Issue 4628 - Added metadata versioning for Apache DataSketches

NOTICES

@@ -461,6 +461,13 @@ testing_logger (https://github.com/brucechapman/rust_testing_logger)
 
 - The 3-Clause BSD License
 
+cargo_metadata (https://github.com/oli-obk/cargo_metadata)
+
+- MIT License
+
+serde_json (https://github.com/serde-rs/json)
+
+- Apache 2 License
 
 
 The build pipeline uses the following GitHub Actions from the marketplace.

rust/Cargo.toml

@@ -31,6 +31,7 @@ aws-config = { version = "1.6.1" } # Credential loading
 aws-credential-types = { version = "1.2.2" } # Credential provider types
 aws-types = { version = "1.3.5" } # for Region
 bytes = { version = "1.10.1" } # Byte buffer for S3 uploading
+cargo_metadata = { version = "0.19.2" } # Access cargo metadata programmtically
 chrono = { version = "0.4.40" } # Log helper
 clap = { version = "4.5.35" } # Cmd line args processing
 color-eyre = { version = "0.6.3" } # Colourised version of `anyhow`
@@ -48,6 +49,7 @@ object_store = { version = "0.11.2" } # Remote cloud storage access
 openssl = { version = "0.10.72" } # Use vendored feature to build from source for cross compilation
 owo-colors = { version = "4.2.0" } # Colourised output
 rand = { version = "0.9.0" }
+serde_json = { version = "1.0.140" } # Process metadata JSON
 tempfile = { version = "3.19.1" } # Create temporary files
 test-log = { version = "0.2.17" }
 testing_logger = { version = "0.1.1" } # Assert on logger output in tests
@@ -56,6 +58,9 @@ tokio = { version = "1.44.2", features = ["full"] } # Async runtime
 tokio-test = { version = "0.4.4" } # Doc tests
 url = { version = "2.5.4" } # URL processing for S3
 
+[workspace.metadata.datasketches]
+git_repository_tag = "5.2.0"
+
 [profile.release]
 incremental = true
 lto = false

rust/rust_sketch/Cargo.toml

@@ -30,6 +30,8 @@ cxx = { workspace = true }
 [build-dependencies]
 cxx-build = { workspace = true }
 git2 = { workspace = true }
+cargo_metadata = { workspace = true }
+serde_json = { workspace = true }
 
 [target.'cfg(target_os = "macos")'.build-dependencies]
 # Mac often won't have openssl library installed in place easy discoverable, if at all

rust/rust_sketch/build.rs

@@ -13,6 +13,11 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */
+use cargo_metadata::MetadataCommand;
+use serde_json::Value;
+
+pub const DEFAULT_TAG: &str = "5.2.0";
+
 fn main() {
     println!("cargo:rerun-if-changed=src/include");
     println!("cargo:rerun-if-changed=src/quantiles.rs");
@@ -26,8 +31,8 @@ fn main() {
     let url = std::env::var("RUST_SKETCH_DATASKETCH_URL")
         .unwrap_or(String::from("https://github.com/apache/datasketches-cpp"));
 
-    // look to see if the repo tag has been overridden
-    let tag = std::env::var("RUST_SKETCH_DATASKETCH_TAG").unwrap_or(String::from("5.2.0"));
+    // look up tag to use
+    let tag = get_repo_tag();
 
     // try to open repository in case it already exists
     if git2::Repository::open(path.join("datasketches-cpp")).is_err() {
@@ -55,3 +60,55 @@ fn main() {
         ])
         .compile("rust_sketch");
 }
+
+/// Retrieve git repository tag for Apache `DataSketches` library to retrieve.
+///
+/// Attempt to determine the git repository tag to retrieve for the Apache `DataSketches`
+/// library. Checks in order:
+/// 1. `RUST_SKETCH_DATASKETCH_TAG` environment variable.
+/// 2. Cargo workspace metadata `workspace.metadata.dataketches` for a `git_repository_tag` key.
+/// 3. Default tag in [`DEFAULT_TAG`].
+fn get_repo_tag() -> String {
+    // 1. Check environment variable
+    if let Ok(env_tag) = std::env::var("RUST_SKETCH_DATASKETCH_TAG") {
+        println!(
+            "cargo:warning=DataSketches repository tag taken from environment variable {env_tag}"
+        );
+        return env_tag;
+    }
+
+    // 2. Check cargo metadata
+    let mut command = MetadataCommand::new();
+    let Ok(metadata) = command.no_deps().exec() else {
+        println!(
+            "cargo:warning=Couldn't execute cargo metadata command. Using default {DEFAULT_TAG} tag."
+        );
+        return String::from(DEFAULT_TAG);
+    };
+
+    let Value::Object(workspace_data) = metadata.workspace_metadata else {
+        println!(
+            "cargo:warning=Couldn't find workspace metadata. Using default {DEFAULT_TAG} tag."
+        );
+        return String::from(DEFAULT_TAG);
+    };
+
+    let Some(Value::Object(sketch_data)) = workspace_data.get("datasketches") else {
+        println!(
+            "cargo:warning=Couldn't find \"datasketches\" metadata section. Using default {DEFAULT_TAG} tag."
+        );
+        return String::from(DEFAULT_TAG);
+    };
+
+    if let Some(Value::String(repo_tag)) = sketch_data.get("git_repository_tag") {
+        println!(
+            "cargo:warning=DataSketches repository tag taken from cargo metadata variable {repo_tag}"
+        );
+        repo_tag.clone()
+    } else {
+        println!(
+            "cargo:warning=Couldn't find \"git_repository_tag\" metadata key. Using default {DEFAULT_TAG} tag."
+        );
+        String::from(DEFAULT_TAG)
+    }
+}