Manifest enrichment

services/madoc-ts/src/extensions/enrichment/extension.ts

@@ -1,6 +1,6 @@
 import { Topic, TopicType, TopicTypeListResponse } from '../../types/schemas/topics';
 import { BaseDjangoExtension } from './base-django-extension';
-import { EnrichmentIndexPayload } from './types';
+import { EnrichmentIndexPayload, EnrichmentPlaintext, EnrichmentTask } from './types';
 import { ApiKey } from '../../types/api-key';
 import { SearchQuery, SearchResponse } from '../../types/search';
 import {
@@ -79,7 +79,28 @@ export class EnrichmentExtension extends BaseDjangoExtension {
   }
 
   getEnrichmentTask(id: string) {
-    return this.api.request(`/api/enrichment/task_log/${id}`);
+    return this.api.request<EnrichmentTask>(`/api/enrichment/task_log/${id}`);
+  }
+  getEnrichmentPlaintext(id: string) {
+    return this.api.request<EnrichmentPlaintext>(`/api/enrichment/plaintext/${id}`);
+  }
+
+  enrichManifest(id: number) {
+    return this.api.request<EnrichmentTask>(`/api/madoc/iiif/manifests/${id}/enrichment`, {
+      method: 'POST',
+    });
+  }
+
+  enrichManifestInternal(id: number, callback?: string) {
+    return this.api.request<EnrichmentTask>(`/api/enrichment/tasks/madoc_manifest_enrichment_pipeline`, {
+      method: 'POST',
+      body: {
+        task: {
+          subject: `urn:madoc:manifest:${id}`,
+          parameters: [{ callback_url: callback }],
+        },
+      },
+    });
   }
 
   allTasks = [

services/madoc-ts/src/extensions/enrichment/types.ts

@@ -1,4 +1,5 @@
 import { Canvas, Collection, InternationalString, Manifest } from '@iiif/presentation-3';
+import { BaseTask } from '../../gateway/tasks/base-task';
 
 export type EnrichmentIndexPayload = {
   madoc_id: string;
@@ -16,3 +17,37 @@ export interface DjangoPagination<T> {
   previous: string;
   results: T[];
 }
+
+export interface EnrichmentTaskSnippet {
+  url: string;
+  id: string;
+  created: string;
+  modified: string;
+  name: string;
+  state: any;
+  status: number;
+  status_text: string;
+  task_type: string;
+  task_class: string;
+}
+
+export interface EnrichmentTask extends Omit<BaseTask, 'parent_task' | 'type' | 'subtasks'> {
+  url: string;
+  task_type: string; // Type
+  parent_task: EnrichmentTaskSnippet;
+  errors: string[];
+  child_tasks: EnrichmentTaskSnippet[];
+  task_class: string;
+}
+
+export interface EnrichmentPlaintext {
+  url: string;
+  id: string;
+  created: string;
+  modified: string;
+  source: string;
+  ocr_backend: string;
+  ocr_format: string;
+  plaintext: string;
+  plaintext_list: string;
+}

services/madoc-ts/src/frontend/admin/pages/content/manifests/manifest-search-index.tsx

@@ -32,7 +32,7 @@ export const ManifestSearchIndex = createUniversalComponent<ManifestSearchIndexT
 
     const api = useApi();
     const [invokeEnrichment, { isLoading: enrichLoading }] = useMutation(async () => {
-      await api.triggerSearchIndex(Number(id), 'manifest');
+      await api.enrichment.enrichManifest(Number(id));
       await refetch();
     });
 

services/madoc-ts/src/gateway/api.ts

@@ -153,14 +153,14 @@ export class ApiClient {
     // Enrichment
     this.authority = new AuthorityExtension(this);
     this.enrichment = new EnrichmentExtension(this);
+    this.webhooks = new WebhookExtension(this);
     this.search = new SearchExtension(this);
 
     if (options.withoutExtensions) {
       this.crowdsourcing = new CrowdsourcingApi(this, null, captureModelDataSources);
       return;
     }
 
-    this.webhooks = new WebhookExtension(this);
     this.pageBlocks = new PageBlockExtension(this, getDefaultPageBlockDefinitions());
     this.media = new MediaExtension(this);
     this.system = new SystemExtension(this);
@@ -1077,6 +1077,13 @@ export class ApiClient {
     return this.request<{ found: boolean; transcription: string }>(`/api/madoc/iiif/canvases/${id}/plaintext`);
   }
 
