ai-contract-platform/scripts/legal_library_monitoring.sql

-- 法律库监控SQL脚本
-- 用于监控法律库使用情况和健康状况

-- ============================================
-- 1. 法律库基础统计
-- ============================================

CREATE OR REPLACE VIEW legal_library_stats AS
SELECT
    COUNT(*) as total_articles,
    COUNT(DISTINCT source_id) as total_laws,
    COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END) as vectorized_articles,
    COUNT(CASE WHEN semantic_vector_json IS NULL THEN 1 END) as pending_vectorization,
    ROUND(COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric / COUNT(*)::numeric * 100, 2) as vectorization_rate,
    MAX(updated_at) as last_update,
    MAX(semantic_vector_updated_at) as last_vectorization
FROM legal_articles;

-- ============================================
-- 2. 法律类型分布统计
-- ============================================

CREATE OR REPLACE VIEW legal_library_type_stats AS
SELECT
    source_type,
    COUNT(DISTINCT source_id) as law_count,
    COUNT(*) as article_count,
    ROUND(COUNT(*)::numeric * 100.0 / SUM(COUNT(*)) OVER ()::numeric, 2) as percentage
FROM legal_articles
GROUP BY source_type
ORDER BY article_count DESC;

-- ============================================
-- 3. 向量化状态监控
-- ============================================

CREATE OR REPLACE VIEW legal_library_vectorization_status AS
SELECT
    semantic_vector_version,
    COUNT(*) as article_count,
    MIN(semantic_vector_updated_at) as first_vectorization,
    MAX(semantic_vector_updated_at) as last_vectorization
FROM legal_articles
WHERE semantic_vector_json IS NOT NULL
GROUP BY semantic_vector_version
ORDER BY article_count DESC;

-- ============================================
-- 4. 法律库健康检查
-- ============================================

CREATE OR REPLACE VIEW legal_library_health_check AS
SELECT
    'total_articles' as metric,
    (SELECT COUNT(*) FROM legal_articles)::text as value,
    CASE
        WHEN (SELECT COUNT(*) FROM legal_articles) >= 700000 THEN 'healthy'
        ELSE 'warning'
    END as status
UNION ALL
SELECT
    'vectorization_rate' as metric,
    ROUND((SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
           / NULLIF((SELECT COUNT(*) FROM legal_articles), 0) * 100), 2)::text || '%' as value,
    CASE
        WHEN (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
              / NULLIF((SELECT COUNT(*) FROM legal_articles), 0)) >= 0.99 THEN 'healthy'
        WHEN (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
              / NULLIF((SELECT COUNT(*) FROM legal_articles), 0)) >= 0.90 THEN 'warning'
        ELSE 'critical'
    END as status
UNION ALL
SELECT
    'pending_vectorization' as metric,
    (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL)::text as value,
    CASE
        WHEN (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL) = 0 THEN 'healthy'
        WHEN (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL) < 1000 THEN 'warning'
        ELSE 'critical'
    END as status
UNION ALL
SELECT
    'last_update' as metric,
    COALESCE(TO_CHAR((SELECT MAX(updated_at) FROM legal_articles), 'YYYY-MM-DD HH24:MI:SS'), 'N/A') as value,
    CASE
        WHEN (SELECT MAX(updated_at) FROM legal_articles) >= NOW() - INTERVAL '30 days' THEN 'healthy'
        WHEN (SELECT MAX(updated_at) FROM legal_articles) >= NOW() - INTERVAL '90 days' THEN 'warning'
        ELSE 'critical'
    END as status;

-- ============================================
-- 5. 索引状态检查
-- ============================================

CREATE OR REPLACE VIEW legal_library_index_status AS
SELECT
    indexname as index_name,
    indexdef as definition,
    'active' as status
FROM pg_indexes
WHERE tablename = 'legal_articles'
ORDER BY indexname;

-- ============================================
-- 6. 存储空间使用
-- ============================================

CREATE OR REPLACE VIEW legal_library_storage AS
SELECT
    pg_size_pretty(pg_total_relation_size('legal_articles')) as total_size,
    pg_size_pretty(pg_relation_size('legal_articles')) as table_size,
    pg_size_pretty(pg_total_relation_size('legal_articles') - pg_relation_size('legal_articles')) as index_size,
    (SELECT COUNT(*) FROM legal_articles) as row_count;

-- ============================================
-- 7. 核心法律完整性检查
-- ============================================

CREATE OR REPLACE VIEW legal_library_core_laws_check AS
SELECT
    source_id,
    source_title,
    COUNT(*) as article_count,
    CASE
        WHEN COUNT(*) > 0 THEN 'present'
        ELSE 'missing'
    END as status
FROM legal_articles
WHERE source_id IN (
    'LAW-CIVIL-CODE-2020',
    'LAW-COMPANY-2023',
    'LAW-LABOR-CONTRACT-2012',
    'LAW-PIPL-2021',
    'LAW-DATA-SECURITY-2021',
    'LAW-E-SIGNATURE-2019'
)
GROUP BY source_id, source_title
ORDER BY source_id;

