initial service

src/models/generated/common.d.ts

@@ -370,6 +370,10 @@ export interface SolrServerNamespaceConfiguration {
    * Solr index name
    */
   index: string;
+  /**
+   * Version of the data schema used in the index. Optional.
+   */
+  schemaVersion?: string;
 }
 export interface ProxyConfig {
   /**
@@ -514,4 +518,8 @@ export interface SolrServerNamespaceConfiguration {
    * Solr index name
    */
   index: string;
+  /**
+   * Version of the data schema used in the index. Optional.
+   */
+  schemaVersion?: string;
 }

src/models/generated/schemas.d.ts

@@ -879,36 +879,18 @@ export interface Image {
    * The unique identifier of the image
    */
   uid: string;
+  /**
+   * The unique identifier of the issue that the image belongs to.
+   */
+  issueUid: string;
   /**
    * The unique identifier of the content item that the image belongs to.
    */
-  contentItemUid: string;
+  contentItemUid?: string;
   /**
    * The URL of the image preview
    */
   previewUrl: string;
-  /**
-   * The URL of the IIIF info.json file
-   */
-  iiifInfoUrl: string;
-  permissions?: ContentPermissions;
-}
-/**
- * Content item permissions
- */
-export interface ContentPermissions {
-  /**
-   * Bitmap representing the 'explore' permissions of the content item
-   */
-  exploreBitmap?: number;
-  /**
-   * Bitmap representing the 'get transcript' permissions of the content item
-   */
-  getTranscriptBitmap?: number;
-  /**
-   * Bitmap representing the 'get images' permissions of the content item
-   */
-  getImagesBitmap?: number;
 }
 
 

src/models/generated/schemasPublic.d.ts

@@ -374,6 +374,29 @@ export interface Freeform {
 }
 
 
+/**
+ * An image from a content item
+ */
+export interface Image {
+  /**
+   * The unique identifier of the image
+   */
+  uid: string;
+  /**
+   * The unique identifier of the issue that the image belongs to.
+   */
+  issueUid: string;
+  /**
+   * The unique identifier of the content item that the image belongs to.
+   */
+  contentItemUid?: string;
+  /**
+   * The URL of the image preview
+   */
+  previewUrl: string;
+}
+
+
 /**
  * A media source is what a content item belongs to. This can be a newspaper, a TV or a radio station, etc.
  */

src/models/generated/solr.d.ts

@@ -0,0 +1,41 @@
+
+/* eslint-disable */
+/**
+ * This file was automatically generated by json-schema-to-typescript.
+ * DO NOT MODIFY IT BY HAND. Instead, modify the source JSONSchema file,
+ * and run json-schema-to-typescript to regenerate this file.
+ */
+
+
+/**
+ * Image Solr document in Impresso v2
+ */
+export interface Image {
+  id?: string;
+  meta_journal_s?: string;
+  meta_year_i?: number;
+  meta_day_i?: number;
+  meta_ed_s?: string;
+  meta_issue_id_s?: string;
+  meta_date_dt?: string;
+  linked_ci_s?: string;
+  reading_order_i?: number;
+  caption_txt?: string[];
+  item_type_s?: string;
+  page_nb_is?: number[];
+  coords_is?: number[];
+  front_b?: boolean;
+  iiif_url_s?: string;
+  cc_b?: boolean;
+  rights_data_domain_s?: string;
+  rights_copyright_s?: string;
+  rights_perm_use_explore_plain?: string[];
+  rights_perm_use_get_tr_plain?: string[];
+  rights_perm_use_get_img_plain?: string[];
+  rights_bm_explore_l?: number;
+  rights_bm_get_tr_l?: number;
+  rights_bm_get_img_l?: number;
+  dinov2_emb_v1024?: number[];
+  openclip_emb_v768?: number[];
+  _version_?: number;
+}

src/schema/common/solrConfiguration.json

@@ -90,6 +90,10 @@
         "index": {
           "type": "string",
           "description": "Solr index name"
+        },
+        "schemaVersion": {
+          "type": "string",
+          "description": "Version of the data schema used in the index. Optional."
         }
       },
       "required": ["namespaceId", "serverId", "index"]

src/schema/schemas/Image.json

@@ -9,6 +9,10 @@
       "type": "string",
       "description": "The unique identifier of the image"
     },
+    "issueUid": {
+      "type": "string",
+      "description": "The unique identifier of the issue that the image belongs to."
+    },
     "contentItemUid": {
       "type": "string",
       "description": "The unique identifier of the content item that the image belongs to."
@@ -17,15 +21,7 @@
       "type": "string",
       "format": "uri",
       "description": "The URL of the image preview"
-    },
-    "iiifInfoUrl": {
-      "type": "string",
-      "format": "uri",
-      "description": "The URL of the IIIF info.json file"
-    },
-    "permissions": {
-      "$ref": "./ContentPermissions.json"
     }
   },
-  "required": ["uid", "contentItemUid", "previewUrl", "iiifInfoUrl"]
+  "required": ["uid", "issueUid", "previewUrl"]
 }

src/schema/schemasPublic/Image.json

@@ -0,0 +1,27 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "title": "Image",
+  "description": "An image from a content item",
+  "type": "object",
+  "additionalProperties": false,
+  "properties": {
+    "uid": {
+      "type": "string",
+      "description": "The unique identifier of the image"
+    },
+    "issueUid": {
+      "type": "string",
+      "description": "The unique identifier of the issue that the image belongs to."
+    },
+    "contentItemUid": {
+      "type": "string",
+      "description": "The unique identifier of the content item that the image belongs to."
+    },
+    "previewUrl": {
+      "type": "string",
+      "format": "uri",
+      "description": "The URL of the image preview"
+    }
+  },
+  "required": ["uid", "issueUid", "previewUrl"]
+}

src/schema/solr/Image.json

@@ -0,0 +1,60 @@
+{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "type": "object",
+  "title": "Image",
+  "description": "Image Solr document in Impresso v2",
+  "properties": {
+    "id": { "type": "string" },
+    "meta_journal_s": { "type": "string" },
+    "meta_year_i": { "type": "integer" },
+    "meta_day_i": { "type": "integer" },
+    "meta_ed_s": { "type": "string" },
+    "meta_issue_id_s": { "type": "string" },
+    "meta_date_dt": { "type": "string", "format": "date-time" },
+    "linked_ci_s": { "type": "string" },
+    "reading_order_i": { "type": "integer" },
+    "caption_txt": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+    "item_type_s": { "type": "string" },
+    "page_nb_is": {
+      "type": "array",
+      "items": { "type": "integer" }
+    },
+    "coords_is": {
+      "type": "array",
+      "items": { "type": "integer" }
+    },
+    "front_b": { "type": "boolean" },
+    "iiif_url_s": { "type": "string", "format": "uri" },
+    "cc_b": { "type": "boolean" },
+    "rights_data_domain_s": { "type": "string" },
+    "rights_copyright_s": { "type": "string" },
+    "rights_perm_use_explore_plain": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+    "rights_perm_use_get_tr_plain": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+    "rights_perm_use_get_img_plain": {
+      "type": "array",
+      "items": { "type": "string" }
+    },
+    "rights_bm_explore_l": { "type": "integer" },
+    "rights_bm_get_tr_l": { "type": "integer" },
+    "rights_bm_get_img_l": { "type": "integer" },
+    "dinov2_emb_v1024": {
+      "type": "array",
+      "items": { "type": "number" }
+    },
+    "openclip_emb_v768": {
+      "type": "array",
+      "items": { "type": "number" }
+    },
+    "_version_": { "type": "number" }
+  },
+  "additionalProperties": false
+}

src/scripts/generate-types.js

@@ -12,7 +12,7 @@ const banner = `
 
 const basePath = './src/schema'
 const outputPath = './src/models/generated'
-const schemaBits = ['schemas', 'schemasPublic', 'shared', 'parameters', 'requestBodies', 'responses', 'common']
+const schemaBits = ['schemas', 'schemasPublic', 'shared', 'parameters', 'requestBodies', 'responses', 'common', 'solr']
 const directories = fs
   .readdirSync(basePath)
   .filter(item => {

