Add full text search index by entity annotations

docs/CHANGELOG.rst

@@ -20,6 +20,12 @@ Added
 -----
 - Add unaccented full text search configuration ``simple_unaccent``
 
+Changed
+-------
+- Use ``simple_unaccent`` full text search configuration instead of ``simple``
+- Use ``simple_unaccent`` full text search configuration instead of ``simple``
+- Create ``simple_unaccent`` full text search index in ``PostgreSQL``
+
 
 ===================
 38.3.2 - 2024-01-25

resolwe/flow/migrations/0023_create_unaccent_full_text_search_config.py

@@ -11,7 +11,15 @@ class Migration(migrations.Migration):
 
     operations = [
         migrations.RunSQL(
-            "CREATE TEXT SEARCH CONFIGURATION simple_unaccent( COPY = simple );"
+            """
+            DO
+            $$BEGIN
+                CREATE TEXT SEARCH CONFIGURATION simple_unaccent( COPY = simple );
+            EXCEPTION
+                WHEN unique_violation THEN
+                    NULL;  -- ignore error
+            END;$$;
+            """
         ),
         migrations.RunSQL(
             "ALTER TEXT SEARCH CONFIGURATION simple_unaccent "

resolwe/flow/migrations/0024_use_unaccent_full_text_search.py

@@ -0,0 +1,35 @@
+# Generated by Django 4.2.9 on 2024-02-13 07:33
+
+import os
+
+from django.db import connection, migrations
+
+
+def load_triggers(apps, schema_editor):
+    file_names = [
+        "utils.sql",
+        "triggers_collection.sql",
+        "triggers_entity.sql",
+        "triggers_data.sql",
+    ]
+    with connection.cursor() as c:
+        for file_name in file_names:
+            file_path = os.path.join(os.path.dirname(__file__), file_name)
+            with open(file_path) as fh:
+                sql_statement = fh.read()
+            c.execute(sql_statement)
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("flow", "0023_create_unaccent_full_text_search_config"),
+    ]
+
+    operations = [
+        migrations.RunPython(load_triggers),
+        # Update existing entries.
+        migrations.RunSQL("UPDATE flow_entity SET id=id;", migrations.RunSQL.noop),
+        migrations.RunSQL("UPDATE flow_collection SET id=id;", migrations.RunSQL.noop),
+        migrations.RunSQL("UPDATE flow_data SET id=id;", migrations.RunSQL.noop),
+    ]

resolwe/flow/migrations/triggers_collection.sql

@@ -30,33 +30,33 @@ CREATE OR REPLACE FUNCTION generate_resolwe_collection_search(collection flow_co
 
         SELECT
             -- Collection name.
-            setweight(to_tsvector('simple', collection.name), 'A') ||
-            setweight(to_tsvector('simple', get_characters(collection.name)), 'B') ||
-            setweight(to_tsvector('simple', get_numbers(collection.name)), 'B') ||
+            setweight(to_tsvector('simple_unaccent', collection.name), 'A') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(collection.name)), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(collection.name)), 'B') ||
             -- Collection description.
-            setweight(to_tsvector('simple', collection.description), 'B') ||
+            setweight(to_tsvector('simple_unaccent', collection.description), 'B') ||
             -- Contributor username.
-            setweight(to_tsvector('simple', contributor.usernames), 'B') ||
-            setweight(to_tsvector('simple', get_characters(contributor.usernames)), 'C') ||
-            setweight(to_tsvector('simple', get_numbers(contributor.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', contributor.usernames), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(contributor.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(contributor.usernames)), 'C') ||
             -- Contributor first name.
-            setweight(to_tsvector('simple', contributor.first_names), 'B') ||
+            setweight(to_tsvector('simple_unaccent', contributor.first_names), 'B') ||
             -- Contributor last name.
-            setweight(to_tsvector('simple', contributor.last_names), 'B') ||
+            setweight(to_tsvector('simple_unaccent', contributor.last_names), 'B') ||
             -- Owners usernames. There is no guarantee that it is not NULL.
-            setweight(to_tsvector('simple', COALESCE(owners.usernames, '')), 'B') ||
-            setweight(to_tsvector('simple', get_characters(owners.usernames)), 'C') ||
-            setweight(to_tsvector('simple', get_numbers(owners.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', COALESCE(owners.usernames, '')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(owners.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(owners.usernames)), 'C') ||
             -- Owners first names. There is no guarantee that it is not NULL.
-            setweight(to_tsvector('simple', COALESCE(owners.first_names, '')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', COALESCE(owners.first_names, '')), 'B') ||
             -- Owners last names. There is no guarantee that it is not NULL.
-            setweight(to_tsvector('simple', COALESCE(owners.last_names, '')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', COALESCE(owners.last_names, '')), 'B') ||
             -- Collection tags.
-            setweight(to_tsvector('simple', array_to_string(collection.tags, ' ')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', array_to_string(collection.tags, ' ')), 'B') ||
             -- Collection descriptor.
-            setweight(to_tsvector('simple', flat_descriptor), 'C') ||
-            setweight(to_tsvector('simple', get_characters(flat_descriptor)), 'D') ||
-            setweight(to_tsvector('simple', get_numbers(flat_descriptor)), 'D')
+            setweight(to_tsvector('simple_unaccent', flat_descriptor), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(flat_descriptor)), 'D') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(flat_descriptor)), 'D')
 
