multi calc outputs (reuse file IO)

[tooptoop4]

how to get 2 computations done on 1 disk read?
ie instead of sum_result=$(cat file | sum column1); crc32_result=$(cat file | crc32 column1); can both the sum and crc32 of the column be done off single disk read?

[jqnatividad]

Hi @tooptoop4 ,
Yes you can, with the luau command.

qsv/src/cmd/luau.rs

Lines 50 to 51 in 97b56c9

	Map multiple new columns in one pass
	$ qsv luau map newcol1,newcol2,newcol3 "{cola + 1, colb + 2, colc + 3}"

And you can even have a full blown luau script:

qsv/tests/test_luau.rs

Lines 1608 to 1696 in 97b56c9

	#[test]
	fn luau_aggregation_with_embedded_begin_end_using_file_random_access_multiple_columns() {
	let wrk = Workdir::new("luau_embedded_multiple_columns");
	wrk.create_indexed(
	"data.csv",
	vec![
	svec!["letter", "Amount"],
	svec!["a", "13"],
	svec!["b", "24"],
	svec!["c", "72"],
	svec!["d", "7"],
	],
	);
	wrk.create_from_string(
	"testbeginend.luau",
	r#"
	BEGIN {
	-- this is the BEGIN block, which is executed once at the beginning
	-- where we typically initialize variables
	running_total = 0;
	grand_total = 0;
	amount_array = {};
	-- note how we use the qsv_log function to log to the qsv log file
	qsv_log("debug", " _INDEX:", _INDEX, " _ROWCOUNT:", _ROWCOUNT);
	-- start from the end of the CSV file, set _INDEX to _LASTROW
	_INDEX = _LASTROW;
	}!
	----------------------------------------------------------------------------
	-- this is the MAIN script, which is executed for the row specified by _INDEX
	-- As we are doing random access, to exit this loop, we need to set
	-- _INDEX to less than zero or greater than _LASTROW
	amount_array[_INDEX] = Amount;
	running_total = running_total + Amount;
	grand_total = grand_total + running_total;
	qsv_log("warn", "logging from Luau script! running_total:", running_total, " _INDEX:", _INDEX);
	-- we modify _INDEX to do random access on the CSV file, in this case going backwards
	-- the MAIN script ends when _INDEX is less than zero or greater than _LASTROW
	_INDEX = _INDEX - 1;
	-- running_total is the value we "map" to the "Running Total" column of each row
	return {running_total, running_total + (running_total * 0.1)};
	----------------------------------------------------------------------------
	END {
	-- and this is the END block, which is executed once at the end
	-- note how we use the _ROWCOUNT special variable to get the number of rows
	min_amount = math.min(unpack(amount_array));
	max_amount = math.max(unpack(amount_array));
	return (`Min/Max: {min_amount}/{max_amount} Grand total of {_ROWCOUNT} rows: {grand_total}`);
	}!
	"#,
	);
	let mut cmd = wrk.command("luau");
	cmd.arg("map")
	.arg("Running Total, Running Total with 10%")
	.arg("file:testbeginend.luau")
	.arg("data.csv");
	let got: Vec<Vec<String>> = wrk.read_stdout(&mut cmd);
	let expected = vec![
	svec![
	"letter",
	"Amount",
	"Running Total",
	"Running Total with 10%"
	],
	svec!["d", "7", "7", "7.7"],
	svec!["c", "72", "79", "86.9"],
	svec!["b", "24", "103", "113.3"],
	svec!["a", "13", "116", "127.6"],
	];
	assert_eq!(got, expected);
	let end = wrk.output_stderr(&mut cmd);
	let expected_end = "Min/Max: 7/72 Grand total of 4 rows: 305\n".to_string();
	assert_eq!(end, expected_end);
	wrk.assert_success(&mut cmd);
	}