[SOLVED] force to grep adjacent words

udiubu · 07-04-2018, 11:48 AM

Dear experts,

A very simple query regarding how to force grep to extract necessarily adjacent words in a while loop.

I have an input.txt file:

blabla
qq det|theb-la n|book
blabla
axc det|thbl-abl n|ball
blabla
aqq det|thublabla adj|blue n|book

and a list of words to be searched for (WORD.txt):

book
ball

I need to extract all exact adjacent occurrences of "det|" immediately followed by n|WORD. "det|" and "n|WORD" are separated by a single space. I also need to get the line before match and line number. Please note that more alphanumeric chars can follow "det|", before the single space occurs.

Expected output:

1: blabla
2: qq det|theb-la n|book
3: blabla
4: axc det|thbl-abl n|ball

So far my code below, which however also gets line 5 and 6.
How to tell "grep only adjacent match"?
And how to tell "any character following det, before space"? I am not sure the star there is really working.

Code:

while read WORD; 
do
grep --no-group-separator -n -B1 " det|t.* n|${WORD}" input.txt
done < WORD.txt

Any help would be highly appreciated.

Sincerely,
Udiubu

Turbocapitalist · 07-04-2018, 12:12 PM

It looks like the -B is not needed.

Then change the span (.*) so that it does not include spaces.

Code:

grep --no-group-separator -n " det|t[^[:space:]]* n|${WORD}" input.txt

If you need fancier and have GNU's grep then you can use perl regular expressions instead and take advantage of lookahead or lookbehind matches.

udiubu · 07-04-2018, 12:30 PM

This works beautifully!
Thanks Turbocapitalist!