USE zhiyuantong;
SET NAMES utf8mb4;

-- 当前版本指针与快照记录分离：切换指针即可发布或回滚，历史快照保持不可变。
CREATE TABLE IF NOT EXISTS current_dataset_publications (
  province_name VARCHAR(64) NOT NULL DEFAULT '' COMMENT '数据适用省份，空字符串表示全国数据',
  admission_year SMALLINT UNSIGNED NOT NULL COMMENT '数据所属招生年度',
  dataset_name VARCHAR(128) NOT NULL COMMENT '公开数据集业务名称',
  publication_id BIGINT UNSIGNED NOT NULL COMMENT '当前对外生效的发布版本ID',
  activated_by VARCHAR(128) NOT NULL COMMENT '切换当前版本的人员姓名或账号标识',
  activated_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '当前版本生效时间',
  PRIMARY KEY (province_name, admission_year, dataset_name),
  UNIQUE KEY uk_current_dataset_publication_id (publication_id),
  CONSTRAINT fk_current_dataset_publication
    FOREIGN KEY (publication_id) REFERENCES dataset_publications(publication_id)
) ENGINE=InnoDB COMMENT='各省年度数据集当前公开版本指针';

CREATE TABLE IF NOT EXISTS public_province_exam_configs (
  publication_id BIGINT UNSIGNED NOT NULL COMMENT '所属不可变发布版本',
  exam_config_id BIGINT UNSIGNED NOT NULL COMMENT '来源考试模式配置ID',
  province_name VARCHAR(64) NOT NULL COMMENT '省级行政区名称',
  exam_year SMALLINT UNSIGNED NOT NULL COMMENT '高考年份',
  exam_mode VARCHAR(32) NOT NULL COMMENT '考试模式稳定代码',
  total_score DECIMAL(7,2) NULL COMMENT '普通类高考总分，未逐省核验时为空',
  rank_basis VARCHAR(128) NULL COMMENT '一分一段或位次的主要分组口径',
  source_id BIGINT UNSIGNED NOT NULL COMMENT '核验考试模式所依据的来源ID',
  source_publisher VARCHAR(128) NOT NULL COMMENT '来源发布机构快照',
  source_title VARCHAR(255) NOT NULL COMMENT '来源标题快照',
  source_url VARCHAR(1024) NULL COMMENT '来源发布页地址快照',
  quality_grade VARCHAR(16) NOT NULL DEFAULT 'A' COMMENT '公开记录质量等级',
  created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '快照生成时间',
  PRIMARY KEY (publication_id, exam_config_id),
  UNIQUE KEY uk_public_exam_config_scope (publication_id, province_name, exam_year),
  KEY idx_public_exam_config_lookup (province_name, exam_year, exam_mode),
  CONSTRAINT fk_public_exam_config_publication
    FOREIGN KEY (publication_id) REFERENCES dataset_publications(publication_id),
  CONSTRAINT fk_public_exam_config_source
    FOREIGN KEY (source_id) REFERENCES data_sources(source_id)
) ENGINE=InnoDB COMMENT='带发布版本的省级考试模式公共查询快照';

