src/test/isolation2/global_sh_executor.sh - cloudberry - Git at Google

 #!/bin/bash

 set -eo pipefail
 # Set env var ${NL} because "\n" can not be converted to new line for unknown escaping reason
 export NL="
 "

 # Retrieve cell content from $RAW_STR according to the row and column numbers,
 # and set it to the given variable name.
 # Arg 1 (output): Variable name.
 # Arg 2 (input): Row number.
 # Arg 3 (input): Column number.
 # Example:
 # 1: @post_run 'get_cell USER_NAME 1 2': SELECT id,name,state FROM users;
 # Assume 'SELECT id,name,state FROM users;' returns:
 # | id | name  | state |
 # |----+-------+--------|
 # | 1  | Jonh  | Alive  |
 # | 2  | Alice | Dead   |
 # by calling `get_cell USER_NAME 1 2' will set $USER_NAME to 'Jonh'.
 get_tuple_cell() {
     var_name=$1
     row=$2
     col=$3
     cmd="echo \"\$RAW_STR\" | awk -F '|' 'NR==(($row+2)) {print \$$col}' | awk '{\$1=\$1;print}'"
     output=`eval $cmd`
     eval $var_name="$output"
 }

 # Generate $MATCHSUBS and echo the $RAWSTR based on the given original string and replacement pairs.
 # Arg 1n (input): The original string to be replaced.
 # Arg 2n (input): The replacement string.
 # Example:
 # 1: @post_run 'genereate_match_sub $USER_NAME user1 $USER_ID id1': SELECT id,name,state FROM users;
 # here we assume $USER_NAME and $USER_ID has been set to 'Jonh' and '1' already. Then the above
 # statement will generate $MATCHSUBS section:
 # m/\bJonh\b/
 # s/\bJonh\b/user1/
 # \b here is for matching the whole word. (word boundaries)
 create_match_sub() {
     to_replace=""
     for var in "$@"
         do
         if [ -z "$to_replace" ]
         then
             to_replace=$var
         else
             # \b is trying to match the whole word to make it more stable.
             export MATCHSUBS="${MATCHSUBS}${NL}m/\\b${to_replace}\\b/${NL}s/\\b${to_replace}\\b/${var}/${NL}"
             to_replace=""
         fi
     done
     echo "${RAW_STR}"
 }

 # Generate $MATCHSUBS and Trim the Tailing spaces.
 # This is similar to create_match_sub() but dealing with the tailing spaces.
 # Sometimes we have variable length cells, like userid:
 # | username | userid | gender |
 # |----------+--------+--------|
 # | john     | 12     | male   |
 # we need to match the 12 with a var $USERID which has been set by get_call().
 # The output source will be something like:
 # | username | userid | gender |
 # |----------+--------+--------|
 # | john     | userid1     | male   |
 # to match it: match_sub userid1 $USERID
 # but the problem here is the userid may change for different test executions. If we
 # get a 3 digits userid like '123', the diff will fail since we have one more space than
 # the actual sql output.
 # To deal with it, use match_sub_tt userid $USERID
 # And make the output source like (note: append one space to the replaced string):
 # | username | userid  | gender |
 # |----------+---------+--------|
 # | john     | userid1 | male   |
 # Notice here that there is no space following userid1 since we replace the whole userid with
 # its tailing spaces with 'userid1'. Like '123   ' -> 'userid'.
 create_match_sub_with_spaces() {
     to_replace=""
     for var in "$@"
         do
         if [ -z "$to_replace" ]
         then
             to_replace=$var
         else
             # \b is trying to match the whole word to make it more stable.
             export MATCHSUBS="${MATCHSUBS}${NL}m/\\b${to_replace}\\b/${NL}s/\\b${to_replace} */${var} /${NL}"
             to_replace=""
         fi
     done
     echo "${RAW_STR}"
 }

 # Substitute in the $RAW_STR and echo the result.
 # Multi substitution pairs can be passed as arguments, like:
 # sub "to_replace_1" "replacement_1" "to_replace_2" "replacement_2"
 # This could be useful for both @pre_run and @post_run. e.g.:
 # @pre_run 'sub @TOKEN1 ${TOKEN1}': SELECT state FROM GP_ENDPOINTS_STATUS_INFO() WHERE token='@TOKEN1';
 # Assume the $TOKEN has value '01234', The SQL will become:
 # SELECT state FROM GP_ENDPOINTS_STATUS_INFO() WHERE token='01234';
 create_sub() {
     to_replace=""
     for var in "$@"
         do
         if [ -z "$to_replace" ]
         then
             to_replace=$var
         else
             RAW_STR=$(echo "$RAW_STR" | sed -E "s/${to_replace}/${var}/g")
             to_replace=""
         fi
     done
     echo "${RAW_STR}"
 }

 # Parse the endpoint state info output and save them into environment variables for @post_run.
 # Usage: parse_endpoint_info <postfix> <endpoint_col> <token_col> <host_col> <port_col>
 # Output(environment variables):
 #   "TOKEN$postfix"
 #   "ENDPOINT_NAME$postfix[]"
 #   "ENDPOINT_TOKEN$postfix[]"
 #   "ENDPOINT_HOST$postfix[]"
 #   "ENDPOINT_PORT$postfix[]"
 # e.g.:
 # For the given SQL result:
 #      endpointname     |             token                |  hostname | port  | state
 # ----------------------+----------------------------------+-------------+-------+--------
 #  c1_00001507_00000000 | 071500004015dc6da471b20417afed65 | host_1111 | 25432 | READY
 #  c1_00001507_00000001 | 071500004015dc6da471b20417afed65 | host_1112 | 25433 | READY
 #  c1_00001507_00000002 | 071500004015dc6da471b20417afed65 | host_1113 | 25434 | READY
 # (3 rows)
 # parse_endpoint_info 1 1 2 3 4 will setup below variables:
 # TOEKN1='071500004015dc6da471b20417afed65'
 # ENDPOINT_NAME1[0]='c1_00001507_00000000'
 # ENDPOINT_TOKEN1[0]='071500004015dc6da471b20417afed65'
 # ENDPOINT_HOST1[0]='host_1111'
 # ENDPOINT_PORT1[0]='25432'
 # ENDPOINT_NAME1[1]='c1_00001507_00000001'
 # ENDPOINT_TOKEN1[1]='071500004015dc6da471b20417afed65'
 # ENDPOINT_HOST1[1]='host_1112'
 # ENDPOINT_PORT1[1]='25433'
 # ENDPOINT_NAME1[2]='c1_00001507_00000002'
 # ENDPOINT_TOKEN1[2]='071500004015dc6da471b20417afed65'
 # ENDPOINT_HOST1[2]='host_1113'
 # ENDPOINT_PORT1[2]='25434'
 parse_endpoint_info() {
     local postfix=$1
     local endpoint_name_col=$2
     local token_col=$3
     local host_col=$4
     local port_col=$5
     export CURRENT_ENDPOINT_POSTFIX="${postfix}"

     eval "ENDPOINT_NAME${postfix}=()"
     eval "ENDPOINT_TOKEN${postfix}=()"
     eval "ENDPOINT_HOST${postfix}=()"
     eval "ENDPOINT_PORT${postfix}=()"
     while IFS= read -r line ; do
         local name=""
         name="$(echo "${line}" | awk -F '|' "{print \$${endpoint_name_col}}" | awk '{$1=$1;print}')"
         local token=""
         token="$(echo "${line}" | awk -F '|' "{print \$${token_col}}" | awk '{$1=$1;print}')"
         local host=""
         host="$(echo "${line}" | awk -F '|' "{print \$${host_col}}" | awk '{$1=$1;print}' )"
         local port=""
         port="$(echo "${line}" | awk -F '|' "{print \$${port_col}}" | awk '{$1=$1;print}' )"
         eval "ENDPOINT_NAME${postfix}+=(${name})"
         eval "ENDPOINT_TOKEN${postfix}+=(${token})"
         eval "ENDPOINT_HOST${postfix}+=(${host})"
         eval "ENDPOINT_PORT${postfix}+=(${port})"

         eval "TOKEN${postfix}=${token}"
         export RETRIEVE_TOKEN=${token}

         create_match_sub_with_spaces  "${name}" "endpoint_id${postfix}" \
             "${port}" port_id \
             "${token}" token_id \
             "${host}" host_id > /dev/null

     # Filter out the first two lines and the last line.
     done <<<"$(echo "$RAW_STR" | sed '1,2d;$d')"
     # Ignore first 2 lines(table header) since hostname length may affect the diff result.
     echo "${RAW_STR}" | sed '1,2d'
 }

