awk

Awk is a text processing language named named after the initials of its authors, Aho, Weinberger, Kernighan.

Attributes

website

http://awk.info/

website

http://cm.bell-labs.com/cm/cs/awkbook/index.html

Documentation

The Awk Programming Language , Aho, Kernighan, Weinberger, 1988

The GNU Awk User's Guide

the guide for one of the most widely-used Awk implementations

An Awk Primer , Greg Goebel, updated 2014-04-01

a comprehensive intro to awk, suitable to beginners.

Famous Awk One-Liners Explained, Part I: File Spacing, Numbering and Calculations , Peteris Krumins, 2008-09-27

a handy FAQ for common tasks.

Community

the usenet newsgroup news:comp.lang.awk

Description

Awk is a standard program on most (all?) Unix-like operating systems.

Awk's ability to scan through a file and manipulate the contents predates Perl's functions to do this, and frankly awk's abilities, while cruder in many ways, are also simpler (simpler even than Tcl!).

Tcl is a good candidate for any tasks that might be programmed in Awk, but is a more general language, and in terms of sheer speed, is outperformed by Awk at many text processing and data extraction tasks. Core team member Alexandre Ferrieux is interested in erasing this performance difference, so maybe that will change.

LV: Awk is one of my favorite languages in which to write...

Using Awk from Tcl

LV: A common question is:

How can I invoke awk scripts from Tcl?

Here is how not to do it:

$ tclsh 
% set a [exec awk {'{print $1}'} /etc/motd]
awk: cmd. line:1: '{print $1}'
awk: cmd. line:1: ^ invalid char ''' in expression

RS answers: This is not an Awk problem, but improper shell quoting: single quotes in a shell have the effect as braces in Tcl - group in one word, don't substitute on contents. Solution here: You have outer braces already, so just drop the single quotes (the inner brace pair is awk syntax, not seen by Tcl):

% set a [exec awk {{print $1}} /etc/motd]

Look Ma, No Awk!

RS 2007-02-07: I love this few-liner that allows tests in a subset of Awk notation (in fact, the common subset of Awk and expr, plus a shortcut for regexp):

proc awktest {filter 0} {
    if {[regexp {^/(.+)/$} $filter -> re]} {return [regexp $re $0]}
    set i 0
    foreach field $0 {set [incr i] $field}
    expr $filter
}

e.g.:

awktest {$1==$2} {foo bar grill} ;#->  0.

$0 is just the input list :^) The following shortcut is also cute:

interp alias {} ~ {} regexp

RS 2007-11-08: When porting Awk scripts, it's also helpful to have the numbered variables:

proc awksplit {list sep} {
    set i 0
    foreach field [split $list $sep] {uplevel 1 [list set [incr i] $field]}
    upvar 1 [list set NF $i]
}

where for example awksplit {foo;bar;grill} {;} assigns foo to $1, bar to $2, grill to $3.

RS 2007-02-28: Here's another piece of awk emulation:

proc substr {str from length} {
    string range $str [expr {$from-1}] [expr {$from-2+$length}]
}

% substr hello 2 2
el
% substr hello 2 99
ello

And this is trivial, but over 50% shorter to type:

interp alias {} length {} string length

SW 2020-11-29: I've written tawk , an awk clone that uses tcl instead.

$ awk '{ print $1 }' input.txt
$ tawk 'line { print $F(1) }' input.txt

Tools

Tclforawk

a graphical front-end for managing and running Awk scripts. It acts like a combined file selector, text editor and terminal window, which makes the process of running Awk simpler if you have lots of scripts.

See Also

• how can I do this awk like operation in Tcl
• Scan and modify text files
• Text processing tips
• exec: includes an example of invoking Awk, something often requested.
• BOOK Mastering Regular Expressions: covers regular expressions in perl, Awk, and tcl.
• owh - a fileless tclsh: a Tcl variation on awk functions
• env changing arguments: how to pass optional parameters to a script, awk-like
• Sqawk: An Awk-like program with SQL queries written in Tcl with SQLite

Reference

• awk, The Open Group Base Specifications Issue 7, 2013 - login required.
• AWK at Wikipedia
• AWK at RosettaCode