shuttle1987 / gilbert-frontmatter-markdown Goto Github PK

The frontmatter-loader stores metadata values as a list of strings and can't be rendered in the templates the same way as the metadata values from the yaml loader, which are stored as any type.

Take for example this line in our default.html template, which takes the value found in title and inserts it into the title tag:

{% block title %}{{ this.title }}{% endblock %}

For any .yaml extension being rendered the output is <title>Blog</title> with the value in the tag, but for any .md extension the rendered output is empty <title></title>

I added some print lines to see the difference in metadata. In frontmatter_markdown.py:

class FrontmatterMarkdownPage(Page):
    extras: list = ['meta']

    @oneshot
    def content(self):
        extras = self.extras
        if not extras:
            extras = self.site.config.get('content_type', {}).get('MarkdownPage', [])
        md = markdown.Markdown(extensions=extras, output_format='html5')
        processed = md.convert(self.data)
        print('md meta: ', md.Meta)
        return processed

gilbert render

Rendering index.md ...
md meta:  {'title': ['Home'], 'in_menu': ['False']}

I realised this from the python markdown docs

Note that the keys are all lowercase and the values consist of a list of strings where each item is one line for that key.

Where as in yaml.py:

def load_yaml(path: Path):
    with path.open() as fin:
        loader = yaml.Loader(fin)
        meta = loader.get_data()
        # PyYAML Reader greedily consumes chunks from the stream.
        # We must recover any un-consumed data, as well as what's left in the stream.
        if loader.buffer:
            data = loader.buffer[loader.pointer:-1]
        else:
            data = ''
        data += fin.read()
        print('yaml meta: ', meta)
    return data, meta

gilbert render

Found 17 content objects.
yaml meta:  {'title': 'Blog', 'template': 'blog.html'}`

In PyYAML

PyYAML allows you to construct a Python object of any type.

Would be useful to publish this plugin on PyPi

Take for example this schema for a page:

class Page(Templated, Content):
    in_menu: bool = True
    call_to_action_text: str

This type of page data fails validation:

---
title: Home
in_menu: false
---

We give high-performing companies the tools they need to compete in the digital age, from skills development and automation to communications and strategy expertise.

the title and in_menu items both get passed over as strings here

As of #4 if a markdown file has no frontmatter it will be loaded with the meta dictionary being empty. This isn't the same as what happens with the default markdown loader. Perhaps this should be changed?