Wrong span reported for HTML with encoded characters

[RobAley]

Check for existing issues

• Completed

Environment

vale version 2.29.0, Ubuntu 22.04, direct download

Describe the bug / provide steps to reproduce it

Hi,

(Thanks for a very useful tool!)

When linting HTML with encoded entities, the wrong Span is given for errors (apparently using their decoded equivalents).

E.g. For the following HTML :

<p>You know it&rsquo;s great software</p>

Vale reports :

{
  "test.html": [
    {
      "Action": {
        "Name": "",
        "Params": null
      },
      "Span": [
        13,
        16
      ],
      "Check": "write-good.E-Prime",
      "Description": "",
      "Link": "",
      "Message": "Try to avoid using 'it's'.",
      "Severity": "suggestion",
      "Match": "it's",
      "Line": 1
    }
  ]
}

The Span is 13 - 16 (4 characters) corresponding to "it's", rather than 13 - 22 corresponding to "it’s" .

I would expect the span to be reported in relation to the input text.

Many thanks.