Packages: Regular expressions for syntaxes: not work ^ after ?<=

[Kristinita]

I wrote my own small syntax for tables.

_My syntax_: https://gist.github.com/Kristinita/4105b1b48fb3dfc7ba13a0c09f9fb216
_My color scheme for syntax_: https://github.com/Kristinita/SashaSublime/blob/master/SashaSublime.tmTheme
_Example file, when use this syntax_: https://gist.github.com/Kristinita/39286e4165fa74ea54b52aa41b14ef81

In syntax not works pattern, that work correctly on regex.101.com site:

_Pattern_: (?<=^\| )[0-9]{1,2}(?= )
_regex101_: https://regex101.com/r/gR8lV6/1

I want consume digits between | and space only when | is in the beginning of a line. If I don't write a symbol ^, figures will be consume in other places, see https://regex101.com/r/gR8lV6/2:

I do not want to consume digits in other places as in the screen 2. But in Sublime Text pattern

- match: (?<=^\| )[0-9]{1,2}(?= )
  scope: kira.numbers

all digits don't highlighted. Pattern (?<=\| )[0-9]{1,2}(?= ) is work for me in Sublime Text.

Thanks.

[jfcherng]

iirc, ^ means the position of current seek rather than line beginning in ST syntax parser.
If it's true, in that case, maybe maintain an extra context called first-column to tackle with it.

BTW, this post is not official-package-related, isn't it? More appropriate to post this in ST forum?

[jfcherng]

Just read your syntax def. ST matches a rule as early as possible (better performance) rather than try to match all rules and select the longest match.

So, move

  - match: (?<=^\| )[0-9]{1,2}(?= )
    scope: kira.numbers

to the first rule would work in this case.

And your color scheme looks really nice :+1:

[Kristinita]

@jfcherng , thanks, it works! But this is probably a bug, I think, it should not be that in what place there is a rule.

Where should I write, if you have any problems with my syntax?

(I'm sorry, what program you have made your screenshots? What is the effect that you have used?)

Thanks.

[wbond]

The oniguruma regex engine used when lookbehinds are present does not seem to support anchors in lookbehinds: http://stackoverflow.com/questions/3391450/ruby-1-9-regex-lookbehind-assertion-anchors.

Generally you can replace all look-behinds with a context that is entered when you find the pattern you were looking behind for.

For example, create a rule that matches ^\| and enter a new context. In that context, if you match \s+(\d+), highlight is specially. If not (via a blank match or a lookahead that matches a non-space char), pop out of the special context.

You can see this pattern in action at: https://github.com/sublimehq/Packages/blob/master/Ruby/Ruby.sublime-syntax#L267-L281.

[Kristinita]

@wbond , is not Oniguruma problem: see comments of Vince Fitz and K.Kosako — developers of Oniguruma.

Thanks.