yams/tests/unit/daemon/components/config_resolver_test.cpp

1031 lines
38 KiB
C++

// Copyright (c) 2025 YAMS Contributors
// SPDX-License-Identifier: GPL-3.0-or-later
//
// Unit tests for ConfigResolver component (PBI-090)
//
// Catch2 migration from GTest (yams-3s4 / yams-zns)
#include <catch2/catch_test_macros.hpp>
#include <catch2/matchers/catch_matchers_string.hpp>
#include <yams/compat/unistd.h>
#include <yams/daemon/components/ConfigResolver.h>
#include <yams/daemon/components/TuneAdvisor.h>
#include <chrono>
#include <cstdlib>
#include <filesystem>
#include <fstream>
#include <string>
#include <yams/vector/document_chunker.h>
using namespace yams::daemon;
using Catch::Matchers::ContainsSubstring;
namespace {
struct ConfigResolverFixture {
std::filesystem::path tempDir;
ConfigResolverFixture() {
tempDir = std::filesystem::temp_directory_path() /
("yams_config_test_" +
std::to_string(std::chrono::steady_clock::now().time_since_epoch().count()));
std::filesystem::create_directories(tempDir);
}
~ConfigResolverFixture() {
std::error_code ec;
std::filesystem::remove_all(tempDir, ec);
}
std::filesystem::path writeToml(const std::string& filename, const std::string& content) {
auto path = tempDir / filename;
std::ofstream out(path);
out << content;
return path;
}
};
class ProfileGuard {
yams::daemon::TuneAdvisor::Profile prev_;
public:
explicit ProfileGuard(yams::daemon::TuneAdvisor::Profile profile)
: prev_(yams::daemon::TuneAdvisor::tuningProfile()) {
yams::daemon::TuneAdvisor::setTuningProfile(profile);
}
~ProfileGuard() { yams::daemon::TuneAdvisor::setTuningProfile(prev_); }
ProfileGuard(const ProfileGuard&) = delete;
ProfileGuard& operator=(const ProfileGuard&) = delete;
};
struct EnvGuard {
std::string name;
std::optional<std::string> originalValue;
explicit EnvGuard(const std::string& envName, const std::string& newValue) : name(envName) {
if (const char* orig = std::getenv(name.c_str())) {
originalValue = orig;
}
#ifdef _WIN32
_putenv_s(name.c_str(), newValue.c_str());
#else
setenv(name.c_str(), newValue.c_str(), 1);
#endif
}
~EnvGuard() {
if (originalValue) {
#ifdef _WIN32
_putenv_s(name.c_str(), originalValue->c_str());
#else
setenv(name.c_str(), originalValue->c_str(), 1);
#endif
} else {
#ifdef _WIN32
_putenv_s(name.c_str(), "");
#else
unsetenv(name.c_str());
#endif
}
}
EnvGuard(const EnvGuard&) = delete;
EnvGuard& operator=(const EnvGuard&) = delete;
};
} // namespace
TEST_CASE("ConfigResolver::resolveEmbeddingChunkingPolicy defaults are embedding-safe",
"[daemon][components][config][chunking][catch2]") {
// Ensure config file discovery doesn't accidentally pick up a user config during local dev.
EnvGuard cfg("YAMS_CONFIG_PATH", "");
auto policy = ConfigResolver::resolveEmbeddingChunkingPolicy();
CHECK(policy.strategy == yams::vector::ChunkingStrategy::PARAGRAPH_BASED);
CHECK_FALSE(policy.config.preserve_sentences);
CHECK_FALSE(policy.config.use_token_count);
}
TEST_CASE("ConfigResolver::resolveEmbeddingChunkingPolicy supports all embed strategies",
"[daemon][components][config][chunking][catch2]") {
struct Case {
const char* value;
yams::vector::ChunkingStrategy expected;
};
const std::vector<Case> cases = {
{"fixed", yams::vector::ChunkingStrategy::FIXED_SIZE},
{"sentence", yams::vector::ChunkingStrategy::SENTENCE_BASED},
{"paragraph", yams::vector::ChunkingStrategy::PARAGRAPH_BASED},
{"recursive", yams::vector::ChunkingStrategy::RECURSIVE},
{"sliding_window", yams::vector::ChunkingStrategy::SLIDING_WINDOW},
{"markdown", yams::vector::ChunkingStrategy::MARKDOWN_AWARE},
};
const std::string sample = "# Title\n\n"
"Intro paragraph with enough text to chunk reasonably. "
"Second sentence here. Third sentence here.\n\n"
"## Section\n"
"- bullet one\n- bullet two\n\n"
"Final paragraph.";
for (const auto& tc : cases) {
DYNAMIC_SECTION("strategy=" << tc.value) {
EnvGuard g("YAMS_EMBED_CHUNK_STRATEGY", tc.value);
auto policy = ConfigResolver::resolveEmbeddingChunkingPolicy();
CHECK(policy.strategy == tc.expected);
auto chunker = yams::vector::createChunker(policy.strategy, policy.config, nullptr);
REQUIRE(chunker);
auto chunks = chunker->chunkDocument(sample, "hash");
// Some strategies may return empty and rely on caller fallback, but for this
// representative input we expect at least one chunk.
