My pattern for "in-place edit" for many files

My pattern for "search and replace many files" is to first grab the filenames, usually with File::Find, and then use "inplace editing mode" to do the reading and writing while I concentrate on the editing part.

use File::Find;
{
  my @starters = @ARGV ? @ARGV : qw(/home/merlyn);
  @ARGV = ();
  find sub {
    push @ARGV, $File::Find::name if /\.asp$/; # or whatever condition
+ here
  }, @starters;
}
{
  local $/; # slurp it
  local $^I = ".bak";
  while (<>) {
    # $_ now contains the entire file, alter at will
    s/foo/bar/g;
    print;
  }
}
[download]

-- Randal L. Schwartz, Perl hacker

Comment on My pattern for "in-place edit" for many files Download Code

Replies are listed 'Best First'.
Re: My pattern for "in-place edit" for many files by C-Keen (Monk) on Sep 29, 2001 at 13:56 UTC
This question is a little bit offtopic maybe, but this is puzzling me since I have seen it in a couple of other nodes too. Why are you using two different code blocks in this example? Has it something to do with the lexical scope of the variables? Thank You very much! Regards, C-Keen	[reply]
On scoping temporaries or global settings by merlyn (Sage) on Sep 29, 2001 at 20:03 UTC
I've found that limiting the scope of my temporaries has reduced my debugging time. So when I need a "scratch variable" for a few steps, or need to set something dangerously global like `$/` for a few steps, I pop open a block and do the work there. Doing so makes it very clear to the reader that this variable doesn't need a long lifetime, and even ensures that the action won't collide with a larger-scoped variable. It's a slick trick, and works nicely. I've also recently started using the idiom of a one-iteration `foreach` loop to "name" a value rather than setting up a temporary, like: `for ($some->complicated(@value)->{$to_fetch}) { print; # print it print LOGFILE $_; # and to the logfile substr($_, 0, 1, ""); # remove its first character }` [download] It's a nice idiom, and is also the explanation of why Larry intentially allows `foreach` to be spelled "f-o-r". And the "backwards `foreach`" makes it even easier: `do_this($_), do_that($_, @moreargs) for $some{$complicated}{$value};` [download] -- Randal L. Schwartz, Perl hacker	[reply] [d/l] [select]
For - Loop construct 'cheaper' than local variable? by C-Keen (Monk) on Sep 29, 2001 at 23:28 UTC
Doesn't the for-loop construct use more internal resources than a local variable? As comming from the C programming section a waste of resources looks somehow awkwardly to me. But it sure is more readable this way. Regards, C-Keen	[reply]
Re: For - Loop construct 'cheaper' than local variable? by Brovnik (Hermit) on Sep 30, 2001 at 02:44 UTC
Re: My pattern for "in-place edit" for many files by tphyahoo (Vicar) on Aug 09, 2005 at 13:40 UTC
And some bat tricks, and perl tricks, from davidrw, ikegami, and others, to do multi edit in place from the dos command line at Inplace editing, dos one liner, without a backup file?.	[reply]
Re: My pattern for "in-place edit" for many files by tphyahoo (Vicar) on Aug 09, 2005 at 14:40 UTC
I like this, but in the version I just added to my toolbox I made one small change: die if there's nothing to replace in @ARGV. Otherwise this script just sits there waiting for STDIN. Being not all that wise about the diamond operator it took me a while to figure out what was going on. `my @starters = @ARGV ? @ARGV : ('./beginning_includes'); #current di +rectory @ARGV = (); find sub { push @ARGV, $File::Find::name if $File::Find::name =~ /.txt$/i; # +or whatever condition here }, @starters; die "nothing to replace" unless @ARGV; }` [download]	[reply] [d/l]
Re: My pattern for "in-place edit" for many files by tphyahoo (Vicar) on Aug 19, 2005 at 11:14 UTC
Another possibility would be to use File::Searcher. UPDATE: Doesn't work too well on windows though. After mucking around a bit I decided to keep using simpler search and replace with $^I.	[reply]