Line:column for embedded languages

Question: how to address line:column mismatches for embedded languages?

For example, I have the following file:

---
foo: bar
---

Baz

in FrontMatter format and the following parser, working with fm.body:

let P = require("parsimmon")

let s = `
Baz
`

let lang = P.createLanguage({
  Baz: function (L) {
    return P.string("Baz+").trim(P.optWhitespace)
  },
})

console.log(lang.Baz.tryParse(s))

The error says:

Error: expected 'Baz+' at line 2 column 1, got '...Baz

While the real line, as the user perceives it is 5. The solution of catching and parsing exception messages with .replace(/line (\d+)/, "...") is obviously a last resort hack. Those messages aren’t an API part and can be changed any day.

What should we do instead, ideally? Accept some parser options?

[ivan-kleshnin]

Yeah, I thought of this version but wanted to avoid writing a YAML parser.

Now I see I can create just-an-FM parser like:

P.seq(
  P.string("---")).skip(P.nl),
  P.any.notFollowedBy(P.string("---")).many().tie(),
  P.nl.then(P.string("---")),
)

combined with body parser as you said.

Then parse YAML with a 3-rd party lib after that, on a separate step. That would raise the original problem again, now for the YAML, which would be 1 line shifted from it’s initial position. But then I can spoof it by adding one empty line to the YAML before parsing. Problem solved 😃

[wavebeem]

Like instead of doing:

• Parse front matter, get raw body text
• Parse raw body text with parser based on file extension

You could do like:

let MarkdownParser = P.createLanguage({ ... });
let FrontmatterParser = P.createLanguage({ ... });
let EnhancedMarkdownParser = P.seq(FrontmatterParser.main, MarkdownParser.main);

EnhancedMarkdownParser.tryParse(fileTextContent);

or something like that?