是否存在可以从文件中特定的一组行中删除一行的正则表达式，Perl或sed？

Question

So for example if I have something like:

<Mansions \>
 windows
 walls
 floors
<\Mansions>

and I want to remove lets say walls and keep the other lines there how do I do that with regex or is there a better option than regex?

So I need it to look like:

<Mansions \>
 windows
 floors
<\Mansions>

and the lines up above will still be there after I remove walls. So the goal is to search for those group of lines in the file and remove one word and still keep the other words there and in the group.

Keep in my mind I'm just changing this group in the file, not an instance of the word in the file because that would be a simpler expression.

I have tried using a regex that goes:

^(<Mansion\s*"/.*/.*">)((?:\n(?!.*(?:walls\s(floor|windows)+.*\n|</Mansion>)).*)*\n</Mansion>)$

but that is not working when I remove a single word like walls

So is it possible to use regex for something like this or no?

Answer 1

Here's a Perl one-liner that uses the flip flop operator ( .. ) to only operate on text within two regex matches. It skips the line if it matches /wall/ and is within the opening and closing text.

perl -ne '/walls/ && next if /Mansions \\/ .. /<\\Mansions>/; print' file.txt

Answer 2

$ awk '/<\\Mansions>/{f=0} f && /walls/{next} /<Mansions \\>/{f=1} 1' file
<Mansions \>
 windows
 floors
<\Mansions>

The above will work using any awk in any shell on any UNIX box.

Answer 3

In Perl you could try the following:

use feature qw(say);
use strict;
use warnings;

my $str = '
<Mansions \>
 walls
 windows
 floors
<\Mansions>';

my $word = 'walls';

$str =~ s/^(\s*<Mansions\s*\\>)(.*?)(<\\Mansions>)/do_replace($1, $2, $3, $word)/mseg;

say $str;

sub do_replace {
    my ( $header, $group, $closer, $word ) = @_;
    $header. ($group =~ s/\n?^\s*\Q$word\E\s*$//mr) . $closer;
}

Output :

<Mansions \>
 windows
 floors
<\Mansions>

Answer 4

If the groups are not grouped inside each other, a sed script might work:

script.sed

/<Mansions \\>/,/<\\Mansions>/ {
    /^[[:blank:]]*walls[[:blank:]]*$/ d
}

You run it like sed -Ef script.sed yourfile .

Explanation

• /<Mansions \\\\>/,/<\\\\Mansions>/ { ... } takes care that the inner expression is only applied to lines between <Mansions \\> and <\\Mansions> . This will not work when Mansions are grouped inside other Mansions.
• /^[[:blank:]]*walls[[:blank:]]*$/ ... selects lines that consists of the word walls with optional leading and trailings blanks.
• d means delete such lines.

Answer 5

Why are you making it hard? As described your problem requires no code or regex or anything other than pure vanilla find and replace functionality. If there is something else hidden in your question that requires a code answer I can't find it.

1. Open Find and Replace.

2. In Find use this:

   <Mansions \\> windows walls floors <\\Mansions>

3. In Replace use this:

   <Mansions \\> windows floors <\\Mansions>

4. Click Replace All .

Mission accomplished.