Punctiation regexp is incomplete

[rlidwka]

I came across a discrepancy between cmark and commonmark.js output:

$ echo '**。**话' | ./cmark/build/src/cmark 
<p>**。**话</p>
$ echo '**。**话' | ./commonmark.js/bin/commonmark 
<p><strong>。</strong>话</p>

So, according to spec v26,

A punctuation character is an ASCII punctuation character or anything in the Unicode classes Pc, Pd, Pe, Pf, Pi, Po, or Ps.

Character "。" or U+3002 belongs to a class Punctuation, Other [Po] (see http://www.fileformat.info/info/unicode/char/3002/index.htm), but it's not included here:

https://github.com/jgm/commonmark.js/blob/3587c91c62128e54a236648ff1ac4a1ad1cd5ad8/lib/inlines.js#L41

For the reference, here's the regexp from unicode-8.0.0 package (we're using that in markdown-it), which includes this character (and appears to be a lot larger):

https://github.com/mathiasbynens/unicode-8.0.0/blob/master/General_Category/Punctuation/regex.js

[rlidwka]

PS: that said, the example above is a fragment from chinese text here:

**无论你是在打网球比赛，在打牌还是在选股票，有经验的人总会猜测对手会怎么做。**这样的话

I don't know the language, but they don't seem to use whitespace at all. Maybe commonmark rules don't quite work with chinese.

Still, commonmark.js works differently from whatever is written in the spec, so I opened bugreport here.

[jgm]

Thanks!

[puzrin]

@jgm i'd recommend to require() files from unicode-* package, instead of hardcoding. Or you can use our proxy https://github.com/markdown-it/uc.micro to not update anything when unicode-10.+ released.

[cinty8b]

It seems the problem is still there.

[jgm]

markdown-it does not use commonmark.js. cinty8b <notifications@github.com> writes:

…

It seems the problem is still there.  -- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: #108 (comment)

[puzrin]

markdown-it does not use commonmark.js.

That's correct, but it uses existing spec for tests. Current sample from commit pass without issues.

@cinty8b, i'd recommend you to make reports more detailed. Screenshots are very inconvenient, and almost useless:

1. Try https://spec.commonmark.org/dingus/, to understand if your issue is related to commonmark.js/spec or not. Use permalinks (not screenshots) for samples.
2. If you are absolutely sure, that official dingus is ok, try https://markdown-it.github.io/ and report to it's tracker, not here (with repmalink too).

May be that's new issue.

@jgm IMHO existing fix with hardcoded regexp is not human-readable and not maintainable - nobody knows data source and how actual is it. If you don't like to use external packages - it worth add comments with link to original & unicode version number.

[cinty8b]

Sorry for the screenshot.
I'm just a user, not a developer, so I am not familiar with these terminologies and conventional bug report process. I appologize for any inconvenience from that and for the wrong place to report.
I tracked the problem from Emphasis cannot be recognized when it has no space with following words #285, and @rlidwka started this issue from there.

Tried https://spec.commonmark.org/dingus/ , it works the same as markdown-it.
Here is the sample
The fox. .fox 狐狸。 。狐狸 parts are not italic as expected.

[puzrin]

@cinty8b thanks for details. As far as i understand, your case is not asian language specific, because is reproducible in english too. So, it's not related to this issue. IMHO worth create a new one to not be lost.

@jgm could you advice better place where to forward this?

[jgm]

This behavior accords with the spec. (So, it is "expected" in that sense.) brown*fox.*jumps Here the second `*` delimiter is not "right-flanking." Notice that this gives you emphasis: brown*fox.*.jumps because now the second `*` is both right- and left- flanking. If you really need the `fox.` to be emphasized here, you could try inserting a zero-width space: the brown*fox.*&#x200b;jumps If you think this is a flaw in the spec, you could bring it up on talk.commonmark.org. But be aware that there are always tradeoffs; the question is whether there's an improvement that could be made without messing up other things we currently get right.

[cinty8b]

@jgm @puzrin Thanks for your explanation and possible workaround.
I think there's something more to say.

In English there are always spaces and punctuations to seperate words, so it is rare that a part of a continuous string has to be emphasized or italic with an ending or starting punctuation, like brown*fox.*jumps. In other words, it is always word as a whole to be emphasized, not just part of it.

But it's different in Chinese. We seldom use spaces in sentences. Punctuations do almost all the seperating work in a paragraph. As a result, it's common in Chinese that I want to emphasize a sentence with its period together (。 is period in Chinese), like 棕色狐狸。**黄色狐狸。**黑色狐狸。 But commonmark could not emphasize the sentence in the middle.
Of course a zero-width space will solve the problem, but it is a little complicated for new users to remember and to understand why a special symble has to be inserted. And I think it causes another problem: The zero-width space is visually invisible. I may look for 黄色狐狸。黑色狐狸。 in the text, but 黄色狐狸。​黑色狐狸。 does not match. It can be confusing.

I hope I make it clear.

[puzrin]

@cinty8b AFAIK, there are some known spec issues with asian languages (no spaces), without good resolution. With high probability this one was discussed at commonmark forum. Try to post there. Probably, it worth to kick such topic again.