From 6eeef70dd8adeda64fb9884522eca755ed02e6f9 Mon Sep 17 00:00:00 2001
From: Paul Pilone <paul@element84.com>
Date: Tue, 23 Aug 2022 22:23:58 -0400
Subject: [PATCH] Hacky fix for DOI searches.

---
 api/lambdas/search-by-keywords.test.ts | 23 ++++++++++++-----------
 api/lib/search-query-builder.ts        | 14 +++++++++++---
 lib/documents-mapping.ts               |  3 +++
 lib/open-search-client.test.ts         |  3 +++
 4 files changed, 29 insertions(+), 14 deletions(-)

diff --git a/api/lambdas/search-by-keywords.test.ts b/api/lambdas/search-by-keywords.test.ts
index 07d87bf9..48e62913 100644
--- a/api/lambdas/search-by-keywords.test.ts
+++ b/api/lambdas/search-by-keywords.test.ts
@@ -62,8 +62,7 @@ describe('search-by-keywords.handler', () => {
     const uuid1 = randomUUID();
     const uuid2 = randomUUID();
     const uuid3 = randomUUID();
-
-    const doiUUID = randomUUID();
+    const uuid4 = randomUUID();
 
     beforeAll(async () => {
       const doc1 = {
@@ -81,28 +80,30 @@ describe('search-by-keywords.handler', () => {
         uuid: uuid3,
         dc_title: 'This is a document with author text.',
         dc_contributor_author: 'Ocean Sea',
-        dc_identifier_uri: [
-          'http://hdl.handle.net/11329/1029',
+        dc_identifier_doi: [
           'http://dx.doi.org/10.25607/OBP-561',
         ],
       };
 
+      const doc4 = {
+        uuid: uuid4,
+        dc_identifier_doi: [
+          'http://dx.doi.org/10.25607/OBP-765',
+        ],
+      };
+
       await osClient.addDocument(esUrl, documentsIndexName, doc1);
       await osClient.addDocument(esUrl, documentsIndexName, doc2);
       await osClient.addDocument(esUrl, documentsIndexName, doc3);
+      await osClient.addDocument(esUrl, documentsIndexName, doc4);
       await osClient.refreshIndex(esUrl, documentsIndexName);
     });
 
     afterAll(async () => {
-      await osClient.deleteByQuery(
-        esUrl,
-        documentsIndexName,
-        { match: { uuid: doiUUID } }
-      );
-
       await osClient.deleteByQuery(esUrl, documentsIndexName, { match: { uuid: uuid1 } });
       await osClient.deleteByQuery(esUrl, documentsIndexName, { match: { uuid: uuid2 } });
       await osClient.deleteByQuery(esUrl, documentsIndexName, { match: { uuid: uuid3 } });
+      await osClient.deleteByQuery(esUrl, documentsIndexName, { match: { uuid: uuid4 } });
       await osClient.refreshIndex(esUrl, documentsIndexName);
     });
 
@@ -199,7 +200,7 @@ describe('search-by-keywords.handler', () => {
     test('should find documents with the DOI metadata field', (done) => {
       const proxyEvent = {
         queryStringParameters: {
-          keywords: ':dc_identifier_uri:10.25607/OBP-561',
+          keywords: ':dc_identifier_doi:10.25607/OBP-561',
         },
       };
 
diff --git a/api/lib/search-query-builder.ts b/api/lib/search-query-builder.ts
index 11836812..bb16919f 100644
--- a/api/lib/search-query-builder.ts
+++ b/api/lib/search-query-builder.ts
@@ -36,7 +36,12 @@ export const nestedQuery = (termPhrase: unknown) => ({
 // All special characters: + - = && || > < ! ( ) { } [ ] ^ " ~ * ? : \ /
 // Characters we currently want to escape: + - = && || ! ( ) { } [ ] : \ /
 const queryStringSpecialCharacters = /\+|-|=|&{2}|\|{2}|!|\(|\)|{|}|\[|]|:|\/|\\/g;
-const encodeQueryStringTermComp = (term: string): string => term.replace(queryStringSpecialCharacters, '\\$&');
+const encodeQueryStringTerm = (term: string, field: string): string => {
+  const encodedTerm = term.replace(queryStringSpecialCharacters, '\\$&');
+
+  // FIXME: This is a hack to make doi searches work. Fix this better please.
+  return field === 'dc_identifier_doi' ? `*${encodedTerm}` : encodedTerm;
+};
 
 const formatKeywordComp = (keywordComp: SearchKeywordComps) => {
   let openSearchOperator;
@@ -52,8 +57,11 @@ const formatKeywordComp = (keywordComp: SearchKeywordComps) => {
       openSearchOperator = 'OR';
   }
 
-  const escapedKeywordCompTerm = encodeQueryStringTermComp(keywordComp.term);
-  return `${openSearchOperator} ${keywordComp.field}:(${escapedKeywordCompTerm})`;
+  const encodedKeywordCompTerm = encodeQueryStringTerm(
+    keywordComp.term,
+    keywordComp.field
+  );
+  return `${openSearchOperator} ${keywordComp.field}:(${encodedKeywordCompTerm})`;
 };
 
 /**
diff --git a/lib/documents-mapping.ts b/lib/documents-mapping.ts
index cc03ca1b..6995bb57 100644
--- a/lib/documents-mapping.ts
+++ b/lib/documents-mapping.ts
@@ -85,6 +85,9 @@ export const documentsMapping = {
         type: 'text',
         analyzer: 'english',
       },
+      dc_identifier_doi: {
+        type: 'keyword',
+      },
     },
   },
 };
diff --git a/lib/open-search-client.test.ts b/lib/open-search-client.test.ts
index d8666d21..32cfe512 100644
--- a/lib/open-search-client.test.ts
+++ b/lib/open-search-client.test.ts
@@ -404,6 +404,9 @@ describe('open-search-client', () => {
           uuid: {
             type: 'keyword',
           },
+          dc_identifier_doi: {
+            type: 'keyword',
+          },
         },
       });
     });