diff --git a/src/backend/distributed/planner/multi_logical_planner.c b/src/backend/distributed/planner/multi_logical_planner.c index ebc71e2a691..662d4050312 100644 --- a/src/backend/distributed/planner/multi_logical_planner.c +++ b/src/backend/distributed/planner/multi_logical_planner.c @@ -1170,7 +1170,8 @@ HasComplexRangeTableType(Query *queryTree) if (rangeTableEntry->rtekind != RTE_RELATION && rangeTableEntry->rtekind != RTE_SUBQUERY && rangeTableEntry->rtekind != RTE_FUNCTION && - rangeTableEntry->rtekind != RTE_VALUES) + rangeTableEntry->rtekind != RTE_VALUES && + !IsJsonTableRTE(rangeTableEntry)) { hasComplexRangeTableType = true; } diff --git a/src/backend/distributed/planner/query_pushdown_planning.c b/src/backend/distributed/planner/query_pushdown_planning.c index 65de8680ccb..fe89284c0b8 100644 --- a/src/backend/distributed/planner/query_pushdown_planning.c +++ b/src/backend/distributed/planner/query_pushdown_planning.c @@ -61,7 +61,8 @@ typedef enum RecurringTuplesType RECURRING_TUPLES_FUNCTION, RECURRING_TUPLES_EMPTY_JOIN_TREE, RECURRING_TUPLES_RESULT_FUNCTION, - RECURRING_TUPLES_VALUES + RECURRING_TUPLES_VALUES, + RECURRING_TUPLES_JSON_TABLE } RecurringTuplesType; /* @@ -347,7 +348,8 @@ IsFunctionOrValuesRTE(Node *node) RangeTblEntry *rangeTblEntry = (RangeTblEntry *) node; if (rangeTblEntry->rtekind == RTE_FUNCTION || - rangeTblEntry->rtekind == RTE_VALUES) + rangeTblEntry->rtekind == RTE_VALUES || + IsJsonTableRTE(rangeTblEntry)) { return true; } @@ -700,6 +702,13 @@ DeferErrorIfFromClauseRecurs(Query *queryTree) "the FROM clause contains VALUES", NULL, NULL); } + else if (recurType == RECURRING_TUPLES_JSON_TABLE) + { + return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED, + "correlated subqueries are not supported when " + "the FROM clause contains JSON_TABLE", NULL, + NULL); + } /* @@ -1204,7 +1213,8 @@ DeferErrorIfUnsupportedTableCombination(Query *queryTree) */ if (rangeTableEntry->rtekind == RTE_RELATION || rangeTableEntry->rtekind == RTE_SUBQUERY || - rangeTableEntry->rtekind == RTE_RESULT) + rangeTableEntry->rtekind == RTE_RESULT || + IsJsonTableRTE(rangeTableEntry)) { /* accepted */ } @@ -1372,6 +1382,13 @@ DeferErrorIfUnsupportedUnionQuery(Query *subqueryTree) "VALUES is not supported within a " "UNION", NULL); } + else if (recurType == RECURRING_TUPLES_JSON_TABLE) + { + return DeferredError(ERRCODE_FEATURE_NOT_SUPPORTED, + "cannot push down this subquery", + "JSON_TABLE is not supported within a " + "UNION", NULL); + } return NULL; } @@ -1477,6 +1494,11 @@ RecurringTypeDescription(RecurringTuplesType recurType) return "a VALUES clause"; } + case RECURRING_TUPLES_JSON_TABLE: + { + return "a JSON_TABLE"; + } + case RECURRING_TUPLES_INVALID: { /* @@ -1673,7 +1695,8 @@ DeferredErrorIfUnsupportedLateralSubquery(PlannerInfo *plannerInfo, * strings anyway. */ if (recurType != RECURRING_TUPLES_VALUES && - recurType != RECURRING_TUPLES_RESULT_FUNCTION) + recurType != RECURRING_TUPLES_RESULT_FUNCTION && + recurType != RECURRING_TUPLES_JSON_TABLE) { recurTypeDescription = psprintf("%s (%s)", recurTypeDescription, recurringRangeTableEntry->eref-> @@ -1750,6 +1773,26 @@ ContainsRecurringRangeTable(List *rangeTable, RecurringTuplesType *recurType) } +/* + * IsJsonTableRTE checks whether the RTE refers to a JSON_TABLE + * table function, which was introduced in PostgreSQL 15. + */ +bool +IsJsonTableRTE(RangeTblEntry *rte) +{ +#if PG_VERSION_NUM >= PG_VERSION_17 + if (rte == NULL) + { + return false; + } + return (rte->rtekind == RTE_TABLEFUNC && + rte->tablefunc->functype == TFT_JSON_TABLE); +#endif + + return false; +} + + /* * HasRecurringTuples returns whether any part of the expression will generate * the same set of tuples in every query on shards when executing a distributed @@ -1811,6 +1854,11 @@ HasRecurringTuples(Node *node, RecurringTuplesType *recurType) *recurType = RECURRING_TUPLES_VALUES; return true; } + else if (IsJsonTableRTE(rangeTableEntry)) + { + *recurType = RECURRING_TUPLES_JSON_TABLE; + return true; + } return false; } diff --git a/src/include/distributed/query_pushdown_planning.h b/src/include/distributed/query_pushdown_planning.h index e0d4f25ddc5..47a34cee072 100644 --- a/src/include/distributed/query_pushdown_planning.h +++ b/src/include/distributed/query_pushdown_planning.h @@ -46,6 +46,7 @@ extern DeferredErrorMessage * DeferErrorIfCannotPushdownSubquery(Query *subquery bool outerMostQueryHasLimit); extern DeferredErrorMessage * DeferErrorIfUnsupportedUnionQuery(Query *queryTree); +extern bool IsJsonTableRTE(RangeTblEntry *rte); #endif /* QUERY_PUSHDOWN_PLANNING_H */ diff --git a/src/test/regress/expected/json_table.out b/src/test/regress/expected/json_table.out new file mode 100644 index 00000000000..7bc91026987 --- /dev/null +++ b/src/test/regress/expected/json_table.out @@ -0,0 +1,562 @@ +-- +-- JSON_TABLE +-- PG17 has added basic JSON_TABLE() functionality +-- JSON_TABLE() allows JSON data to be converted into a relational view +-- and thus used, for example, in a FROM clause, like other tabular +-- data. We treat JSON_TABLE the same as correlated functions (e.g., recurring tuples). +-- In the end, for multi-shard JSON_TABLE commands, we apply the same +-- restrictions as reference tables (e.g., cannot perform a lateral outer join +-- when a distributed subquery references a (reference table)/JSON_TABLE etc.) +-- Relevant PG commit: +-- https://github.com/postgres/postgres/commit/de3600452 +-- +SHOW server_version \gset +SELECT substring(:'server_version', '\d+')::int >= 17 AS server_version_ge_17 +\gset +\if :server_version_ge_17 +\else +\q +\endif +CREATE SCHEMA json_table; +SET search_path TO json_table; +SET citus.next_shard_id TO 1687000; +CREATE TABLE test_table(id bigserial, value text); +SELECT create_distributed_table('test_table', 'id'); + create_distributed_table +--------------------------------------------------------------------- + +(1 row) + +INSERT INTO test_table (value) SELECT i::text FROM generate_series(0,100)i; +CREATE TABLE my_films(id bigserial, js jsonb); +SELECT create_distributed_table('my_films', 'id'); + create_distributed_table +--------------------------------------------------------------------- + +(1 row) + +INSERT INTO my_films(js) VALUES ( +'{ "favorites" : [ + { "kind" : "comedy", "films" : [ { "title" : "Bananas", "director" : "Woody Allen"}, + { "title" : "The Dinner Game", "director" : "Francis Veber" } ] }, + { "kind" : "horror", "films" : [{ "title" : "Psycho", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "thriller", "films" : [{ "title" : "Vertigo", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "drama", "films" : [{ "title" : "Yojimbo", "director" : "Akira Kurosawa" } ] } + ] }'); +INSERT INTO my_films(js) VALUES ( +'{ "favorites" : [ + { "kind" : "comedy", "films" : [ { "title" : "Bananas2", "director" : "Woody Allen"}, + { "title" : "The Dinner Game2", "director" : "Francis Veber" } ] }, + { "kind" : "horror", "films" : [{ "title" : "Psycho2", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "thriller", "films" : [{ "title" : "Vertigo2", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "drama", "films" : [{ "title" : "Yojimbo2", "director" : "Akira Kurosawa" } ] } + ] }'); +-- a router query +SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt + WHERE my_films.id = 1 + ORDER BY 1,2,3,4; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa +(5 rows) + +-- router query with an explicit LATEREL SUBQUERY +SELECT sub.* +FROM my_films, + lateral(SELECT * FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt) as sub +WHERE my_films.id = 1; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa +(5 rows) + +-- router query with an explicit LATEREL SUBQUERY and LIMIT +SELECT sub.* +FROM my_films, + lateral(SELECT * FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt ORDER BY id DESC LIMIT 1) as sub +WHERE my_films.id = 1; + id | kind | title | director +--------------------------------------------------------------------- + 4 | drama | Yojimbo | Akira Kurosawa +(1 row) + +-- set it DEBUG1 in case the plan changes +-- we can see details +SET client_min_messages TO DEBUG1; +-- a mult-shard query +SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt + ORDER BY 1,2,3,4; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | Bananas2 | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 1 | comedy | The Dinner Game2 | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 2 | horror | Psycho2 | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 3 | thriller | Vertigo2 | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa + 4 | drama | Yojimbo2 | Akira Kurosawa +(10 rows) + +-- recursively plan subqueries that has JSON_TABLE +SELECT count(*) FROM +( + SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt + LIMIT 1) as sub_with_json, test_table +WHERE test_table.id = sub_with_json.id; +DEBUG: push down of limit count: 1 +DEBUG: generating subplan XXX_1 for subquery SELECT jt.id, jt.kind, jt.title, jt.director FROM "json_table".my_films, LATERAL JSON_TABLE(my_films.js, '$."favorites"[*]' AS json_table_path_0 COLUMNS (id FOR ORDINALITY, kind text PATH '$."kind"', NESTED PATH '$."films"[*]' AS json_table_path_1 COLUMNS (title text PATH '$."title"', director text PATH '$."director"'))) jt LIMIT 1 +DEBUG: Plan XXX query after replacing subqueries and CTEs: SELECT count(*) AS count FROM (SELECT intermediate_result.id, intermediate_result.kind, intermediate_result.title, intermediate_result.director FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(id integer, kind text, title text, director text)) sub_with_json, "json_table".test_table WHERE (test_table.id OPERATOR(pg_catalog.