Inplace editing the elegant way

[jrpfinch]

I would like to develop a script that can inplace edit a large file
(potentially 1Gb), omitting some lines according to a regex and
replacing some lines according to another regex. Here is an example
of a script that does this with a couple of hardcoded patterns:

{
local ($^I, @ARGV) = ('', ($full_path_to_config_file));
while (<>)
{
s/this/that/;
s/foo/bar/;
print unless ((/delete me/) or (/Delete Me Too/));
}
}

What is the best way to scale this for n patterns? When I say best, I
mean a good balance of speed and readability. Can I put all the
omission patterns in an array and all the search/replace patterns in
another array and somehow wedge them into the example above?

Any help appreciated.

Jon

 

[Jürgen Exner]

I would like to develop a script that can inplace edit a large file
(potentially 1Gb), omitting some lines according to a regex and
replacing some lines according to another regex. Here is an example
of a script that does this with a couple of hardcoded patterns:

{
local ($^I, @ARGV) = ('', ($full_path_to_config_file));
while (<>)
{
s/this/that/;
s/foo/bar/;
print unless ((/delete me/) or (/Delete Me Too/));
}
}

What is the best way to scale this for n patterns? When I say best, I
mean a good balance of speed and readability. Can I put all the
omission patterns in an array and all the search/replace patterns in
another array and somehow wedge them into the example above?

I'd probably store the REs pattern and replacement text as key/value
pairs in a hash(*) and just loop through them.
And a separate hash or array for the "delete" patterns.

*: that should work, because REs are just strings, too.

jue

 

[Hartmut Camphausen]

I'd probably store the REs pattern and replacement text as key/value
pairs in a hash(*) and just loop through them.
And a separate hash or array for the "delete" patterns.

*: that should work, because REs are just strings, too.

I'd consider to store the /compiled/ patterns + replacement strings in
an array - if efficiency is an issue.

Else each pattern would have been recompiled for each chunk of data.


jm2p + mfg, Hartmut

 

[Jürgen Exner]

I'd consider to store the /compiled/ patterns + replacement strings in
an array - if efficiency is an issue.

Nice idea!

jue

 

[sln]

I would like to develop a script that can inplace edit a large file
(potentially 1Gb), omitting some lines according to a regex and
replacing some lines according to another regex. Here is an example
of a script that does this with a couple of hardcoded patterns:

{
local ($^I, @ARGV) = ('', ($full_path_to_config_file));
while (<>)
{
s/this/that/;
s/foo/bar/;
print unless ((/delete me/) or (/Delete Me Too/));
}
}

What is the best way to scale this for n patterns? When I say best, I
mean a good balance of speed and readability. Can I put all the
omission patterns in an array and all the search/replace patterns in
another array and somehow wedge them into the example above?

Any help appreciated.

Jon

This is fine for simple line based parsing.
How would you handle replacements that span lines?

sln