unknown geom type: reading (mixed) Unknown fails

[mdsumner]

vect() can't read polygon TAB, because (see link below)

the mapinfo driver deliberately reports an unknown geometry type for polygon files

terra::vect("/vsicurl/https://github.com/OSGeo/gdal/raw/master/autotest/ogr/data/mitab/poly_indexed.tab")
Error: [vect] cannot read this geometry type: Unknown (any)

## other interfaces work
##sf::read_sf("/vsicurl/https://github.com/OSGeo/gdal/raw/master/autotest/ogr/data/mitab/poly_indexed.tab")

https://lists.osgeo.org/pipermail/gdal-dev/2008-June/017418.html

Will PR if I can, found this exploring use of SQLITE spatial functions in query() which seem to also return mixed type.

dsn <- system.file("ex/lux.shp", package="terra")
layer <- "lux"
dialect <- "SQLITE"
geomname <- "ConvexHull(GEOMETRY) AS GEOMETRY"  
vrt <- glue::glue('<OGRVRTDataSource>
        <OGRVRTLayer name="polygons_as_hulls">
          <SrcDataSource>{dsn}</SrcDataSource>
          <SrcSQL dialect="{dialect}">SELECT NAME_1, {geomname} FROM {layer}</SrcSQL>
    </OGRVRTLayer>
</OGRVRTDataSource>')

terra::vect(vrt)
# Error: [vect] cannot read this geometry type: Unknown (any)

[mdsumner]

possibly related to #1084 because

terra::vect("/vsicurl/https://filesamples.com/samples/code/kml/sample3.kml")
Error: [vect] cannot read this geometry type: Unknown (any)
In addition: Warning message:
[vect] Z coordinates ignored 

[rhijmans]

The bigger problem in the context of "unknown" or "collection" types is that a SpatVector is assumed to have a single data type. Allowing multiple types requires a major overhaul. I will start by adding a method that can read these into a SpatVectorCollection; and then we can take it from there.

[rhijmans]

There now is a svc<character> method that can read MapInfo polygons

library(terra)
v <- vect(system.file("ex/lux.shp", package="terra"))
v$POP <- v$POP * 1
f <-  "mi_pols.tab"
writeVector(v,f)

x <- svc(f)
x[1]
# class       : SpatVector 
# geometry    : polygons 
# dimensions  : 12, 6  (geometries, attributes)
# extent      : 5.74414, 6.528252, 49.44781, 50.18162  (xmin, xmax, ymin, ymax)
# coord. ref. : +proj=longlat +datum=WGS84 +no_defs 
# names       :  ID_1   NAME_1  ID_2   NAME_2  AREA       POP
# type        : <num>    <chr> <num>    <chr> <num>     <num>
# values      :     1 Diekirch     1 Clervaux   312 1.808e+04
#                   1 Diekirch     2 Diekirch   218 3.254e+04
#                   1 Diekirch     3  Redange   259 1.866e+04

[mdsumner]

good to know ty, I'll PR something for the documentation 🙏

[rhijmans]

You can now also read polygon TAB with terra::vect

library(terra)
v <- vect(system.file("ex/lux.shp", package="terra"))
v$POP <- v$POP * 1
f <-  "mi_pols.tab"
writeVector(v,f, overwrite=TRUE)
x <- vect(f)
x
# class       : SpatVector 
# geometry    : polygons 
# dimensions  : 12, 6  (geometries, attributes)
# extent      : 5.74414, 6.528252, 49.44781, 50.18162  (xmin, xmax, ymin, ymax)
# source      : mi_pols.tab
# coord. ref. : +proj=longlat +datum=WGS84 +no_defs 
# names       :  ID_1   NAME_1  ID_2   NAME_2  AREA       POP
# type        : <num>    <chr> <num>    <chr> <num>     <num>
# values      :     1 Diekirch     1 Clervaux   312 1.808e+04
#                   1 Diekirch     2 Diekirch   218 3.254e+04
#                   1 Diekirch     3  Redange   259 1.866e+04

If there are multiple geometries in a file, the highest level is returned (polygons > lines > points), with a warning suggesting to use svc to get all geometry types.

[mdsumner]

awesome, thanks so much, confirmed with all examples above