Shell Programming and Scripting

Hello,

I have some tab delimited text data that I am processing. The second column looks like,

Code:

NAME;pyrimidine-2,4-diol;cpd;2;line;37

I need to clean this up to just the name,

Code:

pyrimidine-2,4-diol

All lines have the same format,
NAME;text;cpd;int;line;int followed by tab

I have tried something like,

Code:

sed 's/NAME;//g' | \
sed 's/;cpd;*;line;*\t/\t/g' | \

To replace NAME; with nothing and ;cpd;*;line;*\t with \t

The NAME; replaces fine, but the second part doesn't .

Next I tried

Code:

sed 's/;cpd;[0-9]*;line;[0-9]*//g' | \

for the second part and that does seem to work.

Am I going about this the right way? It seems like I should just be able to use awk with ; as FS.

LMHmedchem

Code:

awk 'BEGIN{FS=";"}{print $2}' filename

hope this helps

Code:

awk -F "\t" '{split($2, a, ";"); $2 = a[2]}1' file

@OP: Regarding your attempts with sed, you need to use .* instead of * ( .* in a regular expression means zero or more characters). When you use the asterisk like this: ;* it means zero or more semi-colons.

What you did in the last example was [0-9]* which means zero or more digits, so that is correct..

sed solution

Code:

sed 's/\tNAME;\([^;]*\);[^\t]*\t/\t\1\t/' file

---------- Post updated at 09:03 AM ---------- Previous update was at 02:55 AM ----------

Perl solution

Code:

perl -F"\t" -lane 'BEGIN{$"="\t"}
  {@A = split(/;/, $F[1], 3); $F[1] = $A[1]; print "@F"}' file

I am sorry for the very long delay, one of my servers has been acting up and I needed to get it fixed.

I always seem to forget that regex is not the same as using a wildcard in bash. I guess it doesn't help that there are some many flavors of regex. I spent a few days a while ago trying to decipher some regex in ruby.

Could you provide an interpretation of this? Is there a difference between -F "" and FS=""? It looks like you used -F "\t" to indicate that it is a tab delimited file, but then split on ; to split column 2. Is that right? That would be at least three different ways to specify a delimiter.

As indicated, this works properly,
sed 's/;cpd;[0-9]*;line;[0-9]*//g' | \

LMHmedchem

awk's standard field splitting is done according to the FS variable. There are several ways to set that variable, one is with the -F directive. The split() function also uses the FS variable, but it can be overridden with a 3rd field to the function, which is done here..

These are just two different approaches, you can use sed's regex to specify the pattern that needs to be substituted (this could also be done with regex in awk for that matter), or you can use awk's field splitting capabilities (which technically also uses regex BTW)..

---
SriniShoo's solution would also need to set the OFS variable to '\t', for it to work properly:

Code:

awk -F "\t" '{split($2, a, ";"); $2 = a[2]}1' OFS='\t' file

Otherwise the end result would be space-separated rather than TAB-separated..