find a specific string pattern from a file in Unix Shell Scripting

Question

I have the below command.

 unzip -p GLP.K4C.S06F5.BG57218-rdf.zip | grep  ":taxonomies-" | head -1

which gives me the output as,

    <j.2:Taxo_Version rdf:resource="refmat:taxonomies-8.2.0"/>

However, I need to extract only taxonomies-8.2.0 instead of full string as above.

Answer 1

One way is to use grep's -o option, combined with the power of PCREs (-P):

   -o, --only-matching
          Print  only  the  matched  (non-empty) parts of a matching line,
          with each such part on a separate output line.
   -P, --perl-regexp
          Interpret  PATTERN  as  a  Perl  regular  expression  (PCRE, see
          below).  This is highly experimental and grep  -P  may  warn  of
          unimplemented features.

So, you could do

 unzip -p GLP.K4C.S06F5.BG57218-rdf.zip | grep -oP ':\Ktaxonomies-[^"]*' | head -1

The \K causes anything matched up to that point to be ignored (so the : is not printed) and [^"]*" means "match as many non-" characters as possible.

Other options include:

1. sed

   unzip -p GLP.K4C.S06F5.BG57218-rdf.zip | 
       sed -n 's/.*:\(taxonomies-[^"]*\).*/\1/p' | head -1
   

   The -n causes sed to print nothing unless explicitly told to and the s/// is the substitution operator. It will replace everything on the line with the part of the line between the parentheses (\1). The p causes the resulting line to be printed.

2. Perl

   unzip -p GLP.K4C.S06F5.BG57218-rdf.zip | 
     perl -lne 's/.*:(taxonomies-[^"]).*/$1/ && print' | head -1
   

   The same basic idea as the sed. If the substitution was successful, the line is printed. An alternative would be

   unzip -p GLP.K4C.S06F5.BG57218-rdf.zip | 
     perl -lne '/.*:(taxonomies-[^"])/ && print $1' | head -1
   

Answer 2

If you know the occurence of : character in your input, you could do something like this.

echo " <j.2:Taxo_Version rdf:resource="refmat:taxonomies-8.2.0"/>" | 
awk -F\: '{print $4}' | sed 's/..$//'

The awk command prints the 4^th string after the : delimiter and the sed command is used to remove the last 2 characters to get the desired output.

However, if this method works or not depends on your input as terdon points out in his comments.

EDIT

The final pipe to sed could very well be avoided if we use the solution as suggested by jasonwryan in the comments. So, the command would effectively be rephrased as,

 echo " <j.2:Taxo_Version rdf:resource="refmat:taxonomies-8.2.0"/>" | 
 awk -F: '{sub(/\/>/,""); print $4}'

Another solution just using cut and rev can be framed as,

echo " <j.2:Taxo_Version rdf:resource="refmat:taxonomies-8.2.0"/>" | 
cut -d ':' -f4 | rev | cut -c 3- | rev

Again the specifying of delimiter is dependent on the input file and from the example you have provided the characters that I need to extract occur after the 4^th position of the delimiter. I use cut to extract the substring after this 4^th delimiter and use good old rev technique to reverse the string and remove the last 3 characters and again apply rev on it to get the actual string.