ClickHouse/dbms/tests/queries/0_stateless/01060_avro.sh

71 lines
4.6 KiB
Bash
Raw Normal View History

2020-01-10 09:05:20 +00:00
#!/usr/bin/env bash
set -e
CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
. $CUR_DIR/../shell_config.sh
DATA_DIR=$CUR_DIR/data_avro
# input
echo === input
echo = primitive
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/primitive.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a_bool UInt8, b_int Int32, c_long Int64, d_float Float32, e_double Float64, f_bytes String, g_string String' -q 'select * from table'
cat $DATA_DIR/primitive.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a_bool UInt8, c_long Int64, g_string String' -q 'select * from table'
cat $DATA_DIR/primitive.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'g_string String, c_long Int64, a_bool UInt8' -q 'select * from table'
cat $DATA_DIR/primitive.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'g_string String' -q 'select * from table'
2020-01-10 09:05:20 +00:00
echo = complex
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/complex.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "a_enum_to_string String, b_enum_to_enum Enum('t' = 1, 'f' = 0), c_array_string Array(String), d_array_array_string Array(Array(String)), e_union_null_string Nullable(String), f_union_long_null Nullable(Int64), g_fixed FixedString(32)" -q 'select * from table'
cat $DATA_DIR/complex.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "g_fixed FixedString(32)" -q 'select * from table'
2020-01-10 09:05:20 +00:00
echo = logical_types
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/logical_types.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "a_date Date, b_timestamp_millis DateTime64(3, 'UTC'), c_timestamp_micros DateTime64(6, 'UTC')" -q 'select * from table'
cat $DATA_DIR/logical_types.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a_date Int32, b_timestamp_millis Int64, c_timestamp_micros Int64' -q 'select * from table'
2020-01-10 09:05:20 +00:00
2020-02-01 04:13:12 +00:00
echo = references
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/references.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "a String, c String" -q 'select * from table'
2020-01-10 09:05:20 +00:00
echo = compression
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/simple.null.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a Int64' -q 'select count() from table'
cat $DATA_DIR/simple.deflate.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a Int64' -q 'select count() from table'
#snappy is optional
2020-02-03 00:04:08 +00:00
#cat $DATA_DIR/simple.snappy.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a Int64' -q 'select count() from table'
2020-01-10 09:05:20 +00:00
echo = other
#no data
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/empty.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a Int64' -q 'select count() from table'
2020-01-10 09:05:20 +00:00
# type mismatch
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/simple.null.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'a Int32' -q 'select count() from table'
2020-01-10 09:05:20 +00:00
# field not found
2020-02-03 00:04:08 +00:00
cat $DATA_DIR/simple.null.avro | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S 'b Int64' -q 'select count() from table' 2>&1 | grep -i 'not found' -o
2020-01-10 09:05:20 +00:00
# output
echo === output
echo = primitive
S1="a_bool UInt8, b_int Int32, c_long Int64, d_float Float32, e_double Float64, f_bytes String, g_string String"
2020-02-03 00:04:08 +00:00
echo '1,1,2,3.4,5.6,"b1","s1"' | ${CLICKHOUSE_LOCAL} --input-format CSV -S "$S1" -q "select * from table format Avro" | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "$S1" -q 'select * from table'
2020-01-10 09:05:20 +00:00
echo = complex
S2="a_enum_to_string String, b_enum_to_enum Enum('t' = 1, 'f' = 0), c_array_string Array(String), d_array_array_string Array(Array(String)), e_union_null_string Nullable(String), f_union_long_null Nullable(Int64), g_fixed FixedString(32)"
2020-02-03 00:04:08 +00:00
echo "\"A\",\"t\",\"['s1','s2']\",\"[['a1'],['a2']]\",\"s1\",\N,\"79cd909892d7e7ade1987cc7422628ba\"" | ${CLICKHOUSE_LOCAL} --input-format CSV -S "$S2" -q "select * from table format Avro" | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "$S2" -q 'select * from table'
2020-01-10 09:05:20 +00:00
echo = logical_types
2020-01-10 22:44:33 +00:00
S3="a_date Date, b_timestamp_millis DateTime64(3, 'UTC'), c_timestamp_micros DateTime64(6, 'UTC')"
2020-02-03 00:04:08 +00:00
echo '"2019-12-20","2020-01-10 07:31:56.227","2020-01-10 07:31:56.227000"' | ${CLICKHOUSE_LOCAL} --input-format CSV -S "$S3" -q "select * from table format Avro" | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "$S3" -q 'select * from table'
2020-01-10 09:05:20 +00:00
echo = other
S4="a Int64"
2020-02-03 00:04:08 +00:00
${CLICKHOUSE_LOCAL} -q "select toInt64(number) as a from numbers(0) format Avro" | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "$S4" -q 'select count() from table'
${CLICKHOUSE_LOCAL} -q "select toInt64(number) as a from numbers(1000) format Avro" | ${CLICKHOUSE_LOCAL} --input-format Avro --output-format CSV -S "$S4" -q 'select count() from table'
2020-01-10 09:05:20 +00:00
2020-01-23 02:51:49 +00:00
# type supported via conversion
${CLICKHOUSE_LOCAL} -q "select toInt16(123) as a format Avro" | wc -c