On Wed, Feb 19, 2020 at 11:37:03PM +0100, Julia Lawall wrote:
> 
> 
> --- Please note the new email address ---
> 
> 
> On Wed, 19 Feb 2020, Ville Syrjälä wrote:
> 
> > Hi all,
> >
> > I was trying to do the following transformation:
> > @@
> > identifier M;
> > expression E;
> > @@
> > - struct foo M = E;
> > + struct foo M;
> > + copy_struct(&M, &E);
> >
> > but without inserting the function call in the middle of the
> > declarations and instead pushing it past them.
> >
> > This is one attempt:
> > @decl@
> > identifier M;
> > expression E;
> > @@
> > - struct foo M = E;
> > + struct foo M;
> >
> > @copy@
> > identifier decl.M;
> > expression decl.E;
> > declaration D;
> > statement S;
> > @@
> > struct foo M;
> > ...
> > D
> > + copy_struct(&M, &E);
> > S
> >
> > The copy rule fails to match anything when I have != 1
> > declarations after the struct. So the ... doesn't seem
> > to eat the extra declarations for some reason.
> 
> @copy@
> identifier decl.M;
> expression decl.E;
> declaration D;
> statement S;
> @@
> struct foo M;
> ... when any
> D
> + copy_struct(&M, &E);
> S
> 
> Otherwise, ... doesn't match code that includes what is before or after
> it, ie struct foo M or D.

This one still inserts the code after all declarations in nested
blocks. Also misses cases where there are no extra declarations.
So not quite what I need.

> 
> Aternatively:
> 
> @copy@
> identifier decl.M;
> expression decl.E;
> declaration D;
> statement S,S1;
> @@
> struct foo M;
> ... when != S1
> + copy_struct(&M, &E);
> S

This on the other hand seems to work very nicely. Thanks.

Earlier I tried to play around with some '... when !=' stuff,
but this double statement trick did not occur to me.

-- 
Ville Syrjälä
Intel
_______________________________________________
Cocci mailing list
[email protected]
https://systeme.lip6.fr/mailman/listinfo/cocci

Reply via email to