Pinboard popular page in Emacs
Ok, this one is still piping hot in my
*scratch* buffer so remember that before you criticize the rough edges. I've long wanted to quickly filter through the popular links page on pinboard.in without jumping over to a browser. Ideally this page would have an RSS feed or API end point but alas it has neither. I love pulling things into Emacs (yay for text-based interfaces!); so I set out to pull these links into Emacs and display them via Ivy.
Feel free to skip ahead and checkout the repo if you're so inclined.
curled the page down and pasted the HTML into a buffer in Emacs so I could start pulling things apart. My first attempt was to use the excellent
elquery library but that was choking on the DOM structure and I never could quite pin down where (sorry Adam, I need to post an issue about that!). My next attempt was to use a more "manual" scripting approach. Since Emacs possesses so many ways to manipulate buffer text I was sure there could be a programmatic way to do this (without doing a bunch of regexing which I'm terrible at and even more so in Emacs). The basic series of steps boiled down to this:
- Use the new-to-me
keep-linesfunction to trim everything out of the buffer except the popular links. I used the classname of the links (
bookmark_title) to identify them.
loop-for-each-lineto well loop over each line. Inside this loop I would need to pull out the
hrefand the title of each anchor link.
- I used
re-search-forwardto move from target to target to figure out the point boundaries for my href and my link text.
I ended up breaking this functionality out into it's own function since I used it more than once and it was cumbersome to type it all out.
(defun re-capture-between(re-start re-end) "Return the string between two regexes." (let (start end) (setq start (re-search-forward re-start)) (setq end (re-search-forward re-end)) (buffer-substring-no-properties start end)))
- The final step was dumping the
hrefand the link text into a plist and
pushthat into a larger collection I could pass onto
ivy-readto generate my interface.
One gotcha I ran into is that my titles would occasionally contain odd character sequences like
\302. Turns out those are punctuation and various text symbols. The solution is to run the title through
'utf-8 to generate the proper characters.
Emacs is so powerful for this type of text manipulation because you can interactively work each step of a program out in an actual buffer and then put those steps together into a function. Each potential solution I thought of was quickly trialed via
M: and frequent looks at documentation. Hopefully this has been a helpful look into how I approach solving problems in Emacs. Thanks for reading!