This is the third and final instalment on searching. In the first instalment we learned how to search for text within files and streams using egrep. In the second we learned to search for files based on all sorts of criteria with the find command. In this final instalment we’ll start by looking at one last feature of find, its a ability to execute commands on the files it finds. Then we’ll end by looking at an OS X-only alternative to find that makes use of the Spotlight search index to really speed up searches.

Listen Along: Taming the Terminal Podcast Episode 21

Executing Commands with find

The final trick find has up its sleeves is that it can be used to execute a given command on all the files it finds. Depending on the command you choose to execute, this could be VERY dangerous indeed, so tread carefully! Definitely think twice before using find‘s execute feature to delete files!

To execute commands on the results of find you need to use the -exec flag in conjunction with the {} flag and either the ';' or '+' flags. The reason there are two end flags is that there are two modes in which find can execute commands. It can either execute the same command once on each file that was found, or, it can execute the given command once only with all the found files as arguments.

Executing a Command on Each File/Folder Found

To execute a command repeatedly, once on each file or folder found, you use the following construct:

The -exec flag means we are about to start specifying a command to execute, and everything up to the ';' will be a part of that command. {} is the point in the command where the found file should be inserted.

This sounds quite complicated, but hopefully a practical example will help. A real-world problem you may encounter is that you need to make all perl scripts in a given set of nested folders executable. As we learned in instalment 6, we can make a file executable with the command:

Perl scripts have the extension .pl, so using what we learned in the previous instalment we can find all perl scripts in a given path with the command:

If we assume all our perl files will be in a folder called scripts in our Documents folder, we can put all this together to find and chown all perl files in a single command like so:

Executing A Command Once with All Files/Folders Found as Arguments

Sometimes we may want to run a command once with all the found files and/or folders as arguments, you can do that with the following construct:

Like before, {} indicates where the found files and folders should be included in the command, and the '+' on the end indicates the end of the command, and that the command should be run once with all the results as arguments.

Again, a practical example might be helpful. Lets say you’re doing up a report on just how much work went into a coding project. You want to know how many lines of code you wrote, and you know that all your code is in perl scripts and perl modules, i.e. you know all relevant files have .pl and .pm file extensions.

We know from last time that we can find all relevant files with the command:

We also know from instalment 15 that we can use the wc command with the -l flag to count the number of lines in a one or more files:

Assuming our code is still in a folder called scripts in the Documents folder, the command to find the total number of lines in all the perl files would be:

Note that we have to group the two conditions in bracket operators so the scope of the -or is clear.

If you installed the XKPasswd 2 perl module via the quick install instructions at , you can use the example below to see how many lines of code and documentation went into that project:

One final example that might depress you – the command below tells you how much disk space you are wasting with large files (>500MiB) in your downloads folder:

If you are VERY careful you can use find -exec to do things like clean up cache folders by deleting anything older than an given amount of time, and other cool and powerful things. But – ALWAYS be careful when using -exec to invoke a command that alters or destroys files or folders. My approach is to run the command without the -exec first, to be sure ONLY the files and folders you expected to be found are being returned.

Finally – the nerd joke I think I promised during the last instalment:

(If you don’t get it, this might help.)

OS X Only – Spotlight from the Terminal

On OSX the operating system indexes the files on your computer so you can quickly search them with spotlight. Rather than having to search through all the files, spotlight just searches the index, which makes it much faster than find when searching through big chunks of the file system.

Apple have very kindly exposed the power of spotlight to the command line with the very useful mdfind command.

If you can type it into the spotlight text box in the top-left of a Finder window and get back a list of files, you can pass the same query to mdfind, and it will find the same results. Note that I said spotlight in the Finder, and not in the menubar – that was not by accident. mdfind is only for finding files, not for all the other fancy stuff you can do with spotlight in the menu bar like the calculator function or the ability to get dictionary definitions. Another important caveat is that it mdfind can only find files in folders indexed by spotlight. If you add a drive or a folder to spotlight’s ignore list, mdfind can never find files on those drives or folders.

The mdfind command is very easy to use:

For example, to find all PDFs on your computer you could use:

To find all PDFs that contain the word ‘internet’ you could use:

To find every PDF you have edited today you could use:

I haven’t been able to find a definitive list of all possible spotlight search commands, but googling for ‘spotlight syntax’ will lead to useful articles like this one.

mdfind supports a number of arguments, and it can also search based on deep metadata, but I’m going to leave most of that as an exercise for the user – you can get the full documentation through the manual:

However, there are two flags I do want to draw attention to.

By default mdfind will search the entire spotlight index, but you can ask it to only return results contained within a given folder using the -onlyin flag, for example, to find all PDFs in your Documents folder you could use:

Note that the search is always recursive, even when using the -onlyin flag.

The other flag I want to mention is -count, if this flag is set the number of matching files will be returned rather than the files themselves. So, to see how many music files you have in your Music folder you could use:

Or, to answer the eternal question of just how many apps you have installed:

Final Thoughts

We have now seen how to filter streams and search files with egrep, and we’ve learned how to search for files with find and mdfind. That brings us to the end of the searching topic, at least for now. The next big topic will be networking, but before we start into such a big topic we’ll take a break for a fun little tips and tricks instalment.