From 298ceb3804d60486956e4dfc6551677380754699 Mon Sep 17 00:00:00 2001
From: Karen Shaw <karendid@gmail.com>
Date: Wed, 9 Oct 2024 19:59:24 +0000
Subject: [PATCH] Allow _source_includes and _source_excludes for GET search
 requests

---
 docs/docs/spec/types.yaml            | 14 ++++++++
 node/src/handlers/search-runner.js   | 21 +++++++++++-
 node/test/integration/search.test.js | 50 ++++++++++++++++++++++++++++
 3 files changed, 84 insertions(+), 1 deletion(-)

diff --git a/docs/docs/spec/types.yaml b/docs/docs/spec/types.yaml
index 41050ebb..119bdeea 100644
--- a/docs/docs/spec/types.yaml
+++ b/docs/docs/spec/types.yaml
@@ -60,6 +60,20 @@ components:
       description: Comma-delimited list of fields to sort search results (e.g. "create_date:asc,modified_date:desc")
       schema:
         type: string
+    _source_excludes:
+      name: _source_excludes
+      in: query
+      required: false
+      description: Comma-delimited list of fields to exclude from search results (e.g. "embedding,embedding_text_length")
+      schema:
+        type: string
+    _source_includes:
+      name: _source_includes
+      in: query
+      required: false
+      description: Comma-delimited list of fields to include in search results (e.g. "title,accession_number")
+      schema:
+        type: string
     as:
       name: as
       in: query
diff --git a/node/src/handlers/search-runner.js b/node/src/handlers/search-runner.js
index f7063214..c5b3a491 100644
--- a/node/src/handlers/search-runner.js
+++ b/node/src/handlers/search-runner.js
@@ -96,6 +96,26 @@ const constructSearchContext = async (event) => {
   searchContext.size = queryStringParameters.size || searchContext.size || 10;
   searchContext.from = queryStringParameters.from || searchContext.from || 0;
 
+  if (
+    queryStringParameters?._source_excludes ||
+    searchContext._source?.exclude
+  ) {
+    searchContext._source = searchContext._source || {};
+    searchContext._source.exclude =
+      queryStringParameters?._source_excludes.split(",") ||
+      searchContext._source.exclude;
+  }
+
+  if (
+    queryStringParameters?._source_includes ||
+    searchContext._source?.include
+  ) {
+    searchContext._source = searchContext._source || {};
+    searchContext._source.include =
+      queryStringParameters?._source_includes.split(",") ||
+      searchContext._source.include;
+  }
+
   if (queryStringParameters?.sort || searchContext.sort)
     searchContext.sort =
       parseSortParameter(queryStringParameters) || searchContext.sort;
@@ -104,7 +124,6 @@ const constructSearchContext = async (event) => {
     const page = Number(queryStringParameters.page || 1);
     searchContext.from = (page - 1) * searchContext.size;
   }
-
   return searchContext;
 };
 
diff --git a/node/test/integration/search.test.js b/node/test/integration/search.test.js
index febe9576..1fbaa25d 100644
--- a/node/test/integration/search.test.js
+++ b/node/test/integration/search.test.js
@@ -206,5 +206,55 @@ describe("Search routes", () => {
         "?sort=create_date%3Aasc%2Cmodified_date%3Adesc"
       );
     });
+
+    it("allows excluding fields via query string parameters for GET requests", async () => {
+      const originalQuery = {
+        query: { query_string: { query: "*" } },
+        _source: { exclude: ["title"] },
+      };
+      const event = helpers
+        .mockEvent("GET", "/search")
+        .queryParams({ _source_excludes: "title" })
+        .render();
+      const authQuery = new RequestPipeline(originalQuery)
+        .authFilter(helpers.preprocess(event))
+        .toJson();
+
+      mock
+        .post("/dc-v2-work/_search", authQuery)
+        .reply(200, helpers.testFixture("mocks/search.json"));
+
+      const result = await handler(event);
+      expect(result.statusCode).to.eq(200);
+      const resultBody = JSON.parse(result.body);
+      expect(resultBody.pagination.query_url).to.contain(
+        "?_source_excludes=title"
+      );
+    });
+
+    it("allows including fields via query string parameters for GET requests", async () => {
+      const originalQuery = {
+        query: { query_string: { query: "*" } },
+        _source: { include: ["title", "accession_number"] },
+      };
+      const event = helpers
+        .mockEvent("GET", "/search")
+        .queryParams({ _source_includes: "title,accession_number" })
+        .render();
+      const authQuery = new RequestPipeline(originalQuery)
+        .authFilter(helpers.preprocess(event))
+        .toJson();
+
+      mock
+        .post("/dc-v2-work/_search", authQuery)
+        .reply(200, helpers.testFixture("mocks/search.json"));
+
+      const result = await handler(event);
+      expect(result.statusCode).to.eq(200);
+      const resultBody = JSON.parse(result.body);
+      expect(resultBody.pagination.query_url).to.contain(
+        "?_source_includes=title%2Caccession_number"
+      );
+    });
   });
 });