delete lines between match

csegau · 05-15-2011, 10:02 AM

Hi All,

My problem is like this

I have to delete all lines between two pattern match
example- suppose below is the content of the file
then i have to delete all lines between text1 and text2
...
text1
abc
def
ghi
text2
jkl
mno
text1
pqr
xyz
text2
...

I only want to remove text between text1 and text2
so the output i want is like this

text1
text2
jkl
mno
text1
text2

Thanks in advance

EricTRA · 05-15-2011, 10:12 AM

Hello,

If I understand you correctly then you want to delete all lines except the ones that contain jkl and mno? Regardless of what's in the two blocks between text1 and text2?
Kind regards,

Eric

druuna · 05-15-2011, 10:13 AM

Hi,

Code:

sed '/text1/,/text2/d' infile
...
jkl
mno
...

You can set a range (/text1/,/text2/) in sed.

Hope this helps.

csegau · 05-15-2011, 10:23 AM

Quote:

Originally Posted by druuna

Hi,

Code:

sed '/text1/,/text2/d' infile
...
jkl
mno
...

You can set a range (/text1/,/text2/) in sed.

Hope this helps.

i want to remove text between text1 and text2 not between text2 and text1

Nominal Animal · 05-15-2011, 11:10 AM

There is most likely a sed stanza that does this, but with awk,

Code:

awk '(!s){print} /text1/{s=1} /text2/{if(s)print;s=0}' infile

should work. If you wish to omit the text1 and text2 lines, then

Code:

awk '/text1/{s=1;next} /text2/{s=0;next} (!s)' infile

druuna · 05-15-2011, 11:21 AM

Hi,

Quote:

Originally Posted by csegau

i want to remove text between text1 and text2 not between text2 and text1

Code:

sed '/text1/,/text2/d' infile

This does print what's between text1 and text2, what it doesn't do is print the search tokens

If you want the tokens you search for be part of the output things get a bit more complicated. Have a look at this:

Code:

$ cat foo.sh
#!/bin/bash

awk 'BEGIN { seent1 = 0 }
/text1/ { seent1 = 1 ; print }
{ if ( seent1 == 0 ) { print } }
/text2/ { seent1 = 0 ; print }
' infile

$ ./foo.sh 
...
text1
text2
jkl
mno
text1
text2
...

$ cat infile
...
text1
abc
def
ghi
text2
jkl
mno
text1
pqr
xyz
text2
...

I just noticed Nominal Animal's reply: Maybe better (at least shorter and a more manageable one-liner).

Anyway, hope this helps.

MTK358 · 05-15-2011, 11:34 AM

Quote:

Originally Posted by Nominal Animal

There is most likely a sed stanza that does this

This deletes everything outside of text1 and text2, and does not delete the delimiters:

sed -n '/text1/,/text2/p'

EDIT: maybe this isn't right, I got really confused, I'll read the thread again...

MTK358 · 05-15-2011, 11:36 AM

Quote:

Originally Posted by csegau

i want to remove text between text1 and text2

That's exactly what druuna's sed command does. Maybe you meant you want to leave alone the text between text1 and text2?

druuna · 05-15-2011, 11:39 AM

@MTK358: My solution does do that, but the OP wants the search tokens to be shown as well (posts #5 and #6 tackle that issue).

MTK358 · 05-15-2011, 11:44 AM

Quote:

Originally Posted by druuna

@MTK358: My solution does do that, but the OP wants the search tokens to be shown as well (posts #5 and #6 tackle that issue).

I don't know how to do that with sed.

But still, no matter how this comment:

Quote:

Originally Posted by csegau

i want to remove text between text1 and text2 not between text2 and text1

could mean that. Your sed command did not delete the text between text2 and text1.

druuna · 05-15-2011, 11:50 AM

@MTK358: Take a look at the (edited) in and out examples the OP posted (post #1) and my original output in post #4. I'm assuming/guessing the OP did not fully understand the command I used when (re-)describing the problem it generated.

BTW: I'm also not aware of a sed solution (there's probably one, but the awk solutions posted are the better way to go).

Lets wait and see what the OP has to say

MTK358 · 05-15-2011, 11:57 AM

Quote:

Originally Posted by druuna

@MTK358: Take a look at the (edited) in and out examples the OP posted (post #1) and my original output in post #4. I'm assuming/guessing the OP did not fully understand the command I used when (re-)describing the problem it generated.

It acted exactly like the OP wanted, except it removed the delimiters. It did not delete between the wrong delimiters, as the OP's post claimed.

grail · 05-15-2011, 07:09 PM

Well not a sed guru, but this seems to work:

Code:

sed -n -e '/text[12]/p' -e '/text1/,/text2/d;p' file

And same idea in awk:

Code:

awk '/text1/,/text2/{if(/text[12]/)print;next}1' file

vikas027 · 05-16-2011, 09:55 AM

Quote:

Originally Posted by grail

Well not a sed guru, but this seems to work:

Code:

sed -n -e '/text[12]/p' -e '/text1/,/text2/d;p' file

And same idea in awk:

Code:

awk '/text1/,/text2/{if(/text[12]/)print;next}1' file

Hi Grail,

Could you please explain these codes. Thanks in advance.

grail · 05-16-2011, 10:22 AM

Hi vikas

Basically they are doing the same thing:

Code:

sed -n -e '/text[12]/p' -e '/text1/,/text2/d;p' file

1. '/text[12]/p' - print any lines containing the string 'text1' or 'text2'

2. '/text1/,/text2/d;p' - delete anything from text1 to text2 inclusive and print the rest

Code:

awk '/text1/,/text2/{if(/text[12]/)print;next}1' file

1. /text1/,/text2/ - search between text1 and text2 inclusive, when true go to step 2

2. if(/text[12]/)print;next - if line contains strings 'text1' or 'text2' then print. Whether previous is true or not go to the next record

3. 1 - print all lines not in the boundary test from step 1

Hope that helps