75

I know that by using the "-A NUM" switch I can print specific number of trailing lines after each match. I am just wondering if it's possible to print trailing lines until a specific word is found after each match. e.g. When I search for "Word A" I want to see the line containing "Word A" and also the lines after it until the one containing "Word D".

context:

Word A
Word B
Word C
Word D
Word E
Word F

command:

grep -A10 'Word A'

I need this output:

Word A
Word B
Word C
Word D
Improve this question

3 Answers 3

Reset to default
111

It seems that you want to print lines between 'Word A' and 'Word D' (inclusive). I suggest you to use sed instead of grep. It lets you to edit a range of input stream which starts and ends with patterns you want. You should just tell sed to print all lines in range and no other lines:

sed -n -e '/Word A/,/Word D/ p' file
Improve this answer
5
  • 2
    Any tips on how to make it exclusive (for any generic situation, not just OP's)?
    – 2rs2ts
    Commented Jun 18, 2013 at 17:18
  • 7
    @2rs2ts to make it exclusive, just add | sed -e '1d;$d', that is remove first and last line
    – holroy
    Commented Jun 26, 2015 at 9:54
  • 2
    And if you want to exclude all lines between -- that is, you want to just see lines that are outside the two matches -- then: sed -n -e '/pattern A/,/pattern D/d; p' This says, "delete from pattern A to pattern D and print everything else". Unfortunately, this match is greedy. That means if patterns A and D appear more than once, you're going to match from the first pattern A to the last pattern D. I'm afraid Perl or Python are your friends if that's the case.
    – fbicknel
    Commented Dec 19, 2018 at 19:07
  • 2
    How to achieve this with a shell variable inside the pattern ? It is not possible to embed a variable inside an awk/sed pattern enclosed by // (cause it treats everything inside as a regex) as this thread and this one confirm. You need another way in order to fetch the value of a shell variable. However this also seems like the only way to print a range of lines between two patterns. Anyone knows of a solution ?
    – Atralb
    Commented May 21, 2020 at 3:30
  • 1
    @Atralb use double quotes rather than single quotes like sed -n -e "/$pattern1/,/$pattern2/ p" file. see askubuntu.com/questions/76808/…
    – enharmonic
    Commented Jan 28, 2021 at 18:16
25

why not use awk ?

awk '/Word A/,/Word D/' filename
Improve this answer
4
  • 2
    sed appears to be able to do this much more efficiently if a large number of files are involved. awk may be easier to remember, but sed seems to be worth a sticky note in my brain.
    – JimNim
    Commented Nov 24, 2014 at 16:02
  • 1
    awk is superior if Word D falls on the same line as Word A, and possibly elsewhere, eg Word A Word D\nWord D with sed will show two lines where as awk will show one. Additionally one can still use variables with awk, eg awk -v _word="Word A" '$0 ~ _word,/Word D/' "/some/file", so sed maybe more efficient but when dealing with searches of two strings that may or may-not fall on the same line, awk definitely be more reliable.
    – S0AndS0
    Commented Oct 26, 2018 at 20:50
  • it seems that awk better handles the case that the file contains sequences of A and B and you only want to capture what's between each such sequence. it looks like sed doesn't follow these semantics in my case at least.
    – matanox
    Commented Oct 28, 2019 at 18:51
  • This works amazing when piping from tail. Eg. tail -5000f log/development.log |awk '/Started.GET.../,/Parameters/' which is useful in Ruby on Rails application logs for getting the lines associated with the top of the request (the original HTTP request through to the parsed parameters). What might improve this is if there is a way to add a separator similar to greps -A2 or -C2 flags.
    – Peter P.
    Commented Apr 6, 2020 at 21:41
12 
perl -lne 'print if /Word A/ .. /Word D/' file

or

cat file | perl -lne 'print if /Word A/ .. /Word D/'
Improve this answer
2
  • 2
    +1 to counter the drive-by downvote. I'd still use sed for this, unless you need the power of Perl regular expressions to select the delimiting lines.
    – tripleee
    Commented Jul 2, 2013 at 9:32
  • People who wrote sed in the 70s must be thankful :-)
    – matanox
    Commented Oct 28, 2019 at 18:41

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .