Possible to ignore words in magic commands?

[ShawnHymel]

We use pyspelling in a GitHub Actions workflow to check our markdown prior to publishing to GitBook. In GitBook, we can embed videos with the following magic command:

{% embed url="https://www.youtube.com/watch?v=0Uga3GVXgMw" %}

I see that the action can ignore markdown links as per rojopolis/spellcheck-github-actions#24 , but it seems to miss magic commands such as this. I get the following output:

Using pyspelling on configuration outlined in >.spellcheck.yaml<
Checking files matching specified outlined in >.spellcheck.yaml<
----------------------------------------------------------------
Misspelled words:
<htmlcontent> edge-impulse-python-sdk/overview.md: html>body>p
--------------------------------------------------------------------------------
GVXgMw
Uga
--------------------------------------------------------------------------------

!!!Spelling check failed!!!
Error: Files in repository contain spelling errors

Is there any way to specify ignore strings between characters/strings, for example, between {% and %}?

[facelessuser]

I imagine it is possible. You can probably create a context filter to ignore them. We do similar things to ignore content in our source code: https://github.com/facelessuser/pyspelling/blob/master/.pyspelling.yml#L116.

[facelessuser]

So, was any of this helpful? The Context filter documentation is here if you need more info: https://facelessuser.github.io/pyspelling/filters/context/.

Moving to discussions.

[facelessuser]

Yes, it includes what I'm looking for, notice the string 6442885e561b84053a93ccac. We have what looks to be some hash. Aspell will break up a numbers from letters, so it then processes ccac.

b'title: "model_metadata.h"\nhidden: false\ncategory: 6442885e561b84053a93ccac '
Context: C++ Inference SDK Library/macros/model_metadatah.md: html>body>p
Misspelled words:
<context> C++ Inference SDK Library/macros/model_metadatah.md: html>body>p
--------------------------------------------------------------------------------
ccac
--------------------------------------------------------------------------------

I'm curious, how is that specified in your Markdown? What source causes that hash to be generated?

[ShawnHymel]

It's generated from the ReadMe system. I have to create a "category" (directory) in their system, then I get the hash ID that I assign to the category key in the header in each page. That hash will be arbitrary, depending on what ReadMe gives me. Is it possible to simply ignore those headers at the top of each page? They are guaranteed to be between --- delimiters. For example, here's what the top of that markdown page looks like:

---
title: "model_metadata.h"
hidden: false
category: 6442885e561b84053a93ccac
---

The following list gives information about the most important `#define` macros found in _model-parameters/model_metadata.h_. Note that not all macros are listed--just the ones you'll probably care about. 

**Important!** _model_metadata.h_ is automatically generated by the Edge Impulse Studio. You should not modify it.

...

[facelessuser]

I feel like what you really need to do is just just remove the meta data from the output with a metadata extension: https://python-markdown.github.io/extensions/meta_data/. Then you don't have to try and filter it out after the fact. But if something like that doesn't work for you , you can probably apply pattern to remove it before converting it to HTML.

[facelessuser]

@ShawnHymel You never posted any actual config, so I have no idea what you are actually doing. I am assuming you are using the markdown extension in Pyspelling, so I will continue answering with this assumption. I tested with the metadata extension and it should remove the frontmatter from your documents eliminating your current complaint.

  pipeline:
  - pyspelling.filters.markdown:
      markdown_extensions:
      - meta

[ShawnHymel]

@facelessuser I posted my .spellcheck.yaml above--was that not the config you were looking for? Either way, adding the meta markdown extension did work. Thank you!

[facelessuser]

I'm sorry, I probably just missed it, but glad you got it working!