Ubuntu – Spaces in commands with redirection

bashcommand lineredirect

program 2> error.log 
program &> filename 
program >> filename 2>&1
program 2>&1

I was able to figure out what these lines mean.

However I don't quite understand where I should put the spaces. I also worry that it actually doesn't matter where to put spaces. Thanks for reading.

Best Answer

Yes, spacing between words and redirections doesn't matter. That's the short answer.

The details lie in the fact that shell ( for simplicity let's just refer to bash only) treates certain characters and groups of characters as "words" and "metacharacters". From bash 4.3 manual:

metacharacter

A character that, when unquoted, separates words. One of the following:
|  & ; ( ) < > space tab

And

word A sequence of characters considered as a single unit by the shell. Also known as a token.

So when we do:

$ echo "hello world">/dev/null

$ echo "hello world" > /dev/null

that's still 3 words ("hello world" can be considered a single shell word because it's quoted), with one > meta character and couple spaces. Shell will see it, and perform redirection first (so it looks for meta characters first), and then runs commands in accordance with its standard behavior.

Order of redirections, however, matters a lot, especially when you're duplicating file descriptors with something like 2>&1. Say you want to get send both stderr and stdin to same location. Here's a wrong way to do it:

$ stat ./non-existent file 2>&1  > /dev/null
stat: cannot stat './non-existent': No such file or directory
stat: cannot stat 'file': No such file or directory

You're making file descriptor 2 output to same location as 1, which is your terminal, but it was already doing so. That's why stderr shows up.

If you redirect stdout first, and only then change where 2 points - then it'll work:

$ stat ./non-existent file > /dev/null 2>&1

Related Solutions

Ubuntu – Up arrow and .bash_history have different commands, so where does the up arrow get its data

The ~/.bash_history file is only updated when you close/exit from the terminal. While you are in the terminal you can browse backwards in commands which was entered in that session. So if you exit from that terminal session your history will be updated with the missing command.

Also if you put a space at the beginning of a line, before the command, that command won't be saved in the history. This is a good practice if you issue a command which you don't want to be included in the history.

Ubuntu – Use sed command to apply a condition to check if a pattern has been matched

grep does that job well enough.

$ echo "::=BEGIN" > testfile1.txt

$ grep "::=BEGIN" -q testfile1.txt && echo "FOUND"
FOUND

$ grep "::=BEGIN" -q testfile1.txt && echo "FOUND" || echo "NOTFOUND"
FOUND

$ grep "whever" -q testfile1.txt && echo "FOUND" || echo "NOTFOUND"
NOTFOUND

What the code does is simply running a quiet search within subshell. && , or logical AND operand, checks if the first command succeeded, and if it did, then it runs echo "FOUND" . || checks if that echo has been run, i.e., if we found anything. Second echo runs only if the first command failed.

Here's an awk version:

$ awk '{i="NOTFOUND";if($0~/::=BEGIN/) i="FOUNDIT";exit } END { print i}' testfile1.txt
FOUNDIT

$ awk '{i="NOTFOUND";if($0~/whatevs/) i="FOUNDIT";exit } END { print i}' testfile1.txt
NOTFOUND

Basic idea here is to set i to NOTFOUND, and if we find the string - change it to FOUNDIT . At the end after the first set of has finished processing file, we will print i, and it will tell us if we found something or not.

Edit: In the comment you mentioned that you want to place the result into a variable. Shell already provides a variable that reports exit status of the previous command, $0. Exit status 0 means success, everything else - fail. For instance,

$ grep "::=BEGIN" -q testfile1.txt  
$ echo $?
0

If you want to use exit status in a variable, you can do it like so : MYVAR=$(echo $?). Remember , $? report exit status of previous command. So you must use this right after the grep command.

If you want to still use my earlier awk and grep one-liners, you can use same trick MYVAR=$() and place whatever command you want between the braces. The output will get assigned to variable MYVAR.

Best Answer

Related Solutions

Ubuntu – Up arrow and .bash_history have different commands, so where does the up arrow get its data

Ubuntu – Use sed command to apply a condition to check if a pattern has been matched

Related Question