> From: kamaraju kusumanchi [mailto:raju.mailingli...@gmail.com] > Sent: Wednesday, March 04, 2009 3:24 PM > Subject: OT: matched string of a regular expression > > Consider the following file. > > > cat junk.txt > a(i)1b > a(j)1b > a(ij)1b > a(ji)1b > a(ijk)1b > a(jik)1b > a(ikj)1b > a(jki)1b > a(kij)1b > a(kji)1b > > In general the file is very long, might contain some other text. Now, > I would like to replace all the occurrences of > > a(ijk)1b with a(ijk)23b > a(jik)1b with a(jik)23b > a(ikj)1b with a(ikj)23b > and so on for all the strings such as a(???) > > Because of the size of the files involved, the number of files on > which I have to perform this operation I decided to use sed (instead > of doing it manually in vim) > > However my sed script currently involves 6 lines (for all the > combinations of ijk) such as > > s/a(ijk)1b/a(ijk)23b/g > s/a(jik)1b/a(jik)23b/g > s/a(ikj)1b/a(ikj)23b/g > and so on > > This method is very cumbersome, not scalable. If I have to do similar > operations on a(ijklm) the script would be 120 lines! Is there any way > to write something like > > s/a(???)1b/a(???)23b/g > > where the second ??? is the string matched by the first regular > expression. > > In general, how can I obtain the string that is matched by a regular > expression (in a shell script)? > > BTW, Is sed the right tool for this kind of job? If not, can you > suggest any other tool that will get the job done in less time? > > PS: Please include my email in the CC. > > thanks > raju
I am not 100% certain this is what you are looking for, but here is my attempt. $ cat /tmp/junk.txt a(i)1b a(j)1b a(ij)1b a(ji)1b a(ijk)1b a(jik)1b a(ikj)1b a(jki)1b a(kij)1b a(kji)1b $ sed -e '/[ijk][ijk][ijk]/s/1b/23b/g' /tmp/junk.txt a(i)1b a(j)1b a(ij)1b a(ji)1b a(ijk)23b a(jik)23b a(ikj)23b a(jki)23b a(kij)23b a(kji)23b Also, this website is pure awesome. :-D http://student.northpark.edu/pemente/sed/sed1line.txt Hope this helps. Have fun! ~Stack~ -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
