Add IIIF content search route

Author: kdid

api/src/api/opensearch.js

@@ -27,6 +27,7 @@ async function getWorkFileSets(workId, opts = {}) {
   const {
     allowPrivate = false,
     allowUnpublished = false,
+    annotationsQuery = null,
     role = null,
     source = null,
     sortBy = null,
@@ -52,6 +53,9 @@ async function getWorkFileSets(workId, opts = {}) {
   if (role) {
     mustClauses.push({ term: { role: role } });
   }
+  if (annotationsQuery) {
+    mustClauses.push({ match: { "annotations.content": annotationsQuery } });
+  }
 
   const searchBody = {
     size: 10000,

api/src/api/response/iiif/manifest.js

@@ -368,6 +368,12 @@ async function transform(response, options = {}) {
       }
     }
 
+    jsonManifest.service = [
+      {
+        id: `${dcApiEndpoint()}/works/${source.id}/search?as=iiif`,
+        type: "SearchService2",
+      },
+    ];
     jsonManifest.provider = [provider];
     jsonManifest.logo = [nulLogo];
     const navPlace = buildNavPlace(source);

api/src/api/response/iiif/presentation-api/items.js

@@ -157,8 +157,10 @@ module.exports = {
   buildImageService,
   buildSupplementingAnnotation,
   buildTranscriptionAnnotation,
+  getTranscriptionContent,
   isAltFormat,
   isAudioVideo,
   isImage,
   isPDF,
+  normalizeLanguages,
 };

api/src/api/response/iiif/search.js

@@ -0,0 +1,101 @@
+const { dcApiEndpoint } = require("../../../environment");
+const { getWorkFileSets } = require("../../opensearch");
+const {
+  getTranscriptionContent,
+  normalizeLanguages,
+} = require("./presentation-api/items");
+
+function extractSnippet(content, q, contextChars = 100) {
+  const idx = content.toLowerCase().indexOf(q.toLowerCase());
+  if (idx === -1) return null;
+  const start = Math.max(0, idx - contextChars);
+  const end = Math.min(content.length, idx + q.length + contextChars);
+  let snippet = content.slice(start, end).trim();
+  if (start > 0) snippet = "..." + snippet;
+  if (end < content.length) snippet = snippet + "...";
+  return snippet;
+}
+
+function buildSearchAnnotationBody(annotation, snippet) {
+  const body = {
+    type: "TextualBody",
+    value: snippet,
+    format: "text/plain",
+  };
+  const languages = normalizeLanguages(annotation.language);
+  if (languages.length === 1) {
+    body.language = languages[0];
+  } else if (languages.length > 1) {
+    body.language = languages;
+  }
+  return body;
+}
+
+async function transform(workId, q, opts = {}) {
+  const { allowPrivate = false, allowUnpublished = false } = opts;
+
+  const manifestId = `${dcApiEndpoint()}/works/${workId}?as=iiif`;
+  const searchId = `${dcApiEndpoint()}/works/${workId}/search?as=iiif&q=${encodeURIComponent(q)}`;
+
+  const response = await getWorkFileSets(workId, {
+    allowPrivate,
+    allowUnpublished,
+    annotationsQuery: q,
+    role: "Access",
+    source: ["id", "annotations", "group_with"],
+    sortBy: "rank",
+  });
+
+  const fileSets =
+    response.statusCode === 200
+      ? JSON.parse(response.body).hits.hits.map((h) => h._source)
+      : [];
+
+  // Replicate manifest.js grouping: ungrouped file sets use their own id as key
+  const fileSetGroups = {};
+  fileSets.forEach((fs) => {
+    const key = fs.group_with || fs.id;
+    if (!fileSetGroups[key]) fileSetGroups[key] = [];
+    fileSetGroups[key].push(fs);
+  });
+
+  const items = [];
+
+  Object.entries(fileSetGroups).forEach(([groupKey, groupFileSets], index) => {
+    const canvasId = `${manifestId}/canvas/${index}`;
+
+    // Primary file set is the one whose id matches the group key (same as manifest.js)
+    const primary =
+      groupFileSets.find((fs) => fs.id === groupKey) || groupFileSets[0];
+    if (!primary?.annotations) return;
+
+    primary.annotations
+      .filter((ann) => ann.type === "transcription")
+      .forEach((ann) => {
+        const content = getTranscriptionContent(ann);
+        const snippet = extractSnippet(content, q);
+        if (!snippet) return;
+
+        items.push({
+          id: `${canvasId}/annotation/${ann.id}`,
+          type: "Annotation",
+          motivation: "supplementing",
+          body: buildSearchAnnotationBody(ann, snippet),
+          target: canvasId,
+        });
+      });
+  });
+
+  return {
+    statusCode: 200,
+    headers: { "content-type": "application/json" },
+    body: JSON.stringify({
+      "@context": "http://iiif.io/api/search/2/context.json",
+      id: searchId,
+      type: "AnnotationPage",
+      items,
+    }),
+  };
+}
+
+module.exports = { transform };

