On this Page

Jawk 6.4.00

Jawk CLI Quickstart

Jawk CLI behaves like AWK, but runs entirely on the JVM. You can pass an inline program, read a script from a file, feed input through standard input or filenames, assign variables, load explicit extensions, dump syntax or tuples, precompile tuples, and switch to a sandboxed runtime when needed.

Warning

Shell quoting differs by platform. The examples below use separate command forms where quoting is meaningfully different. If a script works in one shell but not another, the quoting is usually the first thing to check.

Run an Inline Script

Linux/UNIX

Linux/UNIX

$ echo "hello world" | java -jar jawk-6.4.00-standalone.jar '{ print $2 ", " $1 "!" }'
world, hello!
Windows

Windows

C:\> echo hello world | java -jar jawk-6.4.00-standalone.jar "{ print $2 "", "" $1 ""!"" }"
world, hello!

If you do not pass -f or -L, Jawk expects the next non-option argument to be the AWK program itself.

Run a Script File

Store the AWK program in a file:

totals.awk:

BEGIN {
  total = 0
}
{
  total += $2
}
END {
  print total
}

And point Jawk at it with -f:

$ java -jar jawk-6.4.00-standalone.jar -f totals.awk values.txt

You can repeat -f to combine multiple script sources before execution.

Read from stdin

If you do not provide input filenames after the script, Jawk reads from standard input:

$ printf "alpha\nbeta\n" | java -jar jawk-6.4.00-standalone.jar '{ print NR ":" $0 }'
1:alpha
2:beta

Read Input Files

Input filenames passed after the script become runtime operands and are processed as AWK input files:

$ java -jar jawk-6.4.00-standalone.jar -F : '{ print FILENAME ":" FNR ":" $1 }' /etc/passwd /etc/group

Jawk follows the usual AWK distinction between the script itself and the remaining operands. Files and name=value operands after the script are visible through ARGV and ARGC.

Pass Variables

Use -v for variables that must exist before BEGIN runs:

$ java -jar jawk-6.4.00-standalone.jar -v prefix=hello 'BEGIN { print prefix }'
hello

Use name=value operands after the script for AWK-style file-list assignments that take effect before the next input file is consumed:

$ java -jar jawk-6.4.00-standalone.jar '{ print mode ":" $0 }' mode=csv data.txt

Set the field separator with -F when you want Jawk to split text records differently:

$ java -jar jawk-6.4.00-standalone.jar -F : '{ print $1 "," $6 }' /etc/passwd

Use --posix when you want POSIX-oriented compile-time behavior from the CLI. In particular, it disables gawk-style nested arrays, so classic multi-dimensional subscripts such as a[i, j] remain valid while a[i][j] is rejected:

$ java -jar jawk-6.4.00-standalone.jar --posix 'BEGIN { a[1,2] = 42; print a[1,2] }'
42

Because -L loads an already compiled tuples file, Jawk rejects --posix together with -L instead of pretending that it can re-apply compile-time restrictions after the fact.

Load Extensions

List the currently registered extension identifiers:

$ java -jar jawk-6.4.00-standalone.jar --list-ext

Then load one explicitly:

$ java -jar jawk-6.4.00-standalone.jar -l stdin -f script.awk

Jawk accepts the registered extension name, the simple class name, or the fully qualified class name. Additional extension classes can be placed on the JVM classpath before launching Jawk.

Enable Sandbox Mode

Use -S or --sandbox to switch to the sandboxed tuple compiler and runtime:

$ java -jar jawk-6.4.00-standalone.jar -S -f script.awk input.txt

Sandbox mode disables system(), input and output redirection, command pipelines, and related runtime features that are intentionally unsafe in a restricted host environment.

See Also

cli awk command line gawk bwk mawk awk
Links:
  • [1] cli-reference.html
  • [2] java.html
  • [3] extensions.html
  • [4] compatibility.html
Searching...
No results.