yams/tests/unit/app/services/grep_service_test.cpp

#include <catch2/catch_test_macros.hpp>
#include <yams/api/content_store_builder.h>
#include <yams/app/services/retrieval_path_policy.hpp>
#include <yams/app/services/services.hpp>
#include <yams/compat/unistd.h>
#include <yams/metadata/connection_pool.h>
#include <yams/metadata/database.h>
#include <yams/metadata/knowledge_graph_store.h>
#include <yams/metadata/metadata_repository.h>
#include <yams/metadata/migration.h>
#include <yams/metadata/query_helpers.h>

#include <algorithm>
#include <filesystem>
#include <fstream>
#include <ranges>

#include "common/test_helpers_catch2.h"

#if defined(_WIN32) && __has_include(<onnxruntime_c_api.h>)
#include <onnxruntime_c_api.h>
#define YAMS_ORT_API_VERSION ORT_API_VERSION
#else
#define YAMS_ORT_API_VERSION 0
#endif

#ifdef _WIN32
TEST_CASE("GrepService - Windows disabled due to ONNX runtime instability",
          "[grep][windows][skip]") {
    SUCCEED("GrepService tests are disabled on Windows pending ONNX Runtime cleanup fixes; this "
            "keeps the suite green.");
}
#else

#define SKIP_GREP_ON_WINDOWS() ((void)0)

using namespace yams;
using namespace yams::app::services;
using namespace yams::metadata;
using namespace yams::api;

namespace {

class FlakyMetadataRepository final : public MetadataRepository {
public:
    explicit FlakyMetadataRepository(ConnectionPool& pool) : MetadataRepository(pool) {}

    void setGetAllMetadataFailures(std::size_t count) { getAllMetadataFailures_ = count; }
    void setQueryDocumentsFailures(std::size_t count) { queryFailures_ = count; }
    void setFindDocumentsByTagsFailures(std::size_t count) { findDocumentsByTagsFailures_ = count; }

    Result<std::unordered_map<std::string, MetadataValue>>
    getAllMetadata(int64_t documentId) override {
        if (yams::test::consume_failure(getAllMetadataFailures_)) {
            return Error{ErrorCode::NotInitialized, "metadata warming up"};
        }
        return MetadataRepository::getAllMetadata(documentId);
    }

    Result<std::vector<DocumentInfo>>
    queryDocuments(const metadata::DocumentQueryOptions& options) override {
        if (yams::test::consume_failure(queryFailures_)) {
            return Error{ErrorCode::DatabaseError, "database is locked"};
        }
        return MetadataRepository::queryDocuments(options);
    }

    Result<std::unordered_map<int64_t, std::unordered_map<std::string, MetadataValue>>>
    getMetadataForDocuments(std::span<const int64_t> documentIds) override {
        if (yams::test::consume_failure(getAllMetadataFailures_)) {
            return Error{ErrorCode::NotInitialized, "metadata warming up"};
        }
        return MetadataRepository::getMetadataForDocuments(documentIds);
    }

    Result<std::vector<DocumentInfo>> findDocumentsByTags(const std::vector<std::string>& tags,
                                                          bool matchAll = false) override {
        if (yams::test::consume_failure(findDocumentsByTagsFailures_)) {
            return Error{ErrorCode::NotInitialized, "tag index warming up"};
        }
        return MetadataRepository::findDocumentsByTags(tags, matchAll);
    }

    Result<std::optional<DocumentInfo>> findDocumentByExactPath(const std::string& path) override {
        return MetadataRepository::findDocumentByExactPath(path);
    }

private:
    std::size_t getAllMetadataFailures_{0};
    std::size_t queryFailures_{0};
    std::size_t findDocumentsByTagsFailures_{0};
};

class RetrieveBytesFailStore final : public IContentStore {
public:
    explicit RetrieveBytesFailStore(std::shared_ptr<IContentStore> inner)
        : inner_(std::move(inner)) {}

    Result<StoreResult> store(const std::filesystem::path& path, const ContentMetadata& metadata,
                              api::ProgressCallback progress) override {
        return inner_->store(path, metadata, progress);
    }

    Result<RetrieveResult> retrieve(const std::string& hash,
                                    const std::filesystem::path& outputPath,
                                    api::ProgressCallback progress) override {
        return inner_->retrieve(hash, outputPath, progress);
    }

    Result<StoreResult> storeStream(std::istream& stream, const ContentMetadata& metadata,
                                    api::ProgressCallback progress) override {
        return inner_->storeStream(stream, metadata, progress);
    }

    Result<RetrieveResult> retrieveStream(const std::string& hash, std::ostream& output,
                                          api::ProgressCallback progress) override {
        return inner_->retrieveStream(hash, output, progress);
    }

    Result<StoreResult> storeBytes(std::span<const std::byte> data,
                                   const ContentMetadata& metadata) override {
        return inner_->storeBytes(data, metadata);
    }

    Result<std::vector<std::byte>> retrieveBytes(const std::string&) override {
        return Error{ErrorCode::NotFound, "retrieveBytes disabled for test"};
    }

    Result<std::vector<std::byte>> retrieveBytesPrefix(const std::string&, std::size_t) override {
        return Error{ErrorCode::NotFound, "retrieveBytesPrefix disabled for test"};
    }

    Result<RawContent> retrieveRaw(const std::string& hash) override {
        return inner_->retrieveRaw(hash);
    }

    std::future<Result<RawContent>> retrieveRawAsync(const std::string& hash) override {
        return inner_->retrieveRawAsync(hash);
    }

    Result<bool> exists(const std::string& hash) const override { return inner_->exists(hash); }

    Result<bool> remove(const std::string& hash) override { return inner_->remove(hash); }

    Result<ContentMetadata> getMetadata(const std::string& hash) const override {
        return inner_->getMetadata(hash);
    }

    Result<void> updateMetadata(const std::string& hash, const ContentMetadata& metadata) override {
        return inner_->updateMetadata(hash, metadata);
    }

    std::vector<Result<StoreResult>>
    storeBatch(const std::vector<std::filesystem::path>& paths,
               const std::vector<ContentMetadata>& metadata) override {
        return inner_->storeBatch(paths, metadata);
    }

    std::vector<Result<bool>> removeBatch(const std::vector<std::string>& hashes) override {
        return inner_->removeBatch(hashes);
    }

    ContentStoreStats getStats() const override { return inner_->getStats(); }

    HealthStatus checkHealth() const override { return inner_->checkHealth(); }

    Result<void> verify(api::ProgressCallback progress) override {
        return inner_->verify(progress);
    }

    Result<void> compact(api::ProgressCallback progress) override {
        return inner_->compact(progress);
    }

    Result<void> garbageCollect(api::ProgressCallback progress) override {
        return inner_->garbageCollect(progress);
    }

private:
    std::shared_ptr<IContentStore> inner_;
};

struct GrepFixture {
    GrepFixture() {
        tmpDir_ =
            std::filesystem::temp_directory_path() / ("grep_test_" + std::to_string(::getpid()));
        std::filesystem::create_directories(tmpDir_);

        db_ = std::make_unique<Database>();
        auto dbPath = tmpDir_ / "yams.db";
        REQUIRE(db_->open(dbPath.string(), ConnectionMode::Create));

        pool_ = std::make_unique<ConnectionPool>(dbPath.string(), ConnectionPoolConfig{});
        repo_ = std::make_shared<MetadataRepository>(*pool_);

        KnowledgeGraphStoreConfig kgConfig;
        kgConfig.enable_wal = false;
        auto kgResult = makeSqliteKnowledgeGraphStore(*pool_, kgConfig);
        REQUIRE(kgResult.has_value());
        kgStore_ = std::shared_ptr<KnowledgeGraphStore>(std::move(kgResult).value());
        repo_->setKnowledgeGraphStore(kgStore_);

        MigrationManager mm(*db_);
        REQUIRE(mm.initialize());
        mm.registerMigrations(YamsMetadataMigrations::getAllMigrations());
        REQUIRE(mm.migrate());

