Markdown metadata format

Is there a standard or convention for embedding metadata in a Markdown formatted post, such as the publication date or post author for conditional rendering by the renderer?

Looks like this Yaml metadata format might be it.

There are all kinds of strategies, e.g. an accompanying file mypost.meta.edn, but I'm hoping to keep it all in one file.

There are two common formats that look very similar but are actually different in some very specific ways. And a third which is very different.

YAML Front Matter

The Jekyll static site generator popularized YAML front matter which is deliminated by YAML section markers. Yes, the dashes are actually part of the YAML syntax. And the metadata is defined using any valid YAML syntax. Here is an example from the Jekyll docs:

> --- > layout: post > title: Blogging Like a Hacker > ---

Note that YAML front matter is not parsed by the Markdown parser, but is removed prior to parsing by Jekyll (or whatever tool you're using) and could actually be used to request a different parser than the default Markdown parser for that page (I don't recall if Jekyll does that, but I have seen some tools which do).

MultiMarkdown Metadata

The older and simpler MultiMarkdown Metadata is actually incorporated into a few Markdown parsers. While it has more recently been updated to optionally support YAML deliminators, traditionally, the metadata ends and the Markdown document begins upon the first blank line (if the first line was blank, then no metadata). And while the syntax looks very similar to YAML, only key-value pairs are supported with no implied types. Here is an example from the MultiMarkdown docs:

> Title: A Sample MultiMarkdown Document
> Author: Fletcher T. Penney
> Date: February 9, 2011
> Comment: This is a comment intended to demonstrate
> metadata that spans multiple lines, yet
> is treated as a single value.
> CSS: http://example.com/standard.css

The MultiMarkdown parser includes a bunch of additional options which are unique to that parser, but the key-value metadata is used across multiple parsers. Unfortunately, I have never seen any two which behaved exactly the same. Without the Markdown rules defining such a format everyone has done their own slightly different interpretation resulting in a lot of variety.

The one thing that is more common is the support for YAML deliminators and basic key-value definitions.

Pandoc Title Block

For completeness there is also the Pandoc Title Block. If has a very different syntax and is not easily confused with the other two. To my knowledge, it is only supported by Pandoc (if enabled), and it only supports three types of data: title, author, and date. Here is an example from the Pandoc documentation:

% title
% author(s) (separated by semicolons)
% date

Note that Pandoc Title Blocks are one of two style supported by Pandoc. Pandoc also supports YAML Metadata as described above.

A workaround use standard syntax and compatible with all other viewers.

I was also looking for a way to add application specific metadata to markdown files while make sure the existing viewers such as vscode and github page will ignore added metadata. Also to use extended markdown syntax is not a good idea because I want to make sure my files can be rendered correctly on different viewers.

So here is my solution: at beginning of markdown file, use following syntax to add metadata:

[metadata:author]:- "daveying"
[metadata:tags]:- "markdown metadata"

This is the standard syntax for link references, and they will not be rendered while your application can extract these data out.

The - after : is just a placeholder for url, I don't use url as value because you cannot have space in urls, but I have scenarios require array values.

Most Markdown renderers seem to support this YAML format for metadata at the top of the file:

---
layout: post
published-on: 1 January 2000
title: Blogging Like a Boss
---

Content goes here.

The most consistent form of metadata that I've found for Markdown is actually HTML meta tags, since most Markdown interpreters recognize HTML tags and will not render meta tags, meaning that metadata can be stored in a way that will not show up in rendered HTML.

<title>Hello World</title>
<meta name="description" content="The quick brown fox jumped over the lazy dog.">
<meta name="author" content="John Smith">

## Heading
Markdown content begins here

You can try this in something like GitHub Gist or StackEdit.

Correct.

Use the yaml front matter key-value syntax — like MultiMarkdown supports — but (ab)use the official markdown URL syntax to add your metadata.

… my workaround looks like this:

---
[//]: # (Title: My Awesome Title)
[//]: # (Author: Alan Smithee)
[//]: # (Date: 2018-04-27)
[//]: # (Comment: This is my awesome comment. Oh yah.) 
[//]: # (Tags: #foo, #bar)  
[//]: # (CSS: https://path-to-css)  
---

Put this block at the top of your .md doc, with no blank line between the top of the doc and the first ---.

Your fake yaml won't be included when you render to HTML, etc. … it only appears in the .md.

You can also use this technique for adding comments in the body of a markdown doc.

This is not a standard way, but works with Markdown Extra.

I wanted something that worked in the parser, but also didn't leave any clutter when I browse the files on Bitbucket where I store the files.

So I use Abbreviations from the Markdown Extra syntax.

*[blog-date]: 2018-04-27
*[blog-tags]: foo,bar

then I parse them with regexp:

 ^\*\[blog-date\]:\s*(.+)\s*$

As long as I don't write the exact keywords in the text, they leave no trace. So use some prefix obscure enough to hide them.

I haven't seen this mentioned elsewhere here or in various blogs discussing the subject, but in a project for my personal website, I've decided to use a simple JSON object at the top of each markdown file to store metadata. It's a little more cumbersome to type compared to some of the more textual formats above, but it's super easy to parse. Basically I just do a regex such as ^\s*({.*?})\s*(.*)$ (with the s option on to treat . as \n) to capture the json and markdown content, then parse the json with the language's standard method. It allows pretty easily for arbitrary meta fields.