How to parse and serialize custom tags

[miffels]

Hey,

I have a situation where I am transforming a YAML document that has some custom tags in it. I would like to preserve the tags, but I haven't found a way to write them out the same way I received them. So in essence, the flow is YAML.parse --> modify object --> YAML.stringify. Note that I have read https://eemeli.org/yaml/v1/#custom-data-types.

YAML document:

markdown_extensions:
  - pymdownx.emoji:
      emoji_index: !!python/name:materialx.emoji.twemoji
      emoji_generator: !!python/name:materialx.emoji.to_svg

If I fail to supply any custom_tags in the YAML options for YAML.parse, YAML.stringify turns this into

markdown_extensions:
  - pymdownx.emoji:
      emoji_index: ""
      emoji_generator: ""

I tried to supply custom types like so

  const twemojiType = {
    identify: () => false,
    tag: 'tag:yaml.org,2002:python/name:materialx.emoji.twemoji',
    resolve: () => '!!python/name:materialx.emoji.twemoji',
    stringify: () => '!!python/name:materialx.emoji.twemoji',
  };
  const toSvgType = {
    identify: () => false,
    tag: 'tag:yaml.org,2002:python/name:materialx.emoji.to_svg',
    resolve: () => '!!python/name:materialx.emoji.to_svg',
    stringify: () => '!!python/name:materialx.emoji.to_svg',
  };

  const yamlOpts = {
    customTags: [
      twemojiType,
      toSvgType,
    ],
  };
  const config = YAML.parse(yamlString, yamlOpts);
  // ...
  YAML.stringify(config);

But then the tag is just rendered as a string:

markdown_extensions:
  - pymdownx.emoji:
      emoji_index: "!!python/name:materialx.emoji.twemoji"
      emoji_generator: "!!python/name:materialx.emoji.to_svg"

Any ideas?

[eemeli]

Have you tried working with the Document interface that yaml provides? That's kind of designed for the sort of use case you appear to have.

const src = `
markdown_extensions:
  - pymdownx.emoji:
      emoji_index: !!python/name:materialx.emoji.twemoji
      emoji_generator: !!python/name:materialx.emoji.to_svg
`

const doc = YAML.parseDocument(src)
doc.getIn(['markdown_extensions', 0, 'pymdownx.emoji', 'emoji_index'], true)
> Scalar {
    value: '',
    range: [ 97, 97, 97 ],
    source: '',
    type: 'PLAIN',
    tag: 'tag:yaml.org,2002:python/name:materialx.emoji.twemoji'
  }

doc.toString()
> 'markdown_extensions:\n' +
  '  - pymdownx.emoji:\n' +
  '      emoji_index: !!python/name:materialx.emoji.twemoji ""\n' +
  '      emoji_generator: !!python/name:materialx.emoji.to_svg ""\n'

That's using yaml@2, but the behaviour is pretty much the same for yaml@1.

[miffels]

Thank you so much! Had to rewrite a bunch but this API fixed the issue for me (note (to self) that the trailing "" appears to be semantically equivalent to the source, which doesn't have it at all)".

[eemeli]

If you'd like to fix the node value, then you'll need to use a custom tag for it; the default behaviour is to parse a scalar value as a string, and the trailing "" is there as the default representation of an empty string.