# name: test/sql/copy/parquet/parquet_metadata.test
# description: Test parquet metadata function
# group: [parquet]

require parquet

statement ok
SELECT * FROM parquet_metadata('data/parquet-testing/lineitem-top10000.gzip.parquet');

statement ok
SELECT * FROM parquet_schema('data/parquet-testing/lineitem-top10000.gzip.parquet');

query I
SELECT COUNT(*) > 0 FROM parquet_metadata('data/parquet-testing/lineitem-top10000.gzip.parquet');
----
true

query I
SELECT COUNT(*) > 0 FROM parquet_schema('data/parquet-testing/lineitem-top10000.gzip.parquet');
----
true

statement ok
select * from parquet_schema('data/parquet-testing/decimal/decimal_dc.parquet');

statement ok
select * from parquet_schema('data/parquet-testing/decimal/int64_decimal.parquet');

# with globs
statement ok
select * from parquet_metadata('data/parquet-testing/glob/*.parquet');

statement ok
select * from parquet_schema('data/parquet-testing/glob/*.parquet');

# list parameters
statement ok
select * from parquet_schema(['data/parquet-testing/decimal/int64_decimal.parquet', 'data/parquet-testing/decimal/int64_decimal.parquet']);

query III
SELECT name, type, duckdb_type FROM parquet_schema('data/parquet-testing/lineitem-top10000.gzip.parquet') WHERE type IS NOT NULL;
----
l_orderkey	INT64	BIGINT
l_partkey	INT64	BIGINT
l_suppkey	INT64	BIGINT
l_linenumber	INT32	INTEGER
l_quantity	INT32	INTEGER
l_extendedprice	DOUBLE	DOUBLE
l_discount	DOUBLE	DOUBLE
l_tax	DOUBLE	DOUBLE
l_returnflag	BYTE_ARRAY	VARCHAR
l_linestatus	BYTE_ARRAY	VARCHAR
l_shipdate	BYTE_ARRAY	VARCHAR
l_commitdate	BYTE_ARRAY	VARCHAR
l_receiptdate	BYTE_ARRAY	VARCHAR
l_shipinstruct	BYTE_ARRAY	VARCHAR
l_shipmode	BYTE_ARRAY	VARCHAR
l_comment	BYTE_ARRAY	VARCHAR

# column_id
query II
SELECT column_id, name FROM parquet_schema('data/parquet-testing/lineitem-top10000.gzip.parquet') ORDER BY column_id;
----
0	spark_schema
1	l_orderkey
2	l_partkey
3	l_suppkey
4	l_linenumber
5	l_quantity
6	l_extendedprice
7	l_discount
8	l_tax
9	l_returnflag
10	l_linestatus
11	l_shipdate
12	l_commitdate
13	l_receiptdate
14	l_shipinstruct
15	l_shipmode
16	l_comment

query III
WITH per_file AS (
    SELECT file_name, COUNT(*) AS rows_per_file
    FROM parquet_schema('data/parquet-testing/glob3/**/*.parquet')
    GROUP BY file_name
)
SELECT
    SUM(rows_per_file) AS total_rows,
    MAX(rows_per_file) AS max_rows_per_filename,
    (SELECT COUNT(DISTINCT column_id) FROM parquet_schema('data/parquet-testing/glob3/**/*.parquet')) AS distinct_column_ids
FROM per_file;
----
9	3	3