REQUIRE_FALSE(chunks.empty());
for (const auto& c : chunks) {
CHECK(c.strategy_used == tc.expected);
CHECK_FALSE(c.content.empty());
}
}
}
}
TEST_CASE("ConfigResolver::resolveEmbeddingChunkingPolicy reads from config file",
"[daemon][components][config][chunking][catch2]") {
ConfigResolverFixture fx;
auto configPath = fx.writeToml("config.toml",
R"TOML(
[embeddings.chunking]
strategy = "fixed"
target = 256
min = 64
max = 512
overlap = 0
use_tokens = 0
preserve_sentences = 0
)TOML");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveEmbeddingChunkingPolicy();
CHECK(policy.strategy == yams::vector::ChunkingStrategy::FIXED_SIZE);
CHECK(policy.config.target_chunk_size == 256);
CHECK(policy.config.min_chunk_size == 64);
CHECK(policy.config.max_chunk_size == 512);
CHECK(policy.config.overlap_size == 0);
CHECK(policy.config.overlap_percentage == 0.0);
CHECK_FALSE(policy.config.use_token_count);
CHECK_FALSE(policy.config.preserve_sentences);
}
TEST_CASE("ConfigResolver::envTruthy correctly parses truthy values",
"[daemon][components][config][catch2]") {
SECTION("truthy values") {
CHECK(ConfigResolver::envTruthy("1"));
CHECK(ConfigResolver::envTruthy("true"));
CHECK(ConfigResolver::envTruthy("TRUE"));
CHECK(ConfigResolver::envTruthy("True"));
CHECK(ConfigResolver::envTruthy("yes"));
CHECK(ConfigResolver::envTruthy("YES"));
CHECK(ConfigResolver::envTruthy("Yes"));
CHECK(ConfigResolver::envTruthy("on"));
CHECK(ConfigResolver::envTruthy("ON"));
CHECK(ConfigResolver::envTruthy("On"));
}
SECTION("falsy values") {
CHECK_FALSE(ConfigResolver::envTruthy("0"));
CHECK_FALSE(ConfigResolver::envTruthy("false"));
CHECK_FALSE(ConfigResolver::envTruthy("FALSE"));
CHECK_FALSE(ConfigResolver::envTruthy("no"));
CHECK_FALSE(ConfigResolver::envTruthy("NO"));
CHECK_FALSE(ConfigResolver::envTruthy("off"));
CHECK_FALSE(ConfigResolver::envTruthy("OFF"));
CHECK_FALSE(ConfigResolver::envTruthy(""));
CHECK_FALSE(ConfigResolver::envTruthy(nullptr));
}
SECTION("unrecognized values are truthy (not in falsey list)") {
// The implementation treats anything NOT in {0, false, off, no} as truthy
CHECK(ConfigResolver::envTruthy("garbage"));
CHECK(ConfigResolver::envTruthy("maybe"));
CHECK(ConfigResolver::envTruthy("123"));
CHECK(ConfigResolver::envTruthy("yep"));
CHECK(ConfigResolver::envTruthy("nope"));
}
}
TEST_CASE_METHOD(ConfigResolverFixture, "ConfigResolver parseSimpleTomlFlat parses TOML files",
"[daemon][components][config][catch2]") {
SECTION("basic TOML parsing with sections") {
auto configPath = writeToml("test.toml", R"(
[daemon]
socket_path = "/tmp/test.sock"
log_level = "debug"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["daemon.socket_path"] == "/tmp/test.sock");
CHECK(config["daemon.log_level"] == "debug");
}
SECTION("missing file returns empty map") {
auto config = ConfigResolver::parseSimpleTomlFlat(tempDir / "nonexistent.toml");
CHECK(config.empty());
}
SECTION("empty file returns empty map") {
auto configPath = writeToml("empty.toml", "");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config.empty());
}
SECTION("comments are ignored") {
auto configPath = writeToml("comments.toml", R"(
# This is a comment
[section]
# Another comment
key = "value"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["section.key"] == "value");
CHECK(config.size() == 1);
}
SECTION("multiple sections parsed correctly") {
auto configPath = writeToml("multi.toml", R"(
[section1]
key1 = "value1"
[section2]
key2 = "value2"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["section1.key1"] == "value1");
CHECK(config["section2.key2"] == "value2");
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver resolveDefaultConfigPath searches standard paths",
"[daemon][components][config][catch2]") {
SECTION("YAMS_CONFIG_PATH env var takes precedence") {
auto configPath = writeToml("custom.toml", "[daemon]\nlog_level = \"trace\"\n");
EnvGuard guard("YAMS_CONFIG_PATH", configPath.string());
auto resolved = ConfigResolver::resolveDefaultConfigPath();
CHECK(resolved == configPath);
}
SECTION("returns empty path if no config found") {
// Clear env var to test fallback
EnvGuard guard("YAMS_CONFIG_PATH", "");
// Note: this test may find system config, so we just check it returns a valid result
auto resolved = ConfigResolver::resolveDefaultConfigPath();
// Either found or empty - both are valid outcomes
CHECK((resolved.empty() || std::filesystem::exists(resolved) || !resolved.empty()));
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver resolveEmbeddingBackend defaults to simeon for new installs",
"[daemon][components][config][catch2]") {
EnvGuard backendGuard("YAMS_EMBED_BACKEND", "");
auto configPath = writeToml("embedding_backend_default.toml", R"(
[core]
data_dir = "~/.yams/data"
)");
EnvGuard configGuard("YAMS_CONFIG_PATH", configPath.string());
CHECK(ConfigResolver::resolveEmbeddingBackend() == "simeon");
CHECK(ConfigResolver::resolveEmbeddingBackend("auto") == "auto");
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver resolveEmbeddingBackend reads and normalizes configured backend",
"[daemon][components][config][catch2]") {
EnvGuard backendGuard("YAMS_EMBED_BACKEND", "");
SECTION("reads daemon from config") {
auto configPath = writeToml("embedding_backend_daemon.toml", R"(
[embeddings]
backend = "daemon"
)");
EnvGuard configGuard("YAMS_CONFIG_PATH", configPath.string());
CHECK(ConfigResolver::resolveEmbeddingBackend() == "daemon");
}
SECTION("normalizes uppercase simeon from config") {
auto configPath = writeToml("embedding_backend_simeon_upper.toml", R"(
[embeddings]
backend = "SIMEON"
)");
EnvGuard configGuard("YAMS_CONFIG_PATH", configPath.string());
CHECK(ConfigResolver::resolveEmbeddingBackend() == "simeon");
}
SECTION("maps legacy onnx backend names to onnxruntime") {
auto configPath = writeToml("embedding_backend_legacy_onnx.toml", R"(
[embeddings]
backend = "local_onnx"
)");
EnvGuard configGuard("YAMS_CONFIG_PATH", configPath.string());
CHECK(ConfigResolver::resolveEmbeddingBackend() == "onnxruntime");
}
SECTION("normalizes explicit onnxruntime from config") {
auto configPath = writeToml("embedding_backend_onnxruntime.toml", R"(
[embeddings]
backend = "ONNX-RUNTIME"
)");
EnvGuard configGuard("YAMS_CONFIG_PATH", configPath.string());
CHECK(ConfigResolver::resolveEmbeddingBackend() == "onnxruntime");
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver resolveEmbeddingBackend maps ONNX env aliases to onnxruntime",
"[daemon][components][config][catch2]") {
SECTION("maps onnx env alias") {
EnvGuard backendGuard("YAMS_EMBED_BACKEND", "onnx");
CHECK(ConfigResolver::resolveEmbeddingBackend() == "onnxruntime");
}
SECTION("maps local_onnx env alias") {
EnvGuard backendGuard("YAMS_EMBED_BACKEND", "local_onnx");
CHECK(ConfigResolver::resolveEmbeddingBackend() == "onnxruntime");
}
SECTION("maps ort env alias") {
EnvGuard backendGuard("YAMS_EMBED_BACKEND", "ort");
CHECK(ConfigResolver::resolveEmbeddingBackend() == "onnxruntime");
}
SECTION("keeps daemon env selection distinct") {
EnvGuard backendGuard("YAMS_EMBED_BACKEND", "daemon");
CHECK(ConfigResolver::resolveEmbeddingBackend() == "daemon");
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveTopologyTunerPolicy defaults are empty when missing",
"[daemon][components][config][topology_tuner][catch2]") {
auto configPath = writeToml("config.toml", R"TOML(
[daemon]
socket_path = "/tmp/test.sock"
)TOML");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveTopologyTunerPolicy();
CHECK_FALSE(policy.enabled.has_value());
CHECK_FALSE(policy.cooldownMinutes.has_value());
CHECK_FALSE(policy.docCountDelta.has_value());
CHECK_FALSE(policy.rewardAlphaSingleton.has_value());
CHECK_FALSE(policy.rewardBetaGiantCluster.has_value());
CHECK_FALSE(policy.rewardGammaGiniDeviation.has_value());
CHECK_FALSE(policy.rewardDeltaIntraEdge.has_value());
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveEmbeddingSelectionPolicy reads semantic graph ingest flag",
"[daemon][components][config][embeddings][catch2]") {
auto configPath = writeToml("config.toml", R"TOML(
[embeddings.selection]
update_semantic_graph_during_ingest = false
)TOML");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveEmbeddingSelectionPolicy();
CHECK(policy.updateSemanticGraphDuringIngest == false);
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveTopologyTunerPolicy reads populated TOML block",
"[daemon][components][config][topology_tuner][catch2]") {
auto configPath = writeToml("config.toml", R"TOML(
[topology.tuner]
enabled = true
cooldown_minutes = 30
doc_count_delta = 250
[topology.tuner.reward]
alpha_singleton = 0.5
beta_giant_cluster = 0.3
gamma_gini_deviation = 0.15
delta_intra_edge = 0.05
)TOML");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveTopologyTunerPolicy();
REQUIRE(policy.enabled.has_value());
CHECK(*policy.enabled == true);
REQUIRE(policy.cooldownMinutes.has_value());
CHECK(*policy.cooldownMinutes == 30u);
REQUIRE(policy.docCountDelta.has_value());
CHECK(*policy.docCountDelta == 250u);
REQUIRE(policy.rewardAlphaSingleton.has_value());
CHECK(*policy.rewardAlphaSingleton == 0.5);
REQUIRE(policy.rewardBetaGiantCluster.has_value());
CHECK(*policy.rewardBetaGiantCluster == 0.3);
REQUIRE(policy.rewardGammaGiniDeviation.has_value());
CHECK(*policy.rewardGammaGiniDeviation == 0.15);
REQUIRE(policy.rewardDeltaIntraEdge.has_value());
CHECK(*policy.rewardDeltaIntraEdge == 0.05);
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveTopologyTunerPolicy honors disabled flag explicitly",
"[daemon][components][config][topology_tuner][catch2]") {
auto configPath = writeToml("config.toml", R"TOML(
[topology.tuner]
enabled = false
)TOML");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveTopologyTunerPolicy();
REQUIRE(policy.enabled.has_value());
CHECK(*policy.enabled == false);
}
TEST_CASE("ConfigResolver vector sentinel operations",
"[daemon][components][config][catch2][sentinel]") {
auto tempDir = std::filesystem::temp_directory_path() /
("yams_sentinel_test_" +
std::to_string(std::chrono::steady_clock::now().time_since_epoch().count()));
std::filesystem::create_directories(tempDir);
SECTION("readVectorSentinelDim returns nullopt for missing file") {
auto dim = ConfigResolver::readVectorSentinelDim(tempDir);
CHECK_FALSE(dim.has_value());
}
SECTION("writeVectorSentinel then readVectorSentinelDim round-trips") {
ConfigResolver::writeVectorSentinel(tempDir, 384, "test_table", 1);
auto dim = ConfigResolver::readVectorSentinelDim(tempDir);
REQUIRE(dim.has_value());
CHECK(dim.value() == 384);
}
SECTION("different dimensions are preserved") {
for (size_t testDim : {128, 256, 384, 768, 1024}) {
ConfigResolver::writeVectorSentinel(tempDir, testDim, "test", 1);
auto dim = ConfigResolver::readVectorSentinelDim(tempDir);
REQUIRE(dim.has_value());
CHECK(dim.value() == testDim);
}
}
std::error_code ec;
std::filesystem::remove_all(tempDir, ec);
}
TEST_CASE_METHOD(ConfigResolverFixture, "ConfigResolver parses [tuning] section",
"[daemon][components][config][catch2]") {
SECTION("efficient profile parses correctly") {
auto configPath = writeToml("efficient.toml", R"(
[tuning]
profile = "efficient"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["tuning.profile"] == "efficient");
}
SECTION("balanced profile parses correctly") {
auto configPath = writeToml("balanced.toml", R"(
[tuning]
profile = "balanced"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["tuning.profile"] == "balanced");
}
SECTION("aggressive profile parses correctly") {
auto configPath = writeToml("aggressive.toml", R"(
[tuning]
profile = "aggressive"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["tuning.profile"] == "aggressive");
}
SECTION("conservative alias for efficient") {
auto configPath = writeToml("conservative.toml", R"(
[tuning]
profile = "conservative"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["tuning.profile"] == "conservative");
}
SECTION("case-insensitive profile names") {
auto configPath = writeToml("case_test.toml", R"(
[tuning]
profile = "EFFICIENT"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["tuning.profile"] == "EFFICIENT");
}
SECTION("missing tuning section returns empty") {
auto configPath = writeToml("no_tuning.toml", R"(
[daemon]
socket_path = "/tmp/test.sock"
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config.find("tuning.profile") == config.end());
}
SECTION("multiple tuning options parsed together") {
auto configPath = writeToml("multi_tuning.toml", R"(
[tuning]
profile = "aggressive"
pool_cooldown_ms = 250
worker_poll_ms = 100
)");
auto config = ConfigResolver::parseSimpleTomlFlat(configPath);
CHECK(config["tuning.profile"] == "aggressive");
CHECK(config["tuning.pool_cooldown_ms"] == "250");
CHECK(config["tuning.worker_poll_ms"] == "100");
}
}
TEST_CASE("Tuning profile from config affects TuneAdvisor methods",
"[daemon][components][config][catch2]") {
SECTION("efficient profile scales post-ingest concurrency down") {
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Efficient);
EnvGuard maxThreadsGuard("YAMS_MAX_THREADS", "0");
EnvGuard postIngestGuard("YAMS_POST_INGEST_TOTAL_CONCURRENT", "0");
yams::daemon::TuneAdvisor::setHardwareConcurrencyForTests(8);
// Efficient profile scale is 0.0, old totalBudget = 2
// With active-stage floor: total = max(2, 6 active stages) = 6
// Each stage gets 1 slot; postIngestBatchSize = max(1, floor(8.0 * 0.0)) = 1
CHECK(TuneAdvisor::postExtractionConcurrent() == 1u);
CHECK(TuneAdvisor::postKgConcurrent() == 1u);
CHECK(TuneAdvisor::postSymbolConcurrent() == 1u);
CHECK(TuneAdvisor::postEntityConcurrent() == 1u);
CHECK(TuneAdvisor::postTitleConcurrent() == 1u);
CHECK(TuneAdvisor::postEmbedConcurrent() == 1u);
CHECK(TuneAdvisor::postIngestBatchSize() == 1u);
}
SECTION("balanced profile uses medium values") {
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Balanced);
EnvGuard maxThreadsGuard("YAMS_MAX_THREADS", "0");
EnvGuard postIngestGuard("YAMS_POST_INGEST_TOTAL_CONCURRENT", "0");
yams::daemon::TuneAdvisor::setHardwareConcurrencyForTests(8);
// Balanced profile scale is 0.5, old totalBudget = 2
// With active-stage floor: total = max(2, 6 active stages) = 6
// Each stage gets 1 slot; postIngestBatchSize = max(1, floor(8.0 * 0.5)) = 4
CHECK(TuneAdvisor::postExtractionConcurrent() == 1u);
CHECK(TuneAdvisor::postKgConcurrent() == 1u);
CHECK(TuneAdvisor::postSymbolConcurrent() == 1u);
CHECK(TuneAdvisor::postEntityConcurrent() == 1u);
CHECK(TuneAdvisor::postTitleConcurrent() == 1u);
CHECK(TuneAdvisor::postEmbedConcurrent() == 1u);
CHECK(TuneAdvisor::postIngestBatchSize() == 4u);
}
SECTION("aggressive profile uses maximum values") {
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Aggressive);
EnvGuard maxThreadsGuard("YAMS_MAX_THREADS", "0");
EnvGuard postIngestGuard("YAMS_POST_INGEST_TOTAL_CONCURRENT", "0");
yams::daemon::TuneAdvisor::setHardwareConcurrencyForTests(8);
// Aggressive profile scale is 1.0, old totalBudget = 3
// With active-stage floor: total = max(3, 6 active stages) = 6
// Each stage gets 1 slot; postIngestBatchSize = max(1, floor(8.0 * 1.0)) = 8
CHECK(TuneAdvisor::postExtractionConcurrent() == 1u);
CHECK(TuneAdvisor::postKgConcurrent() == 1u);
CHECK(TuneAdvisor::postSymbolConcurrent() == 1u);
CHECK(TuneAdvisor::postEntityConcurrent() == 1u);
CHECK(TuneAdvisor::postTitleConcurrent() == 1u);
CHECK(TuneAdvisor::postEmbedConcurrent() == 1u);
CHECK(TuneAdvisor::postIngestBatchSize() == 8u);
}
SECTION("profile affects cpuBudgetPercent") {
{
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Efficient);
CHECK(TuneAdvisor::cpuBudgetPercent() == 40u);
}
{
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Balanced);
CHECK(TuneAdvisor::cpuBudgetPercent() == 50u);
}
{
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Aggressive);
// 40.0 + 1.0 * 20.0 = 60
CHECK(TuneAdvisor::cpuBudgetPercent() == 60u);
}
}
SECTION("profile affects poolCooldownMs") {
{
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Efficient);
CHECK(TuneAdvisor::poolCooldownMs() == 750u);
}
{
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Balanced);
CHECK(TuneAdvisor::poolCooldownMs() == 500u);
}
{
ProfileGuard guard(yams::daemon::TuneAdvisor::Profile::Aggressive);
CHECK(TuneAdvisor::poolCooldownMs() == 250u);
}
}
}
TEST_CASE("YAMS_TUNING_PROFILE env var overrides config", "[daemon][components][config][catch2]") {
auto clearProfileOverride = []() {
yams::daemon::TuneAdvisor::tuningProfileOverride_.store(0, std::memory_order_relaxed);
};
SECTION("efficient profile from env var") {
clearProfileOverride();
EnvGuard envGuard("YAMS_TUNING_PROFILE", "efficient");
auto profile = yams::daemon::TuneAdvisor::tuningProfile();
CHECK(profile == yams::daemon::TuneAdvisor::Profile::Efficient);
}
SECTION("aggressive profile from env var") {
clearProfileOverride();
EnvGuard envGuard("YAMS_TUNING_PROFILE", "aggressive");
auto profile = yams::daemon::TuneAdvisor::tuningProfile();
CHECK(profile == yams::daemon::TuneAdvisor::Profile::Aggressive);
}
SECTION("conservative alias maps to efficient") {
clearProfileOverride();
EnvGuard envGuard("YAMS_TUNING_PROFILE", "conservative");
auto profile = yams::daemon::TuneAdvisor::Profile::Efficient;
CHECK(yams::daemon::TuneAdvisor::tuningProfile() == profile);
}
SECTION("invalid env var falls back to balanced") {
clearProfileOverride();
EnvGuard envGuard("YAMS_TUNING_PROFILE", "invalid_profile");
auto profile = yams::daemon::TuneAdvisor::tuningProfile();
CHECK(profile == yams::daemon::TuneAdvisor::Profile::Balanced);
}
SECTION("empty env var falls back to balanced") {
clearProfileOverride();
EnvGuard envGuard("YAMS_TUNING_PROFILE", "");
auto profile = yams::daemon::TuneAdvisor::tuningProfile();
CHECK(profile == yams::daemon::TuneAdvisor::Profile::Balanced);
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolvePostIngestCaps reads [tuning.post_ingest]",
"[daemon][components][config][post_ingest][catch2]") {
SECTION("missing section returns all nullopt") {
auto configPath = writeToml("no_post_ingest.toml", R"(
[tuning]
profile = "balanced"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto caps = ConfigResolver::resolvePostIngestCaps();
CHECK_FALSE(caps.totalConcurrent.has_value());
CHECK_FALSE(caps.embedConcurrent.has_value());
CHECK_FALSE(caps.extractionConcurrent.has_value());
CHECK_FALSE(caps.kgConcurrent.has_value());
CHECK_FALSE(caps.symbolConcurrent.has_value());
CHECK_FALSE(caps.entityConcurrent.has_value());
CHECK_FALSE(caps.titleConcurrent.has_value());
}
SECTION("all keys populate the struct") {
auto configPath = writeToml("post_ingest.toml", R"(
[tuning.post_ingest]
total_concurrent = 12
embed_concurrent = 4
extraction_concurrent = 5
kg_concurrent = 6
symbol_concurrent = 3
entity_concurrent = 2
title_concurrent = 2
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto caps = ConfigResolver::resolvePostIngestCaps();
REQUIRE(caps.totalConcurrent.has_value());
CHECK(*caps.totalConcurrent == 12u);
REQUIRE(caps.embedConcurrent.has_value());
CHECK(*caps.embedConcurrent == 4u);
REQUIRE(caps.extractionConcurrent.has_value());
CHECK(*caps.extractionConcurrent == 5u);
REQUIRE(caps.kgConcurrent.has_value());
CHECK(*caps.kgConcurrent == 6u);
REQUIRE(caps.symbolConcurrent.has_value());
CHECK(*caps.symbolConcurrent == 3u);
REQUIRE(caps.entityConcurrent.has_value());
CHECK(*caps.entityConcurrent == 2u);
REQUIRE(caps.titleConcurrent.has_value());
CHECK(*caps.titleConcurrent == 2u);
}
SECTION("out-of-range values are dropped (nullopt)") {
auto configPath = writeToml("oor.toml", R"(
[tuning.post_ingest]
total_concurrent = 0
embed_concurrent = 99
extraction_concurrent = 500
entity_concurrent = 17
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto caps = ConfigResolver::resolvePostIngestCaps();
CHECK_FALSE(caps.totalConcurrent.has_value());
CHECK_FALSE(caps.embedConcurrent.has_value());
CHECK_FALSE(caps.extractionConcurrent.has_value());
CHECK_FALSE(caps.entityConcurrent.has_value());
}
SECTION("partial population — only set keys populate") {
auto configPath = writeToml("partial.toml", R"(
[tuning.post_ingest]
embed_concurrent = 3
kg_concurrent = 5
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto caps = ConfigResolver::resolvePostIngestCaps();
CHECK_FALSE(caps.totalConcurrent.has_value());
REQUIRE(caps.embedConcurrent.has_value());
CHECK(*caps.embedConcurrent == 3u);
CHECK_FALSE(caps.extractionConcurrent.has_value());
REQUIRE(caps.kgConcurrent.has_value());
CHECK(*caps.kgConcurrent == 5u);
CHECK_FALSE(caps.symbolConcurrent.has_value());
CHECK_FALSE(caps.entityConcurrent.has_value());
CHECK_FALSE(caps.titleConcurrent.has_value());
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveSimeonEncoderPolicy reads [embeddings.simeon]",
"[daemon][components][config][simeon][catch2]") {
SECTION("missing section returns all nullopt") {
auto configPath = writeToml("no_simeon.toml", R"(
[daemon]
log_level = "info"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
for (auto* env :
{"YAMS_SIMEON_NGRAM_MODE", "YAMS_SIMEON_NGRAM_MIN", "YAMS_SIMEON_NGRAM_MAX",
"YAMS_SIMEON_SKETCH_DIM", "YAMS_SIMEON_OUTPUT_DIM", "YAMS_SIMEON_PROJECTION",
"YAMS_SIMEON_PQ_BYTES"}) {
unsetenv(env);
}
auto policy = ConfigResolver::resolveSimeonEncoderPolicy();
CHECK_FALSE(policy.ngramMode.has_value());
CHECK_FALSE(policy.ngramMin.has_value());
CHECK_FALSE(policy.ngramMax.has_value());
CHECK_FALSE(policy.sketchDim.has_value());
CHECK_FALSE(policy.outputDim.has_value());
CHECK_FALSE(policy.projection.has_value());
CHECK_FALSE(policy.l2Normalize.has_value());
CHECK_FALSE(policy.pqBytes.has_value());
}
SECTION("populated TOML section fills every field") {
auto configPath = writeToml("simeon_encoder.toml", R"(
[embeddings.simeon]
ngram_mode = "char_and_word"
ngram_min = 3
ngram_max = 5
sketch_dim = 4096
output_dim = 384
projection = "fwht"
l2_normalize = true
pq_bytes = 64
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
for (auto* env :
{"YAMS_SIMEON_NGRAM_MODE", "YAMS_SIMEON_NGRAM_MIN", "YAMS_SIMEON_NGRAM_MAX",
"YAMS_SIMEON_SKETCH_DIM", "YAMS_SIMEON_OUTPUT_DIM", "YAMS_SIMEON_PROJECTION",
"YAMS_SIMEON_PQ_BYTES"}) {
unsetenv(env);
}
auto policy = ConfigResolver::resolveSimeonEncoderPolicy();
REQUIRE(policy.ngramMode.has_value());
CHECK(*policy.ngramMode == "char_and_word");
REQUIRE(policy.ngramMin.has_value());
CHECK(*policy.ngramMin == 3u);
REQUIRE(policy.ngramMax.has_value());
CHECK(*policy.ngramMax == 5u);
REQUIRE(policy.sketchDim.has_value());
CHECK(*policy.sketchDim == 4096u);
REQUIRE(policy.outputDim.has_value());
CHECK(*policy.outputDim == 384u);
REQUIRE(policy.projection.has_value());
CHECK(*policy.projection == "fwht");
REQUIRE(policy.l2Normalize.has_value());
CHECK(*policy.l2Normalize);
REQUIRE(policy.pqBytes.has_value());
CHECK(*policy.pqBytes == 64u);
}
SECTION("env vars override TOML values") {
auto configPath = writeToml("override_encoder.toml", R"(
[embeddings.simeon]
projection = "achlioptas_sparse"
sketch_dim = 2048
output_dim = 256
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
EnvGuard proj("YAMS_SIMEON_PROJECTION", "fwht");
EnvGuard sketch("YAMS_SIMEON_SKETCH_DIM", "8192");
EnvGuard out("YAMS_SIMEON_OUTPUT_DIM", "1024");
auto policy = ConfigResolver::resolveSimeonEncoderPolicy();
REQUIRE(policy.projection.has_value());
CHECK(*policy.projection == "fwht");
REQUIRE(policy.sketchDim.has_value());
CHECK(*policy.sketchDim == 8192u);
REQUIRE(policy.outputDim.has_value());
CHECK(*policy.outputDim == 1024u);
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveSimeonBm25Policy reads [embeddings.simeon.bm25]",
"[daemon][components][config][simeon][catch2]") {
SECTION("missing section returns all nullopt") {
auto configPath = writeToml("no_bm25.toml", R"(
[daemon]
log_level = "info"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
for (auto* env : {"YAMS_SIMEON_BM25_ENABLED", "YAMS_SIMEON_BM25_VARIANT",
"YAMS_SIMEON_BM25_SUBWORD_GAMMA", "YAMS_SIMEON_BM25_MAX_CORPUS_DOCS"}) {
unsetenv(env);
}
auto policy = ConfigResolver::resolveSimeonBm25Policy();
CHECK_FALSE(policy.enabled.has_value());
CHECK_FALSE(policy.variant.has_value());
CHECK_FALSE(policy.subwordGamma.has_value());
CHECK_FALSE(policy.maxCorpusDocs.has_value());
}
SECTION("populated TOML section fills every field") {
auto configPath = writeToml("bm25.toml", R"(
[embeddings.simeon.bm25]
enabled = true
variant = "sab_smooth"
subword_gamma = 5.0
max_corpus_docs = 200000
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
for (auto* env : {"YAMS_SIMEON_BM25_ENABLED", "YAMS_SIMEON_BM25_VARIANT",
"YAMS_SIMEON_BM25_SUBWORD_GAMMA", "YAMS_SIMEON_BM25_MAX_CORPUS_DOCS"}) {
unsetenv(env);
}
auto policy = ConfigResolver::resolveSimeonBm25Policy();
REQUIRE(policy.enabled.has_value());