         INTO search;
 
@@ -79,11 +79,14 @@ CREATE OR REPLACE FUNCTION collection_biut()
     END;
     $$;
 
-CREATE TRIGGER collection_biut
-    BEFORE INSERT OR UPDATE
-    ON flow_collection
-    FOR EACH ROW EXECUTE PROCEDURE collection_biut();
-
+DO $$ BEGIN
+    CREATE TRIGGER collection_biut
+        BEFORE INSERT OR UPDATE
+        ON flow_collection
+        FOR EACH ROW EXECUTE PROCEDURE collection_biut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 -- Trigger after update/insert/delete user permission object.
 CREATE OR REPLACE FUNCTION handle_userpermission_collection(user_permission permissions_permissionmodel)
@@ -114,10 +117,14 @@ CREATE OR REPLACE FUNCTION userpermission_collection_aiut()
     END;
     $$;
 
-CREATE TRIGGER userpermission_collection_aiut
-    AFTER INSERT OR UPDATE
-    ON permissions_permissionmodel
-    FOR EACH ROW EXECUTE PROCEDURE userpermission_collection_aiut();
+DO $$ BEGIN
+    CREATE TRIGGER userpermission_collection_aiut
+        AFTER INSERT OR UPDATE
+        ON permissions_permissionmodel
+        FOR EACH ROW EXECUTE PROCEDURE userpermission_collection_aiut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 CREATE OR REPLACE FUNCTION userpermission_collection_adt()
     RETURNS TRIGGER
@@ -129,12 +136,15 @@ CREATE OR REPLACE FUNCTION userpermission_collection_adt()
     END;
     $$;
 
-CREATE TRIGGER userpermission_collection_adt
-    AFTER DELETE
-    -- ON guardian_userobjectpermission
-    ON permissions_permissionmodel
-    FOR EACH ROW EXECUTE PROCEDURE userpermission_collection_adt();
-
+DO $$ BEGIN
+    CREATE TRIGGER userpermission_collection_adt
+        AFTER DELETE
+        -- ON guardian_userobjectpermission
+        ON permissions_permissionmodel
+        FOR EACH ROW EXECUTE PROCEDURE userpermission_collection_adt();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 -- Trigger after update contributor.
 CREATE OR REPLACE FUNCTION collection_contributor_aut()
@@ -149,11 +159,14 @@ CREATE OR REPLACE FUNCTION collection_contributor_aut()
     END;
     $$;
 
