ClickHouse/tests/queries/0_stateless/02732_rename_after_processing.sh

130 lines
4.2 KiB
Bash
Executable File
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

#!/usr/bin/env bash
CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
# shellcheck source=../shell_config.sh
. "$CURDIR"/../shell_config.sh
# see 01658_read_file_to_stringcolumn.sh
CLICKHOUSE_USER_FILES_PATH=$(clickhouse-client --query "select _path, _file from file('nonexist.txt', 'CSV', 'val1 char')" 2>&1 | grep Exception | awk '{gsub("/nonexist.txt","",$9); print $9}')
# Prepare data
unique_name=${CLICKHOUSE_TEST_UNIQUE_NAME}
tmp_dir=${CLICKHOUSE_USER_FILES_PATH}/${unique_name}
mkdir -p $tmp_dir
rm -rf ${tmp_dir:?}/*
chmod 777 ${tmp_dir}
echo '"id","str","int","text"' > ${tmp_dir}/tmp.csv
echo '1,"abc",123,"abacaba"' >> ${tmp_dir}/tmp.csv
echo '2,"def",456,"bacabaa"' >> ${tmp_dir}/tmp.csv
echo '3,"story",78912,"acabaab"' >> ${tmp_dir}/tmp.csv
echo '4,"history",21321321,"cabaaba"' >> ${tmp_dir}/tmp.csv
chmod 777 ${tmp_dir}/tmp.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp1.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp2.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp3_1.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp3_2.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp4.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp5.csv
cp ${tmp_dir}/tmp.csv ${tmp_dir}/tmp6.csv
### Checking that renaming works
# simple select
${CLICKHOUSE_CLIENT} --rename-files-after-processing="processed_%f%e" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp1.csv')"
if [ -e "${tmp_dir}/processed_tmp1.csv" ]; then
echo "processed_tmp1.csv"
fi
if [ ! -e "${tmp_dir}/tmp1.csv" ]; then
echo "!tmp1.csv"
fi
# select with multiple file() calls
${CLICKHOUSE_CLIENT} --rename-files-after-processing="processed_%f%e" --multiline -q """
SELECT
sum(a.id) as aid,
sum(b.id) as bid
FROM file('${unique_name}/tmp2.csv') AS a
INNER JOIN file('${unique_name}/tmp2.csv') AS b
ON a.text = b.text
"""
if [ -e "${tmp_dir}/processed_tmp2.csv" ]; then
echo "processed_tmp2.csv"
fi
if [ ! -e "${tmp_dir}/tmp2.csv" ]; then
echo "!tmp2.csv"
fi
# rename multiple files
${CLICKHOUSE_CLIENT} --rename-files-after-processing="processed_%f%e" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp3*.csv')"
if [ -e "${tmp_dir}/processed_tmp3_1.csv" ]; then
echo "processed_tmp3_1.csv"
fi
if [ -e "${tmp_dir}/processed_tmp3_2.csv" ]; then
echo "processed_tmp3_2.csv"
fi
if [ ! -e "${tmp_dir}/tmp3_1.csv" ]; then
echo "!tmp3_1.csv"
fi
if [ ! -e "${tmp_dir}/tmp3_2.csv" ]; then
echo "!tmp3_2.csv"
fi
# check timestamp placeholder
${CLICKHOUSE_CLIENT} --rename-files-after-processing="processed_%f_%t.csv" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp4.csv')"
# ls ${tmp_dir} | grep -E "^processed_tmp4_[0-9]+\.csv$" > /dev/null && echo "OK"
rg="processed_tmp4_[0-9]+\.csv"
for x in "${tmp_dir}"/processed*; do
if [[ $x =~ $rg ]]; then
echo "OK"
break
fi;
done
### Checking errors
# cannot overwrite an existing file
${CLICKHOUSE_CLIENT} --rename-files-after-processing="tmp.csv" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp5.csv')" \
2>&1| grep "already exists" > /dev/null && echo "OK"
if [ -e "${tmp_dir}/tmp5.csv" ]; then
echo "tmp5.csv"
fi
# сannot move file outside user_files
${CLICKHOUSE_CLIENT} --rename-files-after-processing="../../%f%e" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp5.csv')" \
2>&1| grep "is not inside" > /dev/null && echo "OK"
if [ -e "${tmp_dir}/tmp5.csv" ]; then
echo "tmp5.csv"
fi
# check invalid placeholders
# unknown type of placeholder (%k)
${CLICKHOUSE_CLIENT} --rename-files-after-processing="processed_%f_%k" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp5.csv')" \
2>&1| grep "Allowed placeholders only" > /dev/null && echo "OK"
if [ -e "${tmp_dir}/tmp5.csv" ]; then
echo "tmp5.csv"
fi
# dd number of consecutive percentage signs after replace valid placeholders
${CLICKHOUSE_CLIENT} --rename-files-after-processing="processed_%f_%%%%e" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp5.csv')" \
2>&1| grep "Odd number of consecutive percentage signs" > /dev/null && echo "OK"
if [ -e "${tmp_dir}/tmp5.csv" ]; then
echo "tmp5.csv"
fi
# check full file name placeholder
${CLICKHOUSE_CLIENT} --rename-files-after-processing="%a.processed" -q "SELECT COUNT(*) FROM file('${unique_name}/tmp6.csv')"
if [ -e "${tmp_dir}/tmp6.csv.processed" ]; then
echo "tmp6.csv.processed"
fi
if [ ! -e "${tmp_dir}/tmp6.csv" ]; then
echo "!tmp6.csv"
fi
# Clean
rm -rd $tmp_dir