CHECK(*policy.enabled);
REQUIRE(policy.variant.has_value());
CHECK(*policy.variant == "sab_smooth");
REQUIRE(policy.subwordGamma.has_value());
CHECK(*policy.subwordGamma == 5.0f);
REQUIRE(policy.maxCorpusDocs.has_value());
CHECK(*policy.maxCorpusDocs == 200000u);
}
SECTION("env vars override TOML values") {
auto configPath = writeToml("override_bm25.toml", R"(
[embeddings.simeon.bm25]
variant = "atire"
subword_gamma = 3.0
max_corpus_docs = 50000
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
EnvGuard variantEnv("YAMS_SIMEON_BM25_VARIANT", "sab_smooth");
EnvGuard gammaEnv("YAMS_SIMEON_BM25_SUBWORD_GAMMA", "7.5");
EnvGuard capEnv("YAMS_SIMEON_BM25_MAX_CORPUS_DOCS", "123456");
auto policy = ConfigResolver::resolveSimeonBm25Policy();
REQUIRE(policy.variant.has_value());
CHECK(*policy.variant == "sab_smooth");
REQUIRE(policy.subwordGamma.has_value());
CHECK(*policy.subwordGamma == 7.5f);
REQUIRE(policy.maxCorpusDocs.has_value());
CHECK(*policy.maxCorpusDocs == 123456u);
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveRerankerBackendPolicy reads [search]",
"[daemon][components][config][reranker][catch2]") {
SECTION("returns unset when not configured") {
auto configPath = writeToml("reranker_default.toml", R"(
[search]
default_limit = 10
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveRerankerBackendPolicy();
CHECK_FALSE(policy.backend.has_value());
}
SECTION("reads explicit reranker backend") {
auto configPath = writeToml("reranker_backend.toml", R"(
[search]
reranker_backend = "colbert"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveRerankerBackendPolicy();
REQUIRE(policy.backend.has_value());
CHECK(*policy.backend == "colbert");
}
SECTION("normalizes reranker backend to lowercase") {
auto configPath = writeToml("reranker_backend_upper.toml", R"(
[search]
reranker_backend = "ONNX"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
auto policy = ConfigResolver::resolveRerankerBackendPolicy();
REQUIRE(policy.backend.has_value());
CHECK(*policy.backend == "onnx");
}
}
TEST_CASE_METHOD(ConfigResolverFixture,
"ConfigResolver::resolveInstrumentationPolicy applies memory profile",
"[daemon][components][config][instrumentation][catch2]") {
SECTION("auto profile activates when MallocStackLoggingNoCompact is present") {
auto configPath = writeToml("instrumentation_auto.toml", R"(
[daemon]
log_level = "info"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
EnvGuard msl("MallocStackLoggingNoCompact", "1");
EnvGuard mslCompact("MallocStackLogging", "0");
auto policy = ConfigResolver::resolveInstrumentationPolicy();
CHECK(policy.profile == "auto");
CHECK(policy.memoryProfileActive);
CHECK(policy.suppressAutoRepair);
CHECK(policy.suppressSimeonLexicalBuild);
CHECK(policy.suppressVectorIndexBuild);
}
SECTION("normal profile disables auto MSL suppression") {
auto configPath = writeToml("instrumentation_normal.toml", R"(
[daemon.instrumentation]
profile = "normal"
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
EnvGuard msl("MallocStackLoggingNoCompact", "1");
auto policy = ConfigResolver::resolveInstrumentationPolicy();
CHECK(policy.profile == "normal");
CHECK_FALSE(policy.memoryProfileActive);
CHECK_FALSE(policy.suppressAutoRepair);
CHECK_FALSE(policy.suppressSimeonLexicalBuild);
CHECK_FALSE(policy.suppressVectorIndexBuild);
}
SECTION("memory profile supports explicit overrides and stack-log threshold") {
auto configPath = writeToml("instrumentation_memory.toml", R"(
[daemon.instrumentation]
profile = "memory"
suppress_auto_repair = false
suppress_simeon_lexical_build = true
suppress_vector_index_build = true
msl_stack_log_warn_mb = 1536
)");
EnvGuard cfg("YAMS_CONFIG_PATH", configPath.string());
EnvGuard msl("MallocStackLogging", "0");
EnvGuard mslCompact("MallocStackLoggingNoCompact", "0");
auto policy = ConfigResolver::resolveInstrumentationPolicy();
CHECK(policy.profile == "memory");
CHECK(policy.memoryProfileActive);
CHECK_FALSE(policy.suppressAutoRepair);
CHECK(policy.suppressSimeonLexicalBuild);
CHECK(policy.suppressVectorIndexBuild);
CHECK(policy.mslStackLogWarnBytes == 1536ULL * 1024ULL * 1024ULL);
}
}