+import type { DatasetOrSnapshot } from "../../utils/datasetOrSnapshot"
+import { latestSnapshot } from "./snapshots"
+import { description } from "../../datalad/description"
+import Metadata from "../../models/metadata"
+import CacheItem, { CacheType } from "../../cache/item"
+import * as Sentry from "@sentry/node"
+import fundedAwards from "../../data/funded_awards.json"
+
+const brainInitiativeMatch = new RegExp("brain.initiative", "i")
+
+const brainInitiativeGrants = fundedAwards.map((award) =>
+  award.field_project_number.replace(/[^a-zA-Z0-9]/g, "")
+)
+
+/**
+ * Check for any Brain Initiative metadata
+ */
+export const brainInitiative = async (
+  dataset: DatasetOrSnapshot,
+  _,
+  context,
+): Promise<boolean> => {
+  const cache = new CacheItem(
+    redis,
+    CacheType.brainInitiative,
+    [dataset.id],
+    86400,
+  )
+  return await cache.get(async () => {
+    try {
+      const metadata = await Metadata.findOne({ datasetId: dataset.id })
+      if (metadata.grantFunderName.match(brainInitiativeMatch)) {
+        return true
+      } else {
+        // Fetch snapshot if metadata didn't match
+        const snapshot = await latestSnapshot(dataset, null, context)
+        const snapshotDescription = await description(snapshot)
+        for (const funding of snapshotDescription.Funding) {
+          if (funding.match(brainInitiativeMatch)) {
+            return true
+          }
+        }
+        // Check for grant ids too - filter to only alphanumeric to improve matching across format differences
+        const identifier = metadata.grantIdentifier.replace(/[^a-zA-Z0-9]/g, "")
+        for (const grant of brainInitiativeGrants) {
+          if (
+            identifier.includes(grant)
+          ) {
+            return true
+          }
+          for (const funding of snapshotDescription.Funding) {
+            if (funding.replace(/[^a-zA-Z0-9]/g, "").includes(grant)) {
+              return true
+            }
+          }
+        }
+      }
+      return false
+    } catch (_err) {
+      Sentry.captureException(_err)
+      return false
+    }
+  })
+}

packages/openneuro-server/src/graphql/resolvers/brainInitiative.ts

@@ -19,6 +19,7 @@ import { reviewers } from "./reviewer"
 import { getDatasetWorker } from "../../libs/datalad-service"
 import { getFileName } from "../../datalad/files"
 import { onBrainlife } from "./brainlife"
+import { brainInitiative } from "./brainInitiative"
 import { derivatives } from "./derivatives"
 import { promiseTimeout } from "../../utils/promiseTimeout"
 import semver from "semver"
@@ -301,6 +302,7 @@ const Dataset = {
   history,
   worker,
   reviewers,
+  brainInitiative,
 }
 
 export default Dataset

packages/openneuro-server/src/graphql/resolvers/dataset.ts

@@ -410,6 +410,8 @@ export const typeDefs = `
     worker: String
     # Anonymous reviewers for this dataset
     reviewers: [DatasetReviewer]
+    # Dataset belongs to Brain Initiative
+    brainInitiative: Boolean
   }
 
   type DatasetDerivatives {

packages/openneuro-server/src/graphql/schema.ts

@@ -9,7 +9,7 @@
     "module": "CommonJS"
   },
   "include": ["./src"],
-  "files": ["./src/lerna.json"],
+  "files": ["./src/lerna.json", "./src/data/funded_awards.json"],
   "references": [
     { "path": "../openneuro-search" }
   ]