SP converter

[Ironholds]

For each class of sp WKT object, there should be a converter that turns them into WKT.

It should:

1. Accept either an sp object, or a List of sp objects (hopefully - not actually checked this is possible yet);
2. Return a character vector of those objects represented as WKT strings.

[sckott]

yeah, just checked, looks like you can have sp classes in a list

and we'd only want the sp Spatial classes supported. There are some additional classes in rgeos, but that requires GEOS lib, which is what we want to avoid dealing with

[Ironholds]

Gotcha! So the classes with a matching WKT name? (e.g. Polygon?)

[sckott]

yeah, so

• sp::SpatialPolygons
• sp::SpatialPolygonsDataFrame (this is the above just with data.frame attached I think

sp doesn't have a multipolygon equiv. i don't think (equiv. i guess is just many polygons within one of the above 2 classes)

would be nice to be able to do more sp classes if easy enough - that way people that only need sp to WKT conversion can avoid GEOS lib - i don't have a sense for how hard this is though

[Ironholds]

Hmn, so output format for this will be...interesting: a SpatialPolygons object can contain multiple polygons, and thence multiple WKT objects. So the options are:

1. Support converting a single SP at a time (into a nice tidy vector), or;
2. Support converting a list of SPs at a time (into a less-tidy list)

Got a preference?

[sckott]

I imagine

• if a single SpatialPolygons object has > 1 polygon, then convert to either MULTIPOLYGON WKT or many POLYGON WKT's
• apply same logic to a list of SpatialPolygons objects

??

[Ironholds]

Makes sense. So then we'd end up with....

sp_convert(sp, group = FALSE) which produces a list of vectors of WKT POLYGONS and group=TRUE producing a mixture of polygons and multipolygons. Sound good?

[sckott]

Yes

[Ironholds]

Can do!

[Ironholds]

So when we have a multiple-nested set of polygons within an SP - e.g. the last one here:

str(SpP@polygons)
List of 3
 $ :Formal class 'Polygons' [package "sp"] with 5 slots
  .. ..@ Polygons :List of 1
  .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. ..@ labpt  : num [1:2] 2.7 3.55
  .. .. .. .. ..@ area   : num 5.5
  .. .. .. .. ..@ hole   : logi FALSE
  .. .. .. .. ..@ ringDir: int 1
  .. .. .. .. ..@ coords : num [1:5, 1:2] 2 1 4 4 2 2 4 5 3 2
  .. ..@ plotOrder: int 1
  .. ..@ labpt    : num [1:2] 2.7 3.55
  .. ..@ ID       : chr "s1"
  .. ..@ area     : num 5.5
 $ :Formal class 'Polygons' [package "sp"] with 5 slots
  .. ..@ Polygons :List of 1
  .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. ..@ labpt  : num [1:2] 3.67 2.33
  .. .. .. .. ..@ area   : num 1.5
  .. .. .. .. ..@ hole   : logi FALSE
  .. .. .. .. ..@ ringDir: int 1
  .. .. .. .. ..@ coords : num [1:4, 1:2] 5 2 4 5 2 2 3 2
  .. ..@ plotOrder: int 1
  .. ..@ labpt    : num [1:2] 3.67 2.33
  .. ..@ ID       : chr "s2"
  .. ..@ area     : num 1.5
 $ :Formal class 'Polygons' [package "sp"] with 5 slots
  .. ..@ Polygons :List of 2
  .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. ..@ labpt  : num [1:2] 6.13 3.93
  .. .. .. .. ..@ area   : num 10
  .. .. .. .. ..@ hole   : logi FALSE
  .. .. .. .. ..@ ringDir: int 1
  .. .. .. .. ..@ coords : num [1:5, 1:2] 4 10 5 4 4 5 5 2 3 5
  .. .. ..$ :Formal class 'Polygon' [package "sp"] with 5 slots
  .. .. .. .. ..@ labpt  : num [1:2] 5.5 3.5
  .. .. .. .. ..@ area   : num 1
  .. .. .. .. ..@ hole   : logi TRUE
  .. .. .. .. ..@ ringDir: int -1
  .. .. .. .. ..@ coords : num [1:5, 1:2] 5 5 6 6 5 4 3 3 4 4

Is it acceptable to collapse the entire object into a list of coords matrices, or is there special handling needed? At that point we're talking polygons inside polygons inside polygons inside a list ;p

[sckott]

At least for use cases for packages we maintain, I don't think we'll need to worry about this level of complexity - I'll leave up to you to decide

[Ironholds]

Collapsing it is!