+  async updateCanvasPlaintext(id: number, plaintext: string) {
+    return this.request<{ success: boolean; empty: boolean }>(`/api/madoc/iiif/canvases/${id}/plaintext`, {
+      method: 'POST',
+      body: { plaintext },
+    });
+  }
+
   async getCanvasDeletionSummary(id: number) {
     return this.request<CanvasDeletionSummary>(`/api/madoc/iiif/canvases/${id}/deletion-summary`);
   }
@@ -1757,6 +1764,12 @@ export class ApiClient {
     });
   }
 
+  async deleteLinkingProperty(id: number) {
+    return this.request(`/api/madoc/iiif/linking/${id}`, {
+      method: 'DELETE',
+    });
+  }
+
   async getStorageRaw(bucket: string, fileName: string, isPublic = false) {
     return this.request<Response>(
       isPublic ? `/api/storage/data/${bucket}/public/${fileName}` : `/api/storage/data/${bucket}/${fileName}`,

services/madoc-ts/src/router.ts

@@ -4,8 +4,10 @@ import { keyRegenerate } from './routes/admin/key-regenerate';
 import { listApiKeys } from './routes/admin/list-api-keys';
 import { getProjectAnnotationStyle } from './routes/annotation-styles/get-project-annotation-style';
 import { annotationStyles } from './routes/annotation-styles/index';
+import { manifestEnrichmentPipeline } from "./routes/enrichment/manifest-enrichment-pipeline";
 import { searchAllUsers } from './routes/global/search-all-users';
 import { systemCheck } from './routes/global/system-check';
+import { addPlaintext } from './routes/iiif/linking/add-plaintext';
 import { getAutomatedUsers } from './routes/manage-site/get-automated-users';
 import { createProjectExport } from './routes/projects/create-project-export';
 import { getProjectRawData } from './routes/projects/get-project-raw-data';
@@ -415,6 +417,7 @@ export const router = new TypedRouter({
   'get-manifest-linking': [TypedRouter.GET, '/api/madoc/iiif/manifests/:id/linking', getLinking],
   'get-manifest-canvas-linking': [TypedRouter.GET, '/api/madoc/iiif/manifests/:id/canvas-linking', getParentLinking],
   'search-index-manifest': [TypedRouter.POST, '/api/madoc/iiif/manifests/:id/index', indexManifest],
+  'search-enrich-manifest': [TypedRouter.POST, '/api/madoc/iiif/manifests/:id/enrichment', manifestEnrichmentPipeline],
 
   // Canvas API
   'list-canvases': [TypedRouter.GET, '/api/madoc/iiif/canvases', listCanvases],
@@ -432,6 +435,7 @@ export const router = new TypedRouter({
   'search-index-canvas': [TypedRouter.POST, '/api/madoc/iiif/canvases/:id/index', indexCanvas],
   'convert-linking-property': [TypedRouter.POST, '/api/madoc/iiif/linking/:id/convert', convertLinking],
   'get-canvas-plaintext': [TypedRouter.GET, '/api/madoc/iiif/canvases/:id/plaintext', getCanvasPlaintext],
+  'update-canvas-plaintext': [TypedRouter.POST, '/api/madoc/iiif/canvases/:id/plaintext', addPlaintext],
   'get-canvas-source': [TypedRouter.GET, '/api/madoc/iiif/canvas-source', getCanvasReference],
   'get-canvas-deletion-summary': [
     TypedRouter.GET,

services/madoc-ts/src/routes/enrichment/manifest-enrichment-pipeline.ts

@@ -0,0 +1,54 @@
+import invariant from 'tiny-invariant';
+import { api } from '../../gateway/api.server';
+import { RouteMiddleware } from '../../types/route-middleware';
+import { parseUrn } from '../../utility/parse-urn';
+import { userWithScope } from '../../utility/user-with-scope';
+import { IncomingWebhook, WebhookEventType } from '../../webhooks/webhook-types';
+
+export const manifestEnrichmentPipeline: RouteMiddleware<{ id: number }> = async context => {
+  const { siteId } = userWithScope(context, ['site.admin']);
+  const site = await context.siteManager.getSiteById(siteId);
+  const siteApi = api.asUser({ siteId });
+
+  // 12-hour token.
+  const webhook = await context.webhookExtension.generateWebhookUrl(
+    site,
+    manifestEnrichmentPipelineEvent.event_id,
+    12 * 3600
+  );
+  context.response.body = await siteApi.enrichment.enrichManifestInternal(context.params.id, webhook);
+};
+
+export const manifestEnrichmentPipelineEvent: WebhookEventType = {
+  event_id: 'manifest-enrichment-pipeline.complete',
+  body_variables: ['id'],
+};
+
+export const manifestEnrichmentHook: IncomingWebhook = {
+  type: 'manifest-enrichment-pipeline-task-ingest',
+  event_id: 'manifest-enrichment-pipeline.complete',
+  is_outgoing: false,
+  execute: async (resp, siteApi) => {
+    invariant(resp.id, 'Expected response to contain `id`');
+
+    const task = await siteApi.enrichment.getEnrichmentTask(resp.id);
+    invariant(task.subject, 'Missing subject on task');
+    invariant(task.status === 3, 'Task is not yet complete');
+
+    if (task.task_type === 'ocr_madoc_resource') {
+      const parsed = parseUrn(task.subject);
+      invariant(parsed, 'Invalid subject');
+      invariant(parsed.type === 'canvas', 'Can only process canvases');
+
+      if (task.state && task.state.ocr_resources && task.state.ocr_resources[0]) {
+        const first = task.state.ocr_resources[0];
+        const enrichmentPlaintext = await siteApi.enrichment.getEnrichmentPlaintext(first);
+        invariant(enrichmentPlaintext, 'Missing plaintext from enrichment');
+        if (enrichmentPlaintext.plaintext) {
+          const canvasId = parsed.id; // ??
+          return await siteApi.updateCanvasPlaintext(canvasId, enrichmentPlaintext.plaintext);
+        }
+      }
+    }
+  },
+};

services/madoc-ts/src/routes/iiif/linking/add-plaintext.ts

@@ -0,0 +1,53 @@
+import { api } from '../../../gateway/api.server';
+import { RouteMiddleware } from '../../../types/route-middleware';
+import { userWithScope } from '../../../utility/user-with-scope';
+import { linkHash } from './convert-linking';
+
+export const addPlaintext: RouteMiddleware<{ id: number }, { plaintext: string }> = async context => {
+  const { siteId } = userWithScope(context, ['site.admin']);
+  const canvasId = Number(context.query.id);
+  const plaintext = context.requestBody.plaintext;
+
+  const siteApi = api.asUser({ siteId });
+  const linking = await siteApi.getCanvasLinking(canvasId);
+
+  if (!plaintext.trim()) {
+    context.response.status = 200;
+    context.response.body = { success: true, empty: true };
+    return;
+  }
+
+  const matchingPlaintexts = linking.linking.filter(singleLink => {
+    return singleLink.property === 'seeAlso' && singleLink.link.format === 'text/plain';
+  });
+
+  if (matchingPlaintexts.length) {
+    for (const matchingPlaintext of matchingPlaintexts) {
+      // Delete the existing one, and continue;
+      await siteApi.deleteLinkingProperty(matchingPlaintext.id);
+    }
+  }
+
+  // Create new plaintext and insert it.
+  const bucket = 'plaintext';
+  const filePath = `public/${canvasId}/${linkHash(plaintext)}.txt`;
+
+  await siteApi.saveStoragePlainText(bucket, filePath, plaintext, true);
+
+  await siteApi.addLinkToResource({
+    label: 'Plaintext',
+    link: {
+      id: `/public/storage/urn:madoc:site:${siteId}/${bucket}/${filePath}`,
+      format: 'text/plain',
+      label: 'Plaintext',
+      type: 'Text',
+      file_path: `public/${canvasId}/${linkHash(plaintext)}.txt`,
+      file_bucket: bucket,
+    },
+    resource_id: canvasId as any,
+    property: 'seeAlso',
+  });
+
+  context.response.status = 200;
+  context.response.body = { success: true, empty: false };
+};

services/madoc-ts/src/routes/iiif/linking/convert-linking.ts

@@ -7,7 +7,7 @@ import { RequestError } from '../../../utility/errors/request-error';
 import { userWithScope } from '../../../utility/user-with-scope';
 import contentType from 'content-type';
 
