[pkg/ottl] Add StripHTML to OTTL converters

[michalpristas]

Description

The StripHTML Converter removes all HTML tags from the given input and returns only the plain text content.

value is a string. If value is another type, an error is returned.

The returned type is string.

Examples:

• StripHTML("<b>Bold Text</b>") returns "Bold Text"
• StripHTML("<div><p>Paragraph</p><br>Line break</div>") returns "ParagraphLine break"
• StripHTML("<img src='image.jpg' alt='An image'>") returns ""
• StripHTML("Plain text without tags") returns "Plain text without tags"

Link to tracking issue

Related: #31930

Testing

Unit tests and e2e test case added

Documentation

Updated pkt/ottl/ottfuncs/README.md with examples

[evan-bradley]

@michalpristas Thanks for opening this. Just to get a better picture of how this fits into the existing suite of OTTL functions:

• How does this fit into our existing suite of Converters that handle XML? Could the same functionality be achieved by using of of them?
• Could this be named ExtractXMLBody or something similar, or is it specific to HTML?

[michalpristas]

out of available xml converters only ParseSimplifiedXML works in a bit similar way. it returns a map but what it does not handle is tags like <img src="img_girl.jpg" alt="Girl in a jacket" width="500" height="600">

all other xml converters we have require us to know xml structure ahead, this may be problematic.

this converter should be rather simple, remove all tags and keep text only

[evan-bradley]

Thanks for the details. I dug into it and I see now that valid HTML isn't necessarily valid XML.

I checked the following statement against the inputs you provided and it has the same functionality as the proposed function:

replace_pattern(body, "(?U)<.*>", "")

I know this is isn't quite as convenient as just calling StripHTML(body), but I'd like to avoid functions where we have the capability to do it easily in OTTL with existing functionality. Would this work for you?

[michalpristas]

I will play a bit with this to find edge cases and come back here

[michalpristas]

found

• <script>console.log('Hello')</script>: we don't want scripts to be part of the output
• <style>body {color: red;}</style> we don't want styling to be part of the output

what i also imagined is an allowed list of tags, that won't be stripped. this could be theoretically be done with regex as well. but we're getting to ugly regexes.

[evan-bradley]

Thanks for looking into it, agreed that the functionality sounds too complex to be cleanly handled with a regex.

I think my remaining concern then is making sure we can justify adding an additional dependency for this functionality. I haven't encountered any users who are working with HTML in their observability pipelines, what use cases have you seen that would benefit from this function? Similarly, are there any other similar common modifications we would want to do to HTML that the bluemonday library would help with?

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 14 days.

[github-actions]

Closed as inactive. Feel free to reopen if this PR is still being worked on.