Travailler avec des données Avro, Arrow et ORC dans ClickHouse

INSERT INTO sometable
FROM INFILE 'data.avro'
FORMAT Avro

SELECT path, hits
FROM file('data.avro', Avro)
ORDER BY hits DESC
LIMIT 5;

┌─path────────────┬──hits─┐
│ Amy_Poehler     │ 62732 │
│ Adam_Goldberg   │ 42338 │
│ Aaron_Spelling  │ 25128 │
│ Absence_seizure │ 18152 │
│ Ammon_Bundy     │ 11890 │
└─────────────────┴───────┘

SELECT * FROM sometable
INTO OUTFILE 'export.avro'
FORMAT Avro;

SELECT
    date,
    toDate(date)
FROM file('data.avro', Avro)
LIMIT 3;

┌──date─┬─toDate(date)─┐
│ 16556 │   2015-05-01 │
│ 16556 │   2015-05-01 │
│ 16556 │   2015-05-01 │
└───────┴──────────────┘

CREATE TABLE some_topic_stream
(
    field1 UInt32,
    field2 String
)
ENGINE = Kafka() SETTINGS
kafka_broker_list = 'localhost',
kafka_topic_list = 'some_topic',
kafka_group_name = 'some_group',
kafka_format = 'AvroConfluent';

INSERT INTO sometable
FROM INFILE 'data.arrow'
FORMAT Arrow

SELECT * FROM sometable
INTO OUTFILE 'export.arrow'
FORMAT Arrow

import sys, pyarrow as pa

with pa.ipc.open_stream(sys.stdin.buffer) as reader:
  print(reader.read_pandas())

clickhouse-client -q "SELECT path, hits FROM some_data LIMIT 3 FORMAT ArrowStream" | python3 arrow.py

                           path  hits
0       b'Akiba_Hebrew_Academy'   241
1           b'Aegithina_tiphia'    34
2  b'1971-72_Utah_Stars_season'     1

arrow-stream | clickhouse-client -q "INSERT INTO sometable FORMAT ArrowStream"

SELECT *
FROM sometable
INTO OUTFILE 'data.orc'
FORMAT ORC;

INSERT INTO sometable
FROM INFILE 'data.orc'
FORMAT ORC;