Package

geotrellis

vectortile

Permalink

package vectortile

This package is experimental. Expect API flux.

Invented by Mapbox, VectorTiles are a combination of the ideas of finite-sized tiles and vector geometries. Mapbox maintains the official implementation spec for VectorTile codecs. The specification is free and open source.

VectorTiles are advantageous over raster tiles in that:

Raw VectorTile data is stored in the protobuf format. Any codec implementing the spec must decode and encode data according to this .proto schema.

What is this package?

geotrellis-vectortile is a high-performance implementation of Version 2.1 of the VectorTile spec. It features:

Using this Package

Modules

Users of this library need only pay attention to geotrellis.vectortile. Any classes in the internal.* submodules are unique to the machinery of VectorTile {de,en}coding, and can be safely ignored.

Types

The central type is the VectorTile class. Its companion object can be used to construct VectorTiles from raw byte data:

import geotrellis.spark.SpatialKey
import geotrellis.spark.tiling.LayoutDefinition
import geotrellis.vector.Extent
import geotrellis.vectortile.VectorTile

val bytes: Array[Byte] = ...  // from some `.mvt` file
val key: SpatialKey = ...  // preknown
val layout: LayoutDefinition = ...  // preknown
val tileExtent: Extent = layout.mapTransform(key)

/* Decode Protobuf bytes. */
val tile: VectorTile = VectorTile.fromBytes(bytes, tileExtent)

/* Encode a VectorTile back into bytes. */
val encodedBytes: Array[Byte] = tile.toBytes

The V* types form a small sum type and are used to represent usually untyped Feature-level metadata. This metadata is equivalent to a JSON Object, where String keys index values of any type. A Scala Map requires more rigidity (for the better), and so we use geotrellis.vectortile.Value to guarantee type safety.

Implementation Assumptions

This particular implementation of the VectorTile spec makes the following assumptions:

Version

2.1

Linear Supertypes
AnyRef, Any
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. vectortile
  2. AnyRef
  3. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. All

Type Members

  1. trait Layer extends Serializable

    Permalink

    A layer, which could contain any number of Features of any Geometry type.

    A layer, which could contain any number of Features of any Geometry type. Here, "Feature" and "Geometry" refer specifically to the GeoTrellis classes of the same names.

    Annotations
    @experimental()
  2. case class LazyLayer(rawLayer: internal.vector_tile.Tile.Layer, tileExtent: Extent) extends Layer with Product with Serializable

    Permalink

    A Layer decoded from Protobuf data.

    A Layer decoded from Protobuf data. All of its Features are decoded lazily, making for very fast extraction of single features/geometries.

    Annotations
    @experimental()
  3. case class StrictLayer(name: String, tileWidth: Int, version: Int, tileExtent: Extent, points: Seq[Feature[Point, Map[String, Value]]], multiPoints: Seq[Feature[MultiPoint, Map[String, Value]]], lines: Seq[Feature[Line, Map[String, Value]]], multiLines: Seq[Feature[MultiLine, Map[String, Value]]], polygons: Seq[Feature[Polygon, Map[String, Value]]], multiPolygons: Seq[Feature[MultiPolygon, Map[String, Value]]]) extends Layer with Product with Serializable

    Permalink

    A Layer crafted through some strict ingest process.

    A Layer crafted through some strict ingest process.

    Annotations
    @experimental()
  4. case class VBool(value: Boolean) extends Value with Product with Serializable

    Permalink

    A wrapper for Boolean to allow all Value subtypes to be stored in the same Map.

  5. case class VDouble(value: Double) extends Value with Product with Serializable

    Permalink

    A wrapper for Double to allow all Value subtypes to be stored in the same Map.

  6. case class VFloat(value: Float) extends Value with Product with Serializable

    Permalink

    A wrapper for Float to allow all Value subtypes to be stored in the same Map.

  7. case class VInt64(value: Long) extends Value with Product with Serializable

    Permalink

    A wrapper for Long to allow all Value subtypes to be stored in the same Map.

  8. case class VSint64(value: Long) extends Value with Product with Serializable

    Permalink

    A wrapper for zig-zag encoded ints to allow all Value subtypes to be stored in the same Map.

  9. case class VString(value: String) extends Value with Product with Serializable

    Permalink

    A wrapper for String to allow all Value subtypes to be stored in the same Map.

  10. case class VWord64(value: Long) extends Value with Product with Serializable

    Permalink

    A wrapper for unsigned, 64-bit ints to allow all Value subtypes to be stored in the same Map.

  11. sealed trait Value extends Serializable

    Permalink

    Feature metadata key/value Maps are completely untyped.

    Feature metadata key/value Maps are completely untyped. All keys and values used by Features across a common parent Layer are stored in that parent. Raw Features themselves only store indices into the parent's key/value lists. So, for an example MultiPoint Feature of fire hydrant locations, its metadata could look like:

    { name: "Hydrants",
      colour: "red",
      model: 5
    }

    That's fine if interpreted as JSON, but bad as Scala, as it doesn't give us a clean Map[String, ConcreteTypeHere]. Furthermore, Features within the same Layer don't have to agree on the Value type for the same key:

    { name: "Stop lights",
      colour: 1,
      model: "ABC-123"
    }

    Nor, actually, do Layers have to agree on key sets for their Features. The sealed trait Value here and its extensions aim to provide some type safety in light of the situation described here.

  12. case class VectorTile(layers: Map[String, Layer], tileExtent: Extent) extends Product with Serializable

    Permalink

    A concrete representation of a VectorTile, as one decoded from Protobuf bytes.

    A concrete representation of a VectorTile, as one decoded from Protobuf bytes. At its simplest, a Tile is just a collection of Layers. We opt to expose each Layer name at the Tile level, as the keys of a Map. This way, if the layer names are known by the user ahead of time, they can search through the Tile quickly.

    import geotrellis.vectortile._
    
    val bytes: Array[Byte] = ...  // from some `.mvt` file
    val key: SpatialKey = ...  // preknown
    val layout: LayoutDefinition = ...  // preknown
    val tileExtent: Extent = layout.mapTransform(key)
    
    val tile: VectorTile = VectorTile.fromBytes(bytes, tileExtent)
    Annotations
    @experimental()

Value Members

  1. object VectorTile extends Serializable

    Permalink
    Annotations
    @experimental()

Inherited from AnyRef

Inherited from Any

Ungrouped