OC-888: Deploying ARI ingests to prod

api/.env.example

@@ -33,4 +33,10 @@ MAIL_SERVER=localhost
 LOCALSTACK_SERVER=http://localhost:4566
 
 QUEUE_URL=http://localhost:4566/000000000000/your-queue-name
-SQS_ENDPOINT=http://localhost:4566
+SQS_ENDPOINT=http://localhost:4566
+
+LIST_USERS_API_KEY=12345
+TRIGGER_ARI_INGEST_API_KEY=12345
+
+SLACK_CHANNEL_EMAIL=example@mailinator.com
+INGEST_REPORT_RECIPIENTS=example.one@mailinator.com,example.two@mailinator.com

api/README.md

@@ -139,7 +139,7 @@ A similar process happens when the database is seeded. After publications are in
 
 ## Integrations
 
-Octopus is built to integrate with some external systems in order to import publications. For more information please read the dedicated [integrations readme](./src/lib/integrations/README.md).
+Octopus is built to integrate with some external systems in order to import publications. For more information please read the dedicated [integrations readme](./src/components/integration/README.md).
 
 ---
 

api/package.json

@@ -82,7 +82,6 @@
         "prisma": "^5.11.0",
         "puppeteer": "^22.12.0",
         "serverless-offline": "^12.0.4",
-        "serverless-offline-ssm": "^6.2.0",
         "serverless-plugin-split-stacks": "^1.13.0",
         "serverless-prune-plugin": "^2.0.2",
         "serverless-webpack": "^5.13.0",

api/scripts/fullAriImport.ts

@@ -5,7 +5,7 @@ import { expand } from 'dotenv-expand';
 // Important to do this so that environment variables are treated the same as in deployed code.
 expand(dotenv.config());
 
-import * as ariUtils from 'lib/integrations/ariUtils';
+import * as ariUtils from 'integration/ariUtils';
 import * as I from 'interface';
 
 const fullAriImport = async (): Promise<string> => {

api/serverless-config-default.yml

@@ -494,9 +494,13 @@ functions:
                 cors: true
     # Integrations
     incrementalAriIngest:
-        handler: src/components/integrations/service.incrementalAriIngest
+        handler: src/components/integration/routes.incrementalAriIngest
         timeout: 900
         events:
             - schedule:
                 rate: cron(0 5 ? * TUE *) # Every Tuesday at 5 a.m.
                 enabled: ${self:custom.scheduledAriIngestEnabled.${opt:stage}, false}
+            - http:
+                path: ${self:custom.versions.v1}/integrations/ari/incremental
+                method: POST
+                cors: true

api/serverless-offline.yml

@@ -5,7 +5,6 @@ frameworkVersion: ${file(./serverless-config-default.yml):frameworkVersion}
 useDotenv: ${file(./serverless-config-default.yml):useDotenv}
 
 plugins:
-    - serverless-offline-ssm
     - serverless-offline
     - serverless-webpack
     - serverless-webpack-prisma
@@ -19,29 +18,33 @@ provider:
     stage: 'local'
     environment:
         STAGE: ${self:provider.stage}
-        ELASTICSEARCH_USER: ${ssm:/elasticsearch_user_${self:provider.stage}_octopus}
-        ELASTICSEARCH_PASSWORD: ${ssm:/elasticsearch_password_${self:provider.stage}_octopus}
-        ELASTICSEARCH_ENDPOINT: ${ssm:/elasticsearch_endpoint_${self:provider.stage}_octopus}
-        ELASTICSEARCH_PROTOCOL: ${ssm:/elastic_search_protocol_${self:provider.stage}_octopus}
-        DATABASE_URL: ${ssm:/db_connection_string_${self:provider.stage}_octopus}
-        JWT_SECRET: ${ssm:/jwt_secret_${self:provider.stage}_octopus}
+        DATABASE_URL: ${env:DATABASE_URL}
+        ORCID_SECRET: ${env:ORCID_SECRET}
+        ORCID_ID: ${env:ORCID_ID}
+        ORCID_AUTH_URL: ${env:ORCID_AUTH_URL}
+        ORCID_MEMBER_API_URL: ${env:ORCID_MEMBER_API_URL}
+        JWT_SECRET: ${env:JWT_SECRET}
+        EMAIL_SENDER_ADDRESS: ${env:EMAIL_SENDER_ADDRESS}
+        BASE_URL: ${env:BASE_URL}
+        AUTHORISATION_CALLBACK_URL: ${env:AUTHORISATION_CALLBACK_URL}
+        ELASTICSEARCH_PROTOCOL: ${env:ELASTICSEARCH_PROTOCOL}
+        ELASTICSEARCH_USER: ${env:ELASTICSEARCH_USER}
+        ELASTICSEARCH_PASSWORD: ${env:ELASTICSEARCH_PASSWORD}
+        ELASTICSEARCH_ENDPOINT: ${env:ELASTICSEARCH_ENDPOINT}
         VALIDATION_CODE_EXPIRY: 10
         VALIDATION_CODE_ATTEMPTS: 3
-        ORCID_ID: ${ssm:/orcid_app_id_${self:provider.stage}_octopus}
-        ORCID_SECRET: ${ssm:/orcid_secret_key_${self:provider.stage}_octopus}
-        ORCID_AUTH_URL: ${ssm:/orcid_auth_url_${self:provider.stage}_octopus}
-        ORCID_MEMBER_API_URL: ${ssm:/orcid_member_api_url_${self:provider.stage}_octopus}
-        DOI_PREFIX: ${ssm:/doi_prefix_${self:provider.stage}_octopus}
-        DATACITE_ENDPOINT: ${ssm:/datacite_endpoint_${self:provider.stage}_octopus}
-        DATACITE_USER: ${ssm:/datacite_user_${self:provider.stage}_octopus}
-        DATACITE_PASSWORD: ${ssm:/datacite_password_${self:provider.stage}_octopus}
-        EMAIL_SENDER_ADDRESS: ${ssm:/email_sender_address_${self:provider.stage}_octopus}
-        BASE_URL: ${ssm:/base_url_${self:provider.stage}_octopus}
-        AUTHORISATION_CALLBACK_URL: ${ssm:/authorization_callback_url_${self:provider.stage}_octopus}
-        LIST_USERS_API_KEY: ${ssm:/list_users_api_key_${self:provider.stage}_octopus}
-        QUEUE_URL: ${ssm:/queue_url_${self:provider.stage}_octopus}
-        SQS_ENDPOINT: ${ssm:/sqs_endpoint_${self:provider.stage}_octopus}
-        MAIL_SERVER: ${ssm:/mail_server_${self:provider.stage}_octopus}
+        DOI_PREFIX: ${env:DOI_PREFIX}
+        DATACITE_ENDPOINT: ${env:DATACITE_ENDPOINT}
+        DATACITE_USER: ${env:DATACITE_USER}
+        DATACITE_PASSWORD: ${env:DATACITE_PASSWORD}
+        MAIL_SERVER: ${env:MAIL_SERVER}
+        LOCALSTACK_SERVER: ${env:LOCALSTACK_SERVER}
+        QUEUE_URL: ${env:QUEUE_URL}
+        SQS_ENDPOINT: ${env:SQS_ENDPOINT}
+        LIST_USERS_API_KEY: ${env:LIST_USERS_API_KEY}
+        TRIGGER_ARI_INGEST_API_KEY: ${env:TRIGGER_ARI_INGEST_API_KEY}
+        SLACK_CHANNEL_EMAIL: ${env:SLACK_CHANNEL_EMAIL}
+        INGEST_REPORT_RECIPIENTS: ${env:INGEST_REPORT_RECIPIENTS}
     deploymentBucket:
         tags:
             Project: Octopus

