node package manager
Love JavaScript? Your insights can make it even better. Take the 2017 JavaScript Ecosystem Survey »



executing unix commands with multi processes


$ npm install -g sqd


$ sqd -c command [--debug] [--exit] [-p nProcess] [-s separator_command] <input file> [output file]

grep with 8 processes

sqd -c "grep -e something" -p 8 input.txt

results are on STDOUT.

sed with 4 processes (default), results to output.txt

sqd -c "sed -e y/ATCG/atcg/" input.txt output.txt

with separator option, we can also handle binary files

sqd -c "samtools view -" -s bam input.bam


sqd -c "node foobar.js" sample.txt --reduce

in foobar.js

if (process.env.sqd_map) {
  process.stdin.on("data", function(data) {
    // do something, separated into multi processes 
else if (process.env.sqd_reduce) {
  process.stdin.on("data", function(data) {
    // do somothing which reduces the results 


  • -p: the number of processes
  • --debug: debug mode (showing time, temporary files)
  • --exit: exits when child processes emit an error or emit to stderr
  • -s: (see separator section)
  • --reduce: reducing the results with the same command, which is given an environmental variable named sqd_reduce with value "1"

additional environment variables in child processes

Be careful that all values are parsed as string.

  • sqd_n: process number named by sqd, differs among child processes
  • sqd_start: start position of the file passed to the child process
  • sqd_end: end position of the file passed to the child process
  • sqd_command: command string (common)
  • sqd_input: input file name (common)
  • sqd_tmpfile: path to the tmpfile used in the child process
  • sqd_debug: debug mode or not. '1' or '0' (common)
  • sqd_hStart:: start position of the header (common)
  • sqd_map:: "1", unless it is spawned for reducing. undefined, otherwise
  • sqd_reduce:: "1", if it is spawned for reducing. undefined, otherwise


sqd requires a separator which separates a given input file into multiple chunks. separator offers the way how sqd separates the file by JSON format.

the JSON keys are

  • positions: start positions of each chunks in the file
"positions": [133, 271, 461, 631]
  • header: range of the header section of the file, null when there is no header section
"header": [0, 133]
  • size: file size (optional)
"size": 34503

available separators

sqdm --much more memory

$ sqdm [memory=4000MB] -c command [--debug] [--exit] [-p nProcess] [-s separator_command] <input file> [output file]

sqd with 8000MB(≒8GB) memory

sqdm 8000 -c "cat" sample.txt