Lazy Grep -P: How to show only to the 1st match from the lines

Ajaib Singh 0
Viewed 49

I need to print only the 1st match from each line.

My file contains text something like this:

cat t.txt
abcsuahrcb
abscuharcb
bsaucharcb
absuhcrcab

He is the command I am trying with:

cat t.txt | grep -oP 'a.*?c'

It gives:

abc
ahrc
absc
arc
auc
arc
absuhc

I need it to return:

abc
absc
auc
absuhc

These are the 1st possible matches from each line.

Any other alternatives like sed and aws will work, but not something which needs to be installed on Ubuntu.

grep pcre awk regex
3 Answers

Perl to the rescue:

perl -lne 'print $1 if /(a.*?c)/' t.txt
  • -n reads the input line by line, running the code for each;
  • -l removes newlines from input lines and adds them to output;
  • The code tries to match a.*?c, if matched, it stores the result in $1;
  • As there's no loop, only one match per line is attempted.
choroba0

Using grep you could write the pattern as matching from the first a to the first c using a negated character class.

Using -P for Perl-compatible regular expressions, you can make use of \K to forget what is matched so far.

Note that you don't have to use cat but you can add the filename at the end.

grep -oP '^[^a]*\Ka[^c]*c' t.txt

The pattern matches:

  • ^ Start of string
  • [^a]* Optionally match any char except a
  • \K Forget what is matched so far
  • a Match literally
  • [^c]* Optionally match any char except c
  • c Match literally

Output

abc
absc
auc
absuhc

Another option with gnu-awk and the same pattern, only now using and printing the capture group 1 value:

awk 'match($0,/^[^a]*(a[^c]*c)/, a) { print a[1]}' t.txt
The fourth bird0

A sed variation on The fourth bird's answer:

$ sed -En 's/^[^a]*(a[^c]*c).*/\1/p' t.txt
abc
absc
auc
absuhc

Where:

  • -En - enable extended regex support, suppress automatic printing of pattern space
  • ^[^a]* - from start of line match all follow-on characters that are not a
  • (a[^c]*c) - (1st capture group) match letter a plus all follow-on characters that are not c followed by a c
  • .* - match rest of line
  • \1/p - print contents of 1st capture group

One awk idea:

$ awk 'match($0,/a[^c]*c/) { print substr($0,RSTART,RLENGTH)}' t.txt
abc
absc
auc
absuhc

Where:

  • if we find a match then the match() call is non-zero (ie, 'true') so ...
  • print the substring defined by the RSTART/RLENGTH variables (which are auto-populated by a successful match() call)
markp-fuso0
Related