Finding pattern in a string linux

Question

I have a file that has different lines of texts, I would like to check if the are duplicates of the same pattern.

In the file:

Blah
Blah
Depends: ssloader, firmware (>= 3.0), firmware (<= 6.0), apta
blah

my aim is to get ">= 3.0" & "<= 6.0" into a file. but bear in mind that sometimes there's only 1 "firmware" dependency.

What i have so far, only grabs the 1st firmware info:

if grep -Fq "firmware (" inputfile #checks if pattern exists
then
 compat=$(look 'Depends:' inputfile) #grab line where pattern is
 compat=${##*firmware (} #remove pattern and other stuff infront
 compat=${compat%%)*} #remove other stuff behind ")"
 echo $compat >> outputfile
fi

I would like to know how to check if there's more than 1 pattern in the same line. Or if there's more than 1 line with the same pattern, how to identify that line can get the firmware value. Thanks

EDIT:

My initial intention is to detect if there are more than one of the same pattern. I am open to ideas. :)

something like this:

if (more than one of same pattern)
 get both values #I am open to ideas to get this done <---
else
 get value of this pattern
fi

EDIT2:

I've got this working by doing it like this;

if grep -Fq "firmware (" ./control
then
    compat=$(look 'Depends:' control)
    compat=${compat#*firmware (}
    compat=${compat%%)*}
    echo -n $compat > ./compatibility.txt
    if [ $(grep -o "firmware (" ./control | wc -l) -eq 2 ]; then

    compat=$(look 'Depends:' control)
    compat=${compat##*firmware (}
    compat=${compat%%)*}
    echo " $compat" >> ./compatibility.txt
    fi
fi

I know it's definitely very layman, and it works only if the pattern is in the "Depends:" label.

Any ideas/inputs?

Is it always going to be firmware, or are you looking for any duplication in a Depends:, or any duplication in any labeled line? — geekosaur
– geekosaur, Commented Apr 27, 2012 at 3:43
its gonna be firmware as at times, other labels like Pre-depends: might have the firmware info. — AlwynIsPat
– AlwynIsPat, Commented Apr 27, 2012 at 3:52
Ok, per discussion in comments below: do you want any occurrence, or multiple (apparently meaning 2?) occurrences? — geekosaur
– geekosaur, Commented Apr 27, 2012 at 4:23

Tim Pote · Accepted Answer · 2012-05-29 16:57:46Z

1

If it's alright to use sed:

sed -n '/firmware (/ { s/[^(]*(\(\([<>]=\|=\|[<>]\)\s\+[0-9]\+\(\.[0-9]\+\)*\))[^(]*/\1 /g; p }' file

Sample input:

Blah
Blah
Depends: ssloader, firmware (>= 3.0), firmware (<= 6.0), firmware (= 5.0), apta
Depends: ssloader, firmware (>= 3.0), firmware (<= 6.0), apta
Depends: ssloader, firmware (<= 6.0), apta
blah

Sample Output:

>= 3.0 <= 6.0 = 5.0 
>= 3.0 <= 6.0 
<= 6.0

edited May 29, 2012 at 16:57

answered Apr 27, 2012 at 4:03

Tim Pote

28.4k6 gold badges66 silver badges68 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

AlwynIsPat Over a year ago

i really like this, but can it be enhanced so that if only takes the value of 'firmware' only? Depends: ssloader, firmware (>= 3.0), subr (0.222-1), firmware (< 6.0), apta. does it work if 'firmware (> 6.0)'? (notice there's no '=')

Tim Pote Over a year ago

@AlwynIsPat See my edits. I used a couple of gsed extentions to make it shorter so let me know if you don't have gsed and I'll use the longer version.

geekosaur · Accepted Answer · 2012-04-27 04:11:29Z

1

Another sed version, which might work better depending on what you're doing:

sed -n 's/.* firmware (\([^)]*\)),.* firmware (\([^)]*\)),.*$/\1 \2/p'

(It is relatively easy to generalize this to multiple packages, by the way.)

answered Apr 27, 2012 at 4:11

geekosaur

61.9k13 gold badges127 silver badges115 bronze badges

5 Comments

Tim Pote Over a year ago

This will fail if there is only one instance of firmware ( on the line. Maybe you could make the second one optional if you take out the .*.

geekosaur Over a year ago

My reading of "more than one pattern" doesn't include 1. Should it? (The question seems to be specifically looking for the repeated information.)

Tim Pote Over a year ago

Hmm, maybe so. I read "the most repeats, 2." to mean \{1,2\} in sed speak. It also appears that their program already accommodates the case where there is one, but they would like it to accommodate up to two. Perhaps the OP should clarify.

AlwynIsPat Over a year ago

i've edited above of how my initial idea of how this code should run. @TimPote is right, I am already able to do if its only one. my case here is, if it happens to be 2 of the same pattern, how can i get both values?

Tim Pote Over a year ago

@AlwynIsPat My answer should work for you then. See my changes to see how it reacts to 1, 2, or more version patterns.

Collectives™ on Stack Overflow

Finding pattern in a string linux

2 Answers 2

2 Comments

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related