 # Find the corresponding endpoint in the environment variables saved by previous
 # parse_endpoint call, and substitute in the SQL. Used by @pre_run.
 # The finding process relies on the $GP_HOSTNAME and $GP_PORT to be set to the
 # current postgres connection.
 # Usage: set_endpoint_variable <ENDPOINT_STR><postfix>
 # e.g.:
 # set_endpoint_variable "@ENDPOINT1"
 # This will replace "@ENDPOINT1" in the SQL statement with the corresponding endpoint name
 # with postfix "1".
 set_endpoint_variable() {
     export CURRENT_ENDPOINT_POSTFIX=""
     CURRENT_ENDPOINT_POSTFIX="$(echo "$1" | sed 's/@ENDPOINT//')"
     eval "local names=(\${ENDPOINT_NAME${CURRENT_ENDPOINT_POSTFIX}[@]})"
     eval "local hosts=(\"\${ENDPOINT_HOST${CURRENT_ENDPOINT_POSTFIX}[@]}\")"
     eval "ports=(\${ENDPOINT_PORT${CURRENT_ENDPOINT_POSTFIX}[@]})"
     local i=0
     for h in "${hosts[@]}" ; do
         if [ "$GP_HOSTNAME" = "$h" ] ; then
             if [ "$GP_PORT" = "${ports[$i]}" ] ; then
                 create_sub "$1" "${names[$i]}"
                 return
             fi
         fi
         i=$((i+1))
     done
     # echo "Cannot find endpoint for postfix '$CURRENT_ENDPOINT_POSTFIX', '$GP_HOSTNAME', '$GP_PORT'."
     echo $RAW_STR
 }

 # Return the retrieve token based on the $CURRENT_ENDPOINT_POSTFIX, $GP_HOSTNAME and $GP_PORT
 get_retrieve_token() {
     eval "local tokens=(\${ENDPOINT_TOKEN${CURRENT_ENDPOINT_POSTFIX}[@]})"
     eval "local hosts=(\"\${ENDPOINT_HOST${CURRENT_ENDPOINT_POSTFIX}[@]}\")"
     eval "ports=(\${ENDPOINT_PORT${CURRENT_ENDPOINT_POSTFIX}[@]})"
     local i=0
     for h in "${hosts[@]}" ; do
         if [ "$GP_HOSTNAME" = "$h" ] ; then
             if [ "$GP_PORT" = "${ports[$i]}" ] ; then
                 echo "${tokens[$i]}"
                 return
             fi
         fi
         i=$((i+1))
     done
     echo "Can not find the token."
 }
	#!/bin/bash

	set -eo pipefail
	# Set env var ${NL} because "\n" can not be converted to new line for unknown escaping reason
	export NL="
	"

	# Retrieve cell content from $RAW_STR according to the row and column numbers,
	# and set it to the given variable name.
	# Arg 1 (output): Variable name.
	# Arg 2 (input): Row number.
	# Arg 3 (input): Column number.
	# Example:
	# 1: @post_run 'get_cell USER_NAME 1 2': SELECT id,name,state FROM users;
	# Assume 'SELECT id,name,state FROM users;' returns:
	# \| id \| name \| state \|
	# \|----+-------+--------\|
	# \| 1 \| Jonh \| Alive \|
	# \| 2 \| Alice \| Dead \|
	# by calling `get_cell USER_NAME 1 2' will set $USER_NAME to 'Jonh'.
	get_tuple_cell() {
	var_name=$1
	row=$2
	col=$3
	cmd="echo \"\$RAW_STR\" \| awk -F '\|' 'NR==(($row+2)) {print \$$col}' \| awk '{\$1=\$1;print}'"
	output=`eval $cmd`
	eval $var_name="$output"
	}

	# Generate $MATCHSUBS and echo the $RAWSTR based on the given original string and replacement pairs.
	# Arg 1n (input): The original string to be replaced.
	# Arg 2n (input): The replacement string.
	# Example:
	# 1: @post_run 'genereate_match_sub $USER_NAME user1 $USER_ID id1': SELECT id,name,state FROM users;
	# here we assume $USER_NAME and $USER_ID has been set to 'Jonh' and '1' already. Then the above
	# statement will generate $MATCHSUBS section:
	# m/\bJonh\b/
	# s/\bJonh\b/user1/
	# \b here is for matching the whole word. (word boundaries)
	create_match_sub() {
	to_replace=""
	for var in "$@"
	do
	if [ -z "$to_replace" ]
	then
	to_replace=$var
	else
	# \b is trying to match the whole word to make it more stable.
	export MATCHSUBS="${MATCHSUBS}${NL}m/\\b${to_replace}\\b/${NL}s/\\b${to_replace}\\b/${var}/${NL}"
	to_replace=""
	fi
	done
	echo "${RAW_STR}"
	}

	# Generate $MATCHSUBS and Trim the Tailing spaces.
	# This is similar to create_match_sub() but dealing with the tailing spaces.
	# Sometimes we have variable length cells, like userid:
	# \| username \| userid \| gender \|
	# \|----------+--------+--------\|
	# \| john \| 12 \| male \|
	# we need to match the 12 with a var $USERID which has been set by get_call().
	# The output source will be something like:
	# \| username \| userid \| gender \|
	# \|----------+--------+--------\|
	# \| john \| userid1 \| male \|
	# to match it: match_sub userid1 $USERID
	# but the problem here is the userid may change for different test executions. If we
	# get a 3 digits userid like '123', the diff will fail since we have one more space than
	# the actual sql output.
	# To deal with it, use match_sub_tt userid $USERID
	# And make the output source like (note: append one space to the replaced string):
	# \| username \| userid \| gender \|
	# \|----------+---------+--------\|
	# \| john \| userid1 \| male \|
	# Notice here that there is no space following userid1 since we replace the whole userid with
	# its tailing spaces with 'userid1'. Like '123 ' -> 'userid'.
	create_match_sub_with_spaces() {
	to_replace=""
	for var in "$@"
	do
	if [ -z "$to_replace" ]
	then
	to_replace=$var
	else
	# \b is trying to match the whole word to make it more stable.
	export MATCHSUBS="${MATCHSUBS}${NL}m/\\b${to_replace}\\b/${NL}s/\\b${to_replace} */${var} /${NL}"
	to_replace=""
	fi
	done
	echo "${RAW_STR}"
	}

	# Substitute in the $RAW_STR and echo the result.
	# Multi substitution pairs can be passed as arguments, like:
	# sub "to_replace_1" "replacement_1" "to_replace_2" "replacement_2"
	# This could be useful for both @pre_run and @post_run. e.g.:
	# @pre_run 'sub @TOKEN1 ${TOKEN1}': SELECT state FROM GP_ENDPOINTS_STATUS_INFO() WHERE token='@TOKEN1';
	# Assume the $TOKEN has value '01234', The SQL will become:
	# SELECT state FROM GP_ENDPOINTS_STATUS_INFO() WHERE token='01234';
	create_sub() {
	to_replace=""
	for var in "$@"
	do
	if [ -z "$to_replace" ]
	then
	to_replace=$var
	else
	RAW_STR=$(echo "$RAW_STR" \| sed -E "s/${to_replace}/${var}/g")
	to_replace=""
	fi
	done
	echo "${RAW_STR}"
	}

	# Parse the endpoint state info output and save them into environment variables for @post_run.
	# Usage: parse_endpoint_info <postfix> <endpoint_col> <token_col> <host_col> <port_col>
	# Output(environment variables):
	# "TOKEN$postfix"
	# "ENDPOINT_NAME$postfix[]"
	# "ENDPOINT_TOKEN$postfix[]"
	# "ENDPOINT_HOST$postfix[]"
	# "ENDPOINT_PORT$postfix[]"
	# e.g.:
	# For the given SQL result:
	# endpointname \| token \| hostname \| port \| state
	# ----------------------+----------------------------------+-------------+-------+--------
	# c1_00001507_00000000 \| 071500004015dc6da471b20417afed65 \| host_1111 \| 25432 \| READY
	# c1_00001507_00000001 \| 071500004015dc6da471b20417afed65 \| host_1112 \| 25433 \| READY
	# c1_00001507_00000002 \| 071500004015dc6da471b20417afed65 \| host_1113 \| 25434 \| READY
	# (3 rows)
	# parse_endpoint_info 1 1 2 3 4 will setup below variables:
	# TOEKN1='071500004015dc6da471b20417afed65'
	# ENDPOINT_NAME1[0]='c1_00001507_00000000'
	# ENDPOINT_TOKEN1[0]='071500004015dc6da471b20417afed65'
	# ENDPOINT_HOST1[0]='host_1111'
	# ENDPOINT_PORT1[0]='25432'
	# ENDPOINT_NAME1[1]='c1_00001507_00000001'
	# ENDPOINT_TOKEN1[1]='071500004015dc6da471b20417afed65'
	# ENDPOINT_HOST1[1]='host_1112'
	# ENDPOINT_PORT1[1]='25433'
	# ENDPOINT_NAME1[2]='c1_00001507_00000002'
	# ENDPOINT_TOKEN1[2]='071500004015dc6da471b20417afed65'
	# ENDPOINT_HOST1[2]='host_1113'
	# ENDPOINT_PORT1[2]='25434'
	parse_endpoint_info() {
	local postfix=$1
	local endpoint_name_col=$2
	local token_col=$3
	local host_col=$4
	local port_col=$5
	export CURRENT_ENDPOINT_POSTFIX="${postfix}"

