Re: occ -p -s

On Thu, 2004-08-05 at 20:49, Stefan Seefeld wrote:
> Gilles J. Seguin wrote:
> > Allo Stefan,
> > 
> > How can I have the equivalent of cli
> > $ occ -p -s tata.cc
> 
> erm, you can't. That's precisely one of my goals: to
> provide similar functionality via python (synopsis) scripting.
> 
> the 'occ' executable does quite a number of things:
> 
> 1) running a preprocessor (already possible in synopsis)
> 2) parsing an input file (already possible in synopsis)

That is the one, I need a command to feed to QMtest.

presently I am try to modify thing

diff -u -r1.2.2.2 Parser.cc
--- Parser.cc   28 Jul 2004 05:44:33 -0000      1.2.2.2
+++ Parser.cc   6 Aug 2004 13:54:27 -0000
@@ -3673,10 +3673,35 @@
                                                PtreeUtil::List(new
Leaf(op), e));
                break;
            }
+       case CONSTCAST :
+       case DYNCAST :
+       case REINTCAST :
+       case STATCAST :
+           Token res, tk1,tk2,tk3,tk4;
+           Ptree *castexp, *tname;
+           t2 = lex->GetToken(res);
+           if(lex->GetToken(tk1) != '<')
+                return false;
+           if(rTypeName(tname))
+                return false;
+           if(lex->GetToken(tk2) != '>')
+                return false;
+           if(lex->GetToken(tk3) != '(')
+                return false;
+           if(rExpression(castexp))
+                return false;
+           if(lex->GetToken(tk4) != ')')
+                return false;
+           exp = new PtreeCxxCastExpr( new LeafReserved(res),
+                                       PtreeUtil::List(new Leaf(tk1),
+                                           tname, new Leaf(tk2), new
Leaf(tk3),+                                           castexp, new
Leaf(tk3)));
+           break;
        default :
            return true;
        }
     }
+    return true;
 }
  
 /*


> 3) manipulating the generated parse tree based on special tokens
>     embedded into the input file (not possible in synopsis)
> 4) generating code from this modified parse tree (not possible in synopsis)
> 5) running a compiler on the code (not done in synopsis)
> 6) rerun itself with plugins loaded that had been compiled by a previous run
>     (not possible in synopsis)
> 
> 
> Point 3 is the interesting bit here, and there are multiple
> ways to do it. Point 4 is trivially done given the existing API.
> Point 5 is also quite simple (even portably) in python.
> I don't know what to do about point 6...I believe this can be
> done in a simpler way with a different approach.
> 
> Again, this is all on my todo list. I have ideas how to get there,
> and I welcome everybody who wants to get involved.