-- ============================================
-- 8. 告警阈值配置
-- ============================================

-- 创建告警配置表
CREATE TABLE IF NOT EXISTS legal_library_alert_thresholds (
    id SERIAL PRIMARY KEY,
    metric_name VARCHAR(100) UNIQUE NOT NULL,
    warning_threshold NUMERIC,
    critical_threshold NUMERIC,
    description TEXT,
    created_at TIMESTAMP DEFAULT NOW(),
    updated_at TIMESTAMP DEFAULT NOW()
);

-- 插入默认阈值
INSERT INTO legal_library_alert_thresholds (metric_name, warning_threshold, critical_threshold, description)
VALUES
    ('vectorization_rate', 99.0, 95.0, '向量化完成率'),
    ('pending_vectorization', 1000, 5000, '待向量化条文数'),
    ('total_articles', 700000, 500000, '法律条文总数')
ON CONFLICT (metric_name) DO NOTHING;

-- ============================================
-- 9. 使用统计表（可选）
-- ============================================

CREATE TABLE IF NOT EXISTS legal_article_access_log (
    id SERIAL PRIMARY KEY,
    article_id VARCHAR(255),
    source_id VARCHAR(255),
    access_type VARCHAR(50), -- 'search', 'review', 'chat', 'view'
    user_id VARCHAR(255),
    tenant_id VARCHAR(255),
    created_at TIMESTAMP DEFAULT NOW()
);

-- 创建访问统计视图
CREATE OR REPLACE VIEW legal_article_access_stats AS
SELECT
    source_id,
    COUNT(*) as access_count,
    COUNT(DISTINCT user_id) as unique_users,
    MAX(created_at) as last_accessed
FROM legal_article_access_log
GROUP BY source_id
ORDER BY access_count DESC
LIMIT 20;

-- ============================================
-- 10. 监控查询示例
-- ============================================

-- 查询法律库整体状态
-- SELECT * FROM legal_library_stats;

-- 查询健康检查结果
-- SELECT * FROM legal_library_health_check;

-- 查询向量化进度
-- SELECT * FROM legal_library_vectorization_status;

-- 查询存储使用情况
-- SELECT * FROM legal_library_storage;

-- 查询核心法律完整性
-- SELECT * FROM legal_library_core_laws_check;

-- ============================================
-- 11. 告警检查函数
-- ============================================

CREATE OR REPLACE FUNCTION check_legal_library_alerts()
RETURNS TABLE (
    metric VARCHAR(100),
    current_value NUMERIC,
    threshold NUMERIC,
    severity VARCHAR(20),
    message TEXT
) AS $$
BEGIN
    -- 检查向量化率
    RETURN QUERY
    SELECT
        'vectorization_rate'::VARCHAR(100),
        (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
         / NULLIF(COUNT(*), 0) * 100
         FROM legal_articles) as current_value,
        t.warning_threshold,
        CASE
            WHEN (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
                  / NULLIF(COUNT(*), 0) * 100 FROM legal_articles) < t.critical_threshold THEN 'critical'
            WHEN (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
                  / NULLIF(COUNT(*), 0) * 100 FROM legal_articles) < t.warning_threshold THEN 'warning'
            ELSE 'ok'
        END::VARCHAR(20),
        CASE
            WHEN (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
                  / NULLIF(COUNT(*), 0) * 100 FROM legal_articles) < t.critical_threshold
            THEN '向量化率严重不足，请立即处理'
            WHEN (SELECT COUNT(CASE WHEN semantic_vector_json IS NOT NULL THEN 1 END)::numeric
                  / NULLIF(COUNT(*), 0) * 100 FROM legal_articles) < t.warning_threshold
            THEN '向量化率偏低，建议检查'
            ELSE '向量化率正常'
        END::TEXT
    FROM legal_library_alert_thresholds t
    WHERE t.metric_name = 'vectorization_rate';

    -- 检查待向量化数量
    RETURN QUERY
    SELECT
        'pending_vectorization'::VARCHAR(100),
        (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL)::numeric,
        t.warning_threshold,
        CASE
            WHEN (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL) > t.critical_threshold THEN 'critical'
            WHEN (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL) > t.warning_threshold THEN 'warning'
            ELSE 'ok'
        END::VARCHAR(20),
        CASE
            WHEN (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL) > t.critical_threshold
            THEN '待向量化条文过多，请立即处理'
            WHEN (SELECT COUNT(*) FROM legal_articles WHERE semantic_vector_json IS NULL) > t.warning_threshold
            THEN '有待向量化条文，建议处理'
            ELSE '向量化完成'
        END::TEXT
    FROM legal_library_alert_thresholds t
    WHERE t.metric_name = 'pending_vectorization';
END;
$$ LANGUAGE plpgsql;

-- 使用示例
-- SELECT * FROM check_legal_library_alerts();