removing unwanted characters from a file

[bcdixit]

suppose I have a file called sample.sql which I open in a .pl file
the file is something like this when opened in VI (unix editor)

šæselect * from a
where tablename = 'x';<some white space characters>
šæselect * from b
where tablename = 'x';<some white space characters>
šæselect * from c
where tablename = 'x';<some white space characters>
šæselect * from d
where tablename = 'x';<some white space characters>
šæselect * from e
where tablename = 'x';<some white space characters>
šæselect * from f;
where tablename = 'x'<some white space characters>
I want to clean this file.
I want to remove the first two non-word characters.
also each line in the file has a some white space characters after the ';' before the new line starts that I want to remove.

How do I use regular expressions to clean the contents of this and WRITE TO THE SAME FILE.


This file is generated by some other part of the code.

I have tried to use the following to read and write to the same file but doesn't seem to work
open(myfile,"+<$myfile");

also any help with the regular expression will be highly appreciated.

thanks..
bcdixit

 

[TrojanWarBlade]

Dude, I've already given you the solution to this one.
Did you even bother to try it?

perl -i.bak -lne 's/^..(.*\w)\s*$/$1/;print' $sample.sql

Trojan.

 

[bcdixit]

where do I put this line of code in my .pl file?

As I had already told you yesterday, I didn't understand
your solution.

Thought I would explain my problem in a better way by starting a new thread.

I am a novice in perl.

Thanks anyway
bcdixit

 

[KevinADC]

I also offered you two suggestions in the other thread. I see now you may not know how to use Tie::File, but did you try the other suggestion I posted for you?

 

[bcdixit]

KevinADC,
Yeah the other suggestion you told me works good.
But I am still having trouble building a reg-expression to clean up the file.

Thanks
bcdixit

 

[KevinADC]

OK, then lets see, try this:

$tables = '/opt/teradata/results.txt';
open(FH, $tables) or die "$!";#open for reading
chomp(my @data = <FH>);
close FH;

open(FH ">$tables") or die "$!";#open for overwriting
for (@data) {
   s/^šæ//;
   s/\s+$//;
   print FH "$_\n";
}
close FH;

 

[bcdixit]

the s/^šæ//; works..
but unfortunately when I looked at the input file there are characters other than 'šæ' at the start of each line.
thats where I am finding trouble.

when i give s/^W+//;...it eliminates only one non word charcter from each line.

 

[duncdude]

s/^\W{2}//;

Kind Regards
Duncan

 

[bcdixit]

your solution seems to eliminate just one wierd character from start of each line.
ok..is there a way I can eliminate all the wierd characters from start of each line until a word charcter appears..maybe using \b or \B ?

thanks
bcdixit

 

[TrojanWarBlade]

You may find that although you only SEE two characters, they may be made up from more than two BYTES.


Trojan.

 

[bcdixit]

I tried
s/^(?:W+)//;
and it worked..

 

[KevinADC]

hmmm.... considering W+ means more than one W I don't see how it works.

 

[bcdixit]

I am new to all this ..but i think if you put just the W+ ,it memorizes the non-word character. after going through a PERL book i found that if you put the ?:, it won't memorize the character.

 

[KevinADC]

yes, ?: tells perl not to store the macth inside the parenthesis in memory, but a bare W without a back-slash in front of it is just a W, not a non-word character class: \W

 

[duncdude]

bcdixit

it does not matter whether it captures it or not - that is simply an issue of code optimization - if a capture is made and it is memorised, then $1 can be used to refer to that capture. This will not affect what the regex finds.

Unescaped (i.e. no backslash before the uppercase W) your regex s/^(?:W+)//; will remove an unlimited number of capital W's from the beginning of a line - and nothing else. I promise you.

Kind Regards
Duncan

 

[duncdude]

i.e. i completely agree with Kevin's comment:

hmmm.... considering W+ means more than one W I don't see how it works.

Kind Regards
Duncan