syntax highlighting issues

[nyurik]

Hi, awesome project, I would really like to use it, but ACE support seems to be partially broken, and might need a bit of love. As an editor, I should always know exactly what each element will be treated as, without knowing HJSON rules by heart. I feel that syntax highlighting should be in line with how JSON itself works in ACE: simply show how each element will be parsed -- structure elements {}[]:, comments, object keys, object string, number, booleans, and null values, and most importantly - errors. While ace seems to have extra support for json parsing, even showing things in a bright red will highlight problems, even if it doesn't cover all of them initially.

• a comment is not shown as such if it has a ":" plus any chars, unless there is a space

{
  //abc:
  #xyz:
}

• a string value with two slashes (e.g. a URL) is shown as a comment, which its not

{
  aa: https://example.com
  bb: 700://comment
}

• in a list, special characters before any letters are shown in a different style, making it harder for novices to tell if there is anything special about them, even though HJson is treating it as simple strings

[
    abc
    $?!@^%^&*(){}[]/=-abc
]

[laktak]

I have to admit that the ACE rules were very hard to write - the whole thing is build on regular expressions that are a pain to debug.

If you want to take a look, the source can be found here

[nyurik]

@taktak I would like to hack on the ACE highlighting rules. Turns out there is a rule tester, which makes the process much easier. Do you have any unit tests for it? It would be good to set up a comprehensive single HJSON test file that has every possible usecase, and ensure that the rules produce what is expected.

Thanks!

[laktak]

@nyurik I'm not familiar with the rule tester but we have a lot of tests in https://github.com/hjson/hjson/tree/master/testCases

These target the parser but we have also used them to check the syntax (although manually as I know of no other highlighters that allow unit tests).

[nyurik]

@laktak ACE has a bunch of tokenization test for many languages - https://github.com/ajaxorg/ace/tree/master/lib/ace/mode/_test - so we can add one for hjson there as well. I guess the .txt file will be a combination of most of those test cases hjson has already defined.

[geekley]

a comment is not shown as such if it has a ":" plus any chars, unless there is a space

The same happens with quoted and unquoted string values.