Script ideas??

[Russ Hancock_1]

I need to extract information from lines in a file that are contained within any number of square brackets...

i.e. INPUT returns OUTPUT
XXXXX[A]XXXXX returns A
XYXYX[ABC]XYX returns ABC
ZXCZX[A][B]ZC returns AB
etc...

Does anyone have any ideas please....
Cheers
Russ

[Steven E. Protter]

I might suggest a start.

awk using [ and or ] to strip the data into a variable

var1=$(awk[ 'print $1')

This is just a course of action, the code doesn't work.

SEP

[H.Merijn Brand (procura]

# perl -nle'@x=/\[([^]]*)\]/g and print@x' file
A
ABC
AB
#

Enjoy, Have FUN! H.Merijn

[RAC_1]

cat input_your_file|tr "\[" ":"|tr "\]" ":"|awk -F : '{print $2}'

Anil

[H.Merijn Brand (procura]

RAC, that will only print the first instance in each line

and it also uses cat where absolutely not neccecary. Useless waste of process space. Use input redirection instead:

NOT # cat file | process
BUT # process < file

Enjoy, Have FUN! H.Merijn

[Kent Ostby]

Create a file called s.awk and put this into it:

{usestring=$0;printstring="";
for (idx1=1;idx1 { usestring=substr(usestring,idx1);
check1=index(usestring,"[");
check2=index(usestring,"]");
if (check1>0)
{use2=substr(usestring,check1+1,check2-check1-1);
printstring=printstring use2; idx1=check2+1;
}
else
{idx1=length($0);}
}
print printstring;
}

Then run:

awk -f s.awk < input > output

I'm sure I could stream line it, but it will work as it is now.

Best regards,

Kent M. Ostby

[RAC_1]

Thanks procura.

I am learing perl from gurus like you.

And yes car=t was required. I did not give that much thought to it. Was very fast.

anil

[Jean-Luc Oudart · Accepted solution]

Cannot compete with Merijn, but an awk solution :

awk '{
i2=index($0,"]");
str=$0;
while(i2 > 0) {
i1=index(str,"[");
printf("%s",substr(str,i1 + 1,i2 - i1 -1));
str=substr(str,i2 + 1, length(str));
i2=index(str,"]");
}
printf("\n");
} '

Regards,
Jean-Luc

[Russ Hancock_1]

Thanks Jean-Luc I used your script...

[H.Merijn Brand (procura]

May I ask why you prefer the use several lines of unmaintainable code over a very short and easy perl line of code?

Do you need explaining of any part?

In the original part you did not state what script language you wanted.

Enjoy, Have FUN! H.Merijn

[Elmar P. Kolkman]

There is more to it... Jean-Luc's code will fail if you have lines containing closing square brackets, but without opening brackets, you will get those lines too.

But if are sure the brackets are always matched, you could also do it this way:

awk '/\[/ { sub("^[^\[]*\[","");
sub("\][^\[]*$","");
sub("\][^\[]*\[","");
print
}'
Same could be done with sed and some replacements using sed.

[Elmar P. Kolkman]

Sorry, forgot one char/thing... Last sub should be a gsub, because you can have more then 2 blocks in square brackets.

But I still think Procura's solution is better and on large files faster.

[Jean-Luc Oudart]

Elmar,
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
There is more to it... Jean-Luc's code will fail if you have lines containing closing square brackets, but without opening brackets, you will get those lines too.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The code is based on the sample from the poster.

Merijn,
No offence, I don't think my code is unmaintainable. Certainly, it takes more than one line, and if don't speak perl, that's where you cannot maintain it , would you agree ?

Regards,
Jean-Luc

[H.Merijn Brand (procura]

Jean-Luc, no offence at all, and sure, it is maintainable, but so verbose that typo's are just a matter of time.

IMHO my perl solution is sooooo darn simple that it just is the best solution. No offence to anyone, and I know I sometimes post rather obfuscated solutions, but this one is so simple that I cannot understand why someone would choose any other solution.

There is a right tool for every problem, and I'm not stating that tool is Perl all the time, but here it outperforms any other in simplicity and shortness

Enjoy, Have FUN! H.Merijn