C++: Build BMN extractor with Bazel

Add a public Bazel target for the Rust build-mode-none extractor, expose its generated crate universe through the standalone module, and resolve test resources through Bazel's TEST_SRCDIR runfiles root.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: paldepind

MODULE.bazel

@@ -98,6 +98,29 @@ use_repo(
     "vendor_py__tree-sitter-graph-0.12.0",
 )
 
+# deps for C/C++ build mode none extractor
+# keep in sync by running `misc/bazel/3rdparty/update_cpp_bmn_deps.sh`
+cpp_bmn_deps = use_extension("//misc/bazel/3rdparty:cpp_bmn_deps_extension.bzl", "b")
+use_repo(
+    cpp_bmn_deps,
+    "vendor_cpp_bmn__all_asserts-2.3.3",
+    "vendor_cpp_bmn__anyhow-1.0.98",
+    "vendor_cpp_bmn__cached-0.54.0",
+    "vendor_cpp_bmn__chrono-0.4.41",
+    "vendor_cpp_bmn__itertools-0.14.0",
+    "vendor_cpp_bmn__num_cpus-1.17.0",
+    "vendor_cpp_bmn__os_str_bytes-7.1.1",
+    "vendor_cpp_bmn__rayon-1.11.0",
+    "vendor_cpp_bmn__regex-1.11.1",
+    "vendor_cpp_bmn__serde-1.0.219",
+    "vendor_cpp_bmn__serde_json-1.0.140",
+    "vendor_cpp_bmn__serde_with-1.14.0",
+    "vendor_cpp_bmn__tempfile-3.20.0",
+    "vendor_cpp_bmn__walkdir-2.5.0",
+    "vendor_cpp_bmn__which-8.0.0",
+    "vendor_cpp_bmn__zip-6.0.0",
+)
+
 # deps for ruby+rust
 # keep in sync by running `misc/bazel/3rdparty/update_cargo_deps.sh`
 tree_sitter_extractors_deps = use_extension("//misc/bazel/3rdparty:tree_sitter_extractors_extension.bzl", "r")

cpp/extractor/bmn/BUILD.bazel

@@ -1,4 +1,33 @@
+load("@rules_rust//rust:defs.bzl", "rust_test")
+load("//misc/bazel:rust.bzl", "codeql_rust_binary")
+load("//misc/bazel/3rdparty/cpp_bmn_deps:defs.bzl", "all_crate_deps")
+
 exports_files([
     "Cargo.lock",
     "Cargo.toml",
 ])
+
+codeql_rust_binary(
+    name = "bmn",
+    srcs = glob(["src/**/*.rs"]),
+    proc_macro_deps = all_crate_deps(proc_macro = True),
+    visibility = ["//visibility:public"],
+    deps = all_crate_deps(normal = True),
+)
+
+rust_test(
+    name = "unit_tests",
+    crate = ":single_arch/bmn",
+    data = glob([
+        "tests/resources/**",
+        "Cargo.toml",
+    ]),
+    env = {
+        "BMN_TEST_MANIFEST": "$(rlocationpath :Cargo.toml)",
+    },
+)
+
+test_suite(
+    name = "tests",
+    tests = [":unit_tests"],
+)

cpp/extractor/bmn/src/test_utils.rs

@@ -1,20 +1,24 @@
 use all_asserts::assert_true;
-use std::fs;
 use std::fs::File;
 use std::path::PathBuf;
-use tempfile::{TempDir, tempdir};
+use std::{env, fs};
+use tempfile::{tempdir, TempDir};
 
 /// This function tries to get the root test resource path.
 /// If `allow_symlinks` is `false` will avoid the bazel off-tree test directory as this contains
 /// symlinks instead of the files, and we will return the folder containing the canonicalized
 /// `Cargo.toml` file.
 fn get_in_tree_resource_root(allow_symlinks: bool) -> PathBuf {
-    // Use CARGO_MANIFEST_DIR runtime environment variable to get the path
-    // This is the path where the resources are expected to be, but it can be the bazel out-of-tree path
-    let manifest_path = std::env::var("CARGO_MANIFEST_DIR")
-        .map(PathBuf::from)
-        .expect("$CARGO_MANIFEST_DIR not set")
-        .join("Cargo.toml");
+    // The below ensures that we support both `bazel test` and `cargo test`.
+    let manifest_path = if let Ok(runfiles_manifest) = env::var("BMN_TEST_MANIFEST") {
+        let runfiles_dir = env::var("TEST_SRCDIR").expect("$TEST_SRCDIR not set");
+        PathBuf::from(runfiles_dir).join(runfiles_manifest)
+    } else {
+        env::var("CARGO_MANIFEST_DIR")
+            .map(PathBuf::from)
+            .expect("$CARGO_MANIFEST_DIR not set")
+            .join("Cargo.toml")
+    };
 
     let final_manifest_path = if !allow_symlinks {
         // If we do not want symlinks ()created by Bazel), we can canonicalize the manifest path to
@@ -120,7 +124,7 @@ mod tests {
         assert_true!(file_maybe_symlinked.exists());
         assert_true!(file_maybe_symlinked.ends_with("tests/resources/dir1/file1.c"));
 
-        let is_bazel = std::env::var("BAZEL_TEST")
+        let is_bazel = env::var("BAZEL_TEST")
             .map(|bazel| bazel == "1")
             .unwrap_or(false);
         if is_bazel {

misc/bazel/3rdparty/cpp_bmn_deps_extension.bzl

@@ -0,0 +1,15 @@
+"""Module extension for using BMN's vendored crates with bzlmod."""
+
+load("//misc/bazel/3rdparty/cpp_bmn_deps:defs.bzl", _crate_repositories = "crate_repositories")
+
+def _crate_repositories_impl(module_ctx):
+    direct_deps = _crate_repositories()
+    return module_ctx.extension_metadata(
+        root_module_direct_deps = [repo.repo for repo in direct_deps],
+        root_module_direct_dev_deps = [],
+    )
+
+# Short name to address Windows path length issues.
+b = module_extension(
+    implementation = _crate_repositories_impl,
+)