CREATE TABLE IF NOT EXISTS public_school_profiles (
  publication_id BIGINT UNSIGNED NOT NULL COMMENT '所属不可变发布版本',
  school_id CHAR(36) NOT NULL COMMENT '全国院校稳定UUID',
  moe_school_code CHAR(10) NULL COMMENT '教育部学校标识码快照',
  official_name VARCHAR(255) NOT NULL COMMENT '学校正式名称快照',
  province_name VARCHAR(64) NOT NULL COMMENT '学校所属省份快照',
  city_name VARCHAR(128) NOT NULL COMMENT '学校所在城市快照',
  authority_name VARCHAR(255) NOT NULL COMMENT '学校主管部门快照',
  education_level VARCHAR(32) NOT NULL COMMENT '办学层次快照',
  ownership_type VARCHAR(32) NOT NULL COMMENT '办学性质快照',
  school_status VARCHAR(32) NOT NULL COMMENT '学校状态快照',
  official_website VARCHAR(1024) NULL COMMENT '学校官网快照',
  admission_website VARCHAR(1024) NULL COMMENT '招生官网快照',
  source_id BIGINT UNSIGNED NOT NULL COMMENT '院校信息来源ID',
  source_title VARCHAR(255) NOT NULL COMMENT '来源标题快照',
  source_url VARCHAR(1024) NULL COMMENT '来源发布页地址快照',
  quality_grade VARCHAR(16) NOT NULL COMMENT '公开记录质量等级',
  created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '快照生成时间',
  PRIMARY KEY (publication_id, school_id),
  KEY idx_public_school_lookup (publication_id, province_name, education_level, official_name),
  CONSTRAINT fk_public_school_publication
    FOREIGN KEY (publication_id) REFERENCES dataset_publications(publication_id),
  CONSTRAINT fk_public_school_source
    FOREIGN KEY (source_id) REFERENCES data_sources(source_id)
) ENGINE=InnoDB COMMENT='带发布版本的院校公共档案快照';

CREATE TABLE IF NOT EXISTS public_school_admission_stats (
  publication_id BIGINT UNSIGNED NOT NULL COMMENT '所属不可变发布版本',
  cutoff_line_id BIGINT UNSIGNED NOT NULL COMMENT '来源投档线记录ID',
  province_name VARCHAR(64) NOT NULL COMMENT '招生省份',
  admission_year SMALLINT UNSIGNED NOT NULL COMMENT '招生年度',
  admission_category VARCHAR(32) NOT NULL COMMENT '招生类别',
  subject_category VARCHAR(64) NOT NULL COMMENT '科类或首选科目来源值',
  education_level VARCHAR(32) NOT NULL COMMENT '培养层次',
  batch_name VARCHAR(255) NOT NULL COMMENT '录取批次',
  segment_code VARCHAR(128) NOT NULL COMMENT '批次或段次代码',
  school_id CHAR(36) NOT NULL COMMENT '全国院校稳定UUID',
  school_name VARCHAR(255) NOT NULL COMMENT '院校名称快照',
  group_code VARCHAR(64) NOT NULL COMMENT '院校专业组代码',
  group_name VARCHAR(255) NOT NULL COMMENT '院校专业组名称',
  cutoff_score DECIMAL(7,2) NULL COMMENT '最低投档分',
  cutoff_rank INT UNSIGNED NULL COMMENT '最低投档位次',
  control_score DECIMAL(7,2) NULL COMMENT '省级控制分',
  score_diff DECIMAL(7,2) NULL COMMENT '线差',
  admit_count INT UNSIGNED NULL COMMENT '实际录取或投档人数',
  source_id BIGINT UNSIGNED NOT NULL COMMENT '来源ID',
  source_title VARCHAR(255) NOT NULL COMMENT '来源标题快照',
  source_url VARCHAR(1024) NULL COMMENT '来源发布页地址快照',
  quality_grade VARCHAR(16) NOT NULL COMMENT '公开记录质量等级',
  created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '快照生成时间',
  PRIMARY KEY (publication_id, cutoff_line_id),
  KEY idx_public_school_admission_lookup (
    publication_id, subject_category, education_level, cutoff_rank, cutoff_score
  ),
  KEY idx_public_school_admission_school (publication_id, school_id),
  CONSTRAINT fk_public_school_admission_publication
    FOREIGN KEY (publication_id) REFERENCES dataset_publications(publication_id),
  CONSTRAINT fk_public_school_admission_source
    FOREIGN KEY (source_id) REFERENCES data_sources(source_id)
) ENGINE=InnoDB COMMENT='带发布版本的院校或专业组投档统计快照';

CREATE TABLE IF NOT EXISTS public_major_admission_stats (
  publication_id BIGINT UNSIGNED NOT NULL COMMENT '所属不可变发布版本',
  major_line_id BIGINT UNSIGNED NOT NULL COMMENT '来源专业录取记录ID',
  province_name VARCHAR(64) NOT NULL COMMENT '招生省份',
  admission_year SMALLINT UNSIGNED NOT NULL COMMENT '招生年度',
  admission_category VARCHAR(32) NOT NULL COMMENT '招生类别',
  subject_category VARCHAR(64) NOT NULL COMMENT '科类或首选科目来源值',
  education_level VARCHAR(32) NOT NULL COMMENT '培养层次',
  batch_name VARCHAR(255) NOT NULL COMMENT '录取批次',
  school_id CHAR(36) NOT NULL COMMENT '全国院校稳定UUID',
  school_name VARCHAR(255) NOT NULL COMMENT '院校名称快照',
  group_code VARCHAR(64) NOT NULL COMMENT '院校专业组代码',
  major_code VARCHAR(64) NOT NULL COMMENT '来源专业代码',
  major_name VARCHAR(255) NOT NULL COMMENT '来源专业或专业类名称',
  subject_requirements VARCHAR(255) NULL COMMENT '选科要求原文',
  min_score DECIMAL(7,2) NULL COMMENT '专业录取最低分',
  min_rank INT UNSIGNED NULL COMMENT '专业录取最低位次',
  max_score DECIMAL(7,2) NULL COMMENT '专业录取最高分',
  avg_score DECIMAL(7,2) NULL COMMENT '专业录取平均分',
  admit_count INT UNSIGNED NULL COMMENT '实际录取人数',
  source_id BIGINT UNSIGNED NOT NULL COMMENT '来源ID',
  source_title VARCHAR(255) NOT NULL COMMENT '来源标题快照',
  source_url VARCHAR(1024) NULL COMMENT '来源发布页地址快照',
  quality_grade VARCHAR(16) NOT NULL COMMENT '公开记录质量等级',
  created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '快照生成时间',
  PRIMARY KEY (publication_id, major_line_id),
  KEY idx_public_major_admission_lookup (
    publication_id, subject_category, education_level, min_rank, min_score
  ),
  KEY idx_public_major_admission_school (publication_id, school_id, major_name),
  CONSTRAINT fk_public_major_admission_publication
    FOREIGN KEY (publication_id) REFERENCES dataset_publications(publication_id),
  CONSTRAINT fk_public_major_admission_source
    FOREIGN KEY (source_id) REFERENCES data_sources(source_id)
) ENGINE=InnoDB COMMENT='带发布版本的专业录取统计快照';

CREATE TABLE IF NOT EXISTS public_score_rank_points (
  publication_id BIGINT UNSIGNED NOT NULL COMMENT '所属不可变发布版本',
  score_rank_id BIGINT UNSIGNED NOT NULL COMMENT '来源一分一段记录ID',
  province_name VARCHAR(64) NOT NULL COMMENT '考试省份',
  admission_year SMALLINT UNSIGNED NOT NULL COMMENT '高考年度',
  admission_category VARCHAR(32) NOT NULL COMMENT '招生类别',
  subject_category VARCHAR(32) NOT NULL COMMENT '科类或首选科目来源值',
  score DECIMAL(7,2) NOT NULL COMMENT '高考分数或综合排序分',
  same_score_count INT UNSIGNED NOT NULL COMMENT '同分人数',
  cumulative_count INT UNSIGNED NOT NULL COMMENT '累计人数',
  rank_from INT UNSIGNED NOT NULL COMMENT '最佳位次',
  rank_to INT UNSIGNED NOT NULL COMMENT '最末位次',
  source_id BIGINT UNSIGNED NOT NULL COMMENT '来源ID',
  source_title VARCHAR(255) NOT NULL COMMENT '来源标题快照',
  source_url VARCHAR(1024) NULL COMMENT '来源发布页地址快照',
  quality_grade VARCHAR(16) NOT NULL COMMENT '公开记录质量等级',
  created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '快照生成时间',
  PRIMARY KEY (publication_id, score_rank_id),
  UNIQUE KEY uk_public_score_rank_scope (
    publication_id, admission_category, subject_category, score
  ),
  KEY idx_public_score_rank_lookup (
    publication_id, admission_category, subject_category, score, rank_from, rank_to
  ),
  CONSTRAINT fk_public_score_rank_publication
    FOREIGN KEY (publication_id) REFERENCES dataset_publications(publication_id),
  CONSTRAINT fk_public_score_rank_source
    FOREIGN KEY (source_id) REFERENCES data_sources(source_id)
) ENGINE=InnoDB COMMENT='带发布版本的一分一段公共查询快照';

-- 下列来源用于核验“2025 年各省考试模式”。网页未在本库保存副本，sha256
-- 是本次人工核验记录（标题、规范网址和发布日期）的稳定指纹，而非网页文件摘要。
INSERT INTO data_sources (
  source_type, publisher, title, landing_page_url, published_on, sha256, notes
) VALUES
  ('人工核验', '中华人民共和国教育部',
   '对十三届全国人大二次会议第5836号建议的答复',
   'https://www.moe.gov.cn/jyb_xxgk/xxgk_jyta/jyta_xueshengsi/201911/t20191126_409732.html',
   '2019-11-26', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256),
   '教育部原文明确6个3+3省市及第三批8个3+1+2省市；网页原文由发布机构保管。'),
  ('人工核验', '中华人民共和国教育部',
   '教育部：全国29个省份已启动高考综合改革',
   'https://www.moe.gov.cn/fbh/live/2022/54835/mtbd/202209/t20220915_661428.html',
   '2022-09-15', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256),
   '教育部发布会信息明确第四批7省区于2024年、第五批8省区于2025年落地新高考。'),
  ('人工核验', '新疆维吾尔自治区教育考试院',
   '新疆维吾尔自治区2025年普通高校招生工作规定',
   'https://www.xjzk.gov.cn/c/2025-05-16/494130.shtml',
   '2025-05-16', SHA2('exam-mode-evidence:xinjiang:2025-admission-rules', 256),
   '2025年普通类仍按文史类、理工类组织排序与录取。'),
  ('人工核验', '西藏自治区人民政府',
   '西藏自治区2025年普通高考分数线出炉',
   'https://www.xizang.gov.cn/xwzx_406/bmkx/202506/t20250626_486252.html',
   '2025-06-26', SHA2('exam-mode-evidence:xizang:2025-control-lines', 256),
   '自治区2025年普通生源控制线仍分别按文史类、理工类公布。')
AS incoming
ON DUPLICATE KEY UPDATE
  publisher=incoming.publisher,
  title=incoming.title,
  landing_page_url=incoming.landing_page_url,
  published_on=incoming.published_on,
  notes=incoming.notes;

