Undefined title

[chrisspen]

Let me just say great job on Pagefind's performance.

Although it was difficult for me to index all my pages (took about an hour), the resulting index is very fast to search and looks great.

However, all my search results are showing up with an "undefined" title. How do I correct this?

My html has a valid <title> tag in every page. Is this not automatically pulled as the title?

[bglw]

Hello! 👋

First a quick Q — how many pages are you indexing that takes an hour? (Also, what system are you running on?¹)

As for the titles, Pagefind pulls the h1 element from the page for the automatic title, not the title element² (docs). The best fix for Pagefind (and general web accessibility) would be to make sure your pages have valid h1 heading tags.

Pagefind UI uses the value of the title metadata for the results, so if having h1 elements is not possible for some reason, you can tag any element as the title to use that instead. For example, to use the title element:

<title data-pagefind-meta="title">My Title</title>

Footnotes

1. The reason I ask is that right now if you're running on an M1 Macbook, the npx pagefind release will be running through Rosetta and might be artificially slow. ↩

2. The cause of this default is that the title element on website usually includes redundant information (like the name of the site). For example, the h1 of this page is Undefined title ꖛ134, and the title is Undefined title · Issue ꖛ134 · CloudCannon/pagefind ↩

[chrisspen]

First a quick Q — how many pages are you indexing that takes an hour? (Also, what system are you running on?)

I have about 100k, and they're compressed. I mirrored my files using named pipes that feed each file into zcat when read, so that pagefind can read them.

I'm running on Ubuntu 20.

As for the titles, Pagefind pulls the h1 element from the page for the automatic title, not the title element

Yeah, that's what I figured. I understand your rationale. Although it would be nice to be able to configure that in Pagefind. In my case, it would be a huge hassle to either regenerate every single html document (each is large and takes about ~5min to render). I could write a script to parse and edit them in-place, but that's also cumbersome. Being able to modify Pagefind's default title lookup would save me hours, or possibly days of coding.

Having redundant stuff in the title isn't a big deal in my case. I don't have much redundant text and using the result processor callback, what little I do can be easily cleaned up on demand.

[bglw]

Sounds good — I'll whip up exposing that as an option in the same PR for gzip handing (#135) which together should provide some great QOL improvements for your setup.

[bglw]

Actually, in the interest of keeping the available options lean I'll get Pagefind to automatically fall back to <title> if no <h1> is present.

[chrisspen]

Thanks. That gzip support looks like it would help tremendously.

[bglw]

Hi @chrisspen 👋

A fallback to the title element has been implemented in Pagefind v0.9.2 — let me know how that goes!