Wished-for awk function

danielbmartin · 02-07-2018, 01:02 PM

awk has a built-in function called substring.

substr(s, a, b)
returns b number of chars from string s, starting at position a.
The parameter b is optional. If b is omitted it defaults to infinity.

I wish awk had a similar function called subword.

subword(s, a, b, d)
returns b number of words from string s, starting at word a.
The parameter b is optional. If b is omitted it defaults to infinity.
The parameter d is a string of characters, all of which are delimiters
which define word boundaries.
If d is omitted it defaults to a space (i.e. blank or tab).

Examples of subword ...
s="now is the time forQallZgood men"
subword(s, 2, 3) returns "is the time"
subword(s, 4, 1) returns "time"
subword(s, 4) returns ""time forQallZgood men"
subword(s, 4, 2) returns "time forQallZgood"
subword(s, 4, 2, " Z") returns "time forQall"

What is the mechanism for suggesting this enhancement to awk?
Does subword already exist in an extension package?

Daniel B. Martin

.

Turbocapitalist · 02-07-2018, 01:52 PM

Quote:

Originally Posted by danielbmartin

What is the mechanism for suggesting this enhancement to awk?

From what I've seen of bugfixes and much smaller tweaks, the method would be to write a patch with the desired functionality and then convince the various awk- amd awk-using projects that the coding style and code quality of the patch is as desirable as the change itself.

syg00 · 02-07-2018, 07:55 PM

Quote:

Originally Posted by danielbmartin

I wish awk had a similar function called subword.

So write it - and include it in any other code that requires it.

As for getting it into the upstream, the documentation has a section on this very subject.

Turbocapitalist · 02-07-2018, 09:48 PM

Quote:

Originally Posted by syg00

As for getting it into the upstream, the documentation has a section on this very subject.

Which version? mawk and gawk don't seem to have anything about extending the specification in their manual pages.

danielbmartin, on a more practical note, if you are running into case like this more than once, or perhaps even once, then you might consider escalating to perl 5. awk is good at a lot of things, but perl 5 takes string manipulation much, much further.

syg00 · 02-07-2018, 09:55 PM

Not the manpage, the users guide from the gnu.org website.

grail · 02-07-2018, 11:35 PM

How is subword any better than setting FS and performing a basic loop to return your desired items? Or if on a string, just use split:

Code:

#!/usr/bin/awk -f

BEGIN{
	s="now is the time forQallZgood men"
	n = split(s,a)
	# subword(s, 2, 3) returns "is the time"
	for(i=2; i < 5; i++)
		out = out (out?" ":"") a[i]
	print out
	#subword(s, 4, 1) returns "time"
	out = ""
	for(i=4; i < 5; i++)
		out = out (out?" ":"") a[i]
	print out
	#subword(s, 4) returns ""time forQallZgood men"
	out = ""
	for(i=4; i <= n; i++)
		out = out (out?" ":"") a[i]
	print out
	#subword(s, 4, 2) returns "time forQallZgood"
	out = ""
	for(i=4; i < 6; i++)
		out = out (out?" ":"") a[i]
	print out
	#subword(s, 4, 2, " Z") returns "time forQall"
	out = ""
	split(s,a,"[ Z]")
	for(i=4; i < 6; i++)
		out = out (out?" ":"") a[i]
	print out
}

Output matches that requested

Of course you can just put that in a function if that is also desired