awk can't open file; file merge attempt

Shell Programming and Scripting
1

Dear all, I have an AWK related issue.

I have two data files; the first, FileA has fewer lines, the second, FileB has more. FileA [75MB] is a subset of FileB [20GB]. Both files are tab delimited.

What I want to do?
When the first two columns for FileA match the first two columns of FileB, I want to print all 24 columns of FileB.
The output file should have a row number == the number of rows in FileA.

I tired this in AWK:

awk 'BEGIN{OFS="\t"} NR==FNR {f1[$1$2] = $0; next} ($1$2 in f1) {printf "%s",f1[$1$2]; for(i=4; i<=NF; i++) printf "\t%s",$i; printf "\n"}' FileA FileB*> NewFile

When I run, it throws the error:

" awk: can't open file FileA*
 input record number 1402341, file FileA*
 source line number 1                          ## input number = row number FileA

I can't figure out why this won't work. If someone can, I'd appreciate the help a lot !!

2

Does FileA have an asterisk as its trailing character... "FileA*" or is it simply "FileA"?

3

Ah! Apologies for the asterisks!

It is just called FileA; there is no trailing asterisk. I think the asterisks may have arisen in conversion to the code boxes that allow for clearer display here.

Anyway, no asterisks either in the files, or in the error readout!

------ Post updated at 06:44 PM ------

Unless I have somehow non-visual characters in the code, which I have mistaken for white space, in which case, that could explain why these asterisks have appeared in each box?

4

One reason why a utility like awk might see an asterisk and include it in a diagnostic message even though it isn't visible when you cat the script onto your terminal is <backspace> characters in your script file. For example, if what you're seeing on your screen as:

FileA FileB*> NewFile

actually contained some other text and <backspace>s as in the following:

FileA FileAB<backspace>* <backspace><backspace><backspace><backspace><backspace><backspace><backspace> B*> NewFile

that would produce the diagnostic message you've shown us from the script you've shown us.

Have you tried using:

od -bc script_name

to see if there are any invisible characters in your script?

5

Other findings (that lead to malfunction but hardly relate to your error message).
1.
You don't make use of the OFS, but your input files are tab-separated. So you certainly need

BEGIN { FS="\t" }

It makes sense to include the FS in the key otherwise $1=1 $2=12 are not distinguishable from $1=11 $2=2

{ key=($1 FS $2) }

And use key in lieu of $1$2 .