x/net/html: add Escape/Unescape transformers

[SamWhited]

The golang.org/x/net/html package contains two functions similar to the ones in the html package for escaping and unescaping HTML entities:

func EscapeString(string) string
func UnescapeString(string) string

unfortunately these require loading entire documents into memory and converting them to a string before attempting to escape them.

It would be nice if there were also an implementation of the Transformer (and SpanningTransformer) interface from golang.org/x/text/transform that could perform escaping / unescaping on long byte streams without requiring buffering the entire stream into memory.

This could either be two functions which return transformers:

// Escaper returns a transformer that escapes special characters.
// See EscapeString for more information.
func Escaper() transform.SpanningTransformer

// Unescaper returns a transformer that unescapes special characters.
// See UnescapeString for more information.
func Unescaper() transform.SpanningTransformer

or a Transformer type which contains all the various helper methods (String, Bytes, etc.) that transformer based packages in the text tree have.

/cc @mpvl

[rsc]

/cc @nigeltao

[nigeltao]

Sounds reasonable to me.

Sorry for the late reply. I also don't have much spare time to work on this myself.

[mpvl]

Sounds reasonable.
Most packages in x/text use a wrapper type with convenience methods. I found this works best for most users.

[SamWhited]

@mpvl @nigeltao do have a preferred API? I'd forgotten about this, and have since come up with a workaround for the project I needed it for, but will attempt to find the time to submit a PR.

[mpvl]

Not at the moment, but something along the lines of (all x/text) width.Transformer or cases.Caser makes sense. Using the encoding.Encoding approach seems overkill, but may be worth it if you have a variety of matching (un)escapers.

The trickiness is in the implementation. NewTransformer(FromFunc) in https://godoc.org/github.com/mpvl/textutil could help get you started. It has an example escaping and unescaping implementation. Not as efficient as implementing from scratch, but not too bad either and definitely handy for prototyping.

[SamWhited]

The trickiness is in the implementation. NewTransformer(FromFunc) in https://godoc.org/github.com/mpvl/textutil could help get you started. It has an example escaping and unescaping implementation. Not as efficient as implementing from scratch, but not too bad either and definitely handy for prototyping.

Sounds good; I've actually started implementing this in the past but had trouble convincing the tests to pass, so if I give it another shot maybe I'll start with that. For me the speed wasn't as important as not buffering (very-) large documents into memory. I've done some much simpler escaping and unescaping in my XMPP address package, and that was quite a pain to get right from the ground up, so hopefully this helps.