Feature Request: ignoreSelector Array

[GeorgeWL]

I think this lib is excellent, but I realised an issue with it that I'd like to see as part of the options object:

There are times where you want to transform all of the contents of a Selector, apart from say one or two internal selectors that are only used on the dynamic version of the website (say something like a element that only shows on hover) without moving them out into another element just so they don't get parsed by HtmlToText

I propose the addition of an array of selectors which defaults to empty.

so something like:

let options = {
        hideLinkHrefIfSameAsText: true,
        ignoreImage: true,
        uppercaseHeadings: false,
        preserveNewLines:true,
        tables: '.content',
        baseElement: ['table.content','table.footer']
        ignoreSelectors:['.hidden','.hover','#hover'] 
//selectors using the same method as baseElement does
    }

which then makes any html elements with those selectors ignored by the parser.

[Niek]

There's a PR here but it's severely outdated: #186

The best would be if tags would allow CSS selectors so you can format: 'skip' them. Is that hard to add @KillyMXI?

[KillyMXI]

@Niek yeah, I'm thinking how to expand the system in this direction. Don't want to rush it and end up with something slow or hard to evolve.

As a workaround it might be possible to make a custom formatter to handle the filtering for you. I thought about including something like this along with skip but couldn't finalize the design so left it out for now.

[mattcobb]

+1 for this feature. Our specific use case is we would like to skip tags of a given class, for example <div class="reflist"> or <ol class="references"> from wikipedia articles.

[KillyMXI]

I can see this being the most demanded feature.
I can't promise it will be the next thing I'll ship but it is on the table and I'll give it more attention.

[mattcobb]

Other than PR 186 being outdated, what issues does it have?

[KillyMXI]

I think I should write a contribution guideline to avoid PRs that are unlikely to be merged.

#186:

• does a couple of things outside of it's own defined scope (minor issue in a grand scheme of things);
• has no opened issue with agreed design proposal;
• provided design is not something I'm happy with, and there are enough breaking changes happen naturally already to consciously publish things that I know I'm going to break soon.

What exactly makes me unhappy about it:

• I'd like to avoid adding ignoredSelectors array and solve it differently. I have a couple of approaches in mind with a bunch of considerations to check;
• it adds it's own limited selectors checking logic. There are a couple of places already in the code with similar but different selectors logic. I want to unify that. And again, there are some connected questions I'll have to explore;
• I think it does way too many checks (performance hit) and I can do better.

So it's really not in a state where I could write it down and delegate. A lot of exploratory work.

[KillyMXI]

Short update:
Solving this issue among with several related ones will be the topic of upcoming version (8.0.0).
I've decided on the design, and done a good portion of work for html-to-text.
Then I realized I can't use a package I was aiming for and have to develop a new one. This is half-way done too.
No ETA though. I don't expect much trouble completing both parts. But most disruptions currently come from elsewhere.

[KillyMXI]

I've pushed new code into a separate branch - selectors
Tracking issue: #228

I'd be grateful if some of you can test this version on real tasks and provide your feedback before release (in a week probably).
(You can install the package directly from GitHub branch. Instruction is in the tracking issue.)

[KillyMXI]

Version 8 is now live.
Arbitrary selectors can be ignored as follows:

{
  selectors: [
    { selector: 'foo.hidden', format: 'skip' }
  ]
}