ZhiYuanTong/database/verify_all_provinces_2025.sql

-- 全国 31 个省级地区 2025 年首批招录数据验收
-- 执行：mysql -u root -p zhiyuantong < database/verify_all_provinces_2025.sql

SELECT collection_status AS '采集状态', COUNT(*) AS '省级地区数'
FROM collection_scopes
WHERE admission_year = 2025
GROUP BY collection_status;

SELECT
    scope.province_name AS '省级地区',
    COALESCE(control_counts.record_count, 0) AS '控制线数',
    COALESCE(score_counts.record_count, 0) AS '一分一段数',
    COALESCE(score_counts.subject_count, 0) AS '一分一段科类数',
    COALESCE(coverage_counts.record_count, 0) AS '覆盖说明数',
    scope.collection_status AS '采集状态'
FROM collection_scopes scope
LEFT JOIN (
    SELECT province_name, COUNT(*) AS record_count
    FROM control_score_lines WHERE admission_year = 2025 GROUP BY province_name
) control_counts ON control_counts.province_name = scope.province_name
LEFT JOIN (
    SELECT province_name, COUNT(*) AS record_count,
           COUNT(DISTINCT subject_category) AS subject_count
    FROM score_rank_points WHERE admission_year = 2025 GROUP BY province_name
) score_counts ON score_counts.province_name = scope.province_name
LEFT JOIN (
    SELECT province_name, COUNT(*) AS record_count
    FROM dataset_coverage WHERE admission_year = 2025 GROUP BY province_name
) coverage_counts ON coverage_counts.province_name = scope.province_name
WHERE scope.admission_year = 2025
ORDER BY scope.province_name;

SELECT
    COUNT(DISTINCT province_name) AS '控制线覆盖省级地区数',
    COUNT(*) AS '控制线总数'
FROM control_score_lines
WHERE admission_year = 2025;

SELECT
    COUNT(DISTINCT province_name) AS '一分一段覆盖省级地区数',
    COUNT(*) AS '一分一段总数'
FROM score_rank_points
WHERE admission_year = 2025;

SELECT scope.province_name AS '一分一段尚待结构化省级地区'
FROM collection_scopes scope
WHERE scope.admission_year = 2025
  AND NOT EXISTS (
      SELECT 1 FROM score_rank_points score
      WHERE score.admission_year = scope.admission_year
        AND score.province_name = scope.province_name
  )
ORDER BY scope.province_name;

SELECT
    coverage_status AS '覆盖状态',
    COUNT(*) AS '覆盖说明数',
    COUNT(DISTINCT province_name) AS '涉及省级地区数'
FROM dataset_coverage
WHERE admission_year = 2025
GROUP BY coverage_status
ORDER BY coverage_status;