src/services/images/images-v1.class.js

@@ -12,7 +12,7 @@ const {
 } = require('../../solr')
 
 export default class Service {
-  constructor({ app = null, name = '' }) {
+  constructor({ app, name = '' }) {
     this.app = app
     this.name = name
     this.sequelizeClient = this.app.get('sequelizeClient')

src/services/images/images.class.ts

@@ -0,0 +1,86 @@
+import { NotFound } from '@feathersjs/errors'
+import { ClientService, Id, Params } from '@feathersjs/feathers'
+import { SimpleSolrClient } from '../../internalServices/simpleSolr'
+import { PublicFindResponse } from '../../models/common'
+import { Image } from '../../models/generated/schemas'
+import { Image as ImageDocument } from '../../models/generated/solr'
+import { SolrNamespaces } from '../../solr'
+
+const DefaultLimit = 10
+const ImageSimilarityVectorField: keyof ImageDocument = 'dinov2_emb_v1024'
+
+export interface FindQuery {
+  similar_to_image_id?: string
+  limit?: number
+  offset?: number
+}
+
+export class Images implements Pick<ClientService<Image, unknown, unknown, PublicFindResponse<Image>>, 'find' | 'get'> {
+  constructor(private readonly solrClient: SimpleSolrClient) {}
+
+  async find(params?: Params<FindQuery>): Promise<PublicFindResponse<Image>> {
+    const limit = params?.query?.limit ?? DefaultLimit
+    const offset = params?.query?.offset ?? 0
+
+    const queryParts: string[] = []
+
+    if (params?.query?.similar_to_image_id) {
+      const referenceId = params.query.similar_to_image_id
+      const referenceImage = await this.getImageDocument(referenceId, [ImageSimilarityVectorField])
+      const vector = referenceImage?.[ImageSimilarityVectorField] as number[]
+
+      if (referenceImage == null || vector == null)
+        return {
+          data: [],
+          pagination: {
+            limit: limit ?? 0,
+            offset: offset ?? 0,
+            total: 0,
+          },
+        }
+
+      queryParts.push(`{!knn f=${ImageSimilarityVectorField} topK=${limit}}${JSON.stringify(vector)}`)
+    }
+
+    const query = queryParts.length > 0 ? queryParts.join(' AND ') : '*:*'
+
+    const results = await this.solrClient.select<ImageDocument>(SolrNamespaces.Images, {
+      body: {
+        query,
+        limit,
+        offset,
+      },
+    })
+
+    return {
+      data: results?.response?.docs?.map(toImage) ?? [],
+      pagination: {
+        limit: 0,
+        offset: 0,
+        total: 0,
+      },
+    }
+  }
+
+  async get(id: Id, params?: Params): Promise<Image> {
+    const imageDoc = await this.getImageDocument(String(id))
+    if (imageDoc == null) throw new NotFound(`Image with id ${id} not found`)
+    return toImage(imageDoc)
+  }
+
+  async getImageDocument(id: string, fields?: (keyof ImageDocument)[]): Promise<ImageDocument | undefined> {
+    const result = await this.solrClient.selectOne<ImageDocument>(SolrNamespaces.Images, {
+      body: { query: `id:${id}`, limit: 1, fields: fields != null ? fields?.join(',') : undefined },
+    })
+    return result
+  }
+}
+
+const toImage = (doc: ImageDocument): Image => {
+  return {
+    uid: doc.id!,
+    ...(doc.linked_ci_s != null ? { contentItemUid: doc.linked_ci_s } : {}),
+    issueUid: doc.meta_issue_id_s!,
+    previewUrl: doc.iiif_url_s!,
+  }
+}

src/services/images/images.hooks.ts

@@ -0,0 +1,10 @@
+import { HookMap } from '@feathersjs/feathers'
+import { authenticateAround as authenticate } from '../../hooks/authenticate'
+import { rateLimit } from '../../hooks/rateLimiter'
+import { AppServices, ImpressoApplication } from '../../types'
+
+export default {
+  around: {
+    all: [authenticate({ allowUnauthenticated: true }), rateLimit()],
+  },
+} satisfies HookMap<ImpressoApplication, AppServices>