[PyKDE] QScintilla Lexers

[Phil Thompson]

> Hello. I am in the middle of writing a text editor in Python using
> PyQt/QScintilla. I would like my editor to include syntax highlighting for
> HTML, XML, XSLT, etc. but I have found that QScintilla does not support
> these
> very well (or at all).
>
> There is a lexer for HTML, but the syntax highlighting seems very
> inaccurate.
> For example, it correctly colourises good syntax like <a
> href="index.html">,
> but then does the same for html-nonsense like <href a="index.html">. There
> is
> also no support for CSS.
>
> I thought I might be able to fix some of this myself by sub-classing
> QextScintillaLexerHTML. However, I soon found that several of the
> functions
> which have to be re-implemented return const char * rather than a QString.
> This means none of the lexers can be usefully sub-classed in Python
> because
> the QextScintilla class expects const char * return types when it
> sets/calls
> the lexers internally.
>
> My next step was to try creating my own lexer classes in Python using the
> low-level API. I had some success at this, but at the expense of features
> like auto-indentation which are only provided by QextScintilla when it's
> using one of its own lexers. If only I could sub-class QextScintillaLexer
> in
> Python!
>
> The next thing was to try learning a bit of C++ and write my own lexers.
> But
> then I had a look at the source code for the HTML lexer and thought again
> -
> all that stuff to handle the scripting languages made my head spin... My
> last
> hope was to try looking at something simpler, like the source code for the
> XML lexer - except that there is no real XML lexer because it is just an
> alias for the HTML lexer!
>
> The real point of all this is to introduce a couple proposals for
> improving
> the lexers in QScintilla. The first one is obvious: make
> QextScintillaLexer
> Python-friendly so it can be fully sub-classed. This will open up all the
> "hidden" languages (like CSS) to Python developers who want to go ahead
> and
> create their own lexer classes (or sub-classes).

It's not clear from your explanation exactly what the problem is. Can you
give some specific examples?

> And the second proposal
> is:
> include a purely markup-orientated XML lexer (i.e. one not based on the
> HTML
> lexer). This would then make it easy to add lexer classes for XHTML, XSLT,
> XSD, etc. by sub-classing from the XML lexer.

QScintilla is a port of Scintilla, not an enhancement of it. If there are
issues with how well lexers work (or if you want a new lexer) then these
should be fixed in Scintilla itself and they will then appear in
QScintilla.

Phil