=) sub_with_json.id) + count +--------------------------------------------------------------------- + 1 +(1 row) + +-- multi-shard query with an explicit LATEREL SUBQUERY +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) + ORDER BY 1,2,3,4; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | Bananas2 | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 1 | comedy | The Dinner Game2 | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 2 | horror | Psycho2 | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 3 | thriller | Vertigo2 | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa + 4 | drama | Yojimbo2 | Akira Kurosawa +(10 rows) + +-- JSON_TABLE can be on the inner part of an outer joion +SELECT sub.* +FROM my_films LEFT JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) + ORDER BY 1,2,3,4; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | Bananas2 | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 1 | comedy | The Dinner Game2 | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 2 | horror | Psycho2 | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 3 | thriller | Vertigo2 | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa + 4 | drama | Yojimbo2 | Akira Kurosawa +(10 rows) + +-- we can pushdown this correlated subquery in WHERE clause +SELECT count(*) +FROM my_films WHERE + (SELECT count(*) > 0 + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000); + count +--------------------------------------------------------------------- + 2 +(1 row) + +-- we can pushdown this correlated subquery in SELECT clause + SELECT (SELECT count(*) > 0 + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt) +FROM my_films; + ?column? +--------------------------------------------------------------------- + t + t +(2 rows) + +-- multi-shard query with an explicit LATEREL SUBQUERY +-- along with other tables +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) JOIN test_table ON(my_films.id = test_table.id) + ORDER BY 1,2,3,4; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | Bananas2 | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 1 | comedy | The Dinner Game2 | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 2 | horror | Psycho2 | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 3 | thriller | Vertigo2 | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa + 4 | drama | Yojimbo2 | Akira Kurosawa +(10 rows) + +-- non-colocated join fails +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) JOIN test_table ON(my_films.id != test_table.id) + ORDER BY 1,2,3,4; +ERROR: complex joins are only supported when all distributed tables are co-located and joined on their distribution columns +-- JSON_TABLE can be in the outer part of the join +-- as long as there is a distributed table +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) LEFT JOIN test_table ON(my_films.id = test_table.id) + ORDER BY 1,2,3,4; + id | kind | title | director +--------------------------------------------------------------------- + 1 | comedy | Bananas | Woody Allen + 1 | comedy | Bananas2 | Woody Allen + 1 | comedy | The Dinner Game | Francis Veber + 1 | comedy | The Dinner Game2 | Francis Veber + 2 | horror | Psycho | Alfred Hitchcock + 2 | horror | Psycho2 | Alfred Hitchcock + 3 | thriller | Vertigo | Alfred Hitchcock + 3 | thriller | Vertigo2 | Alfred Hitchcock + 4 | drama | Yojimbo | Akira Kurosawa + 4 | drama | Yojimbo2 | Akira Kurosawa +(10 rows) + +-- JSON_TABLE can be on the outer side of the join +-- We support outer joins where the outer rel is a recurring one +-- and the inner one is a non-recurring one if we don't reference the outer from the inner +-- https://github.com/citusdata/citus/pull/6512 +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) +LEFT JOIN LATERAL + (SELECT * + FROM my_films) AS foo on(foo.id = a); +DEBUG: recursively planning right side of the left join since the outer side is a recurring rel +DEBUG: recursively planning the distributed subquery since it is part of a distributed join node that is outer joined with a recurring rel +DEBUG: generating subplan XXX_1 for subquery SELECT id, js FROM "json_table".my_films +DEBUG: Plan XXX query after replacing subqueries and CTEs: SELECT "json_table".id, "json_table".column_a, "json_table".column_b, "json_table".a, "json_table".b, "json_table".c, foo.id, foo.js FROM (JSON_TABLE('[{"a": 10, "b": 20}, {"a": 30, "b": 40}]'::jsonb, '$[*]' AS json_table_path_0 COLUMNS (id FOR ORDINALITY, column_a integer PATH '$."a"', column_b integer PATH '$."b"', a integer PATH '$."a"', b integer PATH '$."b"', c text PATH '$."c"')) LEFT JOIN LATERAL (SELECT intermediate_result.id, intermediate_result.js FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(id bigint, js jsonb)) foo ON ((foo.id OPERATOR(pg_catalog.=) "json_table".a))) + id | column_a | column_b | a | b | c | id | js +--------------------------------------------------------------------- + 1 | 10 | 20 | 10 | 20 | | | + 2 | 30 | 40 | 30 | 40 | | | +(2 rows) + +-- However we don't support +-- when we reference the JSON_TABLE from the non-recurring distributed table subquery +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (json_id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) +LEFT JOIN LATERAL + (SELECT * + FROM my_films WHERE id::text LIKE c) AS foo on(foo.id = a); +DEBUG: recursively planning right side of the left join since the outer side is a recurring rel +DEBUG: recursively planning the distributed subquery since it is part of a distributed join node that is outer joined with a recurring rel +ERROR: cannot perform a lateral outer join when a distributed subquery references a JSON_TABLE +-- JSON_TABLE cannot be on the FROM clause alone +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (json_id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) as foo +WHERE b > + (SELECT count(*) + FROM my_films WHERE id = foo.a); +ERROR: correlated subqueries are not supported when the FROM clause contains JSON_TABLE +-- we can recursively plan json_tables on set operations +(SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY)) ORDER BY id ASC LIMIT 1) +UNION +(SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY)) ORDER BY id ASC LIMIT 1) +UNION +(SELECT id FROM test_table ORDER BY id ASC LIMIT 1); +DEBUG: generating subplan XXX_1 for subquery SELECT id FROM JSON_TABLE('[{"a": 10, "b": 20}, {"a": 30, "b": 40}]'::jsonb, '$[*]' AS json_table_path_0 COLUMNS (id FOR ORDINALITY)) ORDER BY id LIMIT 1 +DEBUG: generating subplan XXX_2 for subquery SELECT id FROM JSON_TABLE('[{"a": 10, "b": 20}, {"a": 30, "b": 40}]'::jsonb, '$[*]' AS json_table_path_0 COLUMNS (id FOR ORDINALITY)) ORDER BY id LIMIT 1 +DEBUG: push down of limit count: 1 +DEBUG: generating subplan XXX_3 for subquery SELECT id FROM "json_table".test_table ORDER BY id LIMIT 1 +DEBUG: Plan XXX query after replacing subqueries and CTEs: SELECT intermediate_result.id FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(id integer) UNION SELECT intermediate_result.id FROM read_intermediate_result('XXX_2'::text, 'binary'::citus_copy_format) intermediate_result(id integer) UNION SELECT intermediate_result.id FROM read_intermediate_result('XXX_3'::text, 'binary'::citus_copy_format) intermediate_result(id bigint) + id +--------------------------------------------------------------------- + 1 +(1 row) + +-- LIMIT in subquery not supported when json_table exists +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) +JOIN LATERAL + (SELECT * + FROM my_films WHERE json_table.id = a LIMIT 1) as foo ON (true); +ERROR: cannot push down this subquery +DETAIL: Limit clause is currently unsupported when a lateral subquery references a column from a JSON_TABLE +RESET client_min_messages; +-- test some utility functions on the target list & where clause +select jsonb_path_exists(js, '$.favorites') from my_films; + jsonb_path_exists +--------------------------------------------------------------------- + t + t +(2 rows) + +select bool_and(JSON_EXISTS(js, '$.favorites.films.title')) from my_films; + bool_and +--------------------------------------------------------------------- + t +(1 row) + +SELECT count(*) FROM my_films WHERE jsonb_path_exists(js, '$.favorites'); + count +--------------------------------------------------------------------- + 2 +(1 row) + +SELECT count(*) FROM my_films WHERE jsonb_path_exists(js, '$.favorites'); + count +--------------------------------------------------------------------- + 2 +(1 row) + +SELECT count(*) FROM my_films WHERE JSON_EXISTS(js, '$.favorites.films.title'); + count +--------------------------------------------------------------------- + 