I have next file:
G
H
A
B
C
D
N
Let's delete lines from A to D and we will get next output:
G
H
N
It's very easy to do with sed '/A/,/D/d, but if my file don't have D, then output will be empty. I want if there isn't second pattern (D) do not delete anything and show full file.
Second question - how to delete lines between patterns and next line after (N)? Kind of sed '/A/,+1d, but sed '/A/,/D/+1d will not work.
There is no different for me to use sed, awk or python/bash scripts.
With 2 pass awk you can do this:
# when 2nd pattern is not found
awk -v ps='A' -v pe='P' 'NR==FNR{if ($0 ~ ps) start=FNR; else if ($0 ~ pe) stop=FNR; if (stop) nextfile; else next} !stop || FNR<start || FNR>stop' file file
G
H
A
B
C
D
N
# when 2nd pattern is found
awk -v ps='A' -v pe='D' 'NR==FNR{if ($0 ~ ps) start=FNR; else if ($0 ~ pe) stop=FNR; if (stop) nextfile; else next} !stop || FNR<start || FNR>stop' file file
G
H
N
About your 2nd part you can tweak this awk a bit with another parameter:
awk -v n=2 -v ps='A' -v pe='D' 'NR==FNR {
if ($0 ~ ps)
start=FNR
else if ($0 ~ pe)
stop=FNR+n
if (stop)
nextfile
else
next
}
!stop || FNR<start || FNR>stop' file file
One option out of many that use perl: hold the text in an accumulator once you see A, then print them at the end if you didn't see D. That way you only make one pass through the file (although you use a lot of memory for big files!).
use strict; use warnings;
my $accumulator = ''; # Text we're holding while we wait for a "D"
my $printing = 1; # Are we currently printing lines?
while(<>) {
if(/A/) { # stop printing; start accumulating
$printing = 0;
$accumulator .= $_; # $_ is the current line
next;
}
if(/D/) { # we got a D, so we're back to printing
$accumulator = ''; # discard the text we now know we're deleting
$printing = 1;
next;
}
if($printing) {
print;
} else {
$accumulator .= $_;
}
}
print $accumulator; # which is empty if we had both A and D
I tried this on your testcase, and on your testcase with the D removed. It can also handle files with multiple A/D pairs. I have not tested it on files where the D comes before the A, or on files with a single line including both A and D.
With D:
$ awk '
/A/ || f {
f=1 # flag up
b=b (b==""?"":ORS)$0 # buffer...
if(/D/) { # until D
print b # output...
b=f="" # and reset buffer and flag
}
next
}
END { # output if rows runout before finding D
if(b)
print b
}1' file # output outside the range
G
H
A
B
C
D
N
With missing D:
$ cat file2
G
H
A
B
C
N
$ awk '/A/||f{f=1;b=b (b==""?"":ORS)$0;if(/D/){print b;b=f=""}next}END{if(b)print b}1' file2
G
H
A
B
C
N
This might work for you (GNU sed):
sed '/A/{:a;N;/D/!ba;d}' file
Match on A and then gather up lines in the pattern space until a match on D then delete the pattern space. If a D is not matched the N command will terminate sed processing and by default this prints whatever is found in the pattern space.
To delete as above +1, use:
sed '/A/{:a;N;/D/!ba;N;d}' file
N.B. If the +1 line does not exist, no lines are deleted.
