Parse autolinks for urls and emails.

[hayd]

fixes #9564

cc @one-more-minute

[hayd]

I forgot that this fails due to @ being escaped. It's probably safe to remove that from the chars we escape (although perhaps the mailto link still works?? - I'm not sure).

Unfortunately/interestingly I don't think commonmark implementation actually follows it's spec with regards to scheme-sensitivity. I don't think we're following the spec on what/when to escape chars... similarly the commonmark implementation is also not following the spec there either!

[MikeInnes]

Sorry for taking a ridiculously long time to look at this. Looks good anyway. I don't mind removing @ from the escapes to fix this. The only other thing I'd say to to read until >, [space] or \n – for the sake of efficiency it makes sense to stop reading the stream as early as possible.

[tkelman]

Has this been done in Documenter?

[MichaelHatherly]

Has this been done in Documenter?

Needs to be part of Base.Markdown, so would be nice to get this rebased and finished up (if @hayd has spare time to do it?)

[hayd]

The rebase was over footnotes, seemed to be trivial. Pushed but haven't tested...

I'm yet to action on:

The only other thing I'd say to to read until >, [space] or \n – for the sake of efficiency it makes sense to stop reading the stream as early as possible.

What's a clean way to do that?

[MichaelHatherly]

What's a clean way to do that?

Does Markdown.startswith (with a Regex as the second argument) do what you need?

[mpastell]

I added a suggestion for reading. It would nice to have this merged for 0.6

[mpastell]

Try url = readuntil(stream, '>', match = ' ')

[hayd]

Are we hitting this definition already?

julia/base/markdown/parse/util.jl

Lines 139 to 159 in 550f5d3

	function readuntil(stream::IO, delimiter; newlines = false, match = nothing)
	withstream(stream) do
	buffer = IOBuffer()
	count = 0
	while !eof(stream)
	if startswith(stream, delimiter)
	if count == 0
	return String(take!(buffer))
	else
	count -= 1
	write(buffer, delimiter)
	continue
	end
	end
	char = read(stream, Char)
	char == match && (count += 1)
	!newlines && char == '\n' && break
	write(buffer, char)
	end
	end
	end

I guess I haven't fully grokked what match/count do there, but it looked to me like it was suggesting you'd use < as the match character i.e. matching braces.

[mpastell]

Sorry, I misread it. I think you're right and need to do:

text = readuntil(stream, '>', match = '<')

I originally read it too quickly and thought setting it to ' ' to would stop reading at whitespace.

[mpastell]

I think readuntil returns nothing if the input doesn't have '>'. Try with e.g. Markdown.parse("<link"). Try to add url ≡ nothing && return

[StefanKarpinski]

Bump, would love to have this. Anyone know what's busted on CI here?

[tkelman]

failed doc build on travis and failed tests on appveyor, due to the _is_link function somehow getting a nothing input?

[hayd]

Returns nothing and resets the stream if delim is not found.

So that's a bug, it should be returning false if url is nothing. Perhaps implementation changed somewhere down the line, at any rate will push a fix this evening.

[hayd]

This is now green.

[tkelman]

is taken

[tkelman]

=== nothing

[tkelman]

Remind me to merge this in a day or two if we haven't gotten any other feedback.

[KristofferC]

You should merge this in a day or two if we don't get more feedback.

[tkelman]

I meant more like now.