2 +(1 row) + +-- check constraint with json_exists +create table user_profiles ( + id bigserial, + addresses jsonb, + anyjson jsonb, + check (json_exists( addresses, '$.main' )) +); +select create_distributed_table('user_profiles', 'id'); + create_distributed_table +--------------------------------------------------------------------- + +(1 row) + +insert into user_profiles (addresses) VALUES (JSON_SCALAR('1')); +ERROR: new row for relation "user_profiles_1687008" violates check constraint "user_profiles_addresses_check" +DETAIL: Failing row contains (1, "1", null). +CONTEXT: while executing command on localhost:xxxxx +insert into user_profiles (addresses) VALUES ('{"main":"value"}'); +-- we cannot insert because WITH UNIQUE KEYS +insert into user_profiles (addresses) VALUES (JSON ('{"main":"value", "main":"value"}' WITH UNIQUE KEYS)); +ERROR: duplicate JSON object key value +-- we can insert with +insert into user_profiles (addresses) VALUES (JSON ('{"main":"value", "main":"value"}' WITHOUT UNIQUE KEYS)) RETURNING *; + id | addresses | anyjson +--------------------------------------------------------------------- + 4 | {"main": "value"} | +(1 row) + +TRUNCATE user_profiles; +INSERT INTO user_profiles (anyjson) VALUES ('12'), ('"abc"'), ('[1,2,3]'), ('{"a":12}'); +select anyjson, anyjson is json array as json_array, anyjson is json object as json_object, anyjson is json scalar as json_scalar, +anyjson is json with UNIQUE keys +from user_profiles WHERE anyjson IS NOT NULL ORDER BY 1; + anyjson | json_array | json_object | json_scalar | ?column? +--------------------------------------------------------------------- + "abc" | f | f | t | t + 12 | f | f | t | t + [1, 2, 3] | t | f | f | t + {"a": 12} | f | t | f | t +(4 rows) + +-- use json_query +SELECT i, + json_query('[{"x": "aaa"},{"x": "bbb"},{"x": "ccc"}]'::JSONB, '$[$i].x' passing id AS i RETURNING text omit quotes) +FROM generate_series(0, 3) i +JOIN my_films ON(id = i); + i | json_query +--------------------------------------------------------------------- + 1 | bbb + 2 | ccc +(2 rows) + +-- we can use JSON_TABLE in modification queries as well +-- use log level such that we can see trace changes +SET client_min_messages TO DEBUG1; +--the JSON_TABLE subquery is recursively planned +UPDATE test_table SET VALUE = 'XXX' FROM( +SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt) as foo WHERE foo.id = test_table.id; +DEBUG: generating subplan XXX_1 for subquery SELECT jt.id, jt.kind, jt.title, jt.director FROM "json_table".my_films, LATERAL JSON_TABLE(my_films.js, '$."favorites"[*]' AS json_table_path_0 COLUMNS (id FOR ORDINALITY, kind text PATH '$."kind"', NESTED PATH '$."films"[*]' AS json_table_path_1 COLUMNS (title text PATH '$."title"', director text PATH '$."director"'))) jt +DEBUG: Plan XXX query after replacing subqueries and CTEs: UPDATE "json_table".test_table SET value = 'XXX'::text FROM (SELECT intermediate_result.id, intermediate_result.kind, intermediate_result.title, intermediate_result.director FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(id integer, kind text, title text, director text)) foo WHERE (foo.id OPERATOR(pg_catalog.=) test_table.id) +-- Subquery with JSON table can be pushed down because two distributed tables +-- in the query are joined on distribution column +UPDATE test_table SET VALUE = 'XXX' FROM ( +SELECT my_films.id, jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt) as foo WHERE foo.id = test_table.id; +-- we can pushdown with CTEs as well +WITH json_cte AS +(SELECT my_films.id, jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt) +UPDATE test_table SET VALUE = 'XYZ' FROM json_cte + WHERE json_cte.id = test_table.id; + -- we can recursively with CTEs as well +WITH json_cte AS +(SELECT my_films.id as film_id, jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + id FOR ORDINALITY, + title text PATH '$.title', + director text PATH '$.director'))) AS jt ORDER BY jt.id LIMIT 1) +UPDATE test_table SET VALUE = 'XYZ' FROM json_cte + WHERE json_cte.film_id = test_table.id; +DEBUG: generating subplan XXX_1 for CTE json_cte: SELECT my_films.id AS film_id, jt.kind, jt.id, jt.title, jt.director FROM "json_table".my_films, LATERAL JSON_TABLE(my_films.js, '$."favorites"[*]' AS json_table_path_0 COLUMNS (kind text PATH '$."kind"', NESTED PATH '$."films"[*]' AS json_table_path_1 COLUMNS (id FOR ORDINALITY, title text PATH '$."title"', director text PATH '$."director"'))) jt ORDER BY jt.id LIMIT 1 +DEBUG: push down of limit count: 1 +DEBUG: Plan XXX query after replacing subqueries and CTEs: UPDATE "json_table".test_table SET value = 'XYZ'::text FROM (SELECT intermediate_result.film_id, intermediate_result.kind, intermediate_result.id, intermediate_result.title, intermediate_result.director FROM read_intermediate_result('XXX_1'::text, 'binary'::citus_copy_format) intermediate_result(film_id bigint, kind text, id integer, title text, director text)) json_cte WHERE (json_cte.film_id OPERATOR(pg_catalog.