page.excerpt behaviour changed between 4.0.1 and 4.1.0

[seth-reeser]

Hello, version 4.1.0 introduced an unexpected change in behaviour for page.excerpt per pull request 7642.

I was using page.excerpt as a meta description entry:

<meta name="description" content="{% if page.excerpt %}{{ page.excerpt | strip_html | strip_newlines | truncate: 160 }}{% else %}{{ site.description }}{% endif %}">

In an example of an excerpt not being defined for a page, this would have evaluated to the site.description, but now I'm getting markup generated for me that breaks because of HTML being included.

I would recommend the new implementation approach things similarly by stripping html, stripping newlines, and even maybe truncating.

[ashmaroli]

Is your site public @seth-reeser ? If yes, please provide us with a link to it.

[seth-reeser]

@ashmaroli, unfortunately I cannot share the website publicly, but here's another example. The below snippet loops through each Jekyll page and generates a search.json file for consumption by a custom search mechanism:

Given a page, in this example, a "Terms of Use" page that has no page.excerpt defined in it's markup; I would expect page.excerpt to be blank, but now it is not, nor a value that I want. I would expect this to be an issue for others as well.

Before (Jekyll 4.0.1):

After (Jekyll 4.1.0):

[ashmaroli]

The reason I asked for a link to your website is so that I could see how exactly {{ page.excerpt | strip_html }} resulting in rendering HTML tags.
From your comments, I'm guessing that you do not have posts or collections in your site.

As a workaround to get to the old behavior, you can disable excerpts for all pages by setting front matter excerpt_separator: "". If you too many pages that you can't manually modify, you could use front-matter-defaults instead:

# _config.yml

defaults:
  - scope:
      path: ""
      type: pages
    values:
      excerpt_separator: ""

[seth-reeser]

Here's before and after for the original meta description example. Keep in mind for this example, that there was no page.excerpt set for the page in question. So the fallback was site.description in this example:

<meta name="description" content="{% if page.excerpt %}{{ page.excerpt | strip_html | strip_newlines | truncate: 160 }}{% else %}{{ site.description }}{% endif %}">

Before (Jekyll 4.0.1) site.description fallback:

After (Jekyll 4.1.0) new Jekyll logic now creates a page.excerpt from page content if not defined:

Thank you for the excerpt_separator trick and agree that I can programmatically solve this issue - but I could foresee this be a breaking change for many others.