feat: directly call git to get substrait submodule version

.gitignore

@@ -4,11 +4,11 @@
 Cargo.lock
 
 # Generated by build.rs (substrait submodule version information)
-src/version.in
+gen/
 
 .idea
 
 package.json
 package-lock.json
 node_modules/
-.vscode/
+.vscode/

Cargo.toml

@@ -15,8 +15,8 @@ build = "build.rs"
 include = [
     "LICENSE",
     "build.rs",
+    "gen/",
     "src/**/*.rs",
-    "src/version.in",
     "substrait/LICENSE",
     "substrait/README.md",
     "substrait/extensions/**/*.yaml",
@@ -44,7 +44,6 @@ serde_json = "1.0.114"
 thiserror = { version = "1.0.57", optional = true }
 
 [build-dependencies]
-git2 = { version = "0.18.2", default-features = false }
 heck = "0.5.0"
 pbjson-build = { version = "0.6.2", optional = true }
 prettyplease = "0.2.4"

build.rs

@@ -7,60 +7,68 @@ use std::{
     fs::{self, File},
     io::Write,
     path::{Path, PathBuf},
+    process::Command,
 };
 use walkdir::{DirEntry, WalkDir};
 
+const SUBMODULE_ROOT: &str = "substrait";
 const PROTO_ROOT: &str = "substrait/proto";
 const TEXT_ROOT: &str = "substrait/text";
+const GEN_ROOT: &str = "gen";
 
 /// Add Substrait version information to the build
 fn substrait_version() -> Result<(), Box<dyn Error>> {
-    use git2::{DescribeFormatOptions, DescribeOptions, Repository};
+    let gen_dir = Path::new(GEN_ROOT);
+    fs::create_dir_all(gen_dir)?;
 
-    let substrait_version_file =
-        PathBuf::from(env::var("CARGO_MANIFEST_DIR")?).join("src/version.in");
+    let version_in_file = gen_dir.join("version.in");
 
     // Rerun if the Substrait submodule changed (to allow setting `dirty`)
     println!(
         "cargo:rerun-if-changed={}",
         Path::new("substrait").display()
     );
 
-    // Get the version from the submodule
-    match Repository::open("substrait") {
-        Ok(repo) => {
-            // Rerun if the Substrait submodule HEAD changed (when there is a submodule)
-            println!(
-                "cargo:rerun-if-changed={}",
-                Path::new(".git/modules/substrait/HEAD").display()
-            );
-
-            // Get describe output
-            let mut describe_options = DescribeOptions::default();
-            describe_options.describe_tags();
-            let mut describe_format_options = DescribeFormatOptions::default();
-            describe_format_options.always_use_long_format(true);
-            describe_format_options.dirty_suffix("-dirty");
-            let git_describe = repo
-                .describe(&describe_options)?
-                .format(Some(&describe_format_options))?;
-
-            let mut split = git_describe.split('-');
-            let git_version = split.next().unwrap_or_default();
-            let git_depth = split.next().unwrap_or_default();
-            let git_dirty = git_describe.ends_with("dirty");
-            let git_hash = repo.head()?.peel_to_commit()?.id().to_string();
-            let semver::Version {
-                major,
-                minor,
-                patch,
-                ..
-            } = semver::Version::parse(git_version.trim_start_matches('v'))?;
-
-            fs::write(
-                substrait_version_file,
-                format!(
-                    r#"// SPDX-License-Identifier: Apache-2.0
+    // Check if there is a submodule. This file is not included in the packaged crate.
+    if Path::new(SUBMODULE_ROOT).join(".git").exists() {
+        // Rerun if the Substrait submodule HEAD changed (when there is a submodule)
+        println!(
+            "cargo:rerun-if-changed={}",
+            Path::new(".git/modules/substrait/HEAD").display()
+        );
+
+        // Get the version of the submodule by directly calling `git describe`.
+        let version = String::from_utf8(
+            Command::new("git")
+                .current_dir(SUBMODULE_ROOT)
+                .arg("describe")
+                .arg("--tags")
+                .arg("--long")
+                .arg("--dirty=-dirty")
+                .arg("--abbrev=40")
+                .output()?
+                .stdout,
+        )?;
+
+        // Extract the parts.
+        let mut split = version.split('-');
+        let git_version = split.next().unwrap_or_default();
+        let git_depth = split.next().unwrap_or_default();
+        let git_hash = split.next().unwrap_or_default().trim_end();
+        let git_dirty = version.ends_with("dirty");
+        let version = semver::Version::parse(git_version.trim_start_matches('v'))?;
+
+        let &semver::Version {
+            major,
+            minor,
+            patch,
+            ..
+        } = &version;
+
+        fs::write(
+            version_in_file,
+            format!(
+                r#"// SPDX-License-Identifier: Apache-2.0
 
 // Note that this file is auto-generated and auto-synced using `build.rs`. It is
 // included in `version.rs`.
@@ -79,27 +87,23 @@ pub const SUBSTRAIT_GIT_SHA: &str = "{git_hash}";
 
 /// The `git describe` output of the Substrait submodule used to build this
 /// crate
-pub const SUBSTRAIT_GIT_DESCRIBE: &str = "{git_describe}";
+pub const SUBSTRAIT_GIT_DESCRIBE: &str = "{version}";
 
-/// The amount of commits between the latest tag and this version of the
-/// Substrait module used to build this crate
+/// The amount of commits between the latest tag and the version of the
+/// Substrait submodule used to build this crate
 pub const SUBSTRAIT_GIT_DEPTH: u32 = {git_depth};
 
 /// The dirty state of the Substrait submodule used to build this crate
 pub const SUBSTRAIT_GIT_DIRTY: bool = {git_dirty};
 "#
-                ),
-            )?;
+            ),
+        )?;
+    } else {
+        // If we don't have a version file yet we fail the build.
+        if !version_in_file.exists() {
+            panic!("Couldn't find the substrait submodule. Please clone the submodule: `git submodule update --init`.")
         }
-        Err(e) => {
-            // If this is a package build the `substrait_version_file` should
-            // exist. If it does not, it means this is probably a Git build that
-            // did not clone the substrait submodule.
-            if !substrait_version_file.exists() {
-                panic!("Couldn't open the substrait submodule: {e}")
-            }
-        }
-    };
+    }
 
     Ok(())
 }

src/version.rs

@@ -8,7 +8,7 @@
 
 use crate::proto::Version;
 
-include!("version.in");
+include!("../gen/version.in");
 
 /// Returns the version of Substrait used to build this crate.
 ///