[mapserver-users] WFS GetFeature sometimes returns 'missing' even when there are matching features

[Gertjan van Oosten]

Hi all,

It's been a month since I posted this, but so far I have not seen any
activity towards fixing it. I know it's summer holidays and all, but is
there something I could or should to to get this fixed? We're currently
running a locally patched version of Mapserver, but we'd prefer not to.

Kind regards,
Gertjan


As quoted from Gertjan van Oosten <gertjan at West.NL>:
> Hi,
> 
> A WFS GetFeature request like this one:
> 
>   <wfs:GetFeature xmlns:wfs="http://www.opengis.net/wfs" service="WFS" version="1.1.0" xsi:schemaLocation="http://www.opengis.net/wfs http://schemas.opengis.net/wfs/1.1.0/wfs.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
>     <wfs:Query typeName="feature:Results_wfs">
>       <wfs:PropertyName>FEATURE_ID</wfs:PropertyName>
>     </wfs:Query>
>   </wfs:GetFeature>
> 
> sometimes returns 'missing' like so:
> 
>   <?xml version='1.0' encoding="ISO-8859-1" ?>
>   <wfs:FeatureCollection
>      xmlns:ms="http://mapserver.gis.umn.edu/mapserver"
>      xmlns:gml="http://www.opengis.net/gml"
>      xmlns:wfs="http://www.opengis.net/wfs"
>      xmlns:ogc="http://www.opengis.net/ogc"
>      xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
>      xsi:schemaLocation="http://mapserver.gis.umn.edu/mapserver http://localhost/mapserv?SERVICE=WFS&VERSION=1.1.0&REQUEST=DescribeFeatureType&TYPENAME=feature:Results_wfs&OUTPUTFORMAT=text/xml;%20subtype=gml/3.1.1  http://www.opengis.net/wfs http://schemas.opengis.net/wfs/1.1.0/wfs.xsd">
>      <gml:boundedBy>
> 	<gml:Null>missing</gml:Null>
>      </gml:boundedBy>
>   </wfs:FeatureCollection>
> 
> even when there is a matching feature. This happens with Mapserver 6.4.1
> and Oracle 11.2. I've traced it down to the searchrect in msQueryByRect()
> inside mapquery.c being "empty":
> 
>   {minx=-1, maxx=-1, miny=-1, maxy=-1}
> 
> This is caused by the way the extent is computed:
> 
>    msOracleSpatialLayerGetExtent. Using this Sql to retrieve the extent:
>    SELECT SDO_GEOM.SDO_CONVEXHULL(link_geo, 0.001000) AS GEOM from (...)
> 
> The convex hull is NULL for some features. As the documentation says:
> 
>   http://docs.oracle.com/cd/E11882_01/appdev.112/e11830/sdo_objgeom.htm#SPATL1114
>   "The function returns a null value if geom1 is of point type, has
>   fewer than three points or vertices, or consists of multiple points
>   all in a straight line."
> 
> I indeed have features like that, so in this case no features are
> returned at all. That is obviously not right. If I add these three lines
> it works correctly and my features are returned:
> 
> 
> --- mapserver-6.4.1/mapquery.c.distrib	2014-01-02 13:41:49.000000000 +0100
> +++ mapserver-6.4.1/mapquery.c	2014-07-17 13:58:36.000000000 +0200
> @@ -1055,30 +1055,33 @@
>  
>      searchrect = map->query.rect;
>      if(lp->tolerance > 0) {
>        layer_tolerance = lp->tolerance;
>  
>        if(lp->toleranceunits == MS_PIXELS)
>          tolerance = layer_tolerance * msAdjustExtent(&(map->extent), map->width, map->height);
>        else
>          tolerance = layer_tolerance * (msInchesPerUnit(lp->toleranceunits,0)/msInchesPerUnit(map->units,0));
>  
>        searchrect.minx -= tolerance;
>        searchrect.maxx += tolerance;
>        searchrect.miny -= tolerance;
>        searchrect.maxy += tolerance;
>      }
> +    if (searchrect.minx == -1 && searchrect.maxx == -1 && searchrect.miny == -1 && searchrect.maxy == -1) {
> +      searchrect = map->extent;
> +    }
>  
>      msRectToPolygon(searchrect, &searchshape);
>  
>      /* Raster layers are handled specially. */
>      if( lp->type == MS_LAYER_RASTER ) {
>        if( msRasterQueryByRect( map, lp, searchrect ) == MS_FAILURE)
>          return MS_FAILURE;
>  
>        continue;
>      }
>  
>      /* Paging could have been disabled before */
>      paging = msLayerGetPaging(lp);
>      msLayerClose(lp); /* reset */
>      status = msLayerOpen(lp);
> 
> 
> I.e if the query rectangle can't be computed for some reason, fall back
> to the map extent. I don't know if this is the best fix, but it is the
> simplest that I could find. Is this the right way to do it?


-- 
-- Gertjan van Oosten, Principal Consultant, West Consulting B.V.
-- gertjan at West.NL     +31 15 2191 600       www.west.nl