USE zhiyuantong;

SELECT
  COUNT(*) AS total_schools,
  COUNT(DISTINCT moe_school_code) AS unique_moe_codes,
  COUNT(DISTINCT province_name) AS province_count,
  SUM(education_level='本科') AS undergraduate_schools,
  SUM(education_level='专科') AS vocational_colleges,
  SUM(province_name='甘肃省') AS gansu_schools
FROM schools
WHERE school_status='active';

SELECT
  SUM(moe_school_code NOT REGEXP '^[0-9]{10}$') AS invalid_moe_codes,
  SUM(official_name IS NULL OR official_name='') AS missing_names,
  SUM(province_name IS NULL OR province_name='') AS missing_provinces,
  SUM(city_name IS NULL OR city_name='') AS missing_cities,
  SUM(authority_name IS NULL OR authority_name='') AS missing_authorities,
  SUM(education_level NOT IN ('本科','专科','成人高等教育')) AS invalid_levels
FROM schools
WHERE school_status='active';

SELECT ownership_type, COUNT(*) AS school_count
FROM schools
WHERE school_status='active'
GROUP BY ownership_type
ORDER BY school_count DESC;

SELECT dataset_year, status, rows_read, rows_inserted, rows_updated, rows_rejected
FROM ingestion_runs
ORDER BY started_at DESC
LIMIT 5;