api/serverless.yml

@@ -50,6 +50,8 @@ provider:
         SQS_ENDPOINT: ${ssm:/sqs_endpoint_${self:provider.stage}_octopus}
         MAIL_SERVER: ${ssm:/mail_server_${self:provider.stage}_octopus}
         SLACK_CHANNEL_EMAIL: ${ssm:/slack_channel_email_${self:provider.stage}_octopus}
+        TRIGGER_ARI_INGEST_API_KEY: ${ssm:/trigger_ari_ingest_api_key_${self:provider.stage}_octopus}
+        INGEST_REPORT_RECIPIENTS: ${ssm:/ingest_report_recipients_${self:provider.stage}_octopus}
     deploymentBucket:
         tags:
             Project: Octopus
@@ -105,7 +107,8 @@ custom:
         automatic: true
         number: 3
     scheduledAriIngestEnabled:
-        int: true
+        int: false
+        prod: false
 functions:
     - ${file(./serverless-config-default.yml):functions}
     - ${file(./serverless-config-deploy.yml):functions}

api/src/components/ingestLog/__tests__/ingestLog.test.ts

@@ -26,14 +26,14 @@ describe('Ingest log functions', () => {
         });
     });
 
-    test('Get most recent start time', async () => {
-        const mostRecentStart = await ingestLogService.getMostRecentStartTime('ARI');
-        expect(mostRecentStart).toEqual(new Date('2024-09-11T12:53:00.000Z'));
+    test('Get most recent log', async () => {
+        const mostRecentLog = await ingestLogService.getMostRecentLog('ARI');
+        expect(mostRecentLog?.start).toEqual(new Date('2024-09-11T12:53:00.000Z'));
     });
 
     test('Most recent start is null if no run that ended successfully is present', async () => {
         await client.prisma.ingestLog.update({ where: { id: 'ingest-log-1' }, data: { end: null } });
-        const mostRecentStart = await ingestLogService.getMostRecentStartTime('ARI');
+        const mostRecentStart = await ingestLogService.getMostRecentLog('ARI');
         expect(mostRecentStart).toBeNull();
     });
 });

api/src/components/ingestLog/service.ts

@@ -13,26 +13,14 @@ export const setEndTime = (id: string, end: Date) =>
         }
     });
 
-export const getMostRecentStartTime = async (source: I.PublicationImportSource): Promise<Date | null> => {
-    const mostRecentStartQuery = await client.prisma.ingestLog.findFirst({
+export const getMostRecentLog = (source: I.PublicationImportSource, includeOpenLogs?: boolean) =>
+    client.prisma.ingestLog.findFirst({
         where: {
             source,
-            // Successful runs only.
-            end: {
-                not: null
-            }
+            // By default, get successful (having an end time) logs only.
+            ...(includeOpenLogs ? {} : { end: { not: null } })
         },
         orderBy: {
             start: 'desc'
-        },
-        select: {
-            start: true
         }
     });
-
-    if (mostRecentStartQuery) {
-        return mostRecentStartQuery.start;
-    } else {
-        return null;
-    }
-};

api/src/lib/integrations/README.md → api/src/components/integration/README.md

@@ -32,7 +32,7 @@ On import, ARIs go through a handling flow:
 
 #### How ARI data is mapped to octopus data
 
-Various ARI fields are mapped to octpous ones in the `mapAriQuestionToPublicationVersion` function in [ariUtils.ts](./ariUtils.ts).
+Various ARI fields are mapped to octopus ones in the `mapAriQuestionToPublicationVersion` function in [ariUtils.ts](./ariUtils.ts).
 
 Of particular importance is how ARIs are matched to an owning organisational user account. The mapping process expects a UserMapping to exist associating the `department` field value from the ARI (where the title matches, case insensitive, and the mapping source is 'ARI') with the user ID of an organisational account.
 

api/src/lib/integrations/__tests__/ari.test.ts → api/src/components/integration/__tests__/ari.test.ts

@@ -1,5 +1,6 @@
-import * as ariUtils from 'lib/integrations/ariUtils';
+import * as ariUtils from 'integration/ariUtils';
 import * as I from 'interface';
+import * as ingestLogService from 'ingestLog/service';
 import * as testUtils from 'lib/testUtils';
 
 // This ARI will match a publication in the seed data via the questionId.
@@ -120,6 +121,18 @@ describe('ARI Mapping', () => {
         });
     });
 
+    test('Unrecognised topics are reported', async () => {
+        const mappingAttempt = await ariUtils.mapAriQuestionToPublicationVersion({
+            ...sampleARIQuestion,
+            topics: ['unrecognised topic']
+        });
+        expect(mappingAttempt).toMatchObject({
+            success: true,
+            message: 'Found unrecognised topic(s).',
+            unrecognisedTopics: ['unrecognised topic']
+        });
+    });
+
     test('Department is matched to existing user', async () => {
         const mappingAttempt = await ariUtils.mapAriQuestionToPublicationVersion(sampleARIQuestion);
         expect(mappingAttempt).toMatchObject({
@@ -135,7 +148,8 @@ describe('ARI Mapping', () => {
         expect(mappingAttempt).toMatchObject({
             success: false,
             mappedData: null,
-            message: 'User not found for department: unrecognised department.'
+            message: 'User not found for department: unrecognised department.',
+            unrecognisedDepartment: 'unrecognised department'
         });
     });
 
@@ -221,7 +235,7 @@ describe('ARI handling', () => {
         });
     });
 
-    test('ARI with unrecognised department is skipped', async () => {
+    test('ARI with unrecognised department is skipped and dept name is reported in a field', async () => {
         const handleARI = await ariUtils.handleIncomingARI({
             ...sampleARIQuestion,
             department: 'Unrecognised Department name'
@@ -231,7 +245,8 @@ describe('ARI handling', () => {
             actionTaken: 'none',
             success: false,
             message:
-                'Failed to map ARI data to octopus data. User not found for department: Unrecognised Department name.'
+                'Failed to map ARI data to octopus data. User not found for department: Unrecognised Department name.',
+            unrecognisedDepartment: 'Unrecognised Department name'
         });
     });
 
@@ -281,6 +296,25 @@ describe('ARI handling', () => {
         });
     });
 
+    test('Unrecognised topics are reported', async () => {
+        const handleARI = await ariUtils.handleIncomingARI({
+            ...sampleARIQuestion,
+            topics: [...sampleARIQuestion.topics, 'unrecognised topic']
+        });
+        expect(handleARI).toMatchObject({
+            actionTaken: 'none',
+            success: true,
+            publicationVersion: {
+                topics: [
+                    {
+                        id: 'test-topic-1a'
+                    }
+                ]
+            },
+            unrecognisedTopics: ['unrecognised topic']
+        });
+    });
+
     test('Keywords update when fieldsOfResearch/tags change', async () => {
         const handleARI = await ariUtils.handleIncomingARI({
             ...sampleARIQuestion,
@@ -352,3 +386,32 @@ describe('ARI handling', () => {
         });
     });
 });
+
+describe('ARI import processes', () => {
+    beforeEach(async () => {
+        await testUtils.clearDB();
+        await testUtils.testSeed();
+    });
+
+    test('Incremental import endpoint requires API key', async () => {
+        const triggerImport = await testUtils.agent.post('/integrations/ari/incremental');
+
+        expect(triggerImport.status).toEqual(401);
+        expect(triggerImport.body).toMatchObject({
+            message: "Please provide a valid 'apiKey'."
+        });
+    });
+
+    test('Incremental ingest cancels if already in progress', async () => {
+        // Create an open ended log first.
+        await ingestLogService.create('ARI');
+        const triggerImport = await testUtils.agent
+            .post('/integrations/ari/incremental')
+            .query({ apiKey: process.env.TRIGGER_ARI_INGEST_API_KEY });
+
+        expect(triggerImport.status).toEqual(202);
+        expect(triggerImport.body).toMatchObject({
+            message: 'Cancelling ingest. Either an import is already in progress or the last import failed.'
+        });
+    });
+});