-CREATE TRIGGER collection_contributor_aut
-    AFTER UPDATE
-    ON auth_user
-    FOR EACH ROW EXECUTE PROCEDURE collection_contributor_aut();
-
+DO $$ BEGIN
+    CREATE TRIGGER collection_contributor_aut
+        AFTER UPDATE
+        ON auth_user
+        FOR EACH ROW EXECUTE PROCEDURE collection_contributor_aut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 -- Trigger after update owner.
 CREATE OR REPLACE FUNCTION collection_owner_aut()
@@ -178,7 +191,11 @@ CREATE OR REPLACE FUNCTION collection_owner_aut()
     END;
     $$;
 
-CREATE TRIGGER collection_owner_aut
-    AFTER UPDATE
-    ON auth_user
-    FOR EACH ROW EXECUTE PROCEDURE collection_owner_aut();
+DO $$ BEGIN
+    CREATE TRIGGER collection_owner_aut
+        AFTER UPDATE
+        ON auth_user
+        FOR EACH ROW EXECUTE PROCEDURE collection_owner_aut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;

resolwe/flow/migrations/triggers_data.sql

@@ -1,8 +1,12 @@
 -- Trigger after insert/update Data object.
-CREATE TYPE process_result AS (
-    name text,
-    type text
-);
+DO $$ BEGIN
+    CREATE TYPE process_result AS (
+        name text,
+        type text
+    );
+EXCEPTION
+    WHEN duplicate_object THEN null;
+END $$;
 
 CREATE OR REPLACE FUNCTION generate_resolwe_data_search(data flow_data)
     RETURNS tsvector
@@ -38,37 +42,37 @@ CREATE OR REPLACE FUNCTION generate_resolwe_data_search(data flow_data)
 
         SELECT
             -- Data name.
-            setweight(to_tsvector('simple', data.name), 'A') ||
-            setweight(to_tsvector('simple', get_characters(data.name)), 'B') ||
-            setweight(to_tsvector('simple', get_numbers(data.name)), 'B') ||
+            setweight(to_tsvector('simple_unaccent', data.name), 'A') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(data.name)), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(data.name)), 'B') ||
             -- Contributor username.
-            setweight(to_tsvector('simple', contributor.usernames), 'B') ||
-            setweight(to_tsvector('simple', get_characters(contributor.usernames)), 'C') ||
-            setweight(to_tsvector('simple', get_numbers(contributor.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', contributor.usernames), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(contributor.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(contributor.usernames)), 'C') ||
             -- Contributor first name.
-            setweight(to_tsvector('simple', contributor.first_names), 'B') ||
+            setweight(to_tsvector('simple_unaccent', contributor.first_names), 'B') ||
             -- Contributor last name.
-            setweight(to_tsvector('simple', contributor.last_names), 'B') ||
+            setweight(to_tsvector('simple_unaccent', contributor.last_names), 'B') ||
             -- Owners usernames. There is no guarantee that it is not NULL.
-            setweight(to_tsvector('simple', COALESCE(owners.usernames, '')), 'B') ||
-            setweight(to_tsvector('simple', get_characters(owners.usernames)), 'C') ||
-            setweight(to_tsvector('simple', get_numbers(owners.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', COALESCE(owners.usernames, '')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(owners.usernames)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(owners.usernames)), 'C') ||
             -- Owners first names. There is no guarantee that it is not NULL.
-            setweight(to_tsvector('simple', COALESCE(owners.first_names, '')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', COALESCE(owners.first_names, '')), 'B') ||
             -- Owners last names. There is no guarantee that it is not NULL.
-            setweight(to_tsvector('simple', COALESCE(owners.last_names, '')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', COALESCE(owners.last_names, '')), 'B') ||
             -- Process name.
-            setweight(to_tsvector('simple', process.name), 'B') ||
-            setweight(to_tsvector('simple', get_characters(process.name)), 'C') ||
-            setweight(to_tsvector('simple', get_numbers(process.name)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', process.name), 'B') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(process.name)), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(process.name)), 'C') ||
             -- Process type.
-            setweight(to_tsvector('simple', process.type), 'D') ||
+            setweight(to_tsvector('simple_unaccent', process.type), 'D') ||
             -- Data tags.
-            setweight(to_tsvector('simple', array_to_string(data.tags, ' ')), 'B') ||
+            setweight(to_tsvector('simple_unaccent', array_to_string(data.tags, ' ')), 'B') ||
             -- Data descriptor.
-            setweight(to_tsvector('simple', flat_descriptor), 'C') ||
-            setweight(to_tsvector('simple', get_characters(flat_descriptor)), 'D') ||
-            setweight(to_tsvector('simple', get_numbers(flat_descriptor)), 'D')
+            setweight(to_tsvector('simple_unaccent', flat_descriptor), 'C') ||
+            setweight(to_tsvector('simple_unaccent', get_characters(flat_descriptor)), 'D') ||
+            setweight(to_tsvector('simple_unaccent', get_numbers(flat_descriptor)), 'D')
         INTO search;
 