=) test_table.id) +-- Should succeed (JSON arguments are passed to root and nested paths) +SELECT * +FROM + generate_series(1, 4) x, + generate_series(1, 3) y, + JSON_TABLE(jsonb + '[[1,2,3],[2,3,4,5],[3,4,5,6]]', + 'strict $[*] ? (@[*] < $x)' + PASSING x AS x, y AS y + COLUMNS ( + y text FORMAT JSON PATH '$', + NESTED PATH 'strict $[*] ? (@ >= $y)' + COLUMNS ( + z int PATH '$' + ) + ) + ) jt ORDER BY 4,1,2,3; + x | y | y | z +--------------------------------------------------------------------- + 2 | 1 | [1, 2, 3] | 1 + 3 | 1 | [1, 2, 3] | 1 + 4 | 1 | [1, 2, 3] | 1 + 2 | 1 | [1, 2, 3] | 2 + 2 | 2 | [1, 2, 3] | 2 + 3 | 1 | [1, 2, 3] | 2 + 3 | 1 | [2, 3, 4, 5] | 2 + 3 | 2 | [1, 2, 3] | 2 + 3 | 2 | [2, 3, 4, 5] | 2 + 4 | 1 | [1, 2, 3] | 2 + 4 | 1 | [2, 3, 4, 5] | 2 + 4 | 2 | [1, 2, 3] | 2 + 4 | 2 | [2, 3, 4, 5] | 2 + 2 | 1 | [1, 2, 3] | 3 + 2 | 2 | [1, 2, 3] | 3 + 2 | 3 | [1, 2, 3] | 3 + 3 | 1 | [1, 2, 3] | 3 + 3 | 1 | [2, 3, 4, 5] | 3 + 3 | 2 | [1, 2, 3] | 3 + 3 | 2 | [2, 3, 4, 5] | 3 + 3 | 3 | [1, 2, 3] | 3 + 3 | 3 | [2, 3, 4, 5] | 3 + 4 | 1 | [1, 2, 3] | 3 + 4 | 1 | [2, 3, 4, 5] | 3 + 4 | 1 | [3, 4, 5, 6] | 3 + 4 | 2 | [1, 2, 3] | 3 + 4 | 2 | [2, 3, 4, 5] | 3 + 4 | 2 | [3, 4, 5, 6] | 3 + 4 | 3 | [1, 2, 3] | 3 + 4 | 3 | [2, 3, 4, 5] | 3 + 4 | 3 | [3, 4, 5, 6] | 3 + 3 | 1 | [2, 3, 4, 5] | 4 + 3 | 2 | [2, 3, 4, 5] | 4 + 3 | 3 | [2, 3, 4, 5] | 4 + 4 | 1 | [2, 3, 4, 5] | 4 + 4 | 1 | [3, 4, 5, 6] | 4 + 4 | 2 | [2, 3, 4, 5] | 4 + 4 | 2 | [3, 4, 5, 6] | 4 + 4 | 3 | [2, 3, 4, 5] | 4 + 4 | 3 | [3, 4, 5, 6] | 4 + 3 | 1 | [2, 3, 4, 5] | 5 + 3 | 2 | [2, 3, 4, 5] | 5 + 3 | 3 | [2, 3, 4, 5] | 5 + 4 | 1 | [2, 3, 4, 5] | 5 + 4 | 1 | [3, 4, 5, 6] | 5 + 4 | 2 | [2, 3, 4, 5] | 5 + 4 | 2 | [3, 4, 5, 6] | 5 + 4 | 3 | [2, 3, 4, 5] | 5 + 4 | 3 | [3, 4, 5, 6] | 5 + 4 | 1 | [3, 4, 5, 6] | 6 + 4 | 2 | [3, 4, 5, 6] | 6 + 4 | 3 | [3, 4, 5, 6] | 6 +(52 rows) + +SET client_min_messages TO ERROR; +DROP SCHEMA json_table CASCADE; diff --git a/src/test/regress/expected/json_table_0.out b/src/test/regress/expected/json_table_0.out new file mode 100644 index 00000000000..f8f98b932fc --- /dev/null +++ b/src/test/regress/expected/json_table_0.out @@ -0,0 +1,18 @@ +-- +-- JSON_TABLE +-- PG17 has added basic JSON_TABLE() functionality +-- JSON_TABLE() allows JSON data to be converted into a relational view +-- and thus used, for example, in a FROM clause, like other tabular +-- data. We treat JSON_TABLE the same as correlated functions (e.g., recurring tuples). +-- In the end, for multi-shard JSON_TABLE commands, we apply the same +-- restrictions as reference tables (e.g., cannot perform a lateral outer join +-- when a distributed subquery references a (reference table)/JSON_TABLE etc.) +-- Relevant PG commit: +-- https://github.com/postgres/postgres/commit/de3600452 +-- +SHOW server_version \gset +SELECT substring(:'server_version', '\d+')::int >= 17 AS server_version_ge_17 +\gset +\if :server_version_ge_17 +\else +\q diff --git a/src/test/regress/multi_schedule b/src/test/regress/multi_schedule index 0fa54bb384d..2635d0ce815 100644 --- a/src/test/regress/multi_schedule +++ b/src/test/regress/multi_schedule @@ -66,7 +66,7 @@ test: pg14 test: pg15 test: pg15_jsonpath detect_conn_close test: pg16 -test: pg17 +test: pg17 json_table test: drop_column_partitioned_table test: tableam diff --git a/src/test/regress/sql/json_table.sql b/src/test/regress/sql/json_table.sql new file mode 100644 index 00000000000..cc7eb237a25 --- /dev/null +++ b/src/test/regress/sql/json_table.sql @@ -0,0 +1,341 @@ +-- +-- JSON_TABLE +-- PG17 has added basic JSON_TABLE() functionality +-- JSON_TABLE() allows JSON data to be converted into a relational view +-- and thus used, for example, in a FROM clause, like other tabular +-- data. We treat JSON_TABLE the same as correlated functions (e.g., recurring tuples). +-- In the end, for multi-shard JSON_TABLE commands, we apply the same +-- restrictions as reference tables (e.g., cannot perform a lateral outer join +-- when a distributed subquery references a (reference table)/JSON_TABLE etc.) +-- Relevant PG commit: +-- https://github.com/postgres/postgres/commit/de3600452 +-- + +SHOW server_version \gset +SELECT