Bash: After redirected input file is done, allow user to control application via STDIN 1

Recently, we needed to start an application using a script, which application had its own CLI.
After starting it, we had to feed it with some input to configure it before handing it over to the end user.

The application we used was named dog. We wrote into a plain text file (named food) the commands that we needed to feed the application and then we started the execution using a plain input redirect like so dog < food;.
Doing so, resulted into properly feeding the dog application the food data  but it would cause the application to terminate immediately after the food was done as it would also feed the EOF (End Of File) directive to dog.
Apparently, the application after receiving the EOF, it would then terminate without allowing the end user to take control.

To mitigate the problem, we used the cat command to concatenate the input file along with the stdin stream, permitting us to first feed all the data in the food file and then allow the user to input data using the standard input stream.
Our final command resulted in the following solution as below

cat <(cat food) - | dog;

Where - is the special sign for standard input stdin.
cat food can be of course replaced with other commands that produce output on the standard output (stdout).

A bad side-effect of this solution, is that we lost some functionality of the terminal including, but not limited to, using the backspace and navigation.


Find all git repositories and perform a pull operation on them.

The following command will find all git projects in your home folder and perform a pull operation on them.

find ~ -name ".git" -type d -exec bash -c "echo '{}' && cd '{}'/.. && git pull" \;

The above command is based on finding the .git folders that exist in any clone of a git repository. Once a .git folder is found, it will navigate to its parent folder where it will perform the pull request.

Bonus – Automate the procedure using a cron job

The following entry in crontab allows us to periodically perform a pull request on all the cloned repositories that we have in a specific folder. Specifically, it will perform this operation once every five minutes.

*/5    *    *    *    *    cd /home/bytefreaks/Projects; find . -name ".git" -type d -exec bash -c "echo '{}' && cd '{}'/.. && git pull" \; &> /tmp/bf.git.log

Please note that it would be easier to use an ssh key that does not have a password for this automation.
If you do not, the you will need to either pass the password via this configuration line (not recommended) or have a key agent running to provide the password for the key.

Redirecting standard error (stderr)

The following command will redirect stderr to a different file than the one stdout is redirected to:

command >log.txt 2>errors.txt;

In case you want to redirect stderr to stdout (&1), and then redirect stdout to a file you can use the following setup:

command >mixed-log.txt 2>&1;

The following command will have the same effect as the previous one, the difference between them is the way they are implemented. This time we will redirect both the stdout and stderr to a file:

command &> mixed-log.txt;


Git: fatal: index file smaller than expected

Recently, we got the following error from git

$ git status
 fatal: index file smaller than expected

To fix it, without losing local changes, we removed the corrupted index and reset back to the HEAD using the following commands:

$ rm .git/index
$ git reset HEAD .