UNIX for Dummies Questions & Answers

Hello folks,
I have got the following line in my script:
searchstring=".*_1234</MessageTag>"
count=`find /home/mydir/* -name "$1" | xargs egrep -hc -e $searchstring "$1" | awk '{sum+=$1};END{print sum}'`

where $1 is the name of some file that I am passing at run time.

The problem is that whenever I am running this script by just giving the name of the file instead of giving its path (like ./file1.txt) I am getting the following error:
egrep: can't open file file1.txt.

However if I am giving the path of the file it is giving me no error.
Is there any workaround for this?
Thanks a lot.

>(like ./file1.txt) ...
what about just file1.txt ?

Yep ... when I give it like ./file.txt it works fine. But when I give just file.txt I am getting the egrep : cannot read file file.txt. However it brings me some data alright. But obviously this is not the greatest way to do some scripting!!!
Thanks.

Let's treat your command by step:
find /home/mydir/* -name "$1" | xargs egrep -hc -e $searchstring "$1"

so you looking for all files name $1 from a directory to egrep -e searchstring but in $1
deosent make sence since the output of find will not be the same as $1:
you will find strings (file with absolute path) which do contain $1 - Not the same

I would do this (Im using ksh...)
# egrep -hc -e $searchstring $(find /home/mydir/* -name "$1" | xargs )

I let you figure the rest...

I agree with "vbe", but would like to explain the error message.

The error message is because the xargs generates a string containing the filename from the find and appends that string to the egrep line.

egrep -hc -e $searchstring <filename from $1> <filename from the find>

You end up searching the same file twice if both filenames are identical ... or in your case getting an error message from one of the filenames if they are different.

You need to lose the "$1" from the egrep statement, the filename from find is filled in by xargs. If "$1" does not ever contain wildcards, the find and xargs are surplus.

The example from "vbe" addresses this point.

please forgive me but i am a bit of a numb skull ... if i am doing a find with an absolutely correct filename (just filename with no wild character involved) ... so i'll get the exact file i suppose ... and then i am running the egrep part ... so why do i exactly need to lose the $1 argument?

Answer: Because the xargs put the filename there for you already.
The "$1" came from the original parameter used to call the script and has not been changed.
i.e. The filename is on the egrep command line twice!

xargs does not have special meaning for $1, $2 ... unlike awk. It just reads a stream of lines, removes the linebreaks, and optionally appends this string to anything else on the xargs command line.

To illustrate "xargs" in its simplest form without trying to generate a command:

cat <++++ | xargs
line1
line2
line3
++++

line1 line2 line3


To illustrate using xargs to generate a simple echo command. In the second example the filename appears twice.

myname="filename"
echo $myname|xargs echo

filename

echo $myname|xargs echo $myname

filename filename



Footnote 1: xargs has limited use on large systems for creating lists of filenames for subsequent commands because a large number of filenames can cause the generated command to malfunction if it becomes too long.

Footnote 2: It is best practice to save the parameters used to call a script (or a subroutine) into variables with meaningful names. This makes it clear to the reader whether $1, $2, $3 .... is special syntax (like in awk) or the parameters used to call the shell script.
e.g. FILENAME="${1}"