Remove process env mutation (#177) (#221)

Cargo.toml

@@ -22,9 +22,11 @@ arrow-array = "59.2.0"
 arrow-schema = "59.2.0"
 num-traits = "0.2.19"
 parquet = "59.2.0"
+mockable = "3.0.0"
 
 [workspace.lints.clippy]
 pedantic = { level = "warn", priority = -1 }
+disallowed_methods = "deny"
 
 # 1. hygiene
 allow_attributes                    = "deny"

chutoro-benches/Cargo.toml

@@ -13,6 +13,7 @@ chutoro-core = { path = "../chutoro-core" }
 chutoro-providers-dense = { path = "../chutoro-providers/dense" }
 criterion = { version = "0.5.1", features = ["html_reports"] }
 flate2 = "1.1.9"
+mockable = { workspace = true }
 rand = { version = "0.8.5", features = ["small_rng"] }
 strsim = "0.11.1"
 thiserror = "2.0.17"
@@ -59,6 +60,7 @@ harness = false
 # and updated to stay in sync.
 [lints.clippy]
 pedantic = { level = "warn", priority = -1 }
+disallowed_methods = "deny"
 
 # 1. hygiene
 allow_attributes                    = "deny"

chutoro-benches/benches/hnsw.rs

@@ -9,6 +9,7 @@ use criterion::{
     BatchSize, BenchmarkGroup, BenchmarkId, Criterion, black_box, criterion_main,
     measurement::WallTime,
 };
+use mockable::{DefaultEnv, Env};
 
 use chutoro_benches::{
     criterion_support::{
@@ -257,8 +258,8 @@ fn hnsw_build(c: &mut Criterion) {
     }
 }
 
-fn should_collect_memory_profile() -> bool {
-    if let Ok(value) = std::env::var("CHUTORO_BENCH_HNSW_MEMORY_PROFILE") {
+fn should_collect_memory_profile_with_env(env: &dyn Env) -> bool {
+    if let Some(value) = env.string("CHUTORO_BENCH_HNSW_MEMORY_PROFILE") {
         let normalized = value.trim().to_ascii_lowercase();
         if matches!(normalized.as_str(), "0" | "false" | "off") {
             return false;
@@ -270,17 +271,21 @@ fn should_collect_memory_profile() -> bool {
     !is_benchmark_discovery() && !is_exact_benchmark_probe()
 }
 
-fn memory_report_path() -> PathBuf {
-    std::env::var_os("CHUTORO_BENCH_HNSW_MEMORY_REPORT_PATH")
+fn memory_report_path_with_env(env: &dyn Env) -> PathBuf {
+    env.os_string("CHUTORO_BENCH_HNSW_MEMORY_REPORT_PATH")
         .map_or_else(|| PathBuf::from(MEMORY_REPORT_PATH), PathBuf::from)
 }
 
 fn profile_hnsw_memory_impl() -> Result<Option<PathBuf>, BenchSetupError> {
-    if !should_collect_memory_profile() {
+    profile_hnsw_memory_impl_with_env(&DefaultEnv)
+}
+
+fn profile_hnsw_memory_impl_with_env(env: &dyn Env) -> Result<Option<PathBuf>, BenchSetupError> {
+    if !should_collect_memory_profile_with_env(env) {
         return Ok(None);
     }
 
-    let report_path = memory_report_path();
+    let report_path = memory_report_path_with_env(env);
     let mut records = Vec::new();
 
     for &point_count in POINT_COUNTS {
@@ -330,6 +335,13 @@ fn hnsw_build_with_edges(c: &mut Criterion) {
 }
 
 fn hnsw_build_diverse_sources_impl(c: &mut Criterion) -> Result<(), BenchSetupError> {
+    hnsw_build_diverse_sources_impl_with_env(c, &DefaultEnv)
+}
+
+fn hnsw_build_diverse_sources_impl_with_env(
+    c: &mut Criterion,
+    env: &dyn Env,
+) -> Result<(), BenchSetupError> {
     let mut group = c.benchmark_group("hnsw_build_diverse_sources");
     configure_hnsw_group(&mut group);
 
@@ -368,7 +380,7 @@ fn hnsw_build_diverse_sources_impl(c: &mut Criterion) -> Result<(), BenchSetupEr
         &params,
     );
 
-    if std::env::var("CHUTORO_BENCH_ENABLE_MNIST").as_deref() == Ok("1") {
+    if env.string("CHUTORO_BENCH_ENABLE_MNIST").as_deref() == Some("1") {
         let mnist = SyntheticSource::load_mnist(&MnistConfig::default())?;
         bench_build_source(
             &mut group,

chutoro-benches/benches/hnsw_ef_sweep.rs

@@ -6,6 +6,7 @@
 use std::{num::NonZeroUsize, path::PathBuf, time::Duration, time::Instant};
 
 use criterion::{BatchSize, BenchmarkId, Criterion, criterion_main};
+use mockable::{DefaultEnv, Env};
 
 use chutoro_benches::{
     ef_sweep::{
@@ -82,8 +83,8 @@ fn warn_unrecognised_bool_env(env_var_name: &str, value: &str) {
     );
 }
 
-fn parse_bool_env_var(env_var_name: &str) -> Option<bool> {
-    let value = std::env::var(env_var_name).ok()?;
+fn parse_bool_env_var(env: &dyn Env, env_var_name: &str) -> Option<bool> {
+    let value = env.string(env_var_name)?;
     let normalized = value.trim().to_ascii_lowercase();
     if matches!(normalized.as_str(), "0" | "false" | "off") {
         return Some(false);
@@ -122,24 +123,42 @@ fn ef_sweep_point_counts() -> &'static [usize] {
 }
 
 fn should_collect_recall_report() -> bool {
-    parse_bool_env_var("CHUTORO_BENCH_HNSW_RECALL_REPORT").unwrap_or_else(|| !is_discovery_mode())
+    should_collect_recall_report_with_env(&DefaultEnv)
+}
+
+fn should_collect_recall_report_with_env(env: &dyn Env) -> bool {
+    parse_bool_env_var(env, "CHUTORO_BENCH_HNSW_RECALL_REPORT")
+        .unwrap_or_else(|| !is_discovery_mode())
 }
 
 fn recall_report_path() -> PathBuf {
-    std::env::var_os("CHUTORO_BENCH_HNSW_RECALL_REPORT_PATH")
+    recall_report_path_with_env(&DefaultEnv)
+}
+
+fn recall_report_path_with_env(env: &dyn Env) -> PathBuf {
+    env.os_string("CHUTORO_BENCH_HNSW_RECALL_REPORT_PATH")
         .map_or_else(|| PathBuf::from(RECALL_REPORT_PATH), PathBuf::from)
 }
 
 fn should_collect_cluster_quality_report() -> bool {
-    parse_bool_env_var("CHUTORO_BENCH_HNSW_CLUSTER_QUALITY_REPORT")
+    should_collect_cluster_quality_report_with_env(&DefaultEnv)
+}
+
+fn should_collect_cluster_quality_report_with_env(env: &dyn Env) -> bool {
+    parse_bool_env_var(env, "CHUTORO_BENCH_HNSW_CLUSTER_QUALITY_REPORT")
         .unwrap_or_else(|| !is_discovery_mode())
 }
 
 fn cluster_quality_report_path() -> PathBuf {
-    std::env::var_os("CHUTORO_BENCH_HNSW_CLUSTER_QUALITY_REPORT_PATH").map_or_else(
-        || PathBuf::from(CLUSTERING_QUALITY_REPORT_PATH),
-        PathBuf::from,
-    )
+    cluster_quality_report_path_with_env(&DefaultEnv)
+}
+
+fn cluster_quality_report_path_with_env(env: &dyn Env) -> PathBuf {
+    env.os_string("CHUTORO_BENCH_HNSW_CLUSTER_QUALITY_REPORT_PATH")
+        .map_or_else(
+            || PathBuf::from(CLUSTERING_QUALITY_REPORT_PATH),
+            PathBuf::from,
+        )
 }
 
 /// Returns an evenly-spaced query index for deterministic recall sampling.

chutoro-benches/src/criterion_support.rs

@@ -6,6 +6,7 @@
 use std::{fmt::Display, time::Duration};
 
 use criterion::{BenchmarkGroup, BenchmarkId, Criterion, measurement::WallTime};
+use mockable::{DefaultEnv, Env};
 
 /// Returns whether the current command line includes `flag`.
 ///
@@ -111,9 +112,13 @@ pub fn is_exact_benchmark_probe() -> bool {
 /// ```
 #[must_use]
 pub fn is_nextest_exact_benchmark_probe() -> bool {
+    is_nextest_exact_benchmark_probe_with_env(&DefaultEnv)
+}
+
+fn is_nextest_exact_benchmark_probe_with_env(env: &dyn Env) -> bool {
     is_nextest_exact_benchmark_probe_args(
         std::env::args(),
-        std::env::var_os("NEXTEST_TEST_NAME").is_some(),
+        env.os_string("NEXTEST_TEST_NAME").is_some(),
     )
 }
 

chutoro-benches/src/neighbour_scoring/benchmark_runner.rs

@@ -16,6 +16,7 @@ use chutoro_core::DataSource;
 use criterion::{
     BenchmarkGroup, BenchmarkId, Criterion, Throughput, black_box, measurement::WallTime,
 };
+use mockable::{DefaultEnv, Env};
 
 use super::{
     CandidateBucket, ScoringFixture, benchmark_support::BenchError, benchmark_support::BenchResult,
@@ -34,9 +35,12 @@ fn should_use_short_measurement_value(value: Option<&str>) -> bool {
 }
 
 fn should_use_short_measurement() -> bool {
-    should_use_short_measurement_value(std::env::var(SHORT_MEASUREMENT_ENV).ok().as_deref())
+    should_use_short_measurement_with_env(&DefaultEnv)
 }
 
+fn should_use_short_measurement_with_env(env: &dyn Env) -> bool {
+    should_use_short_measurement_value(env.string(SHORT_MEASUREMENT_ENV).as_deref())
+}
 fn score_candidates(
     scoring_fixture: &ScoringFixture,
     candidates: &[usize],
@@ -109,7 +113,7 @@ fn neighbour_scoring_impl_with(
 ) -> BenchResult<()> {
     let report_parent_dir = report_parent_dir();
     let build_profile_target = build_profile_report_target_value(
-        std::env::var(BUILD_PROFILE_ENV).ok().as_deref(),
+        DefaultEnv.string(BUILD_PROFILE_ENV).as_deref(),
         &report_parent_dir,
     );
     let build_profile_report_dir = build_profile_target