Pun Salad

[Update 2020-07-16: added some new logic to allow link target text to be provided on standard input. Prettified (slightly) the site-string chopping regex.]

First, a bit of background on my environment:

• I use Google Chrome for my browser in Linux.

• I use plain old vim in a terminal window to compose HTML for this blog.

• And what I want to do all the time when composing HTML is to generate a link to the page displayed in the active tab in Chrome's current window.

For example, if I'm looking at this page in Chrome, I might say "Ooh, cool!" and want to insert the following into my HTML:

<a href="https://science.slashdot.org/story/20/07/01/1816253/a-massive-star-has-seemingly-vanished-from-space-with-no-explanation">A Massive Star Has Seemingly Vanished from Space With No Explanation</a>

That's not hard to do by hand: copy the link from Chrome into the terminal window, add in the surrounding code for the a tag, add the target text, don't forget the end tag (</a>), and we're done!

Yeah, it's not hard, but it can be tedious.

I'm sure people—much smarter people—have come up with good solutions for this. But I'm a DIY kind of guy. So eventually (it only took years), I wrote this small (43 68-line) Perl script to do that for me. For historical reasons (by which I mean: arbitrary and silly reasons), I named it ttb. Which stands for "Thing To Blog", and it's installed in my $HOME/bin directory.

My usual use is in vim command mode, bang-bang-ttb:

!!ttb

… which will replace the current line with the HTML link:

<a href="URL">target</a>

where URL is (duh) the URL of the active tab of Chrome's current window.

The target link text is determined by the following logic:

• If the current line contains any (non-whitespace) text, use that for the target text. (After trimming any leading or trailing whitespace.)
• Otherwise, if any command-line arguments are specified, join them together with spaces, using the result as the target text.
• Otherwise, use the HTML title of the displayed page as the target text.

That might look a bit convoluted, but… well, it is. But it works OK for me.

Notes:

• The script assumes you have installed the chromix-too Chrome extension package. Which is easy enough to get. In Fedora, I install the npm package first:

  # dnf install npm

  or equivalent sudo if you prefer that. Then:

  # npm install -g chromix-too

  This package contains a client-server pair: chromix-too and chromix-too-server. The server can be run after Chrome itself starts up. (I run both Chrome and chromix-too-server as startup commands.)

• The script executes the client via the shell command:

  $ chromix-too raw chrome.tabs.query '{ "active":true, "currentWindow":true }'

  which produces JSON output about the active tab in the current window. The JSON perl module (I think it's installed by default in Fedora) is required to decode that into a Perl structure. The decode function returns an "array of hash", but I think the array should always have just one element, so we just pop that.

• Ugly things probably happen if you run this without the browser or the chromix server running. I should probably provide a clean exit in that case.

• I noticed a lot of sites (mostly blogs) have HTML page titles that append a uniform site string. There's an ugly ad hoc regex in the code to chop those off. (Or should that be ad hack?)

That's a dreadful lot of verbiage about such a short script. As usual, this is not earth-shattering code, but I hope someone finds it useful, if only for tutorial purposes.

And if you know of a better way to do this… don't tell me, OK?

The source may be found at GitHub.