UNIX for Dummies Questions & Answers

Hello guys,

should be a very easy questn for you:

I need to delete strings in file1 based on the list of strings in file2.

like file2:
word1_word2_
word3_word5_
word3_word4_
word6_word7_

file1:
word1_word2_otherwords..,word3_word5_others
word3_word4_otherwords....,word6_word7_others

.. and the output would be:

otherwords..,others
otherwords....,others

etc.

I tried it with the "for" of "while read" loops and sed, but no result: I guess the problems might be: 1. if read $line, the line ending is an obstacle, 2. if read $word, the "_" probably is..


thanx in advance

See if this works:

Code:

awk -F'[_,]' 'NR==FNR{A[$1,$2]=1;next}{for (i=1;i<=NF;i++) if(A[$i,$(i+1)])sub($i FS $(i+1) FS,x)}1' file2 file1

Thanks a lot, but this awk solution deletes from file1 only some part of the strings specified in file2.. why would it not recognize the rest?..
I'm attaching a small example of files 1 and 2, and the awk output, maybe that would make thing clearer.

Thanks again

Hi that is because that solution was catered to entries in file2 with two underscores, like in the sample. Try this instead:

Code:

awk 'NR==FNR{A[$1];next}{for(i in A)gsub(i,x)}1' file2 file1

Thanks for help: works fine!
Except in cases where a string is a substring of another one in file2 (eg, BACILLUS_Bacilli_ is a sub of LACTOBACILLUS_Bacilli_), in which cases there are leftovers after the deletion.
Is there a way to make awk find and delete only strings corresponding to entire lines in file2?

You could do a reverse length sort first:

Code:

awk '{print length($1),$1}' file2 | sort -rn | awk '{print $2}' > file2.sort

and then:

Code:

awk 'NR==FNR{A[++n]=$1;next}{for(i=1;i<=n;i++)gsub(A[i],x)}1' file2.sort file1

Thanks indeed - you helped a lot. Cheers