add fts to simple collection/institution suggestions #652

registry-integration-tests/src/test/java/org/gbif/registry/ws/it/collections/service/InstitutionServiceIT.java

@@ -647,6 +647,8 @@ public void testSuggest() {
     assertEquals(2, institutionService.suggest("II").size());
     assertEquals(1, institutionService.suggest("II2").size());
     assertEquals(1, institutionService.suggest("name2").size());
+    assertEquals(1, institutionService.suggest("Institution name2").size());
+    assertDoesNotThrow(() -> institutionService.suggest(""));
   }
 
   @Test

registry-persistence/src/main/resources/liquibase/153-institution-suggest-fts.xml

@@ -0,0 +1,13 @@
+<databaseChangeLog
+  xmlns="http://www.liquibase.org/xml/ns/dbchangelog"
+  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+  xsi:schemaLocation="http://www.liquibase.org/xml/ns/dbchangelog http://www.liquibase.org/xml/ns/dbchangelog/dbchangelog-2.0.xsd">
+
+  <changeSet id="153" author="auzun" runInTransaction="false">
+    <sql splitStatements="false" stripComments="false">
+      <![CDATA[
+        CREATE INDEX idx_institution_fulltext_search ON institution USING gin (to_tsvector('english', code || ' ' || name));
+      ]]>
+    </sql>
+  </changeSet>
+</databaseChangeLog>

registry-persistence/src/main/resources/liquibase/master.xml

@@ -158,4 +158,5 @@
   <include file="liquibase/150-similar-name-list-function.xml" />
   <include file="liquibase/151-indexes-collections-contacts.xml" />
   <include file="liquibase/152-function-search-paths.xml" />
+  <include file="liquibase/153-institution-suggest-fts.xml" />
 </databaseChangeLog>

registry-persistence/src/main/resources/org/gbif/registry/persistence/mapper/collections/CollectionMapper.xml

@@ -560,15 +560,24 @@
 
   <!-- Simple suggest -->
   <select id="suggest" resultType="org.gbif.api.model.registry.search.collections.KeyCodeNameResult">
-    SELECT key,code,name FROM (
-    (SELECT key,code,name FROM collection WHERE code ilike #{q}||'%' AND deleted IS NULL ORDER BY code LIMIT 20)
-    UNION ALL
-    (SELECT key,code,name FROM collection WHERE name ilike #{q}||'%' AND deleted IS NULL ORDER BY name LIMIT 20)
-    UNION ALL
-    (SELECT key,code,name FROM collection WHERE code ilike '%'||#{q}||'%' AND NOT code ilike #{q}||'%' AND deleted IS NULL ORDER BY code LIMIT 20)
-    UNION ALL
-    (SELECT key,code,name FROM collection WHERE name ilike '%'||#{q}||'%' AND NOT name ilike #{q}||'%' AND deleted IS NULL ORDER BY name LIMIT 20)
-    ) t1 LIMIT 20
+    <if test="q != null and !q.isEmpty()">
+      SELECT key, code, name
+      FROM (
+      SELECT key, code, name,
+      to_tsvector('english', code || ' ' || name) AS document,
+      to_tsquery('english', unaccent(regexp_replace(#{q} || ':*', ' ', '&amp;', 'g'))) AS query
+      FROM collection
+      WHERE deleted IS NULL
+      ) AS subquery
+      WHERE document @@ query
+      ORDER BY ts_rank(document, query) DESC
+      LIMIT 20;
+    </if>
+    <if test="q == null or q.isEmpty()">
+      SELECT key, code, name FROM collection
+      WHERE deleted IS NULL
+      LIMIT 20;
+    </if>
   </select>
 
   <!-- COMMENTS -->

registry-persistence/src/main/resources/org/gbif/registry/persistence/mapper/collections/InstitutionMapper.xml

@@ -549,15 +549,24 @@
 
   <!-- Simple suggest -->
   <select id="suggest" resultType="org.gbif.api.model.registry.search.collections.KeyCodeNameResult">
-    SELECT key,code,name FROM (
-    (SELECT key,code,name FROM institution WHERE code ilike #{q}||'%' AND deleted IS NULL ORDER BY code LIMIT 20)
-    UNION ALL
-    (SELECT key,code,name FROM institution WHERE name ilike #{q}||'%' AND deleted IS NULL ORDER BY name LIMIT 20)
-    UNION ALL
-    (SELECT key,code,name FROM institution WHERE code ilike '%'||#{q}||'%' AND NOT code ilike #{q}||'%' AND deleted IS NULL ORDER BY code LIMIT 20)
-    UNION ALL
-    (SELECT key,code,name FROM institution WHERE name ilike '%'||#{q}||'%' AND NOT name ilike #{q}||'%' AND deleted IS NULL ORDER BY name LIMIT 20)
-    ) t1 LIMIT 20
+    <if test="q != null and !q.isEmpty()">
+      SELECT key, code, name
+      FROM (
+        SELECT key, code, name,
+        to_tsvector('english', code || ' ' || name) AS document,
+        to_tsquery('english', unaccent(regexp_replace(#{q} || ':*', ' ', '&amp;', 'g'))) AS query
+        FROM institution
+        WHERE deleted IS NULL
+      ) AS subquery
+      WHERE document @@ query
+      ORDER BY ts_rank(document, query) DESC
+      LIMIT 20;
+    </if>
+    <if test="q == null or q.isEmpty()">
+      SELECT key, code, name FROM institution
+      WHERE deleted IS NULL
+      LIMIT 20;
+    </if>
   </select>
 
   <!-- COMMENTS -->