[#2917] Modify Solr to not use auto-created fields to prevent unneces…

…sary tokenization (#11323)
TEAMMATES · Aug 3, 2021 · c3def1a · c3def1a
1 parent 236913f
commit c3def1a
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 17 deletions.
diff --git a/solr/solr.sh b/solr/solr.sh
@@ -2,8 +2,12 @@
 
 # Create core to run students collection
 bin/solr create -c students -s 2 -rf 2
-curl -X POST -H "Content-Type: application/json" --data-binary '{"add-copy-field": {"source": "*", "dest": "_text_"}}' localhost:8983/solr/students/schema
+bin/solr config -c students -p 8983 -action set-user-property -property update.autoCreateFields -value false
+curl -X POST -H 'Content-type: application/json' --data-binary '{"add-field": {"name": "courseId", "type": "string"}}' localhost:8983/solr/students/schema
+curl -X POST -H 'Content-type: application/json' --data-binary '{"add-field": {"name": "email", "type": "string"}}' localhost:8983/solr/students/schema
 
 # Create core to run instructors collection
 bin/solr create -c instructors -s 2 -rf 2
-curl -X POST -H "Content-Type: application/json" --data-binary '{"add-copy-field": {"source": "*", "dest": "_text_"}}' localhost:8983/solr/instructors/schema
+bin/solr config -c instructors -p 8983 -action set-user-property -property update.autoCreateFields -value false
+curl -X POST -H 'Content-type: application/json' --data-binary '{"add-field": {"name": "courseId", "type": "string"}}' localhost:8983/solr/instructors/schema
+curl -X POST -H 'Content-type: application/json' --data-binary '{"add-field": {"name": "email", "type": "string"}}' localhost:8983/solr/instructors/schema
diff --git a/src/main/java/teammates/storage/search/InstructorSearchDocument.java b/src/main/java/teammates/storage/search/InstructorSearchDocument.java
@@ -22,15 +22,16 @@ class InstructorSearchDocument extends SearchDocument<InstructorAttributes> {
     Map<String, Object> getSearchableFields() {
         Map<String, Object> fields = new HashMap<>();
         InstructorAttributes instructor = attribute;
+        String[] searchableTexts = {
+                instructor.getName(), instructor.getEmail(), instructor.getCourseId(),
+                course == null ? "" : course.getName(),
+                instructor.getGoogleId(), instructor.getRole(), instructor.getDisplayedName(),
+        };
 
         fields.put("id", instructor.getEmail() + "%" + instructor.getCourseId());
-        fields.put("name", instructor.getName());
-        fields.put("email", instructor.getEmail());
+        fields.put("_text_", String.join(" ", searchableTexts));
         fields.put("courseId", instructor.getCourseId());
-        fields.put("courseName", course == null ? "" : course.getName());
-        fields.put("googleId", instructor.getGoogleId());
-        fields.put("role", instructor.getRole());
-        fields.put("displayedName", instructor.getDisplayedName());
+        fields.put("email", instructor.getEmail());
 
         return fields;
     }

diff --git a/src/main/java/teammates/storage/search/StudentSearchDocument.java b/src/main/java/teammates/storage/search/StudentSearchDocument.java
@@ -22,14 +22,16 @@ class StudentSearchDocument extends SearchDocument<StudentAttributes> {
     Map<String, Object> getSearchableFields() {
         Map<String, Object> fields = new HashMap<>();
         StudentAttributes student = attribute;
+        String[] searchableTexts = {
+                student.getName(), student.getEmail(), student.getCourse(),
+                course == null ? "" : course.getName(),
+                student.getTeam(), student.getSection(),
+        };
 
         fields.put("id", student.getId());
-        fields.put("name", student.getName());
-        fields.put("email", student.getEmail());
+        fields.put("_text_", String.join(" ", searchableTexts));
         fields.put("courseId", student.getCourse());
-        fields.put("courseName", course == null ? "" : course.getName());
-        fields.put("team", student.getTeam());
-        fields.put("section", student.getSection());
+        fields.put("email", student.getEmail());
 
         return fields;
     }

diff --git a/src/main/java/teammates/storage/search/StudentSearchManager.java b/src/main/java/teammates/storage/search/StudentSearchManager.java
@@ -85,10 +85,9 @@ public List<StudentAttributes> searchStudents(String queryString, List<Instructo
         QueryResponse response = performQuery(query);
         SolrDocumentList documents = response.getResults();
 
-        // Even though FQ has been applied, it may still match some unwanted results,
-        // e.g. if a course ID specified in FQ is the substring of another valid course.
-        // An additional filtering is done here such that only exact match will be returned.
-        // TODO a better way is to modify the field type in Solr instead of doing this
+        // Sanity check such that the course ID of the students match exactly.
+        // In ideal case, this check is not expected to do anything,
+        // i.e. the resulting list should be the same as the incoming list.
 
         List<SolrDocument> filteredDocuments = documents.stream()
                 .filter(document -> {