Scala Highlighter is broken

[guest]

The scala highlighter in pygments is quite useless.

I attach a patch that fixes most of the obvious problems and even does some feeble attempts at semicolon inference (i.e. nicely formatted code should be highlighted acceptably, but it should be easy to write code that breaks the highlighting). It produces acceptable highlighting on the scalaz code I tested it on.

Florian Hars <florian (at) hars.de>

[thatch]

Hi Florian,

Thanks for the submission. Can you provide example files that didn't work well under the old lexer?

[guest]

I attaced a file showing some of the problems:

• Neither `interface` nor interface is highlighted as in identifier, although both are

• val isn't highlighted as a keyword

• Nested commtents are broken

• Multiline strings are broken

• ⌘ is highlighted as an error, although it is a valid name

• foo_+ is highlighted as an identifier followed by an operator, although it is a single itentifier

• foo_⌬⌬ is totally borked.

More prolems with the old lexer:

• The deprecated type names boolean|byte|char|double|float|int|long|short|void are highligted different from their prefered uppercase versions

• The class names String|Int|Array|HashMap are highlighted different from all other class names

Thing the new lexer doesn't solve:

• Any name can designate almost everything (a type, a class, a method, a variable, an operator), you can only make the decision after type checking. So I make some common sense assumptions:
  • an identifier that starts with an uppercase letter is a Name.class
  • an identifier that starts with any other type of letter is a Name
  • a backticked itentifier is a Name
  • an identifier consisting of opchars is an Operator

This can be as off as you want, but captures conventional practice.

[guest]

Oh, by the way, your CC: seems not to work.

[gbrandl]

Thanks, applied the patch in changeset [703:aeb0be1650c5].