A reader suggested that I add a Table of Contents to the LeanPub page for my latest book, Mojolicious WebClients. I hadn’t thought about it, but once he suggested it, I thought “Why doesn’t LeanPub extract this from my ePub automatically?”. Maybe is does, but it has no instructions of what to do to make that happen. I’ll make that much easier for you.

It the book details page, I can add HTML that LeanPub will insert into my book’s summary page. It’s instructions tell you to look at the HTML source for other book pages, but also that they can change their rules at any time (screenshot at the end).

Rather than comb through the HTML of other book’s details pages, here’s what I did. I have a nested ul lists:

<ul class="toc no-parts mojo-web-clients-toc">
<li>Preface
    <ul>
    <li>Some eBook Notes</li>
    <li>Installing Mojolicious</li>
    <li>Getting Help</li>
    <li>Acknowledgments</li>
    <li>Perl School</li>
    </ul></li>
<li>Introduction
    <ul>
    <li>The Mojo Philosophy</li>
    <li>Be Nice to Servers</li>
    <li>How HTTP Works</li>
    <li>Add to the Request</li>
    <li>httpbin</li>
    </ul></li>
...
</ul>

If you like your browser’s version of “View Source” (Safari’s is designed to make you not want to do that), fine. I downloaded the page source locally so I could view it in BBEdit:

$ curl --silent https://leanpub.com/mojo_web_clients | bbedit

With Mojolicious’s mojo command, that’s a little shorter:

$ mojo get https://leanpub.com/mojo_web_clients | bbedit

That sucks though. Here a Mojolicious one-liner to get as close to the Table of Contents as I can. This CSS selector is obviously fragile, and would have been much shorter if they had added id attributes. If you are telling people to extract something from HTML source, you should label it so make that easy:

$ perl -Mojo -E 'say g(shift)->dom->at(
    q(#scroll-wrapper > div > article > section:nth-child(9) > div > div > div > div > div > div )
    )' https://leanpub.com/mojo_web_clients

Since I added a special class to my ul (LeanPub strips any id element), you can extract my Table of Contents with a class selector:

$ perl -Mojo -E 'say g(shift)->dom->at(
    q( ul.mojo-web-clients-toc )
    )' https://leanpub.com/mojo_web_clients

The toc class might work, but I’m not confident it wouldn’t be repurposed from something else. It’s only there because I cargo-culted it from another book’s page:

$ perl -Mojo -E 'say g(shift)->dom->at( q( ul.toc )
    )' https://leanpub.com/mojo_web_clients

Finally, the screenshot of the instructions: