yams/tests/unit/vector/simeon_backend_catch2_test.cpp

#include <catch2/catch_test_macros.hpp>

#include <yams/compat/unistd.h>
#include <yams/vector/embedding_generator.h>
#include <yams/vector/simeon_embedding_backend.h>

#include <array>
#include <cmath>
#include <cstdlib>
#include <string>
#include <utility>
#include <vector>
#include <catch2/matchers/catch_matchers_string.hpp>

#include "../../common/test_helpers_catch2.h"

using namespace yams::vector;
using Catch::Matchers::StartsWith;

namespace {

EmbeddingConfig make_cfg(size_t dim = 256) {
    EmbeddingConfig cfg;
    cfg.backend = EmbeddingConfig::Backend::Simeon;
    cfg.backend_is_resolved = true;
    cfg.embedding_dim = dim;
    cfg.normalize_embeddings = true;
    cfg.max_sequence_length = 512;
    return cfg;
}

} // namespace

TEST_CASE("SimeonBackend initializes and reports its dimension", "[vector][simeon]") {
    auto backend = makeSimeonBackend(make_cfg(384));
    REQUIRE(backend != nullptr);
    REQUIRE(backend->isAvailable());
    REQUIRE(backend->initialize());
    REQUIRE(backend->isInitialized());
    REQUIRE(backend->getEmbeddingDimension() == 384u);
    REQUIRE(backend->getBackendName() == "Simeon");
}

TEST_CASE("SimeonBackend fixed hash profile exposes the frozen retrieval space",
          "[vector][simeon][identity]") {
    auto cfg = make_cfg(1024);
    cfg.simeon_encoder_profile = EmbeddingConfig::SimeonEncoderProfile::FixedHash384;

    auto backend = makeSimeonBackend(cfg);
    REQUIRE(backend != nullptr);
    REQUIRE(backend->initialize());
    CHECK(backend->getEmbeddingDimension() == 384u);
    CHECK(backend->getEmbeddingSpaceIdentity() == "simeon-v1-384");
    CHECK(simeonEmbeddingSpaceIdentity(cfg) == "simeon-v1-384");
}

TEST_CASE("SimeonBackend produces deterministic, unit-norm embeddings", "[vector][simeon]") {
    auto backend = makeSimeonBackend(make_cfg(256));
    REQUIRE(backend->initialize());

    const std::string text = "The quick brown fox jumps over the lazy dog.";
    auto r1 = backend->generateEmbedding(text);
    auto r2 = backend->generateEmbedding(text);
    REQUIRE(r1);
    REQUIRE(r2);
    REQUIRE(r1.value().size() == 256u);
    REQUIRE(r1.value() == r2.value());

    double acc = 0.0;
    for (float f : r1.value())
        acc += static_cast<double>(f) * static_cast<double>(f);
    REQUIRE(std::fabs(std::sqrt(acc) - 1.0) < 1e-4);
}

TEST_CASE("SimeonBackend batch matches per-item encoding", "[vector][simeon]") {
    auto backend = makeSimeonBackend(make_cfg(128));
    REQUIRE(backend->initialize());

    const std::vector<std::string> texts = {"alpha bravo charlie", "delta echo foxtrot",
                                            "golf hotel india"};
    auto per = std::vector<std::vector<float>>{};
    for (const auto& t : texts) {
        auto r = backend->generateEmbedding(t);
        REQUIRE(r);
        per.push_back(std::move(r).value());
    }
    auto batch = backend->generateEmbeddings(std::span<const std::string>(texts));
    REQUIRE(batch);
    REQUIRE(batch.value().size() == per.size());
    for (size_t i = 0; i < per.size(); ++i) {
        REQUIRE(batch.value()[i] == per[i]);
    }
}

TEST_CASE("EmbeddingGenerator trusts typed backend over ambient environment",
          "[vector][simeon][config]") {
    yams::test::ScopedEnvVar backend{"YAMS_EMBED_BACKEND", std::string{"daemon"}};

    EmbeddingConfig cfg;
    cfg.backend = EmbeddingConfig::Backend::Simeon;
    cfg.embedding_dim = 384;
    cfg.normalize_embeddings = true;
    EmbeddingGenerator gen(cfg);
    REQUIRE(gen.initialize());
    REQUIRE(gen.getBackendName() == "Simeon");
    REQUIRE(gen.getEmbeddingDimension() == 384u);

    auto v = gen.generateEmbedding("hello world");
    REQUIRE(v.size() == 384u);
}

TEST_CASE("EmbeddingGenerator selects Simeon when configured directly", "[vector][simeon]") {
    EmbeddingConfig cfg;
    cfg.backend = EmbeddingConfig::Backend::Simeon;
    cfg.backend_is_resolved = true;
    cfg.embedding_dim = 128;
    cfg.normalize_embeddings = true;

    EmbeddingGenerator gen(cfg);
    REQUIRE(gen.initialize());
    REQUIRE(gen.getBackendName() == "Simeon");
    REQUIRE(gen.getEmbeddingDimension() == 128u);
}

TEST_CASE("EmbeddingGenerator moved-from observers report neutral state",
          "[vector][embedding-generator][move][catch2]") {
    EmbeddingGenerator source(make_cfg(128));
    EmbeddingGenerator destination(std::move(source));

    CHECK_FALSE(source.isInitialized());
    CHECK((source.estimateTokenCount("moved-from") == 0));
    CHECK((source.getLastError().empty()));
    CHECK_FALSE(source.hasError());
    CHECK((destination.getConfig().embedding_dim == 128));
}

TEST_CASE("EmbeddingGenerator maps OnnxRuntime config to Daemon path", "[vector][daemon]") {
    EmbeddingConfig cfg;
    cfg.backend = EmbeddingConfig::Backend::OnnxRuntime;
    cfg.backend_is_resolved = true;

    EmbeddingGenerator gen(cfg);
    REQUIRE_THAT(gen.getBackendName(), StartsWith("Daemon"));
}

TEST_CASE("EmbeddingGenerator maps legacy Hybrid config to Daemon path", "[vector][daemon]") {
    EmbeddingConfig cfg;
    cfg.backend = EmbeddingConfig::Backend::Hybrid;
    cfg.backend_is_resolved = true;

    EmbeddingGenerator gen(cfg);
    REQUIRE_THAT(gen.getBackendName(), StartsWith("Daemon"));
}

TEST_CASE("EmbeddingGenerator selects Daemon when configured directly", "[vector][daemon]") {
    EmbeddingConfig cfg;
    cfg.backend = EmbeddingConfig::Backend::Daemon;
    cfg.backend_is_resolved = true;

    EmbeddingGenerator gen(cfg);
    REQUIRE_THAT(gen.getBackendName(), StartsWith("Daemon"));
}