-function linkHash(uri: string) {
+export function linkHash(uri: string) {
   return createHash('sha1')
     .update(uri)
     .digest('hex');

services/madoc-ts/src/webhooks/routes/execute-webhook-internal.ts

@@ -1,4 +1,5 @@
 import { generateId } from '../../frontend/shared/capture-models/helpers/generate-id';
+import { api } from '../../gateway/api.server';
 import { RouteMiddleware } from '../../types/route-middleware';
 import { optionalUserWithScope } from '../../utility/user-with-scope';
 import { WebhookCallRow } from '../webhook-types';
@@ -14,6 +15,8 @@ export const executeWebhookInternal: RouteMiddleware<{ event_id: string }> = asy
   const results = { success: 0, fail: 0 };
   const callId = generateId();
 
+  const siteApi = api.asUser({ siteId }, {}, true);
+
   const databaseHooks = await context.webhooks.listWebhooksByEvent(eventId, siteId);
 
   for (const databaseHook of databaseHooks) {
@@ -64,7 +67,7 @@ export const executeWebhookInternal: RouteMiddleware<{ event_id: string }> = asy
         continue;
       } else {
         // Do internal thing.
-        result.response = (await hook.execute(body)) || {};
+        result.response = (await hook.execute(body, siteApi)) || {};
       }
       results.success++;
     } catch (e) {

services/madoc-ts/src/webhooks/webhook-events.ts

@@ -1,8 +1,9 @@
+import { manifestEnrichmentPipelineEvent } from '../routes/enrichment/manifest-enrichment-pipeline';
 import { WebhookEventType } from './webhook-types';
 
 const testEvent = {
   event_id: 'test-event',
   body_variables: ['hello'] as const,
 };
 
-export const webhookEvents: WebhookEventType[] = [testEvent];
+export const webhookEvents: WebhookEventType[] = [testEvent, manifestEnrichmentPipelineEvent];

services/madoc-ts/src/webhooks/webhook-server-extension.ts

@@ -3,6 +3,7 @@ import invariant from 'tiny-invariant';
 import { RegistryExtension } from '../extensions/registry-extension';
 import { generateId } from '../frontend/shared/capture-models/helpers/generate-id';
 import { apiGateway, gatewayHost } from '../gateway/api.server';
+import { manifestEnrichmentHook } from '../routes/enrichment/manifest-enrichment-pipeline';
 import { getPem, getPublicPem } from '../utility/get-pem';
 import { IncomingWebhook, OutgoingWebhook } from './webhook-types';
 import { JWK, JWS } from 'jose';
@@ -13,12 +14,14 @@ export class WebhookServerExtension extends RegistryExtension<IncomingWebhook |
       registryName: 'webhook',
     });
 
+    WebhookServerExtension.register(manifestEnrichmentHook);
+
     WebhookServerExtension.register({
       is_outgoing: false,
       type: 'example-test',
       event_id: 'test-event',
       execute: body => {
-        console.log('Did this work?', body);
+        console.log('WebHooks - test event:', body);
       },
     });
   }
@@ -41,7 +44,7 @@ export class WebhookServerExtension extends RegistryExtension<IncomingWebhook |
       code: await this.sign({ eventId, expires, siteId: site.id }),
     };
 
-    return `${internal ? apiGateway : gatewayHost}/s/${site.slug}/madoc/api/webhook?${stringify(query)}}`;
+    return `${internal ? apiGateway : gatewayHost}/s/${site.slug}/madoc/api/webhook?${stringify(query)}`;
   }
 
   getHooksForEvents(eventId: string, siteId: number): Array<IncomingWebhook | OutgoingWebhook> {

services/madoc-ts/src/webhooks/webhook-types.ts

@@ -1,3 +1,5 @@
+import { ApiClient } from '../gateway/api';
+
 export interface WebhookRow {
   id: string;
   event_id: string;
@@ -43,6 +45,6 @@ export interface IncomingWebhook {
   is_outgoing: false;
   type: string;
   event_id: string;
-  execute: (request: any) => any | Promise<any>;
+  execute: (request: any, api: ApiClient) => any | Promise<any>;
   source?: { type: string; id?: string; name: string };
 }