Yesterday I found myself in a situation where I had a text document interspersed with a bunch of email addresses and I wanted to extract those email addresses.

Specifically, I had to copy email addresses from a couple of spreadsheets and other sources and I thought it would be faster to copy lines form the sheets instead of email cells as well as just grabbing text from the other sources and then isolating the email addresses in Emacs.

Now, writing a little "extract emails" script to operate on a text file is easy enough. You can search for an email address using a regular expression. Regular expressions are a text pattern matching language that most programming language support. I used something like the following regular expression to identify an email address:

[a-zA-Z0-9\.]+@[a-zA-Z0-9]+\.[a-zA-Z]{3}

The first [a-zA-Z0-9\.]+ matches a sequence of one or more characters that are letters, numbers or a period (the slash before the period is to escape it - prevent using the . for its special regular expression meaning). That's followed by the @ symbol, a second set of characters, this time without the period, then a dot and then three more characters.

This will match most email addresses.

In Clojure, I'd use the function re-seq which would take a string (presumably my text file) and return a list of all the matches. That is, all the email addresses. I could do the same in Python with re.findall.

In an editor, at least in Emacs, it's a little trickier. You can search for a regular expression but that just finds the next email address. I'd then have to manually copy it and then repeat the process.

I guess a keyboard macro could do the trick but easier would be to just write an elisp function and extend Emacs.

The video shows the whole walkthrough but the core of the routine is using the search-forward-regexp function which searches for the next occurrence of an email match. I then add that email address to a list. Putting it all in a while loop, you get:

(while (search-forward-regexp regexp nil t 1)
  (push (match-string-no-properties 0) matches))

The match-string-no-properties grabs the actual text of the match.

There's a little more detail but basically I wrote an equivalent of re-seq for an Emacs buffer - it grabs everything that matches a regular expression and returns all the matches in a list.

Then, it was a simple matter of looping over that list and inserting the results. I ultimately named the routine extract-emails. So now, I can be in any Emacs buffer, copy over all my source email material, run the command, and I'll have all the emails together one per line ready to be pasted into an email client.

I'm thinking that I should probably give the routine an additional optional parameter - have it extract emails by default but allow the user to pass in any regular expression.

Just another reason why Emacs is such a powerful editor.

Here's the video, enjoy: