GREP: how to specify a line or lines where to search

carolflb · 11-21-2009, 06:24 AM

Hi guys,

I need to find a term but just between certain lines, so I don't want the command 'grep' to search through the whole text file. Is there a way to do that using 'grep'?

If not, would somebody know another way to do that?

cola · 11-21-2009, 06:29 AM

What are you trying to do?
Describe with example.

carolflb · 11-21-2009, 06:47 AM

Hello cola,

I have a .h file where I need to know where all IF-ELSE statements begin and end. I need to do that because I'm counting variables and if a variable is inside of an IF-ELSE I should just count it once, since just the IF or just the ELSE will be runned at a time. So if I have for example:

if (smt) {
var=true;
}
else{
var=false;
}

I need to know where the if begins, where the else begins, but also where the else ends (this is the hard part for me). I tough I can mark it if I try to find the first "{" and the last "}" after the ELSE, to solve this. But since I also need to consider such scenarios:

...more code

if (smt){ ... }
if (smt){ ... }
else { ...
if { ... }
else { ... }
}

That means there might be other IF-ELSEs inside of an IF or ELSE, what means more "{" and "}" that won't be the end of the ELSE I'm looking for. Any idea how to solve this?

pixellany · 11-21-2009, 07:15 AM

You might want to look at the addressing feature in SED.
Example:

Code:

sed -n '/start/,/end/p' filename

This returns all lines from one containing "start" up through one containing "end". You might need more that one chained together to satisfy all of your conditions.

Really goos SED tutorial here:
http://www.grymoire.com/Unix/Sed.html

ghostdog74 · 11-21-2009, 08:09 AM

i think its not that simple just using grep or sed. what if you have nested if else, or if without else. how are you going to count? do you always want to find if/else pairs??
you should describe more clearly what you want to do. If you do not care about these, then grep for "if" and count how many are there should be fine right?

carolflb · 11-21-2009, 08:51 AM

Yes guys,

I also think that 'sed' wouldn't help and with 'grep' it is also lots of work... but I'm trying to find a way marking the 'IF', 'ELSE' and 'brackets' I already found, so that once I run 'grep' I can find the next one.

@ghostdog: What I need to do is to count how many times an specific variable is USED in the .h file (not counting de declaration). So if we have:

bool var=false;

if (smt){ var=true; }

if (smt) {
var=true;
}
else{
var=false;
}

...more code

if (smt){ ... }

...more code

if (smt){
var=true;
}
else {
if {
var=false;
}
else {
var=true;
}
}

The variable counter needs to give me '3' (colored in pink), since the others are inside IF-ELSEs. The 'var' is changed just one time for each IF-ELSE once the code runs. I hope that now it is clear enough

ghostdog74 · 11-21-2009, 09:10 AM

as usual, a gawk solution.

Code:

awk 'NR{
 m=split($0,e,"else")
 for(i=1;i<=m;i++){    
    if( e[i] ~ /if.*var=true/){
        print "$ ==> "e[i]
        ++count
    }
 } 
}END{
    print "total: "count
}' RS= file

output

Code:

$ ./shell.sh
$ ==> if (smt){ var=true; }
$ ==> if (smt) {
var=true;
}

$ ==> if (smt){
var=true;
}

total: 3

carolflb · 11-23-2009, 08:45 AM

Hi ghostdog,

I just tried your solution, but it just counts the 'var's inside of IFs. What could I add in your code so that it also counts the 'var's outside IFs but in the .h file?

Ah! And another question, I saw that I can set a variable for the program in awk, but how do I make to receive a result from the program? I would like to keep the 'count' to use some where else.