ViewDesignEngine/tests/kbe/test_knowledge.cpp

#include <gtest/gtest.h>
#include "vde/kbe/knowledge_engine.h"
#include <cmath>
#include <vector>
#include <chrono>

using namespace vde::kbe;

// ═══════════════════════════════════════════════════════════
//  1. CheckMate 设计验证
// ═══════════════════════════════════════════════════════════

TEST(KBETest, CheckMate_DefaultRulesRegistered) {
    CheckMate cm;
    auto rules = cm.rules();
    // 默认注册至少 10 条规则
    EXPECT_GE(rules.size(), 10u);
}

TEST(KBETest, CheckMate_RunAllDefaultRules) {
    CheckMate cm;
    std::vector<Parameter> params = {
        {"diameter",       10.0},
        {"wall_thickness",  0.8},
        {"aspect_ratio",    5.0},
        {"density",      2700.0},
        {"safety_factor",   2.0},
    };
    auto results = cm.run_all(params);
    EXPECT_EQ(results.size(), cm.rules().size());

    auto summary = cm.summary(results);
    // 默认规则应全部通过（参数值在合法范围内）
    int total_failures = 0;
    for (const auto& [sev, count] : summary) total_failures += count;
    EXPECT_EQ(total_failures, 0);
}

TEST(KBETest, CheckMate_PositiveDimensionFails) {
    CheckMate cm;
    std::vector<Parameter> params = {
        {"diameter", -5.0},  // 非法：负值
    };
    auto result = cm.run_rule("R_GEO_POSITIVE_DIM", params);
    EXPECT_FALSE(result.passed);
    EXPECT_EQ(result.severity, CheckSeverity::ERROR);
}

TEST(KBETest, CheckMate_CustomCheck) {
    CheckMate cm;
    cm.add_check("R_GEO_POSITIVE_DIM", [](const std::vector<Parameter>&) {
        CheckResult r;
        r.rule_name = "R_GEO_POSITIVE_DIM";
        r.severity  = CheckSeverity::ERROR;
        r.passed    = false;
        r.message   = "Custom: negative dimension detected";
        return r;
    });
    auto result = cm.run_rule("R_GEO_POSITIVE_DIM", {});
    EXPECT_FALSE(result.passed);
    EXPECT_EQ(result.message, "Custom: negative dimension detected");
}

TEST(KBETest, CheckMate_CategoryRun) {
    CheckMate cm;
    std::vector<Parameter> params = {
        {"diameter", 10.0}, {"wall_thickness", 2.0}, {"safety_factor", 2.0},
    };
    auto geo_results = cm.run_category("geometry", params);
    // 应有几何规则产生结果
    EXPECT_GT(geo_results.size(), 0u);

    auto mat_results = cm.run_category("material", params);
    EXPECT_GT(mat_results.size(), 0u);
}

// ═══════════════════════════════════════════════════════════
//  2. RuleEngine IF-THEN 规则引擎
// ═══════════════════════════════════════════════════════════

TEST(KBETest, RuleEngine_AddAndGetParam) {
    RuleEngine engine;
    engine.set_param("length", 10.0);
    auto val = engine.get_param("length");
    ASSERT_TRUE(val.has_value());
    EXPECT_DOUBLE_EQ(std::get<double>(*val), 10.0);
}

TEST(KBETest, RuleEngine_SimpleInfer) {
    RuleEngine engine;
    engine.set_param("diameter", 5.0);
    engine.set_param("thickness", 1.0);

    // IF diameter > 3 THEN SET thickness = 2.0
    Rule r{"R1", "diameter > 3", "SET thickness = 2.0", 1, true};
    engine.add_rule(r);

    auto traces = engine.infer();
    EXPECT_GE(traces.size(), 1u);

    auto new_thickness = engine.get_param("thickness");
    ASSERT_TRUE(new_thickness.has_value());
    EXPECT_DOUBLE_EQ(std::get<double>(*new_thickness), 2.0);
}

TEST(KBETest, RuleEngine_PriorityOrder) {
    RuleEngine engine;
    engine.set_param("x", 1.0);

    // 高优先级规则先执行: SET x = 20
    Rule r1{"low",  "x >= 0", "SET x = 10.0", 0, true};   // priority 0
    Rule r2{"high", "x >= 0", "SET x = 20.0", 10, true};  // priority 10
    engine.add_rule(r1);
    engine.add_rule(r2);

    // 高优先级规则先触发，但同一轮迭代中低优先级规则也会触发
    // 最终结果由最后触发的规则决定（按优先级排序，低优先级后执行）
    auto traces = engine.infer();
    EXPECT_GE(traces.size(), 2u);  // 两条规则都触发

    auto val = engine.get_param("x");
    ASSERT_TRUE(val.has_value());
    // 低优先级(priority 0)后执行，最终为 10
    EXPECT_DOUBLE_EQ(std::get<double>(*val), 10.0);
}

TEST(KBETest, RuleEngine_ANDCondition) {
    RuleEngine engine;
    engine.set_param("a", 10.0);
    engine.set_param("b", 5.0);

    Rule r{"R_and", "a > 5 AND b < 10", "SET a = 100.0", 1, true};
    engine.add_rule(r);
    engine.infer();

    auto val = engine.get_param("a");
    ASSERT_TRUE(val.has_value());
    EXPECT_DOUBLE_EQ(std::get<double>(*val), 100.0);
}

TEST(KBETest, RuleEngine_NoInfiniteLoop) {
    RuleEngine engine;
    engine.set_param("counter", 0.0);

    // 自引用规则 — 应被循环检测阻止
    Rule r{"loop", "counter < 1000", "SET counter = counter + 1", 1, true};
    engine.add_rule(r);

    auto traces = engine.infer();
    // 最多执行 100 次迭代（由引擎内部限制）
    EXPECT_LE(traces.size(), 100u);
}

// ═══════════════════════════════════════════════════════════
//  3. DesignTable 设计表
// ═══════════════════════════════════════════════════════════

TEST(KBETest, DesignTable_DefineColumnsAndAddRow) {
    DesignTable table;
    table.define_columns({"length", "width", "height"}, {"mm", "mm", "mm"});
    EXPECT_EQ(table.col_count(), 3u);

    DesignRow row;
    row["length"] = 100.0;
    row["width"]  = 50.0;
    row["height"] = 25.0;
    table.add_row(row);
    EXPECT_EQ(table.row_count(), 1u);

    auto r = table.get_row(0);
    ASSERT_TRUE(r.has_value());
    EXPECT_DOUBLE_EQ(std::get<double>(r->at("length")), 100.0);
}

TEST(KBETest, DesignTable_CSVImportExport) {
    DesignTable table;
    std::string csv =
        "length,width,height\n"
        "100,50,25\n"
        "200,80,40\n"
        "150,60,30\n";

    EXPECT_TRUE(table.import_csv(csv));
    EXPECT_EQ(table.row_count(), 3u);
    EXPECT_EQ(table.col_count(), 3u);

    std::string exported = table.export_csv();
    EXPECT_FALSE(exported.empty());
    EXPECT_NE(exported.find("length"), std::string::npos);
    EXPECT_NE(exported.find("100"), std::string::npos);
}

TEST(KBETest, DesignTable_MixedTypes) {
    DesignTable table;
    std::string csv =
        "name,diameter,material,is_standard\n"
        "Bolt_M10,10,Steel,true\n"
        "Nut_M8,8,Brass,false\n";

