Bash string comparison function.. something probably stupid wrong.

[qwertygeek]

Bash script, simple string comparison, doesn't match on confirmed matches (added debug prints to check).. I know this is going to be a "duh" thing... does someone see something glaringly wrong?

CheckName () {

    local pushIFS=$IFS #add pop it later
    IFS=$'\n'

    recordfound=-1

    for line in $(cat $OUTFILE)
    do
        recname=`echo $line | cut -f1`
        echo "$recname ..... $1"
        if [ "$recname" = "$1" ]
        then
            recordfound=0
            echo "found it"
            return
        elif [ "$recname" != "$1" ]
        then
            echo "not eq"
            continue
        fi
    done

    return
}

read name
CheckName "$name"

[gerard4143]

I'm not really a basher but this sticks out

IFS=$'\n'

Why the $ in front of '\n'

[qwertygeek]

I'm not really a basher but this sticks out

IFS=$'\n'

Why the $ in front of '\n'

Without it, I was getting rather wonky results. I googled around and saw another example where someone used that, so presumably that gets the ASCII code for the newline char?

Interestingly enough, I just tried the exact script I've posted on a Fedora box (without the surrounding code) and it works fine... so it's either the surrounding code or OpenSATAN.. I mean SUSE. *shudder* (I'm a Fedora/Ubuntu guy..) I don't have the rest of the script with me, but I don't know what code before it could be causing the problem really given it's just reading a string and passing it to a function...


EDIT: what I mean by the wonky results:

Here's just an example file it's reading from:

chris   something       3       1       4       1 
bob     latida  5       9       2       6
foo     bar

Here's the output (working, on the Fedora box) if I use $'\n':

$ ./strcmp out
bob
chris ..... bob
not eq
bob ..... bob
found it

Here's the output if I use just '\n':

$ ./strcmp out
bob
chris ..... bob
not eq
g
bob
foo ..... bob
not eq

I think without the $ it doesn't do the value-of return.

[gerard4143]

I tried this on Arch Linux and it work...well once I corrected the data file to have tabs delimiters...Damn copy and paste from web pages..

As you can see it pretty much what you had except I hard coded "bob" as the choice..

#! /bin/sh

echo -n "enter filename->"
read myfile

if [ -f "$myfile" ]
then
	IFS=$'\n'
	for line in $(cat $myfile)
	do
		recname=`echo $line | cut -f1`
		if [ "$recname" = "bob" ]
		then
			echo	$line
		fi
	done
fi

exit 0

Even tried it with a function like below and it worked

#! /bin/sh

echo -n "enter filename->"
read myfile

echo -n "enter name->"
read myname

getit()
{
	for line in $(cat $myfile)
	do
		recname=`echo $line | cut -f1`
		if [ "$recname" = $1 ]
		then
			echo	$line
		fi
	done
}

IFS=$'\n'

if [ -f "$myfile" ]
then
	getit $myname
fi

exit 0

[sixdrift]

First off, the IFS variable is used to control parsing on input and arguments. It defaults to basic white space characters (space, tab, and newline). You have to include the $ in setting it as you have because otherwise it would set IFS to the ' character (single quote). But since you wanted the newline, you have to enclose the symbol for newline in single quotes and then escape the single quote when setting.

But I think the way you are going about reading the list might be a bit convoluted, unless you really want to run "cut" on each pass. The whole for loop is unneeded I think. Consider this version:

#!/bin/bash

myfile="$1"

CheckName() {

    while read name thing eolstuff
    do
    if [ "$name" = "$1" ]; then
        echo "found it"
        return
    else
        echo "not found yet"
    fi
    done < "$myfile"

    return
}

echo -n "enter name to find: "
read name
CheckName "$name"

I ran this on BASH (using your data) and got the following:

$ ./strcmp data
enter name to find: bob
not found yet
found it

$ ./strcmp data
enter name to find: chris
found it

$ ./strcmp data
enter name to find: foo
not found yet
not found yet
found it

$ ./strcmp data
enter name to find: dummy
not found yet
not found yet
not found yet

The use of the "here" document in the read loop gets your one line at a time. The read of different variable names implicitly cuts the input line around white space giving you the pieces of the line without calling cut.

Now if you need to act on the remaining fields of the record you found, you can simple take the "eolstuff" variable and then chomp around on it to break it out into its various pieces.