grep command in bash scripting

Question

grepstr()
{
 grep "$1 $2" $TMP/"ORACLE_SID"_dbmode.txt > /dev/null 2>$1
}

1. What is the meaning of the above command?
2. What are $1 and $2? Why it is necessary?
3. Is there any alternatives for using $1 and $2?

Answer 1

Is that exactly what the code looks like?

As it is, it is a function that takes two arguments, let's say they're arg1 and arg2.

Then it reads a file called ORACLE_SID_dbmode.txt and prints every line that contains arg1 arg2 (joined together with a space in between), and saves the error messages in a file called arg1.

This is quite confusing, so I don't think this is the actual code.

I would imagine it is actually this:

grepstr()
{
    grep "$1" "$2" $TMP/"$ORACLE_SID"_dbmode.txt >/dev/null 2>&1
}

Changes:

• "$1 $2" becomes "$1" "$2"
• ORACLE_SID becomes $ORACLE_SID
• 2>$1 becomes 2>&1

If so, it could be used two ways:

grepstr options string_or_pattern

for example:

grepstr -i something

would make it print (see below) any lines in the file "$ORACLE_SID"_dbmode.txt that include the string something, Something, SOMETHING, etc.

-i means case insensitive and is just an example; any option could be used.

(Also, technically something is not a string, but a pattern, for example [0-9] would match any digit).

The second way is:

grepstr string_or_pattern filename

in which case it will print any lines matching string_or_pattern in filename or "$ORACLE_SID"_dbmode.txt.

That seems less likely, but it is possible, depending how the code is used.

It could also be called like this:

grepstr option1 option2

but that would read from standard input (e.g. the keyboard by default), and print any lines that contained "$ORACLE_SID"_dbmode.txt, which is almost certainly not what was intended, given "$ORACLE_SID"_dbmode.txt looks like a file name, not a string or pattern to match.

If my guess is right, the code isn't >/dev/null 2>$1, instead it's >/dev/null 2>&1, which means output and error messages are not printed.

>/dev/null means that grep's standard output (i.e. the matching lines) is being sent to /dev/null, which is a special file that the operating system just ignores.

2>&1 after >/dev/null means that error messages go there too.

So if the output isn't important, grepstr is probably being called either like this:

if grepstr...

or like this:

grepstr...
if [ $? -ne 0 ]

to only execute some code if there was a match.

In answer to the other part of your question, yes, it could be changed to

grepstr()
{
    grep "$@" $TMP/"$ORACLE_SID"_dbmode.txt > /dev/null 2>&1
}

The "$@" expands to "$1" "$2" "$3"... (i.e. as many arguments as are provided).

I think that way would be better, because it would be more flexible.

On the other hand, if it does require two arguments for some reason I can't think of, then perhaps "$1" "$2" is what the author intended.

Answer 2

$1 and $2 are parameters. Your snippet is defining function grepstr.

For example, running

grepstr first second

actually runs command

grep "first second" $TMP/"ORACLE_SID"_dbmode.txt

and redirects output to /dev/null (which is message sink, everything put there just disappears).

Basically, when you run script (or function), $1, $2, $3 (etc.) are parameters you gave. You can test whether parameter is set with -z, for example if [ -z $3 ]; then ....