[Proj] [AXIS] pj_init parsing issues

[Frank Warmerdam]

RICHARD Didier wrote:
> Hi devs,
> 
> I am preparing a ticket for PROJ4JS for axis handling and looking at the
> code of pj_init.c when parsing +axis= (trunk version) :
> 
> 
>     /* axis orientation */
>     if( (pj_param(ctx, start,"saxis").s) != NULL )
>     {
>         static const char *axis_legal = "ewnsud";
>         const char *axis_arg = pj_param(ctx, start,"saxis").s;
> [1]     if( strlen(axis_arg) != 3 )
>         {
>             pj_ctx_set_errno( ctx, PJD_ERR_AXIS );
>             goto bum_call;
>         }
> 
>         if( strchr( axis_legal, axis_arg[0] ) == NULL
>             || strchr( axis_legal, axis_arg[1] ) == NULL
> [2]         || (axis_arg[2] && strchr( axis_legal, axis_arg[1] ) == NULL))
>         {
>             pj_ctx_set_errno( ctx, PJD_ERR_AXIS );
>             goto bum_call;
>         }
> 
>         /* it would be nice to validate we don't have on axis repeated */
>         strcpy( PIN->axis, axis_arg );
>     }
> 
> 
> it seems to me that :
> 
> [1] implies axis_arg length is 3 : is there any problem with 2D only CRS
> (although pj_transform() and pj_adjust_axis() check that) ?

Didier,

I agree that it is desirable to allow 2D axis definitions in which case
the first two would need to be in [ewns] and the third would be assumed
to be "u" when needed.

> [2] if length is 3 then axis_arg[2] is defined and the comparizon should
> be done versus axis_arg[2] and not axis_arg[1] ?

Yes, this does look like a bug.

Could you file a ticket on this?  Feel free to also fix it if you are
comfortable doing so.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent