i still a newbie of perl script, but want to perl script to make my work easier.

command:
hack.pl input.txt target.txt
,where input.txt is user input file, eg the net to be replaced etc
target.txt is the file i want to modify

method #1,
-read in input.txt & process
-read in target.txt & save all the lines into memory
-read line by line & make changes in the memory if matched.
-after that, print to another file target.txt.hack

method #2
-read in input.txt & process
-copy target.txt tp target.txt.hack
-read the target.txt.hack line by line
-no need to save all the line into memory
-modify if matched
(same principle as like user modify in vi)

which one is better? Is the method #2 doable?

could perl script modidy the file directly? i saw one similar example, it uses method #1

I am not sure I fully understand your question, but modifying a file in place is almost always a truly bad idea regardless of language.

Reading full file into memory or not is an orthogonal issue. You can easily read small files into memory. You can also map files onto Perl arrays: http://search.cpan.org/search?query=map+file&mode=all -> http://search.cpan.org/~leont/File-M...ib/File/Map.pm , etc.

1 members found this post helpful.

I believe in method 1 you're talking of producing a copy of the contents of the target.txt file into memory, and then your substitutions as dictated by your input.txt are performed to this copy, and then the result is written to target.txt.hack.
I fail to see how this is modifying the file directly, and how this matches with your similar example statement where this method was used.

Method 2 seems to require a mapping mechanism as Sergei provides, and then I'd question why you'd only map one line at a time, as most systems using Perl would surely be capable of mapping more/all of the file, even if your substitutions are performed per line.

I'm going to question the need to edit the file directly. IIRC under Linux, any process that reads a file will recieve a copy of the contents, and your changes will not be automatically picked up unless the other program is monitoring for alteration timestamp changes or actual file contents changes. You seem to describe producing a second file, and no mention of replacing the first with the second.

I'll also mention you might want to consider how any subsequent program will know when you're done modifying the .hack file, especially with the 2nd/mapped method. With the 1st method, you'd need only consider if the file exists in the filesystem before all its contents is readable by someone immediately opening it, whether if it starts empty and grows or won't be listed until all output is written.
AFAIK the usual method is write out to a tmp file, then lastly rename to the filename you wish others to use.

1 members found this post helpful.

Yep, that's the usual method.

I accept with method Two will be the most efficient way.But in Perl method one is possible but it may fail in some cases.

open(FILE,"+<hello.dat");
# "+<" option has to be used for read and writing to the file
@lines=<FILE>;
foreach(@lines)
{
chop;
print; # it is same as print $_;
print "\n";
}
print FILE "\n"."karthik";
close(FILE)

This program can help you to understand better.
The file examples in the site:[url]http://linux-forum-karthik.blogspot.com
can give you a better understanding.

0 members found this post helpful.

sounds like the 2nd method is a bad idea (modifying a file directly)

this is about method #1 (simplified one)
is it the common way of writing perl?

No, it is common method for writing "C". Get back to the tutorials -> http://perldoc.perl.org/perlintro.html -> http://perldoc.perl.org/perlsyn.html , look up 'foreach'.

while($line < $INPUT_LINE){

do_something...and change the content in @INPUT_FILE
$line++;
}

Change the above to:

foreach( @INPUT_FILE)
{
do_something...and change the content in @INPUT_FILE
}

open($INPUT, $ARGV[0]) || die('No input_net file');
@INPUT_FILE=<$INPUT>; #save the entire file into memory
$INPUT_LINE=@INPUT_FILE; #total line
close($INPUT);

this step will be put a large file into memory.
is there any other way?

Why do you need your whole input file in memory ?

It is easier to edit. (i know this method, at least the test case work if i load into memory since i edit from memory)
Besides this, I have tried other method which writing to a new file but it does not work


Flow:
-Open input file
-read input file line by line #step2
-open target file
-read target line by line
-if matching --> do replacement
-print the line into a new file
-repeat the step 2

## 2nd iterations
-read input file line by line
-open target file
-read target line by line
-if matching -->do replacement
-print the line in to a new file (the problem is here , it only changes the current edit, the previous changes is not written)
-repeat the step2

(the flow is the simplified one, I use simple replacement as test case.)

I think i got to overwrite target file after printing a new file.

this will change THIS to THAT in file

this will create a backup .bak

or you can add the switches to a more complex script

hi bigearsbilly,
thx for the information.

I still need to improve the flow because my script for my work need some manipulations rather than simple substitution.
For simple substitution, i can manage to achieve by using vi, instead of perl script.

Hi,

you can use Tie::File to edit Files in place


rgds flo