Reporting unique nulls not distinct in analyze schema and assessment report (#2125)

Author: ShivanshGahlot

migtests/tests/analyze-schema/dummy-export-dir/schema/tables/INDEXES_table.sql

@@ -103,4 +103,9 @@ CREATE INDEX idx_udt1 on test_udt(home_address1);
 
 CREATE INDEX idx_enum on test_udt(some_field);
 
-CREATE INDEX "idx&_enum2" on test_udt((some_field::non_public.enum_test));
+CREATE INDEX "idx&_enum2" on test_udt((some_field::non_public.enum_test));
+
+-- Create a unique index on a column with NULLs with the NULLS NOT DISTINCT option
+CREATE UNIQUE INDEX users_unique_nulls_not_distinct_index_email
+    ON users_unique_nulls_not_distinct_index (email)
+    NULLS NOT DISTINCT;

migtests/tests/analyze-schema/dummy-export-dir/schema/tables/table.sql

@@ -429,3 +429,39 @@ CREATE TABLE timestamptz_multirange_table (
     global_event_times tstzmultirange
 );
 
+-- Testing tables with unique nulls not distinct constraints
+
+-- Control case
+CREATE TABLE users_unique_nulls_distinct (
+    id SERIAL PRIMARY KEY,
+    email TEXT,
+    UNIQUE (email)
+);
+
+CREATE TABLE users_unique_nulls_not_distinct (
+    id SERIAL PRIMARY KEY,
+    email TEXT,
+    UNIQUE NULLS NOT DISTINCT (email)
+);
+
+CREATE TABLE sales_unique_nulls_not_distinct (
+    store_id INT,
+    product_id INT,
+    sale_date DATE,
+    UNIQUE NULLS NOT DISTINCT (store_id, product_id, sale_date)
+);
+
+CREATE TABLE sales_unique_nulls_not_distinct_alter (
+	store_id INT,
+	product_id INT,
+	sale_date DATE
+);
+
+ALTER TABLE sales_unique_nulls_not_distinct_alter
+	ADD CONSTRAINT sales_unique_nulls_not_distinct_alter_unique UNIQUE NULLS NOT DISTINCT (store_id, product_id, sale_date);
+
+-- Create a unique index on a column with NULLs with the NULLS NOT DISTINCT option
+CREATE TABLE users_unique_nulls_not_distinct_index (
+    id INTEGER PRIMARY KEY,
+    email TEXT
+);

migtests/tests/analyze-schema/expected_issues.json

@@ -2060,5 +2060,45 @@
             "Suggestion": "Multirange data type is not yet supported in YugabyteDB, no workaround available currently",
             "GH": "",
             "MinimumVersionsFixedIn": null
+        },
+        {
+            "IssueType": "unsupported_features",
+            "ObjectType": "TABLE",
+            "ObjectName": "users_unique_nulls_not_distinct",
+            "Reason": "Unique Nulls Not Distinct",
+            "SqlStatement": "CREATE TABLE users_unique_nulls_not_distinct (\n    id SERIAL PRIMARY KEY,\n    email TEXT,\n    UNIQUE NULLS NOT DISTINCT (email)\n);",
+            "Suggestion": "",
+            "GH": "",
+            "MinimumVersionsFixedIn": null
+        },
+        {
+            "IssueType": "unsupported_features",
+            "ObjectType": "TABLE",
+            "ObjectName": "sales_unique_nulls_not_distinct",
+            "Reason": "Unique Nulls Not Distinct",
+            "SqlStatement": "CREATE TABLE sales_unique_nulls_not_distinct (\n    store_id INT,\n    product_id INT,\n    sale_date DATE,\n    UNIQUE NULLS NOT DISTINCT (store_id, product_id, sale_date)\n);",
+            "Suggestion": "",
+            "GH": "",
+            "MinimumVersionsFixedIn": null
+        },
+        {
+            "IssueType": "unsupported_features",
+            "ObjectType": "TABLE",
+            "ObjectName": "sales_unique_nulls_not_distinct_alter",
+            "Reason": "Unique Nulls Not Distinct",
+            "SqlStatement": "ALTER TABLE sales_unique_nulls_not_distinct_alter\n\tADD CONSTRAINT sales_unique_nulls_not_distinct_alter_unique UNIQUE NULLS NOT DISTINCT (store_id, product_id, sale_date);",
+            "Suggestion": "",
+            "GH": "",
+            "MinimumVersionsFixedIn": null
+        },
+        {
+            "IssueType": "unsupported_features",
+            "ObjectType": "INDEX",
+            "ObjectName": "users_unique_nulls_not_distinct_index_email ON users_unique_nulls_not_distinct_index",
+            "Reason": "Unique Nulls Not Distinct",
+            "SqlStatement": "CREATE UNIQUE INDEX users_unique_nulls_not_distinct_index_email\n    ON users_unique_nulls_not_distinct_index (email)\n    NULLS NOT DISTINCT;",
+            "Suggestion": "",
+            "GH": "",
+            "MinimumVersionsFixedIn": null
         }
 ]

migtests/tests/analyze-schema/summary.json

@@ -26,15 +26,14 @@
             },
             {
                 "ObjectType": "TABLE",
-                "TotalCount": 59,
-                "InvalidCount": 50,
-                "ObjectNames": "public.json_data, employees, image, public.xml_data_example, combined_tbl1, test_arr_enum, public.locations, test_udt, combined_tbl, public.ts_query_table, public.documents, public.citext_type, public.inet_type, public.test_jsonb, test_xml_type, test_xid_type, public.range_columns_partition_test_copy, anydata_test, uritype_test, public.foreign_def_test, test_4, enum_example.bugs, table_abc, anydataset_test, unique_def_test1, test_2, table_1, public.range_columns_partition_test, table_xyz, public.users, test_3, test_5, test_7, foreign_def_test2, unique_def_test, sales_data, table_test, test_interval, test_non_pk_multi_column_list, test_9, test_8, order_details, public.employees4, anytype_test, public.meeting, test_table_in_type_file, sales, test_1, \"Test\", foreign_def_test1, salaries2, test_6, public.pr, bigint_multirange_table, date_multirange_table, int_multirange_table, numeric_multirange_table, timestamp_multirange_table, timestamptz_multirange_table"            },
-
+                "TotalCount": 64,
+                "InvalidCount": 53,
+                "ObjectNames": "test_table_in_type_file, sales_data, salaries2, sales, test_1, test_2, test_non_pk_multi_column_list, test_3, test_4, test_5, test_6, test_7, test_8, test_9, order_details, public.employees4, enum_example.bugs, table_xyz, table_abc, table_1, table_test, test_interval, public.range_columns_partition_test, public.range_columns_partition_test_copy, anydata_test, anydataset_test, anytype_test, uritype_test, \"Test\", public.meeting, public.pr, public.foreign_def_test, public.users, foreign_def_test1, foreign_def_test2, unique_def_test, unique_def_test1, test_xml_type, test_xid_type, public.test_jsonb, public.inet_type, public.citext_type, public.documents, public.ts_query_table, combined_tbl, combined_tbl1, test_udt, test_arr_enum, public.locations, public.xml_data_example, image, public.json_data, employees, bigint_multirange_table, date_multirange_table, int_multirange_table, numeric_multirange_table, timestamp_multirange_table, timestamptz_multirange_table, users_unique_nulls_distinct, users_unique_nulls_not_distinct, sales_unique_nulls_not_distinct, sales_unique_nulls_not_distinct_alter, users_unique_nulls_not_distinct_index"            },
             {
                 "ObjectType": "INDEX",
-                "TotalCount": 43,
-                "InvalidCount": 39,
-                "ObjectNames": "idx1 ON combined_tbl1, idx2 ON combined_tbl1, idx3 ON combined_tbl1, idx4 ON combined_tbl1, idx5 ON combined_tbl1, idx6 ON combined_tbl1, idx7 ON combined_tbl1, idx8 ON combined_tbl1, film_fulltext_idx ON public.film, idx_actor_last_name ON public.actor, idx_name1 ON table_name, idx_name2 ON table_name, idx_name3 ON schema_name.table_name, idx_fileinfo_name_splitted ON public.fileinfo, abc ON public.example, abc ON schema2.example, tsvector_idx ON public.documents, tsquery_idx ON public.ts_query_table, idx_citext ON public.citext_type, idx_inet ON public.inet_type, idx_json ON public.test_jsonb, idx_json2 ON public.test_jsonb, idx_valid ON public.test_jsonb, idx_array ON public.documents, idx1 ON combined_tbl, idx2 ON combined_tbl, idx3 ON combined_tbl, idx4 ON combined_tbl, idx5 ON combined_tbl, idx6 ON combined_tbl, idx7 ON combined_tbl, idx8 ON combined_tbl, idx9 ON combined_tbl, idx10 ON combined_tbl, idx11 ON combined_tbl, idx12 ON combined_tbl, idx13 ON combined_tbl, idx14 ON combined_tbl, idx15 ON combined_tbl, idx_udt ON test_udt, idx_udt1 ON test_udt, idx_enum ON test_udt, \"idx\u0026_enum2\" ON test_udt",
+                "TotalCount": 44,
+                "InvalidCount": 40,
+                "ObjectNames": "film_fulltext_idx ON public.film, idx_actor_last_name ON public.actor, idx_name1 ON table_name, idx_name2 ON table_name, idx_name3 ON schema_name.table_name, idx_fileinfo_name_splitted ON public.fileinfo, abc ON public.example, abc ON schema2.example, tsvector_idx ON public.documents, tsquery_idx ON public.ts_query_table, idx_citext ON public.citext_type, idx_inet ON public.inet_type, idx_json ON public.test_jsonb, idx_json2 ON public.test_jsonb, idx_valid ON public.test_jsonb, idx_array ON public.documents, idx1 ON combined_tbl, idx2 ON combined_tbl, idx3 ON combined_tbl, idx4 ON combined_tbl, idx5 ON combined_tbl, idx6 ON combined_tbl, idx7 ON combined_tbl, idx8 ON combined_tbl, idx9 ON combined_tbl, idx10 ON combined_tbl, idx11 ON combined_tbl, idx12 ON combined_tbl, idx13 ON combined_tbl, idx14 ON combined_tbl, idx15 ON combined_tbl, idx1 ON combined_tbl1, idx2 ON combined_tbl1, idx3 ON combined_tbl1, idx4 ON combined_tbl1, idx5 ON combined_tbl1, idx6 ON combined_tbl1, idx7 ON combined_tbl1, idx8 ON combined_tbl1, idx_udt ON test_udt, idx_udt1 ON test_udt, idx_enum ON test_udt, \"idx\u0026_enum2\" ON test_udt, users_unique_nulls_not_distinct_index_email ON users_unique_nulls_not_distinct_index",
                 "Details": "There are some GIN indexes present in the schema, but GIN indexes are partially supported in YugabyteDB as mentioned in (https://github.com/yugabyte/yugabyte-db/issues/7850) so take a look and modify them if not supported."
             },
             {