Merge branch 'apache:main' into main

Author: pwrliang

.github/workflows/r.yml

@@ -91,21 +91,23 @@ jobs:
         if: matrix.config.os == 'macos-latest'
         run: brew install geos
 
+      # Dynamically set a short target directory on Windows to bypass 260-char MinGW limits,
+      # while keeping paths standard on Mac/Linux.
+      - name: Configure Target Directory
+        shell: bash
+        run: |
+          if [ "$RUNNER_OS" == "Windows" ]; then
+            echo "SEDONADB_TARGET_DIR=D:/a/t" >> $GITHUB_ENV
+            echo "CACHE_WORKSPACE=. -> D:/a/t" >> $GITHUB_ENV
+          else
+            echo "SEDONADB_TARGET_DIR=$(pwd)/r/sedonadb/src/rust/target" >> $GITHUB_ENV
+            echo "CACHE_WORKSPACE=. -> r/sedonadb/src/rust/target" >> $GITHUB_ENV
+          fi
+
       - uses: Swatinem/rust-cache@v2
         with:
-          # Update this key to force a new cache
           prefix-key: "r-v4"
-          # The R package uses its own target directory
-          workspaces: ". -> r/sedonadb/src/rust/target"
-
-      # Explicitly specify for Windows, which otherwise chooses a shorter version
-      # to work around a path length limitation when building from longer starting paths.
-      # This allows us to use a common cache configuration for Windows, MacOS, and Linux
-      - name: Set R/Rust target directory
-        if: matrix.config.os == 'windows-latest'
-        run: |
-          echo "SEDONADB_TARGET_DIR=$(pwd -W)/r/sedonadb/src/rust/target" >> $GITHUB_ENV
-        shell: bash
+          workspaces: ${{ env.CACHE_WORKSPACE }}
 
       - uses: r-lib/actions/setup-r-dependencies@v2
         with:

Cargo.lock

@@ -40,6 +40,7 @@ members = [
     "rust/sedona-raster-functions",
     "rust/sedona-schema",
     "rust/sedona-spatial-join",
+    "rust/sedona-spatial-join-geography",
     "rust/sedona-spatial-join-gpu",
     "rust/sedona-query-planner",
     "rust/sedona-testing",
@@ -91,6 +92,7 @@ datafusion-execution = { version = "51.0.0", default-features = false }
 datafusion-expr = { version = "51.0.0" }
 datafusion-ffi = {  version = "51.0.0" }
 datafusion-functions-nested = { version = "51.0.0" }
+datafusion-optimizer = { version = "51.0.0" }
 datafusion-physical-expr = { version = "51.0.0" }
 datafusion-physical-plan = { version = "51.0.0" }
 datafusion-pruning = { version = "51.0.0" }
@@ -109,7 +111,7 @@ glam = "0.32.0"
 libmimalloc-sys = { version = "0.1", default-features = false }
 log = "^0.4"
 libloading = "0.9"
-lru = "0.16"
+lru = "0.17"
 mimalloc = { version = "0.1", default-features = false }
 num-traits = { version = "0.2", default-features = false, features = ["libm"] }
 object_store = { version = "0.12.4", default-features = false }
@@ -126,7 +128,7 @@ serde_json = { version = "1" }
 serde_with = { version = "3" }
 tempfile = { version = "3"}
 thiserror = { version = "2" }
-tokio = { version = "1.51", features = ["macros", "rt", "sync"] }
+tokio = { version = "1.52", features = ["macros", "rt", "sync"] }
 url = "2.5.7"
 wkb = "0.9.2"
 wkt = "0.14.0"
@@ -149,6 +151,7 @@ sedona-raster-functions = { version = "0.4.0", path = "rust/sedona-raster-functi
 sedona-schema = { version = "0.4.0", path = "rust/sedona-schema" }
 sedona-spatial-join = { version = "0.4.0", path = "rust/sedona-spatial-join" }
 sedona-spatial-join-gpu = { version = "0.4.0", path = "rust/sedona-spatial-join-gpu" }
+sedona-spatial-join-geography = { version = "0.4.0", path = "rust/sedona-spatial-join-geography" }
 sedona-query-planner = { version = "0.4.0", path = "rust/sedona-query-planner" }
 sedona-testing = { version = "0.4.0", path = "rust/sedona-testing" }
 