         RETURN search;
@@ -90,11 +94,14 @@ CREATE OR REPLACE FUNCTION data_biut()
     END;
     $$;
 
-CREATE TRIGGER data_biut
-    BEFORE INSERT OR UPDATE
-    ON flow_data
-    FOR EACH ROW EXECUTE PROCEDURE data_biut();
-
+DO $$ BEGIN
+    CREATE TRIGGER data_biut
+        BEFORE INSERT OR UPDATE
+        ON flow_data
+        FOR EACH ROW EXECUTE PROCEDURE data_biut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 -- Trigger after update/insert/delete user permission object.
 CREATE OR REPLACE FUNCTION handle_userpermission_data(user_permission permissions_permissionmodel)
@@ -125,10 +132,14 @@ CREATE OR REPLACE FUNCTION userpermission_data_aiut()
     END;
     $$;
 
-CREATE TRIGGER userpermission_data_aiut
-    AFTER INSERT OR UPDATE
-    ON permissions_permissionmodel
-    FOR EACH ROW EXECUTE PROCEDURE userpermission_data_aiut();
+DO $$ BEGIN
+    CREATE TRIGGER userpermission_data_aiut
+        AFTER INSERT OR UPDATE
+        ON permissions_permissionmodel
+        FOR EACH ROW EXECUTE PROCEDURE userpermission_data_aiut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 CREATE OR REPLACE FUNCTION userpermission_data_adt()
     RETURNS TRIGGER
@@ -140,10 +151,14 @@ CREATE OR REPLACE FUNCTION userpermission_data_adt()
     END;
     $$;
 
-CREATE TRIGGER userpermission_data_adt
-    AFTER DELETE
-    ON permissions_permissionmodel
-    FOR EACH ROW EXECUTE PROCEDURE userpermission_data_adt();
+DO $$ BEGIN
+    CREATE TRIGGER userpermission_data_adt
+        AFTER DELETE
+        ON permissions_permissionmodel
+        FOR EACH ROW EXECUTE PROCEDURE userpermission_data_adt();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 
 -- Trigger after update contributor.
@@ -159,11 +174,14 @@ CREATE OR REPLACE FUNCTION data_contributor_aut()
     END;
     $$;
 
-CREATE TRIGGER data_contributor_aut
-    AFTER UPDATE
-    ON auth_user
-    FOR EACH ROW EXECUTE PROCEDURE data_contributor_aut();
-
+DO $$ BEGIN
+    CREATE TRIGGER data_contributor_aut
+        AFTER UPDATE
+        ON auth_user
+        FOR EACH ROW EXECUTE PROCEDURE data_contributor_aut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;
 
 -- Trigger after update owner.
 CREATE OR REPLACE FUNCTION data_owner_aut()
@@ -188,7 +206,11 @@ CREATE OR REPLACE FUNCTION data_owner_aut()
     END;
     $$;
 
-CREATE TRIGGER data_owner_aut
-    AFTER UPDATE
-    ON auth_user
-    FOR EACH ROW EXECUTE PROCEDURE data_owner_aut();
+DO $$ BEGIN
+    CREATE TRIGGER data_owner_aut
+        AFTER UPDATE
+        ON auth_user
+        FOR EACH ROW EXECUTE PROCEDURE data_owner_aut();
+EXCEPTION
+  WHEN others THEN null;
+END $$;