Metadata

Meta

The Meta method provides metadata handling capabilities in GeometryBasics. Similarly to remove the metadata and keep only the geometry, use metafree, and for vice versa i.e., remove the geometry and keep the metadata use meta.

Syntax

meta(geometry, meta::NamedTuple)
meta(geometry; meta...)

metafree(meta-geometry)
meta(meta-geometry)

Examples

julia> using GeometryBasics
julia> p1 = Point(2.2, 3.6)2-element Point2{Float64} with indices SOneTo(2):
 2.2
 3.6
julia> poi = meta(p1, city="Abuja", rainfall=1221.2)2-element PointMeta{2, Float64, Point2{Float64}, (:city, :rainfall), Tuple{String, Float64}} with indices SOneTo(2):
 2.2
 3.6

Metadata is stored in a NamedTuple and can be retrieved as such:

julia> meta(poi)(city = "Abuja", rainfall = 1221.2)

Specific metadata attributes can be directly retrieved:

julia> poi.rainfall1221.2
julia> metafree(poi)2-element Point2{Float64} with indices SOneTo(2):
 2.2
 3.6

Metatypes are predefined for geometries:

julia> multipoi = MultiPointMeta([p1], city="Abuja", rainfall=1221.2)1-element MultiPointMeta{Point2{Float64}, MultiPoint{2, Float64, Point2{Float64}, Vector{Point2{Float64}}}, (:city, :rainfall), Tuple{String, Float64}}:
 [2.2, 3.6]

(In the above example we have also used a geometry-specific meta method.)

julia> GeometryBasics.MetaType(Polygon)PolygonMeta
julia> GeometryBasics.MetaType(Mesh)MeshMeta

The metageometry objects are infact composed of the original geometry types.

julia> GeometryBasics.MetaFree(PolygonMeta)Polygon
julia> GeometryBasics.MetaFree(MeshMeta)Mesh

MetaT

In GeometryBasics we can have tabular layout for a collection of meta-geometries by putting them into a StructArray that extends the Tables.jl API.

In practice it's not necessary for the geometry or metadata types to be consistent. For example, a geojson format can have heterogeneous geometries. Hence, such cases require automatic widening of the geometry data types to the most appropriate type. The MetaT method works around the fact that, a collection of geometries and metadata of different types can be represented tabularly whilst widening to the appropriate type.

Syntax

MetaT(geometry, meta::NamedTuple)
MetaT(geometry; meta...)

Returns a MetaT that holds a geometry and its metadata MetaT acts the same as Meta method. The difference lies in the fact that it is designed to handle geometries and metadata of different/heterogeneous types.

For example, while a Point MetaGeometry is a PointMeta, the MetaT representation is MetaT{Point}.

Examples

julia> MetaT(Point(1, 2), city = "Mumbai")MetaT{Point2{Int64}, (:city,), Tuple{String}}([1, 2], (city = "Mumbai",))

For a tabular representation, an iterable of MetaT types can be passed on to a meta_table method.

Syntax

meta_table(iter)

Examples

Create an array of 2 linestrings:

julia> ls = [LineString([Point(i, i+1), Point(i-1,i+5)]) for i in 1:2];
julia> coordinates.(ls)2-element Vector{Vector{Point2{Int64}}}:
 [[1, 2], [0, 6]]
 [[2, 3], [1, 7]]

Create a multi-linestring:

julia> mls = MultiLineString(ls);
julia> coordinates.(mls)2-element Vector{Vector{Point2{Int64}}}:
 [[1, 2], [0, 6]]
 [[2, 3], [1, 7]]

Create a polygon:

julia> poly = Polygon(Point{2, Int}[(40, 40), (20, 45), (45, 30), (40, 40)]);
julia> coordinates(poly)4-element Vector{Point2{Int64}}:
 [40, 40]
 [20, 45]
 [45, 30]
 [40, 40]

Put all geometries into an array:

julia> geom = [ls..., mls, poly];

Generate some random metadata:

julia> prop = [(country_states = "India$(i)", rainfall = (i*9)/2) for i in 1:4]4-element Vector{NamedTuple{(:country_states, :rainfall), Tuple{String, Float64}}}:
 (country_states = "India1", rainfall = 4.5)
 (country_states = "India2", rainfall = 9.0)
 (country_states = "India3", rainfall = 13.5)
 (country_states = "India4", rainfall = 18.0)
julia> feat = [MetaT(i, j) for (i,j) = zip(geom, prop)]; # create an array of MetaT

We can now generate a StructArray / Table with meta_table:

julia> sa = meta_table(feat);

The data can be accessed through sa.main and the metadata through sa.country_states and sa.rainfall. Here we print only the type names of the data items for brevity:

julia> [nameof.(typeof.(sa.main)) sa.country_states sa.rainfall]4×3 Matrix{Any}:
 :LineString       "India1"   4.5
 :LineString       "India2"   9.0
 :MultiLineString  "India3"  13.5
 :Polygon          "India4"  18.0

Disadvantages

The MetaT is pretty generic in terms of geometry types, it's not subtype to geometries. eg : A MetaT{Point, NamedTuple{Names, Types}} is not subtyped to AbstractPoint like a PointMeta is.
This might cause problems on using MetaT with other constructors/methods inside or even outside GeometryBasics methods designed to work with the main Meta types.