Merge pull request #53 from nulib/2392-oai-pmh

Implements the OAI-PMH protocol at the '/oai' endpoint.

src/api/opensearch.js

@@ -62,7 +62,7 @@ function initRequest(path) {
   });
 }
 
-async function search(targets, body) {
+async function search(targets, body, optionsQuery = {}) {
   const endpoint = elasticsearchEndpoint();
 
   const request = new HttpRequest({
@@ -74,9 +74,49 @@ async function search(targets, body) {
     },
     body: body,
     path: `/${targets}/_search`,
+    query: optionsQuery,
   });
 
   return await awsFetch(request);
 }
 
-module.exports = { getCollection, getFileSet, getSharedLink, getWork, search };
+async function scroll(scrollId) {
+  const endpoint = elasticsearchEndpoint();
+
+  const request = new HttpRequest({
+    method: "POST",
+    hostname: endpoint,
+    headers: {
+      Host: endpoint,
+      "Content-Type": "application/json",
+    },
+    body: JSON.stringify({ scroll: "2m" }),
+    path: `_search/scroll/${scrollId}`,
+  });
+  return await awsFetch(request);
+}
+
+async function deleteScroll(scrollId) {
+  const endpoint = elasticsearchEndpoint();
+
+  const request = new HttpRequest({
+    method: "DELETE",
+    hostname: endpoint,
+    headers: {
+      Host: endpoint,
+      "Content-Type": "application/json",
+    },
+    path: `_search/scroll/${scrollId}`,
+  });
+  return await awsFetch(request);
+}
+
+module.exports = {
+  getCollection,
+  getFileSet,
+  getSharedLink,
+  getWork,
+  search,
+  scroll,
+  deleteScroll,
+};

src/handlers/oai.js

@@ -0,0 +1,63 @@
+const { processRequest } = require("./middleware");
+const { baseUrl } = require("../helpers");
+const {
+  getRecord,
+  identify,
+  listIdentifiers,
+  listMetadataFormats,
+  listRecords,
+} = require("./oai/verbs");
+const { invalidOaiRequest } = require("./oai/xml-transformer");
+
+const allowedVerbs = [
+  "GetRecord",
+  "Identify",
+  "ListIdentifiers",
+  "ListMetadataFormats",
+  "ListRecords",
+  "ListSets",
+];
+
+/**
+ * A function to support the OAI-PMH harvesting specfication
+ */
+exports.handler = async (event) => {
+  event = processRequest(event);
+  const url = `${baseUrl(event)}oai`;
+  let verb, identifier, metadataPrefix, resumptionToken;
+  if (event.requestContext.http.method === "GET") {
+    verb = event.queryStringParameters?.verb;
+    identifier = event.queryStringParameters?.identifier;
+    metadataPrefix = event.queryStringParameters?.metadataPrefix;
+    resumptionToken = event.queryStringParameters?.resumptionToken;
+  } else {
+    const body = new URLSearchParams(event.body);
+    verb = body.get("verb");
+    identifier = body.get("identifier");
+    metadataPrefix = body.get("metadataPrefix");
+    resumptionToken = body.get("resumptionToken");
+  }
+
+  if (!verb) return invalidOaiRequest("badArgument", "Missing required verb");
+
+  switch (verb) {
+    case "GetRecord":
+      return await getRecord(url, identifier, metadataPrefix);
+    case "Identify":
+      return await identify(url);
+    case "ListIdentifiers":
+      return await listIdentifiers(url, event, metadataPrefix, resumptionToken);
+    case "ListMetadataFormats":
+      return await listMetadataFormats(url);
+    case "ListRecords":
+      return await listRecords(url, event, metadataPrefix, resumptionToken);
+    case "ListSets":
+      return invalidOaiRequest(
+        "noSetHierarchy",
+        "This repository does not support Sets",
+        401
+      );
+    default:
+      return invalidOaiRequest("badVerb", "Illegal OAI verb");
+  }
+};

src/handlers/oai/search.js

@@ -0,0 +1,56 @@
+const { search } = require("../../api/opensearch");
+const {
+  extractRequestedModels,
+  modelsToTargets,
+} = require("../../api/request/models");
+const fs = require("fs");
+
+async function earliestRecordCreateDate() {
+  const body = {
+    size: 1,
+    _source: "create_date",
+    query: {
+      bool: {
+        must: [
+          { term: { api_model: "Work" } },
+          { term: { published: true } },
+          { term: { visibility: "Public" } },
+        ],
+      },
+    },
+    sort: [{ create_date: "desc" }],
+  };
+  const esResponse = await search(
+    modelsToTargets(extractRequestedModels()),
+    JSON.stringify(body)
+  );
+  const responseBody = JSON.parse(esResponse.body);
+  return responseBody?.hits?.hits[0]?._source?.create_date;
+}
+
+async function oaiSearch() {
+  const body = {
+    size: 5000,
+    query: {
+      bool: {
+        must: [
+          { term: { api_model: "Work" } },
+          { term: { published: true } },
+          { term: { visibility: "Public" } },
+        ],
+      },
+    },
+    sort: [{ create_date: "desc" }],
+  };
+  const esResponse = await search(
+    modelsToTargets(extractRequestedModels()),
+    JSON.stringify(body),
+    { scroll: "2m" }
+  );
+  return {
+    ...esResponse,
+    expiration: new Date(new Date().getTime() + 2 * 60000).toISOString(),
+  };
+}
+
+module.exports = { earliestRecordCreateDate, oaiSearch };