    EXPECT_TRUE(table.import_csv(csv));
    EXPECT_EQ(table.row_count(), 2u);

    auto r0 = table.get_row(0);
    ASSERT_TRUE(r0.has_value());
    EXPECT_EQ(std::get<std::string>(r0->at("name")), "Bolt_M10");
    EXPECT_DOUBLE_EQ(std::get<double>(r0->at("diameter")), 10.0);
    EXPECT_EQ(std::get<bool>(r0->at("is_standard")), true);
}

// ═══════════════════════════════════════════════════════════
//  4. ParametricOptimization 参数优化
// ═══════════════════════════════════════════════════════════

TEST(KBETest, Optimization_GA_Minimization) {
    ParametricOptimization opt;

    // 目标：最小化 (x-5)^2
    opt.set_fitness([](const std::vector<Parameter>& params) {
        for (const auto& p : params) {
            if (p.name == "x" && std::holds_alternative<double>(p.value)) {
                double x = std::get<double>(p.value);
                return -((x - 5.0) * (x - 5.0));  // 负值因为 GA 最大化适应度
            }
        }
        return 0.0;
    });

    std::vector<Parameter> initial = {{"x", 10.0, 0.0, 100.0, "", "", false}};
    opt.set_bounds("x", 0.0, 100.0);

    GAConfig ga_config;
    ga_config.population_size = 50;
    ga_config.generations     = 30;

    auto result = opt.optimize_ga(initial, ga_config);
    EXPECT_TRUE(result.converged);

    // 最优解应接近 x=5.0
    for (const auto& p : result.optimal_params) {
        if (p.name == "x") {
            double val = std::get<double>(p.value);
            EXPECT_NEAR(val, 5.0, 5.0);  // GA 有随机性，容差大些
        }
    }

    // 适应度历史应有记录
    EXPECT_GT(opt.fitness_history().size(), 0u);
}

TEST(KBETest, Optimization_Gradient_Descent) {
    ParametricOptimization opt;

    // 目标：最小化 (x-3)^2，梯度下降直接最小化 fitness
    opt.set_fitness([](const std::vector<Parameter>& params) {
        for (const auto& p : params) {
            if (p.name == "x" && std::holds_alternative<double>(p.value)) {
                double x = std::get<double>(p.value);
                return (x - 3.0) * (x - 3.0);  // 正值，梯度下降最小化
            }
        }
        return 0.0;
    });

    std::vector<Parameter> initial = {{"x", 8.0, 0.0, 100.0, "", "", false}};
    opt.set_bounds("x", -10.0, 100.0);

    GradientConfig grad_config;
    grad_config.learning_rate  = 0.05;
    grad_config.max_iterations = 500;
    grad_config.tolerance      = 1e-6;

    auto result = opt.optimize_gradient(initial, grad_config);

    for (const auto& p : result.optimal_params) {
        if (p.name == "x") {
            double val = std::get<double>(p.value);
            EXPECT_NEAR(val, 3.0, 3.0);  // 梯度下降有收敛性，容差较大
        }
    }
}

TEST(KBETest, Optimization_Constraints) {
    ParametricOptimization opt;

    opt.set_fitness([](const std::vector<Parameter>& params) {
        for (const auto& p : params) {
            if (p.name == "x" && std::holds_alternative<double>(p.value)) {
                double x = std::get<double>(p.value);
                return -x;  // 最小化 x（GA 最大化适应度 = 最小化 x）
            }
        }
        return 0.0;
    });

    std::vector<Parameter> initial = {{"x", 5.0, 0.0, 100.0, "", "", false}};
    // 用 bounds 约束 x 上限为 10
    opt.set_bounds("x", 0.0, 10.0);

    GAConfig ga_config;
    ga_config.population_size = 30;
    ga_config.generations     = 30;

    auto result = opt.optimize_ga(initial, ga_config);

    for (const auto& p : result.optimal_params) {
        if (p.name == "x") {
            double val = std::get<double>(p.value);
            EXPECT_LE(val, 10.0);  // 必须满足 bound 约束
            EXPECT_GT(val, -1.0);  // 不能为负
        }
    }
}