INSERT INTO province_exam_configs (
  province_name, exam_year, exam_mode, total_score, rank_basis,
  verification_status, source_id
)
SELECT seed.province_name, 2025, seed.exam_mode, NULL, seed.rank_basis,
       'verified', source.source_id
FROM (
  SELECT '北京市' province_name, 'three_plus_three' exam_mode,
         '普通类不分文理，按选考组合统一排序' rank_basis,
         SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256) source_sha
  UNION ALL SELECT '天津市', 'three_plus_three', '普通类不分文理，按选考组合统一排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '上海市', 'three_plus_three', '普通类不分文理，按选考组合统一排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '浙江省', 'three_plus_three', '普通类不分文理，按选考组合统一排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '山东省', 'three_plus_three', '普通类不分文理，按选考组合统一排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '海南省', 'three_plus_three', '普通类不分文理，按选考组合统一排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '河北省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '辽宁省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '江苏省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '福建省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '湖北省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '湖南省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '广东省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '重庆市', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2019-11-26:5836', 256)
  UNION ALL SELECT '吉林省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '黑龙江省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '安徽省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '江西省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '广西壮族自治区', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '贵州省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '甘肃省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '山西省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '内蒙古自治区', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '河南省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '四川省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '云南省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '陕西省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '青海省', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '宁夏回族自治区', 'three_plus_one_plus_two', '普通类按物理科目组合、历史科目组合分别排序', SHA2('exam-mode-evidence:moe:2022-09-15:29-provinces', 256)
  UNION ALL SELECT '新疆维吾尔自治区', 'traditional', '普通类按文史类、理工类分别排序', SHA2('exam-mode-evidence:xinjiang:2025-admission-rules', 256)
  UNION ALL SELECT '西藏自治区', 'traditional', '普通类按文史类、理工类分别排序', SHA2('exam-mode-evidence:xizang:2025-control-lines', 256)
) seed
JOIN data_sources source ON source.sha256=seed.source_sha
ON DUPLICATE KEY UPDATE
  exam_mode=VALUES(exam_mode),
  rank_basis=VALUES(rank_basis),
  verification_status='verified',
  source_id=VALUES(source_id);

-- 考试模式是本迁移中逐省核验的数据，可直接生成首个公开版本；其他招录事实数据
-- 必须经审核和发布服务生成快照，迁移不会把待审核记录复制到公共表。
INSERT INTO dataset_publications (
  province_name, admission_year, dataset_name, version_no, publication_status,
  source_record_count, published_record_count, quality_grade,
  validation_summary, published_by, published_at
)
SELECT config.province_name, config.exam_year, 'province-exam-configs',
       '2025.exam-mode.v1', 'published', 1, 1, 'A',
       JSON_OBJECT('verifiedRecordCount', 1, 'sourceAttached', TRUE),
       'system:migration-011', NOW()
FROM province_exam_configs config
WHERE config.exam_year=2025 AND config.verification_status='verified'
ON DUPLICATE KEY UPDATE
  source_record_count=VALUES(source_record_count),
  published_record_count=VALUES(published_record_count),
  quality_grade=VALUES(quality_grade),
  validation_summary=VALUES(validation_summary);

INSERT INTO public_province_exam_configs (
  publication_id, exam_config_id, province_name, exam_year, exam_mode,
  total_score, rank_basis, source_id, source_publisher, source_title,
  source_url, quality_grade
)
SELECT publication.publication_id, config.exam_config_id,
       config.province_name, config.exam_year, config.exam_mode,
       config.total_score, config.rank_basis, source.source_id,
       source.publisher, source.title, source.landing_page_url, 'A'
FROM province_exam_configs config
JOIN dataset_publications publication
  ON publication.province_name=config.province_name
 AND publication.admission_year=config.exam_year
 AND publication.dataset_name='province-exam-configs'
 AND publication.version_no='2025.exam-mode.v1'
JOIN data_sources source ON source.source_id=config.source_id
WHERE config.exam_year=2025 AND config.verification_status='verified'
ON DUPLICATE KEY UPDATE publication_id=VALUES(publication_id);

INSERT INTO current_dataset_publications (
  province_name, admission_year, dataset_name, publication_id,
  activated_by, activated_at
)
SELECT publication.province_name, publication.admission_year,
       publication.dataset_name, publication.publication_id,
       'system:migration-011', NOW()
FROM dataset_publications publication
WHERE publication.admission_year=2025
  AND publication.dataset_name='province-exam-configs'
  AND publication.version_no='2025.exam-mode.v1'
  AND publication.publication_status='published'
ON DUPLICATE KEY UPDATE
  publication_id=VALUES(publication_id),
  activated_by=VALUES(activated_by),
  activated_at=VALUES(activated_at);