api/src/handlers/get-work-search.js

@@ -0,0 +1,29 @@
+const { getWork } = require("../api/opensearch");
+const iiifSearchResponse = require("../api/response/iiif/search");
+const { wrap } = require("./middleware");
+
+exports.handler = wrap(async (event) => {
+  const id = event.pathParameters.id;
+  const { as, q } = event.queryStringParameters;
+
+  const allowPrivate =
+    event.userToken.isSuperUser() ||
+    event.userToken.isReadingRoom() ||
+    event.userToken.hasEntitlement(id);
+  const allowUnpublished =
+    event.userToken.isSuperUser() || event.userToken.hasEntitlement(id);
+
+  if (as !== "iiif" || !q?.trim()) {
+    return {
+      statusCode: 400,
+      body: JSON.stringify({
+        message: "Request must include ?as=iiif&q={query}",
+      }),
+    };
+  }
+
+  const workResponse = await getWork(id, { allowPrivate, allowUnpublished });
+  if (workResponse.statusCode !== 200) return workResponse;
+
+  return iiifSearchResponse.transform(id, q, { allowPrivate, allowUnpublished });
+});

api/template.yaml

@@ -584,6 +584,30 @@ Resources:
             ApiId: !Ref dcApi
             Path: /works/{id}/thumbnail
             Method: HEAD
+  getWorkSearchFunction:
+    Type: AWS::Serverless::Function
+    Condition: DeployAPI
+    Properties:
+      Handler: handlers/get-work-search.handler
+      Description: IIIF Search 2.0 for a Work's transcription annotations.
+      #* Layers:
+      #*   - !Ref apiDependencies
+      Policies:
+        - !Ref SecretsPolicy
+        - !Ref readIndexPolicy
+      Events:
+        WorkApiGet:
+          Type: HttpApi
+          Properties:
+            ApiId: !Ref dcApi
+            Path: /works/{id}/search
+            Method: GET
+        WorkApiHead:
+          Type: HttpApi
+          Properties:
+            ApiId: !Ref dcApi
+            Path: /works/{id}/search
+            Method: HEAD
   getSimilarFunction:
     Type: AWS::Serverless::Function
     Condition: DeployAPI

api/test/integration/get-work-by-id.test.js

@@ -91,6 +91,10 @@ describe("Retrieve work by id", () => {
         "http://iiif.io/api/presentation/3/context.json"
       );
       expect(resultBody.label.none[0]).to.eq("Canary Record TEST 1");
+      expect(resultBody.service).to.deep.include({
+        id: `${process.env.DC_API_ENDPOINT}/works/1234/search?as=iiif`,
+        type: "SearchService2",
+      });
     });
 
     it("will retrieve a private, unpublished work document with an entitlement", async () => {