Remove leading whitespace

Sometimes it’s useful to copy some code from a file and paste it somewhere else as an example. For instance, I like to write code in a jsfiddle and then paste a relevant subset of that code to a Stack Overflow answer. I have also copied code from a project file to paste into HipChat to quickly explain something to a coworker, or to paste into a JIRA ticket as a comment.

If you copy code from the middle of a file, there’s usually some leading whitespace on all lines that you do not want to preserve in the context into which you are pasting the code. The problem is that you don’t want to get rid of all leading whitespace on all lines; you want to keep the indentation intact. I usually get rid of the unwanted whitespace by manually deleting it if it’s only one or two lines or by pasting into a new vim buffer and using << to shift the text over as much as desired.

To illustrate, I have this text:

    var flatten = function(result, next_array) {
        console.log('current result', result);
        return result.concat(next_array);
    };

    [1, [2], [3, 4]]
        .reduce(flatten, []);

and I want this text:

var flatten = function(result, next_array) {
    console.log('current result', result);
    return result.concat(next_array);
};

[1, [2], [3, 4]]
    .reduce(flatten, []);

I knew there must be a way to remove the shortest leading whitespace from all lines programmatically, but I’m not familiar enough with awk, sed, or shell scripting in general to tackle the problem. I asked the question on Stack Overflow and got a few great answers. I ended up accepting the single process awk version.

If you use OS X, the built-in awk will not work with the given solution. If you use Hombrew, fixing that is just a simple matter of brew install gawk and using gawk instead of awk.

The given solution has a great explanation and works fine, but I made one addition. If the input is a single line with no leading whitespace, the script fails. I fixed this with if (!s) s=0; at the beginning of the END block.

The final version of my command looks like this. I’ve added some comments to explain what’s going on.

gawk  -F '\\S.*' \              # The awk field separator is everything after the first non-whitespace character, inclusive
'{                              # The first block of the awk program
    l=length($1);               # The length of the first field, the leading whitespace
    if(l>0)                     # If the length of whitespace is non-zero,
        if(NR==1)               # and this is the first record,
            s=l;                # make 's', the number of whitespace characters, equal to its length
    else s=s>l?l:s;             # otherwise, make s the shorter of itself and the current whitespace
    a[NR]=$0                    # Index the entire line in an array by line number
}                               # End of the first block of the awk program

END{                            # Start the END, printing block of the awk program
    if(!s)s=0;                  # Make sure we always have a value for s
    for(i=1;i<=NR;i++){         # Loop over all records
        sub("^ {"s"}","",a[i]); # Substitute 's' whitespace characters with nothing
        print a[i];             # Print the line after substition
    }                           # End of the loop over records
}'                              # End of the END block of the awk program

It is worth noting that this probably only works on text with spaces for whitespace, not tabs or mixed whitespace.

Integrate with an Alfred workflow

The ability to remove the shortest leading whitespace with a shell command is great, but I really wanted a way to do this quickly with text on the clipboard. Alfred workflows make that possible.

I created a simple three step workflow with a hotkey trigger, a script action, and a clipboard output. You can download it here and use it with Alfred 2.

And now you can copy code in context and paste it with no leading whitespace wherever you want!