substring(:'server_version', '\d+')::int >= 17 AS server_version_ge_17 +\gset +\if :server_version_ge_17 +\else +\q +\endif + +CREATE SCHEMA json_table; +SET search_path TO json_table; + +SET citus.next_shard_id TO 1687000; + +CREATE TABLE test_table(id bigserial, value text); +SELECT create_distributed_table('test_table', 'id'); +INSERT INTO test_table (value) SELECT i::text FROM generate_series(0,100)i; + + +CREATE TABLE my_films(id bigserial, js jsonb); +SELECT create_distributed_table('my_films', 'id'); + +INSERT INTO my_films(js) VALUES ( +'{ "favorites" : [ + { "kind" : "comedy", "films" : [ { "title" : "Bananas", "director" : "Woody Allen"}, + { "title" : "The Dinner Game", "director" : "Francis Veber" } ] }, + { "kind" : "horror", "films" : [{ "title" : "Psycho", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "thriller", "films" : [{ "title" : "Vertigo", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "drama", "films" : [{ "title" : "Yojimbo", "director" : "Akira Kurosawa" } ] } + ] }'); + +INSERT INTO my_films(js) VALUES ( +'{ "favorites" : [ + { "kind" : "comedy", "films" : [ { "title" : "Bananas2", "director" : "Woody Allen"}, + { "title" : "The Dinner Game2", "director" : "Francis Veber" } ] }, + { "kind" : "horror", "films" : [{ "title" : "Psycho2", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "thriller", "films" : [{ "title" : "Vertigo2", "director" : "Alfred Hitchcock" } ] }, + { "kind" : "drama", "films" : [{ "title" : "Yojimbo2", "director" : "Akira Kurosawa" } ] } + ] }'); + +-- a router query +SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt + WHERE my_films.id = 1 + ORDER BY 1,2,3,4; + +-- router query with an explicit LATEREL SUBQUERY +SELECT sub.* +FROM my_films, + lateral(SELECT * FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt) as sub +WHERE my_films.id = 1; + +-- router query with an explicit LATEREL SUBQUERY and LIMIT +SELECT sub.* +FROM my_films, + lateral(SELECT * FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt ORDER BY id DESC LIMIT 1) as sub +WHERE my_films.id = 1; + +-- set it DEBUG1 in case the plan changes +-- we can see details +SET client_min_messages TO DEBUG1; + +-- a mult-shard query +SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt + ORDER BY 1,2,3,4; + +-- recursively plan subqueries that has JSON_TABLE +SELECT count(*) FROM +( + SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt + LIMIT 1) as sub_with_json, test_table +WHERE test_table.id = sub_with_json.id; + + +-- multi-shard query with an explicit LATEREL SUBQUERY +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) + ORDER BY 1,2,3,4; + +-- JSON_TABLE can be on the inner part of an outer joion +SELECT sub.* +FROM my_films LEFT JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) + ORDER BY 1,2,3,4; + +-- we can pushdown this correlated subquery in WHERE clause +SELECT count(*) +FROM my_films WHERE + (SELECT count(*) > 0 + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000); + +-- we can pushdown this correlated subquery in SELECT clause + SELECT (SELECT count(*) > 0 + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt) +FROM my_films; + +-- multi-shard query with an explicit LATEREL SUBQUERY +-- along with other tables +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) JOIN test_table ON(my_films.id = test_table.id) + ORDER BY 1,2,3,4; + +-- non-colocated join fails +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) JOIN test_table ON(my_films.id != test_table.id) + ORDER BY 1,2,3,4; + +-- JSON_TABLE can be in the outer part of the join +-- as long as there is a distributed table +SELECT sub.* +FROM my_films JOIN + lateral + (SELECT * + FROM JSON_TABLE (js, '$.favorites[*]' COLUMNS (id FOR ORDINALITY, + kind text PATH '$.kind', NESTED PATH '$.films[*]' + COLUMNS (title text PATH '$.title', director text PATH '$.director'))) AS jt + LIMIT 1000) AS sub ON (true) LEFT JOIN test_table ON(my_films.id = test_table.id) + ORDER BY 1,2,3,4; + +-- JSON_TABLE can be on the outer side of the join +-- We support outer joins where the outer rel is a recurring one +-- and the inner one is a non-recurring one if we don't reference the outer from the inner +-- https://github.com/citusdata/citus/pull/6512 + +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) +LEFT JOIN LATERAL + (SELECT * + FROM my_films) AS foo on(foo.id = a); + +-- However we don't support +-- when we reference the JSON_TABLE from the non-recurring distributed table subquery +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (json_id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) +LEFT