GeoArrow WKB reader

.gitignore

@@ -1,2 +1,3 @@
+.vscode
 **/target
 **/Cargo.lock

geozero/Cargo.toml

@@ -12,6 +12,7 @@ keywords = ["geo", "geojson", "gdal", "geos", "postgis"]
 
 [features]
 default = ["with-svg", "with-wkt", "with-geo", "with-geojson"]
+with-arrow = ["arrow2"]
 with-csv = ["csv", "with-wkt"]
 with-svg = []
 with-wkt = ["wkt"]
@@ -43,6 +44,7 @@ postgres-types = { version = "0.2", optional = true }
 bytes = { version = "1.0", optional = true }
 prost = { version = "0.10.0", optional = true }
 wkt = { version = "0.10.0", optional = true }
+arrow2 = { version = "0.12", optional = true, features = ["io_ipc"]}
 
 [dev-dependencies]
 seek_bufread = "1.2"

geozero/src/arrow/geoarrow_reader.rs

@@ -0,0 +1,64 @@
+use crate::error::Result;
+use crate::wkb::wkb_reader::{process_wkb_geom_n, read_wkb_header};
+use crate::{GeomProcessor, GeozeroGeometry};
+use arrow2::array::{BinaryArray, Offset};
+
+impl GeozeroGeometry for BinaryArray<i32> {
+    fn process_geom<P: GeomProcessor>(&self, processor: &mut P) -> Result<()> {
+        process_geoarrow_wkb_geom(self, processor)
+    }
+}
+
+impl GeozeroGeometry for BinaryArray<i64> {
+    fn process_geom<P: GeomProcessor>(&self, processor: &mut P) -> Result<()> {
+        process_geoarrow_wkb_geom(self, processor)
+    }
+}
+
+pub fn process_geoarrow_wkb_geom<T: Offset>(
+    array: &BinaryArray<T>,
+    processor: &mut impl GeomProcessor,
+) -> Result<()> {
+    let array_len = array.len();
+    processor.geometrycollection_begin(array_len, 0)?;
+
+    for i in 0..array_len {
+        let raw = &mut array.value(i);
+        let info = read_wkb_header(raw)?;
+        process_wkb_geom_n(raw, &info, read_wkb_header, i, processor)?;
+    }
+
+    processor.geometrycollection_end(array_len - 1)?;
+
+    Ok(())
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+    use crate::wkt::conversion::ToWkt;
+    use arrow2::io::ipc::read;
+    use std::fs::File;
+
+    #[test]
+    fn multipoly_file() -> arrow2::error::Result<()> {
+        let mut file = File::open("tests/data/countries.arrow")?;
+        let metadata = read::read_file_metadata(&mut file)?;
+        let mut reader = read::FileReader::new(file, metadata, None);
+
+        let columns = reader.next().unwrap()?;
+
+        let array = &columns.arrays()[2];
+        let wkbarr = array.as_any().downcast_ref::<BinaryArray<i32>>().unwrap();
+        let wkt = wkbarr.to_wkt().unwrap();
+        assert_eq!(
+            &wkt[0..100],
+            "GEOMETRYCOLLECTION(MULTIPOLYGON(((-59.572095 -80.040179,-59.865849 -80.549657,-60.159656 -81.000327,"
+        );
+        assert_eq!(
+            &wkt[wkt.len()-100..],
+            "-51.5,-58.55 -51.1,-57.75 -51.55,-58.05 -51.9,-59.4 -52.2,-59.85 -51.85,-60.7 -52.3,-61.2 -51.85))))"
+        );
+        Ok(())
+    }
+}

geozero/src/arrow/mod.rs

@@ -0,0 +1,5 @@
+//! GeoArrow conversions.
+//!
+pub(crate) mod geoarrow_reader;
+
+pub use geoarrow_reader::*;

geozero/src/lib.rs

@@ -41,6 +41,9 @@ pub use geometry_processor::*;
 pub use multiplex::*;
 pub use property_processor::*;
 
+#[cfg(feature = "with-arrow")]
+pub mod arrow;
+
 #[cfg(feature = "with-csv")]
 pub mod csv;
 #[cfg(feature = "with-csv")]

geozero/src/wkb/wkb_reader.rs

@@ -22,7 +22,7 @@ impl GeozeroGeometry for Ewkb {
     }
 }
 
-/// GepPackage WKB reader.
+/// GeoPackage WKB reader.
 pub struct GpkgWkb(pub Vec<u8>);
 
 impl GeozeroGeometry for GpkgWkb {
@@ -63,7 +63,7 @@ pub fn process_wkb_type_geom<R: Read, P: GeomProcessor>(
 }
 
 #[derive(Debug)]
-struct WkbInfo {
+pub(crate) struct WkbInfo {
     endian: scroll::Endian,
     base_type: WKBGeometryType,
     has_z: bool,
@@ -75,7 +75,7 @@ struct WkbInfo {
 }
 
 /// OGC WKB header.
-fn read_wkb_header<R: Read>(raw: &mut R) -> Result<WkbInfo> {
+pub(crate) fn read_wkb_header<R: Read>(raw: &mut R) -> Result<WkbInfo> {
     let byte_order = raw.ioread::<u8>()?;
     let endian = if byte_order == WKBByteOrder::XDR as u8 {
         scroll::BE
@@ -177,7 +177,7 @@ fn read_gpkg_header<R: Read>(raw: &mut R) -> Result<WkbInfo> {
 
 // TODO: Spatialite https://www.gaia-gis.it/gaia-sins/BLOB-Geometry.html
 
-fn process_wkb_geom_n<R: Read, P: GeomProcessor>(
+pub(crate) fn process_wkb_geom_n<R: Read, P: GeomProcessor>(
     raw: &mut R,
     info: &WkbInfo,
     read_header: fn(&mut R) -> Result<WkbInfo>,