	eval "ENDPOINT_NAME${postfix}=()"
	eval "ENDPOINT_TOKEN${postfix}=()"
	eval "ENDPOINT_HOST${postfix}=()"
	eval "ENDPOINT_PORT${postfix}=()"
	while IFS= read -r line ; do
	local name=""
	name="$(echo "${line}" \| awk -F '\|' "{print \$${endpoint_name_col}}" \| awk '{$1=$1;print}')"
	local token=""
	token="$(echo "${line}" \| awk -F '\|' "{print \$${token_col}}" \| awk '{$1=$1;print}')"
	local host=""
	host="$(echo "${line}" \| awk -F '\|' "{print \$${host_col}}" \| awk '{$1=$1;print}' )"
	local port=""
	port="$(echo "${line}" \| awk -F '\|' "{print \$${port_col}}" \| awk '{$1=$1;print}' )"
	eval "ENDPOINT_NAME${postfix}+=(${name})"
	eval "ENDPOINT_TOKEN${postfix}+=(${token})"
	eval "ENDPOINT_HOST${postfix}+=(${host})"
	eval "ENDPOINT_PORT${postfix}+=(${port})"

	eval "TOKEN${postfix}=${token}"
	export RETRIEVE_TOKEN=${token}

	create_match_sub_with_spaces "${name}" "endpoint_id${postfix}" \
	"${port}" port_id \
	"${token}" token_id \
	"${host}" host_id > /dev/null

	# Filter out the first two lines and the last line.
	done <<<"$(echo "$RAW_STR" \| sed '1,2d;$d')"
	# Ignore first 2 lines(table header) since hostname length may affect the diff result.
	echo "${RAW_STR}" \| sed '1,2d'
	}

	# Find the corresponding endpoint in the environment variables saved by previous
	# parse_endpoint call, and substitute in the SQL. Used by @pre_run.
	# The finding process relies on the $GP_HOSTNAME and $GP_PORT to be set to the
	# current postgres connection.
	# Usage: set_endpoint_variable <ENDPOINT_STR><postfix>
	# e.g.:
	# set_endpoint_variable "@ENDPOINT1"
	# This will replace "@ENDPOINT1" in the SQL statement with the corresponding endpoint name
	# with postfix "1".
	set_endpoint_variable() {
	export CURRENT_ENDPOINT_POSTFIX=""
	CURRENT_ENDPOINT_POSTFIX="$(echo "$1" \| sed 's/@ENDPOINT//')"
	eval "local names=(\${ENDPOINT_NAME${CURRENT_ENDPOINT_POSTFIX}[@]})"
	eval "local hosts=(\"\${ENDPOINT_HOST${CURRENT_ENDPOINT_POSTFIX}[@]}\")"
	eval "ports=(\${ENDPOINT_PORT${CURRENT_ENDPOINT_POSTFIX}[@]})"
	local i=0
	for h in "${hosts[@]}" ; do
	if [ "$GP_HOSTNAME" = "$h" ] ; then
	if [ "$GP_PORT" = "${ports[$i]}" ] ; then
	create_sub "$1" "${names[$i]}"
	return
	fi
	fi
	i=$((i+1))
	done
	# echo "Cannot find endpoint for postfix '$CURRENT_ENDPOINT_POSTFIX', '$GP_HOSTNAME', '$GP_PORT'."
	echo $RAW_STR
	}

	# Return the retrieve token based on the $CURRENT_ENDPOINT_POSTFIX, $GP_HOSTNAME and $GP_PORT
	get_retrieve_token() {
	eval "local tokens=(\${ENDPOINT_TOKEN${CURRENT_ENDPOINT_POSTFIX}[@]})"
	eval "local hosts=(\"\${ENDPOINT_HOST${CURRENT_ENDPOINT_POSTFIX}[@]}\")"
	eval "ports=(\${ENDPOINT_PORT${CURRENT_ENDPOINT_POSTFIX}[@]})"
	local i=0
	for h in "${hosts[@]}" ; do
	if [ "$GP_HOSTNAME" = "$h" ] ; then
	if [ "$GP_PORT" = "${ports[$i]}" ] ; then
	echo "${tokens[$i]}"
	return
	fi
	fi
	i=$((i+1))
	done
	echo "Can not find the token."
	}