JOIN LATERAL + (SELECT * + FROM my_films WHERE id::text LIKE c) AS foo on(foo.id = a); + +-- JSON_TABLE cannot be on the FROM clause alone +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (json_id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) as foo +WHERE b > + (SELECT count(*) + FROM my_films WHERE id = foo.a); + +-- we can recursively plan json_tables on set operations +(SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY)) ORDER BY id ASC LIMIT 1) +UNION +(SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY)) ORDER BY id ASC LIMIT 1) +UNION +(SELECT id FROM test_table ORDER BY id ASC LIMIT 1); + +-- LIMIT in subquery not supported when json_table exists +SELECT * +FROM json_table('[{"a":10,"b":20},{"a":30,"b":40}]'::JSONB, '$[*]' + COLUMNS (id FOR ORDINALITY, column_a int4 PATH '$.a', column_b int4 PATH '$.b', a int4, b int4, c text)) +JOIN LATERAL + (SELECT * + FROM my_films WHERE json_table.id = a LIMIT 1) as foo ON (true); + +RESET client_min_messages; + +-- test some utility functions on the target list & where clause +select jsonb_path_exists(js, '$.favorites') from my_films; +select bool_and(JSON_EXISTS(js, '$.favorites.films.title')) from my_films; +SELECT count(*) FROM my_films WHERE jsonb_path_exists(js, '$.favorites'); +SELECT count(*) FROM my_films WHERE jsonb_path_exists(js, '$.favorites'); +SELECT count(*) FROM my_films WHERE JSON_EXISTS(js, '$.favorites.films.title'); + +-- check constraint with json_exists +create table user_profiles ( + id bigserial, + addresses jsonb, + anyjson jsonb, + check (json_exists( addresses, '$.main' )) +); +select create_distributed_table('user_profiles', 'id'); +insert into user_profiles (addresses) VALUES (JSON_SCALAR('1')); +insert into user_profiles (addresses) VALUES ('{"main":"value"}'); + +-- we cannot insert because WITH UNIQUE KEYS +insert into user_profiles (addresses) VALUES (JSON ('{"main":"value", "main":"value"}' WITH UNIQUE KEYS)); + +-- we can insert with +insert into user_profiles (addresses) VALUES (JSON ('{"main":"value", "main":"value"}' WITHOUT UNIQUE KEYS)) RETURNING *; + +TRUNCATE user_profiles; +INSERT INTO user_profiles (anyjson) VALUES ('12'), ('"abc"'), ('[1,2,3]'), ('{"a":12}'); +select anyjson, anyjson is json array as json_array, anyjson is json object as json_object, anyjson is json scalar as json_scalar, +anyjson is json with UNIQUE keys +from user_profiles WHERE anyjson IS NOT NULL ORDER BY 1; + +-- use json_query +SELECT i, + json_query('[{"x": "aaa"},{"x": "bbb"},{"x": "ccc"}]'::JSONB, '$[$i].x' passing id AS i RETURNING text omit quotes) +FROM generate_series(0, 3) i +JOIN my_films ON(id = i); + +-- we can use JSON_TABLE in modification queries as well + +-- use log level such that we can see trace changes +SET client_min_messages TO DEBUG1; + +--the JSON_TABLE subquery is recursively planned +UPDATE test_table SET VALUE = 'XXX' FROM( +SELECT jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + id FOR ORDINALITY, + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt) as foo WHERE foo.id = test_table.id; + +-- Subquery with JSON table can be pushed down because two distributed tables +-- in the query are joined on distribution column +UPDATE test_table SET VALUE = 'XXX' FROM ( +SELECT my_films.id, jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt) as foo WHERE foo.id = test_table.id; + +-- we can pushdown with CTEs as well +WITH json_cte AS +(SELECT my_films.id, jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + title text PATH '$.title', + director text PATH '$.director'))) AS jt) +UPDATE test_table SET VALUE = 'XYZ' FROM json_cte + WHERE json_cte.id = test_table.id; + + -- we can recursively with CTEs as well +WITH json_cte AS +(SELECT my_films.id as film_id, jt.* FROM + my_films, + JSON_TABLE ( js, '$.favorites[*]' COLUMNS ( + kind text PATH '$.kind', + NESTED PATH '$.films[*]' COLUMNS ( + id FOR ORDINALITY, + title text PATH '$.title', + director text PATH '$.director'))) AS jt ORDER BY jt.id LIMIT 1) +UPDATE test_table SET VALUE = 'XYZ' FROM json_cte + WHERE json_cte.film_id = test_table.id; + +-- Should succeed (JSON arguments are passed to root and nested paths) +SELECT * +FROM + generate_series(1, 4) x, + generate_series(1, 3) y, + JSON_TABLE(jsonb + '[[1,2,3],[2,3,4,5],[3,4,5,6]]', + 'strict $[*] ? (@[*] < $x)' + PASSING x AS x, y AS y + COLUMNS ( + y text FORMAT JSON PATH '$', + NESTED PATH 'strict $[*] ? (@ >= $y)' + COLUMNS ( + z int PATH '$' + ) + ) + ) jt ORDER BY 4,1,2,3; + +SET client_min_messages TO ERROR; +DROP SCHEMA json_table CASCADE;