case insenitive gsub

I have an awk statment that takes a string from a user and substitutes
the string with the string in color to the display:

awk "{ if (gsub(/[^ ,;]*(${2})[^ ,;]*/,\"${COLOR}&$&#
123;OFF}\"))
{ printf(\"$COLOR$z$OFF  \") ; print } }" $x | less -r

As you can see I select the input string ($2) and any leading and
trailing characters that aren't [ ,;].  The string is any combination of
letters and numbers and [#$-].  I want to be able to perform this
substitution in a case insentive manner.  So $2=test should return Test,
TEst, tEst, etc.  How can I change my regex to accomodate this?

Thanks!

fyb

--
This is my sig

In article <slrnc3s9ig.vkd.fybar@aristotle.hammerdog.org>,
fybar <fybar@aristotle.hammerdog.org> wrote:

%  awk "{ if (gsub(/[^ ,;]*(${2})[^ ,;]*/,\"${COLOR}&
${OFF}\"))
% { printf(\"$COLOR$z$OFF  \") ; print } }" $x | less -r
%
% As you can see I select the input string ($2) and any leading and
% trailing characters that aren't [ ,;].  The string is any combination 
of
% letters and numbers and [#$-].  I want to be able to perform this
% substitution in a case insentive manner.  So $2=test should return Test,
% TEst, tEst, etc.  How can I change my regex to accomodate this?

If you use gawk, there's a special variable called IGNORECASE which can
be set to 1 to give the behaviour you want. This is not available in
standard awk, though.

One approach is to make a copy of the string using tolower(), then match()
against a tolower()ed version of the shell's $2, and rebuild the original
string using substr(). Something like (untested!):

awk -v sd2="$2" -v clr = "$COLOR" on="$ON" off="$OFF" '
BEGIN { re = "/[^ ,;]*(" tolower(sd2) ")[^ ,;]*"
clron =  clr "z" on
clroff = clr "z" off
}
{d0l = tolower($0)
d0r = ""
lrs = 0
while (match(d0l, re)) {
d0r = d0r clron substr($0, lrs+RSTART, RLENGTH) clroff
lrs += RSTART
d0l = substr(d0l, RSTART+RLENGTH)
}
d0r = d0r substr($0, lrs)
print d0r
}'

--

Patrick TJ McPhee
East York  Canada
ptjm@interlog.com

In article <c1lggk$hpc$5@news.eusc.inter.net>, Patrick TJ McPhee wrote:
> In article <slrnc3s9ig.vkd.fybar@aristotle.hammerdog.org>,
> fybar <fybar@aristotle.hammerdog.org> wrote:
>
> %  awk "{ if (gsub(/[^ ,;]*(${2})[^ ,;]*/,\"${COLOR
}&${OFF}\"))
> % { printf(\"$COLOR$z$OFF  \") ; print } }" $x | less -r
> %
> % As you can see I select the input string ($2) and any leading and
> % trailing characters that aren't [ ,;].  The string is any combinatio
n of
> % letters and numbers and [#$-].  I want to be able to perform this
> % substitution in a case insentive manner.  So $2=test should return Test,
> % TEst, tEst, etc.  How can I change my regex to accomodate this?
>
> If you use gawk, there's a special variable called IGNORECASE which can
> be set to 1 to give the behaviour you want. This is not available in
> standard awk, though.
>
> One approach is to make a copy of the string using tolower(), then match()
> against a tolower()ed version of the shell's $2, and rebuild the original
> string using substr(). Something like (untested!):
>
> awk -v sd2="$2" -v clr = "$COLOR" on="$ON" off="$OFF" '
>   BEGIN { re = "/[^ ,;]*(" tolower(sd2) ")[^ ,;]*"
>           clron =  clr "z" on
>           clroff = clr "z" off
>   }
>   {d0l = tolower($0)
>    d0r = ""
>    lrs = 0
>    while (match(d0l, re)) {
>       d0r = d0r clron substr($0, lrs+RSTART, RLENGTH) clroff
>       lrs += RSTART
>       d0l = substr(d0l, RSTART+RLENGTH)
>    }
>    d0r = d0r substr($0, lrs)
>    print d0r
>  }'
>
I have tried this and modified it to work, but I want to try something
else.  If I get a string for re that is all lower case I want to build a
regex that encompasses both cases:

re = "[^ .,;]*"toupper(sd2)"[^ ,;]*"
gsub(/[[:alpha:]]/,"&"tolower("&"),re)

The first statement will make sure all letters are upper case, the
second should make the regex have both.  Well, it doesn't.  It will make
a duplicate of the same case. So a string for sd2 of 89oN will end up
being 89OONN.  I know that I have to massage this more to get
89[Oo][Nn], but first I would like to know why I get what I do from 
the
second statement above.  It's as if the tolower statement doesn't get
executed.

Thanks,

fyb
--
This is my sig

In article <slrnc47mfv.apj.fybar@aristotle.hammerdog.org>,
fybar <fybar@aristotle.hammerdog.org> wrote:

% re = "[^ .,;]*"toupper(sd2)"[^ ,;]*"

toupper() will be executed at the time re is assigned, so it will
have the value
[^ .,;]*{something}[^ ,;]*
where {something} is `sd2' upper-cased.

gsub(/[[:alpha:]]/,"&"tolower("&"),re)

This will globally replace any one letter with itself., twice.
tolower("&") is "&", so "&"tolwoer("&") is the same as "&&".

--

Patrick TJ McPhee
East York  Canada
ptjm@interlog.com

In article <c210qh$rkp$2@news.eusc.inter.net>, Patrick TJ McPhee wrote:
> In article <slrnc47mfv.apj.fybar@aristotle.hammerdog.org>,
> fybar <fybar@aristotle.hammerdog.org> wrote:
>
> % re = "[^ .,;]*"toupper(sd2)"[^ ,;]*"
>
> toupper() will be executed at the time re is assigned, so it will
> have the value
>   [^ .,;]*{something}[^ ,;]*
> where {something} is `sd2' upper-cased.
>
>  gsub(/[[:alpha:]]/,"&"tolower("&"),re)
>
> This will globally replace any one letter with itself., twice.
> tolower("&") is "&", so "&"tolwoer("&") is the same as "&&".
So, the tolower is not perfomed on the "&" to give a lower case?

I am interested in why it is not done, but right now it is immaterial as
I have a working version.  Here is my entire script.  I post it so that
I may get suggestions on how to stream-line it to maybe improve
performance or improve style.

BEGIN {
gsub(/\$/,"\\$",sd2) # put an escape in front of $'s so that
they are not mistaken for variables
re1 = toupper(sd2)
re2 = tolower(sd2)
for ( i = 1; i <= length(re1); i++) {
re1_tmp = substr(re1, i, 1)
re2_tmp = substr(re2, i, 1)
#if square brackets are included in the search string then
#the regex here must be contained
if ( re1_tmp == "[" && re2_tmp == "[" ) {
regex = "true"
}
if ( re1_tmp == "]" && re2_tmp == "]" ) {
regex = "false"
}
if ( re1_tmp != re2_tmp ) {
if ( regex == "true" ) {
string=string re1_tmp re2_tmp
}
else {
regex="false"
string=string "["re1_tmp re2_tmp "]"
}
}
else {
if ( regex == "true" && re1_tmp != "[" &&\
re1_tmp != "]" && re1_tmp !~ /[0-9]/){
string = string re1_tmp "]"
}
else
string = string re1_tmp
}
}
re = "[^ .,;]*("string")[^ ,;]*"
clron =  clr
clroff = off
}
{
if (gsub(re,clron"&"clroff,$0)){
match(FILENAME,/rui[[:digit:]]{2}\
[[:alpha:]]*\..*$/)
name_match = substr(FILENAME, RSTART)
print clron name_match clroff " " $0
#print re #uncomment to see regex
}
#}
}

This will print the filename highlighted by $COLOR followed by the line
with the results of the gsub with my case insensitve regex also
highlighted.

Thanks,

fyb
--
This is my sig