[antlr-interest] Range generation in C++ mode

Ric Klaren klaren at cs.utwente.nl
Mon Sep 6 14:04:03 PDT 2004

On Mon, Sep 06, 2004 at 08:45:03PM +0200, Martin Probst wrote:
> On other terms I now have a problem with my grammar under windows. I'm
> trying to parse UTF-8 from C++ by just ignoring the multi-byte nature of
> UTF-8 chars (e.g. I have special letter tokens composed of multiple
> single chars).
> If I compile & run the code on windows I get an exception that a debug
> assertion has failed in isctype.c (?) if I put UTF-8 chars into my input
> file. Is this a known bug or problem?

It sounds vaguely familiar. Is this using MSVC? What does the assertion
say? It could be some locale related setting that may send isprint
somewhere down a wrong path (if this comes from isprint)? A backtrace of
the crash could also help.

I also got a hack that has better UTF8 support in C++. Using a modified
InputBuffer and another CharScanner superclass. The InputBuffer decodes the
UTF8 and the CharScanner superclass can repackage things in the form you
like (currently UTF8 stored in std::string)


    ---- Ric Klaren ----- j.klaren at utwente.nl ----- +31 53 4893755  ----
     "Evil will always triumph, because Good is dumb." ---  Spaceballs

Yahoo! Groups Links

<*> To visit your group on the web, go to:

<*> To unsubscribe from this group, send an email to:
    antlr-interest-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:

More information about the antlr-interest mailing list