How to search and replace string using perl?

[Elizabeth H]

Hi,

I had a few questions regarding searching for a string and replacing it using perl scripts. I have 100's of files written in Unix shell scripts and I need to edit them all. Its like

function1()
{ 
         variable=xxx
}
 
function 2()
{
         variable='aaa bbb ccc ddd'
}
 

I have to search exactly for function 2 and only replace the variable in function 2. I have to append a new value ggg to the variable only in function 2.

Everytime I try doing this with the perl one liner 's/searchtext/replacetext/g', it inserts in the wrong position and in the wrong function. Can you please help me with this as I am completely new to perl

Thanks

[rovf]

I could imagine two approaches:

(1) Your attempted solution processes one line at a time. If you would instead slurp the whole file into a big string, you could then include the function name into your search.

(2) If you prefer processing line-by-line, you can keep track of the functions being seen, and substitute only after you have seen function2. Perls ".." operator could be handy here.

[rski]

You need to define range and change within it

sed  -i  "/function 2()/,/}/ s/variable='\(.*\)'/variable='\1 ggg'/" 1.txt
 

[Elizabeth H]

That was such a great answer and worked really really well...Thankss a bunch..I have one more issue....In the above same problem..How do I insert the new string "ggg" exactly after the string "ccc" or so...is there a position based substitution?

[rski]

sed  -i  "/function 2()/,/}/ s/variable='\(.* ccc\)\(.*\)'/variable='\1 ggg\2'/" 1.txt
 

[Elizabeth H]

Thnks again...Could you please explain what it does..I especially the 1 and 2

[rski]

The pieces

\(.* ccc\) and \(.*\) 
 

are so called capturing groups.
And \1 and \2 are capturing groups result (so called backreference) so \1 is a string that will match
".* ccc" and \2 is a string that match ".*"
I'm not sure if understand what I mean (is is hard to explain in a few words) so if you google for "regexp backreference" you will find clear explanation.

[Elizabeth H]

Thanks a lot! :)

[Elizabeth H]

I also had another question. In the same scenario above, how do I set two variables VAR1 and VAR2(both used in function1) to blank at the end of function1(). Do i need to write a small script or if there is a one liner which can search for function1() and set those variables to blank at the end of the function. Thanks

[rovf]

> how do I set two variables VAR1 and VAR2(both used in function1) to blank at the end of function1

This is tricky, because you need to recognize the end of function1. It depends what you know about "function1". For example, if you know that your input files are formatted properly, you can parse the file until you encounter /^function1/, then continue parsing until you encounter /^\}/, and insert your statement just before.

[Elizabeth H]

Thanks..lemme try

[Prakash Gnana]

Thanks rski

[Elizabeth H]

Can I have some more clue on this as the function itself has several small blocks of statements with closing curly braces which makes it hard for me to parse exactly till end of function where I need to put my two variables

[rovf]

If we know that the functions are nicely formatted, you just need to have a look at closing braces in column 1.

If this is not the case, you would basically need a Shell script parser, because even counting opening and closing braces would not help (because they might also occur, unmatched, in arbitrary command line arguments, HERE-documents, comments). With other words: If you want to succeed using simple line-by-line reading of your file and checking the lines by regular expressions, you need to specify pretty exactly how the "inside" of your functions might look like. If the answer is "arbitrary shell code", then your analysis program needs to be as powerful as the parser of the shell.