Merge pull request #261 from UoA-eResearch/rc2.6.1

Rc2.6.1
UoA-eResearch · Mar 21, 2022 · ba434d4 · ba434d4
2 parents 9f3905a + 2dc1d38
commit ba434d4
Show file tree

Hide file tree

Showing 62 changed files with 8,939 additions and 12,694 deletions.
diff --git a/Jenkinsfile b/Jenkinsfile
@@ -296,14 +296,21 @@ pipeline {
                         }
                     }
                     steps {
-                        echo 'Testing research-hub-web project'
+                        script {
+                            echo 'Testing research-hub-web project'
+
+                            // need to trim the trailing slash from the graphql server url for the e2e tests
+                            def graphqlServer = env.SCHEMA_PATH.substring(0, env.SCHEMA_PATH.length() - 1)
 
-                        dir("research-hub-web") {
-                            echo 'Running research-hub-web unit tests'
-                            sh 'npm run test-ci'
+                            dir("research-hub-web") {
+                                echo 'Running research-hub-web unit tests'
+                                sh 'npm run test-ci'
 
-                            echo 'Running research-hub-web e2e tests'
-                            sh "npm run e2e-ci"
+                                echo 'Running research-hub-web e2e tests'
+                                // set the graphql server url as an env variable for Cypress
+                                // for intercepting some of the graphql queries and returning mocked data
+                                sh "export cypress_graphql_server=${graphqlServer} && npm run e2e-ci"
+                            }
                         }
                     }
                 }

diff --git a/cer-graphql/package-lock.json b/cer-graphql/package-lock.json
diff --git a/cer-graphql/package.json b/cer-graphql/package.json
@@ -1,6 +1,6 @@
 {
   "name": "cer-graphql",
-  "version": "2.5.1",
+  "version": "2.6.1",
   "description": "A GraphQL server used to proxy and authorise requests to external data sources.",
   "main": "build/index.js",
   "scripts": {
@@ -35,7 +35,7 @@
     "@types/jest": "^27.4.0",
     "@types/jsonwebtoken": "^8.5.8",
     "@types/jwk-to-pem": "^2.0.0",
-    "aws-sdk": "^2.1058.0",
+    "aws-sdk": "^2.1095.0",
     "aws4": "^1.10.1",
     "jest": "^27.4.7",
     "nodemon": "^2.0.15",

diff --git a/hub-search-proxy/handler.js b/hub-search-proxy/handler.js
@@ -200,23 +200,16 @@ module.exports.search = async (event, context) => {
 
       let formattedQueryString = queryString;
 
-      if (queryString.split(' ').length < 5) {
-        // format the query string for fuzzy search if the query contains less than 5 words. 
+      // format the query string for fuzzy search if the query contains less than 5 words.
+      // Also if the query string contains query operators dont make it fuzzy
+      if (queryString.split(' ').length < 5 && !containsOperators(queryString)) {
         // modify degree of fuzziness here. 
         // Note: '~AUTO' is not supported currently despite what the docs say.
-        const fuzziness = '~2'; 
+        const fuzziness = '~2';
 
-        formattedQueryString = queryString.split(' ').map(s => {
-          // if word contains query operators dont make it fuzzy
+        formattedQueryString = queryString.split(' ').map(s => {          
           // if word is less than 3 letters dont make it fuzzy
-          const queryOperators = ['+', '|', '-', '*', '"', '(', ')'];
-          let matchCount = 0;
-          for (let i = 0; i < queryOperators.length; i++) {
-            if (s.indexOf(queryOperators[i]) > -1) {
-            matchCount++;
-            }
-          };
-          return (matchCount > 0 || s.length < 3) ? s : s + fuzziness; 
+          return s.length < 3 ? s : s + fuzziness; 
         }).join(' ');
       }
 
@@ -252,20 +245,34 @@ module.exports.search = async (event, context) => {
           includes: includeInResult
         },
         query: {
-          bool: {
-            must: simpleQuery,
-            minimum_should_match: minimum_should_match,
-            should: queryParts,
-            filter: [
+          function_score: {
+            query: {
+              bool: {
+                must: simpleQuery,
+                minimum_should_match: minimum_should_match,
+                should: queryParts,
+                filter: [
+                  {
+                    term: {
+                      "fields.searchable.en-US": true
+                    }
+                  },
+                  {
+                    terms: {
+                      "sys.contentType.sys.id": contentTypes
+                    }
+                  }
+                ]
+              }
+            },
+            functions: [
               {
-                term: {
-                  "fields.searchable.en-US": true
-                }
+                filter: {"match":{"fields.title.en-US": queryString}},
+                weight: 2 // boost match score for titles that contain the individual query terms
               },
               {
-                terms: {
-                  "sys.contentType.sys.id": contentTypes
-                }
+                filter: {"match":{"fields.title.en-US.raw": queryString}},
+                weight: 2 // boost match score for titles that contain the exact query terms (the 'raw' field is a keyword field meaning it is not tokenised)
               }
             ]
           }
@@ -484,3 +491,8 @@ function formatResponse(status, body) {
       }
   }
 }
+
+function containsOperators(queryString) {
+  const queryOperators = ['+', '|', '-', '*', '"', '(', ')'];
+  return queryOperators.some(operator => queryString.includes(operator));
+}