        ContentStoreBuilder builder;
        auto storeRes = builder.withStoragePath(tmpDir_ / "storage")
                            .withCompression(false)
                            .withDeduplication(false)
                            .build();
        REQUIRE(storeRes);
        store_ = std::shared_ptr<IContentStore>(
            const_cast<std::unique_ptr<IContentStore>&>(storeRes.value()).release());

        ctx_.service_manager = nullptr;
        ctx_.store = store_;
        ctx_.metadataRepo = repo_;
        ctx_.searchEngine = nullptr;
        ctx_.kgStore = kgStore_;

        grepService_ = makeGrepService(ctx_);
    }

    ~GrepFixture() {
        grepService_.reset();
        ctx_.metadataRepo.reset();
        ctx_.store.reset();
        ctx_.searchEngine.reset();
        ctx_.kgStore.reset();
        repo_.reset();
        kgStore_.reset();
        pool_.reset();
        db_.reset();
        store_.reset();
        std::filesystem::remove_all(tmpDir_);
    }

    void addDocument(std::string_view name, std::string_view content) {
        auto path = tmpDir_ / name;
        std::filesystem::create_directories(path.parent_path());
        std::ofstream{path} << content;

        auto docService = makeDocumentService(ctx_);
        StoreDocumentRequest req;
        req.path = path.string();
        REQUIRE(docService->store(req));
    }

    void addHotContent(std::string_view name, std::string_view content,
                       std::string_view mimeType = "text/plain") {
        (void)mimeType;
        auto path = (tmpDir_ / name).string();
        auto docRes = repo_->findDocumentByExactPath(path);
        REQUIRE(docRes.has_value());
        REQUIRE(docRes.value().has_value());
        DocumentContent docContent;
        docContent.documentId = docRes.value()->id;
        docContent.contentText = std::string(content);
        docContent.contentLength = static_cast<int64_t>(docContent.contentText.size());
        docContent.extractionMethod = "test";
        docContent.language = "en";
        REQUIRE(repo_->insertContent(docContent));
    }

    auto grep(const GrepRequest& req) const { return grepService_->grep(req); }

    void seedRelationsForFile(const std::string& filePath, const std::string& hash) const {
        REQUIRE(kgStore_ != nullptr);
        REQUIRE_FALSE(filePath.empty());
        REQUIRE_FALSE(hash.empty());

        KGNode fileNode;
        fileNode.nodeKey = "path:file:" + filePath;
        fileNode.type = std::string("file");
        fileNode.label = filePath;
        auto fileNodeId = kgStore_->upsertNode(fileNode);
        REQUIRE(fileNodeId.has_value());

        KGNode docNode;
        docNode.nodeKey = "doc:" + hash;
        docNode.type = std::string("document");
        docNode.label = filePath;
        auto docNodeId = kgStore_->upsertNode(docNode);
        REQUIRE(docNodeId.has_value());

        KGNode symbolNode;
        symbolNode.nodeKey = "symbol:test:" + hash;
        symbolNode.type = std::string("symbol");
        symbolNode.label = std::string("SeededSymbol");
        auto symbolNodeId = kgStore_->upsertNode(symbolNode);
        REQUIRE(symbolNodeId.has_value());

        KGNode moduleNode;
        moduleNode.nodeKey = "module:test:" + hash;
        moduleNode.type = std::string("module");
        moduleNode.label = std::string("SeededModule");
        auto moduleNodeId = kgStore_->upsertNode(moduleNode);
        REQUIRE(moduleNodeId.has_value());

        KGEdge versionEdge;
        versionEdge.srcNodeId = fileNodeId.value();
        versionEdge.dstNodeId = docNodeId.value();
        versionEdge.relation = "has_version";
        REQUIRE(kgStore_->addEdge(versionEdge).has_value());

        KGEdge usesOne;
        usesOne.srcNodeId = fileNodeId.value();
        usesOne.dstNodeId = symbolNodeId.value();
        usesOne.relation = "uses";
        REQUIRE(kgStore_->addEdge(usesOne).has_value());

        KGEdge usesTwo;
        usesTwo.srcNodeId = fileNodeId.value();
        usesTwo.dstNodeId = moduleNodeId.value();
        usesTwo.relation = "uses";
        REQUIRE(kgStore_->addEdge(usesTwo).has_value());
    }

    std::filesystem::path tmpDir_;
    std::unique_ptr<Database> db_;
    std::unique_ptr<ConnectionPool> pool_;
    std::shared_ptr<MetadataRepository> repo_;
    std::shared_ptr<KnowledgeGraphStore> kgStore_;
    std::shared_ptr<IContentStore> store_;
    AppContext ctx_;
    std::shared_ptr<IGrepService> grepService_;
};

} // namespace

TEST_CASE("GrepService - Basic Functionality", "[grep][service][basic]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("a.txt", "alpha beta gamma\nhello world\nregex target\n");
    fixture.addDocument("b.txt", "semantic related content about programming\n");

    SECTION("Regex-only mode finds pattern matches") {
        GrepRequest req;
        req.pattern = "regex";
        req.regexOnly = true;
        req.lineNumbers = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().totalMatches > 0);
        CHECK(res.value().semanticMatches == 0);
        CHECK(res.value().executionTimeMs >= 0);
        REQUIRE(res.value().searchStats.contains("metadata_operations"));
        CHECK(res.value().searchStats.at("metadata_operations") != "0");
        REQUIRE(res.value().searchStats.contains("latency_ms"));
        REQUIRE(res.value().searchStats.contains("match_render_us"));
        REQUIRE(res.value().searchStats.contains("matches_rendered"));
        CHECK(std::stoll(res.value().searchStats.at("match_render_us")) >= 0);
        CHECK(std::stoll(res.value().searchStats.at("matches_rendered")) > 0);
    }

    SECTION("Relation metadata is added for matched files") {
        auto docRes = fixture.repo_->findDocumentByExactPath((fixture.tmpDir_ / "a.txt").string());
        REQUIRE(docRes.has_value());
        REQUIRE(docRes.value().has_value());
        fixture.seedRelationsForFile(docRes.value()->filePath, docRes.value()->sha256Hash);

        GrepRequest req;
        req.pattern = "regex";
        req.regexOnly = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        auto it = std::find_if(res.value().results.begin(), res.value().results.end(),
                               [](const GrepFileResult& file) { return file.fileName == "a.txt"; });
        REQUIRE(it != res.value().results.end());

        REQUIRE(it->metadata.contains("relation_count"));
        CHECK(it->metadata.at("relation_count") == "3");
        REQUIRE(it->metadata.contains("relation_types"));
        CHECK(it->metadata.at("relation_types").find("uses:2") != std::string::npos);
        REQUIRE(it->metadata.contains("relation_summary"));
        CHECK(it->metadata.at("relation_summary").find("uses(2)") != std::string::npos);
        REQUIRE(it->metadata.contains("primary_relation"));
        CHECK(it->metadata.at("primary_relation") == "uses");

        REQUIRE(res.value().searchStats.contains("relation_results_enriched"));
        CHECK(res.value().searchStats.at("relation_results_enriched") != "0");
    }

    SECTION("Literal short pattern matches with boundaries") {
        fixture.addDocument("c.txt", "alpha beta alpha\nalpha-beta\n");

        GrepRequest req;
        req.pattern = "alpha";
        req.literalText = true;
        req.word = true;
        // Scope grep to only c.txt to avoid counting matches in a.txt
        req.paths = {(fixture.tmpDir_ / "c.txt").string()};

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().totalMatches == 2);
        REQUIRE(res.value().searchStats.contains("retrieval_mode"));
        CHECK(res.value().searchStats.at("retrieval_mode") == "literal");
        CHECK(res.value().searchStats.at("fts_literal_filter") == "off");
    }

    SECTION("Hybrid mode includes semantic suggestions") {
// Skip semantic search on Windows - ONNX initialization hangs
#ifdef _WIN32
        SKIP("Semantic search unavailable on Windows - ONNX init issues");
#endif

        GrepRequest req;

        req.pattern = "how to learn programming";

        req.regexOnly = false;

        req.semanticLimit = 2;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().semanticMatches >= 0);
        REQUIRE(res.value().searchStats.contains("retrieval_mode"));
        CHECK(res.value().searchStats.at("retrieval_mode") == "semantic");
    }

    SECTION("Path filters normalize relative segments") {
        auto res = fixture.grep({
            .pattern = "hello",
            .paths = {(fixture.tmpDir_ / "./a.txt").string()},
        });

        REQUIRE(res);
        REQUIRE_FALSE(res.value().results.empty());
        CHECK(res.value().results.front().fileName == "a.txt");
        CHECK(res.value().totalMatches > 0);
    }

    SECTION("Subpath filters match correctly") {
        std::filesystem::create_directory(fixture.tmpDir_ / "sub");
        fixture.addDocument("sub/c.txt", "subpath content");

        auto res = fixture.grep({
            .pattern = "subpath",
            .paths = {"sub/c.txt"},
        });

        REQUIRE(res);
        REQUIRE_FALSE(res.value().results.empty());
        CHECK(res.value().results.front().fileName == "c.txt");
        CHECK(res.value().totalMatches > 0);
    }
}

TEST_CASE("GrepService - Output Modes", "[grep][service][modes]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("a.txt", "programming content\n");

    SECTION("Paths-only and file-set projections are deterministic") {
        fixture.addDocument("c.txt", "needle content c\n");
        fixture.addDocument("b.txt", "needle content b\n");
        fixture.addDocument("z.txt", "haystack only\n");

        GrepRequest req;
        req.pattern = "needle";
        req.regexOnly = true;
        req.pathsOnly = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        const std::vector<std::string> expectedWith = {
            std::filesystem::weakly_canonical(fixture.tmpDir_ / "b.txt").string(),
            std::filesystem::weakly_canonical(fixture.tmpDir_ / "c.txt").string(),
        };
        CHECK(res.value().filesWith == expectedWith);
        CHECK(res.value().pathsOnly == expectedWith);

        GrepRequest withoutReq;
        withoutReq.pattern = "needle";
        withoutReq.regexOnly = true;
        withoutReq.filesWithoutMatch = true;
        auto withoutRes = fixture.grep(withoutReq);

        REQUIRE(withoutRes);
        const std::vector<std::string> expectedWithout = {
            std::filesystem::weakly_canonical(fixture.tmpDir_ / "a.txt").string(),
            std::filesystem::weakly_canonical(fixture.tmpDir_ / "z.txt").string(),
        };
        CHECK(withoutRes.value().filesWithout == expectedWithout);
    }

    SECTION("Count mode allows semantic suggestions") {
// Skip semantic search on Windows - ONNX initialization hangs
#ifdef _WIN32
        SKIP("Semantic search unavailable on Windows - ONNX init issues");
#endif

        GrepRequest req;

        req.pattern = "programming";

        req.regexOnly = false;

        req.semanticLimit = 2;

        req.count = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().semanticMatches >= 0);
    }

    SECTION("Files-only mode allows semantic suggestions") {
// Skip semantic search on Windows - ONNX initialization hangs
#ifdef _WIN32
        SKIP("Semantic search unavailable on Windows - ONNX init issues");
#endif

        GrepRequest req;

        req.pattern = "programming";

        req.regexOnly = false;

        req.semanticLimit = 2;

        req.filesWithMatches = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().semanticMatches >= 0);
    }

    SECTION("Paths-only mode allows semantic suggestions") {
// Skip semantic search on Windows - ONNX initialization hangs
#ifdef _WIN32
        SKIP("Semantic search unavailable on Windows - ONNX init issues");
#endif

        GrepRequest req;

        req.pattern = "programming";

        req.regexOnly = false;

        req.semanticLimit = 2;

        req.pathsOnly = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().semanticMatches >= 0);
    }
}

TEST_CASE("GrepService - Error Handling", "[grep][service][reliability]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("a.txt", "alpha beta gamma\n");

    SECTION("Retries transient metadata errors") {
        auto flakyRepo = std::make_shared<FlakyMetadataRepository>(*fixture.pool_);
        auto docsRes = metadata::queryDocumentsByPattern(*flakyRepo, "%");
        REQUIRE(docsRes);
        REQUIRE_FALSE(docsRes.value().empty());

        const auto docId = docsRes.value().front().id;
        REQUIRE(flakyRepo->setMetadata(docId, "force_cold", MetadataValue("true")));

        flakyRepo->setGetAllMetadataFailures(1);
        fixture.repo_ = flakyRepo;
        fixture.ctx_.metadataRepo = flakyRepo;
        fixture.grepService_ = makeGrepService(fixture.ctx_);

        GrepRequest req;

        req.pattern = "alpha";

        req.literalText = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().totalMatches > 0);
    }

    SECTION("Propagates tag query failures after retry exhaustion") {
        auto flakyRepo = std::make_shared<FlakyMetadataRepository>(*fixture.pool_);
        auto docsRes = metadata::queryDocumentsByPattern(*flakyRepo, "%");
        REQUIRE(docsRes);
        REQUIRE_FALSE(docsRes.value().empty());

        const auto docId = docsRes.value().front().id;
        REQUIRE(
            flakyRepo->setMetadata(docId, "tag:ready_flag", MetadataValue(std::string("true"))));

        flakyRepo->setFindDocumentsByTagsFailures(8);
        fixture.repo_ = flakyRepo;
        fixture.ctx_.metadataRepo = flakyRepo;
        fixture.grepService_ = makeGrepService(fixture.ctx_);

        auto res = fixture.grep({
            .pattern = "alpha",
            .literalText = true,
            .tags = {"ready_flag"},
        });

        REQUIRE_FALSE(res);
        CHECK(res.error().code == ErrorCode::NotInitialized);
        CHECK_FALSE(res.error().message.empty());
    }

    SECTION("Tag-filtered grep tolerates metadata warming when tag index query succeeds") {
        auto flakyRepo = std::make_shared<FlakyMetadataRepository>(*fixture.pool_);
        auto docsRes = metadata::queryDocumentsByPattern(*flakyRepo, "%");
        REQUIRE(docsRes);
        REQUIRE_FALSE(docsRes.value().empty());

        const auto docId = docsRes.value().front().id;
        REQUIRE(
            flakyRepo->setMetadata(docId, "tag:ready_flag", MetadataValue(std::string("true"))));

        flakyRepo->setGetAllMetadataFailures(8);
        fixture.repo_ = flakyRepo;
        fixture.ctx_.metadataRepo = flakyRepo;
        fixture.grepService_ = makeGrepService(fixture.ctx_);

        auto res = fixture.grep({
            .pattern = "alpha",
            .literalText = true,
            .tags = {"ready_flag"},
        });

        REQUIRE(res);
        CHECK(res.value().totalMatches > 0);
    }
}

TEST_CASE("GrepService - Falls back to extracted content when blob unavailable",
          "[grep][service][session]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("session_hot.txt", "alpha from metadata\n");
    fixture.addHotContent("session_hot.txt", "alpha from metadata\n");

    fixture.store_ = std::make_shared<RetrieveBytesFailStore>(fixture.store_);
    fixture.ctx_.store = fixture.store_;
    fixture.grepService_ = makeGrepService(fixture.ctx_);

    auto coldRes = fixture.grep({
        .pattern = "alpha",
        .literalText = true,
        .useSession = false,
    });

    REQUIRE(coldRes);
    CHECK(coldRes.value().totalMatches == 1);

    auto sessionRes = fixture.grep({
        .pattern = "alpha",
        .literalText = true,
        .useSession = true,
    });

    REQUIRE(sessionRes);
    CHECK(sessionRes.value().totalMatches == 1);
    REQUIRE(sessionRes.value().filesWith.size() == 1);
    CHECK(sessionRes.value().filesWith.front().find("session_hot.txt") != std::string::npos);
}

TEST_CASE("GrepService - Path-scoped grep prefers latest indexed version",
          "[grep][service][versioning]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    yams::test::ScopedEnvVar hotCap("YAMS_GREP_MAX_DOCS_HOT", std::string("1"));
    yams::test::ScopedEnvVar coldCap("YAMS_GREP_MAX_DOCS_COLD", std::string("1"));

    fixture.addDocument("versioned.txt", "old-token only\n");
    fixture.addDocument("versioned.txt", "new-token only\n");

    auto docs = metadata::queryDocumentsByPattern(*fixture.repo_, "%versioned.txt");
    REQUIRE(docs);
    REQUIRE(docs.value().size() >= 2);

    auto res = fixture.grep({
        .pattern = "new-token",
        .paths = {"versioned.txt"},
        .literalText = true,
    });

    REQUIRE(res);
    CHECK(res.value().totalMatches == 1);
    REQUIRE(res.value().filesWith.size() == 1);
    CHECK(res.value().filesWith.front().find("versioned.txt") != std::string::npos);
}

TEST_CASE("GrepService - Broad regex grep does not stop at candidate cap",
          "[grep][service][versioning]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    yams::test::ScopedEnvVar hotCap("YAMS_GREP_MAX_DOCS_HOT", std::string("2"));
    yams::test::ScopedEnvVar coldCap("YAMS_GREP_MAX_DOCS_COLD", std::string("0"));
    yams::test::ScopedEnvVar budget("YAMS_GREP_TIME_BUDGET_MS", std::string("10000"));

    fixture.addDocument("old_match.txt", "rare_token_42\n");
    fixture.addDocument("newer_a.txt", "alpha\n");
    fixture.addDocument("newer_b.txt", "beta\n");
    fixture.addDocument("newer_c.txt", "gamma\n");

    auto res = fixture.grep({
        .pattern = "rare_token_[0-9]+",
        .regexOnly = true,
    });

    REQUIRE(res);
    CHECK(res.value().totalMatches == 1);
    REQUIRE(res.value().filesWith.size() == 1);
    CHECK(res.value().filesWith.front().find("old_match.txt") != std::string::npos);
}

TEST_CASE("GrepService - Workspace scope rejects generated retrieval artifacts",
          "[grep][service][scope]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("src/authored.cpp", "shared_scope_token authored\n");
    fixture.addDocument("meson.build", "shared_scope_token build definition\n");
    fixture.addDocument("bin/tool.py", "shared_scope_token tool\n");
    fixture.addDocument("build/fold/findings.sarif", "shared_scope_token fold\n");
    fixture.addDocument(".pi/tasks/agent.json", "shared_scope_token agent\n");
    fixture.addDocument("audit_artifacts/report.json", "shared_scope_token audit\n");

    auto res = fixture.grep({
        .pattern = "shared_scope_token.*",
        .scopePathPrefix = fixture.tmpDir_.string(),
        .regexOnly = true,
        .useSession = false,
    });

    REQUIRE(res);
    REQUIRE(res.value().results.size() == 3);
    CHECK(std::ranges::none_of(res.value().results, [](const auto& result) {
        return utils::isGeneratedWorkspacePath(result.file);
    }));
    CHECK(res.value().searchStats.at("generated_candidates_rejected") == "3");
}

TEST_CASE("GrepService - Edge Cases", "[grep][service][edge]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("edge.txt", "foo bar\nfoo-bar\nfoo_bar\n(hello) and (foo)\nHELLO world\n");

    SECTION("Word boundary excludes hyphens") {
        GrepRequest req;

        req.pattern = "foo";

        req.word = true;

        req.lineNumbers = true;

        req.withFilename = false;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().totalMatches > 0);

        for (const auto& fileResult : res.value().results) {
            for (const auto& match : fileResult.matches) {
                CHECK(match.line.find("foo-") == std::string::npos);
            }
        }
    }

    SECTION("Literal parentheses not interpreted as regex") {
        auto literalRes = fixture.grep({
            .pattern = "(foo)",
            .literalText = true,
        });

        REQUIRE(literalRes);
        CHECK(literalRes.value().totalMatches > 0);

        auto regexRes = fixture.grep({
            .pattern = "(foo)",
            .literalText = false,
        });

        REQUIRE(regexRes);
        CHECK(regexRes.value().totalMatches > 0);
    }

    SECTION("Ignore case matches uppercase and lowercase") {
        GrepRequest req;

        req.pattern = "HELLO";

        req.ignoreCase = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        CHECK(res.value().totalMatches > 0);
    }
}

TEST_CASE("GrepService - Unicode Support", "[grep][service][unicode]") {
    SKIP_GREP_ON_WINDOWS();

    GrepFixture fixture;
    fixture.addDocument("unicode.txt",
                        "café\nCAFÉ\nnaïve\n(naïve)\nsmile 😊 end\n東京大学 CJK line\n");

    SECTION("Literal unicode patterns match correctly") {
        auto cafeRes = fixture.grep({
            .pattern = "café",
            .literalText = true,
        });
        REQUIRE(cafeRes);
        CHECK(cafeRes.value().totalMatches > 0);

        auto emojiRes = fixture.grep({
            .pattern = "😊",
            .literalText = true,
        });
        REQUIRE(emojiRes);
        CHECK(emojiRes.value().totalMatches > 0);

        auto diacriticRes = fixture.grep({
            .pattern = "(naïve)",
            .literalText = true,
        });
        REQUIRE(diacriticRes);
        CHECK(diacriticRes.value().totalMatches > 0);
    }

    SECTION("Case-insensitive unicode matching (best effort)") {
        GrepRequest req;

        req.pattern = "café";

        req.literalText = true;

        req.ignoreCase = true;

        auto res = fixture.grep(req);

        REQUIRE(res);
        if (res.value().totalMatches == 0) {
            SKIP("Unicode case-folding not available in this build");
        }
        CHECK(res.value().totalMatches > 0);
    }
}

#endif // _WIN32