Cargo.toml

@@ -27,7 +27,16 @@ fn main() {
     // MACOS_DEPLOYMENT_TARGET isn't passed through properly but this can probably
     // be solved by setting the environment variable in CI before building a portable
     // target like a Python package.
-    let dst = cmake::Config::new(".").build();
+    let mut cmake_config = cmake::Config::new(".");
+
+    // Use RelWithDebInfo if building release with debug symbols
+    if std::env::var("PROFILE").unwrap_or_default() == "release"
+        && std::env::var("CARGO_PROFILE_RELEASE_DEBUG").is_ok()
+    {
+        cmake_config.define("CMAKE_BUILD_TYPE", "RelWithDebInfo");
+    }
+
+    let dst = cmake_config.build();
 
     // Link the libraries that are easy to enumerate by hand and whose location
     // we control in CMakeLists.txt. s2geography_c and s2geography are always built

c/sedona-s2geography/build.rs

@@ -38,6 +38,7 @@ use crate::utils::S2GeogCError;
 /// required in a thread safe way (although it may result in faster evaluation
 /// to force a build early in a situation where it is known that a lot of
 /// evaluations are about to happen).
+#[derive(Debug)]
 pub struct Geography<'a> {
     ptr: *mut S2Geog,
     _marker: PhantomData<&'a [u8]>,
@@ -125,9 +126,11 @@ impl GeographyFactory {
         Ok(geog)
     }
 
-    /// Internal wrappers around the actual init. This is the function that should be used
-    /// in the event of looping over WKBs from the same arrow array.
-    fn init_from_wkb(&mut self, wkb: &[u8], geog: &mut Geography) -> Result<(), S2GeogCError> {
+    /// Initialize an existing Geography with a new WKB buffer
+    ///
+    /// This is the function that should be used in the event of looping over
+    /// WKBs from the same arrow array.
+    pub fn init_from_wkb(&mut self, wkb: &[u8], geog: &mut Geography) -> Result<(), S2GeogCError> {
         unsafe {
             s2geog_call!(S2GeogFactoryInitFromWkbNonOwning(
                 self.ptr,

c/sedona-s2geography/s2geography

@@ -51,6 +51,11 @@ impl RectBounder {
         unsafe { s2geog_call!(S2GeogRectBounderBound(self.ptr, geog.as_ptr())) }
     }
 
+    /// Perform the minimum expansion required to satisfy a distance expansion
+    pub fn expand_by_distance(&mut self, distance_meters: f64) {
+        unsafe { S2GeogRectBounderExpandByDistance(self.ptr, distance_meters) }
+    }
+
     /// Check if the bounder is empty (no geometries or only empty geometries
     /// have been added)
     pub fn is_empty(&self) -> bool {
@@ -139,6 +144,13 @@ mod tests {
         assert!(hi_lng >= 10.0);
         assert!(hi_lat >= 20.0);
 
+        bounder.expand_by_distance(100_000.0); // 100km
+        let expanded = bounder.finish().unwrap().unwrap();
+        assert!(expanded.0 < lo_lng);
+        assert!(expanded.1 < lo_lat);
+        assert!(expanded.2 > hi_lng);
+        assert!(expanded.3 > hi_lat);
+
         bounder.clear();
         assert!(bounder.is_empty());
     }

c/sedona-s2geography/src/geography.rs

@@ -97,6 +97,10 @@ unsafe extern "C" {
         geog: *const S2Geog,
         err: *mut S2GeogError,
     ) -> S2GeogErrorCode;
+    pub fn S2GeogRectBounderExpandByDistance(
+        rect_bounder: *mut S2GeogRectBounder,
+        distance_meters: f64,
+    );
     pub fn S2GeogRectBounderIsEmpty(rect_bounder: *mut S2GeogRectBounder) -> u8;
     pub fn S2GeogRectBounderFinish(
         rect_bounder: *mut S2GeogRectBounder,