"""用人工复核映射修正招生来源学校名，并删除确定误入的数据行。"""

import argparse
import csv
import hashlib
import os
import subprocess
import uuid
from pathlib import Path


def sql_string(value):
    if value is None or value == "":
        return "NULL"
    return "'" + str(value).replace("\\", "\\\\").replace("'", "\\'") + "'"


def mysql_run(args, sql, capture=False):
    command = [args.mysql_bin, "--protocol=TCP", "-h", args.host, "-P", str(args.port),
               "-u", args.user, "--default-character-set=utf8mb4", "-D", args.database]
    if capture:
        command.extend(["-N", "-B"])
    result = subprocess.run(command, input=sql, text=True, encoding="utf-8", capture_output=True)
    if result.returncode:
        raise RuntimeError(result.stderr.strip() or result.stdout.strip())
    return result.stdout


def main():
    parser = argparse.ArgumentParser(description="清理2025招生来源学校名称异常")
    parser.add_argument("--corrections-file", required=True)
    parser.add_argument("--invalid-file", required=True)
    parser.add_argument("--mysql-bin", required=True)
    parser.add_argument("--host", default="127.0.0.1")
    parser.add_argument("--port", type=int, default=3306)
    parser.add_argument("--user", default="root")
    parser.add_argument("--database", default="zhiyuantong")
    args = parser.parse_args()
    if not os.environ.get("MYSQL_PWD"):
        raise RuntimeError("必须通过 MYSQL_PWD 环境变量提供数据库密码")

    corrections_path = Path(args.corrections_file)
    invalid_path = Path(args.invalid_file)
    with corrections_path.open(encoding="utf-8-sig", newline="") as stream:
        corrections = list(csv.DictReader(stream))
    with invalid_path.open(encoding="utf-8-sig", newline="") as stream:
        invalid = list(csv.DictReader(stream))
    if not corrections or set(corrections[0]) != {"raw_name", "target_school_name", "correction_type", "notes"}:
        raise ValueError("名称修正规则字段不符合预期")
    if not invalid or set(invalid[0]) != {"raw_name", "reason"}:
        raise ValueError("无效行规则字段不符合预期")
    if len({row["raw_name"] for row in corrections}) != len(corrections):
        raise ValueError("名称修正规则存在重复来源名称")

    targets = sorted({row["target_school_name"] for row in corrections})
    target_sql = ",".join(sql_string(name) for name in targets)
    found_text = mysql_run(
        args,
        f"SELECT official_name,COUNT(*) FROM schools WHERE official_name IN ({target_sql}) GROUP BY official_name;",
        capture=True,
    )
    found = {line.split("\t")[0]: int(line.split("\t")[1]) for line in found_text.strip().splitlines()}
    missing = [name for name in targets if found.get(name) != 1]
    if missing:
        raise ValueError(f"目标学校不存在或不唯一：{missing}")

    digest = hashlib.sha256(corrections_path.read_bytes() + b"\n" + invalid_path.read_bytes()).hexdigest()
    run_id = str(uuid.uuid4())
    correction_values = ",\n".join(
        "(" + ",".join(sql_string(row[key]) for key in [
            "raw_name", "target_school_name", "correction_type", "notes"
        ]) + ")" for row in corrections
    )
    invalid_values = ",\n".join(
        "(" + ",".join(sql_string(row[key]) for key in ["raw_name", "reason"]) + ")"
        for row in invalid
    )
    sql = f"""
    SET NAMES utf8mb4;
    SET time_zone='+08:00';
    START TRANSACTION;
    INSERT INTO data_sources (
      source_type,publisher,title,retrieved_at,sha256,raw_file_path,notes
    ) VALUES (
      'curated_mapping','志愿通数据治理','2025招生来源学校名称异常复核规则',NOW(),
      {sql_string(digest)},{sql_string(str(corrections_path.resolve()) + ';' + str(invalid_path.resolve()))},
      '逐条保留错别字、截断、旧称、二级机构和无效行的复核依据；无效行删除前已有全库备份。'
    ) AS incoming ON DUPLICATE KEY UPDATE retrieved_at=incoming.retrieved_at,
      raw_file_path=incoming.raw_file_path,notes=incoming.notes;
    SET @source_id=(SELECT source_id FROM data_sources WHERE sha256={sql_string(digest)});
    INSERT INTO ingestion_runs (
      ingestion_run_id,source_id,dataset_name,dataset_year,started_at,status,rows_read,rows_rejected
    ) VALUES ({sql_string(run_id)},@source_id,'school_name_anomaly_cleanup',2025,NOW(),'running',
      {len(corrections) + len(invalid)},0);

    CREATE TEMPORARY TABLE tmp_school_name_corrections (
      raw_name VARCHAR(255) CHARACTER SET utf8mb4 COLLATE utf8mb4_bin NOT NULL COMMENT '招生来源中的异常学校名称原值',
      target_school_name VARCHAR(255) NOT NULL COMMENT '人工复核后的学校主体正式名称',
      correction_type VARCHAR(32) NOT NULL COMMENT '异常类型：旧称错字截断标点或二级机构',
      notes VARCHAR(512) NOT NULL COMMENT '人工复核依据',
      PRIMARY KEY(raw_name)
    ) ENGINE=InnoDB COMMENT='招生来源学校名称异常修正临时表';
    INSERT INTO tmp_school_name_corrections VALUES
    {correction_values};

    INSERT IGNORE INTO school_names (
      school_id,school_name,name_type,valid_from_year,valid_to_year,source_id
    )
    SELECT s.school_id,c.raw_name,
      CASE WHEN c.correction_type='former' THEN 'former'
           WHEN c.correction_type='secondary_unit' THEN 'campus_alias'
           ELSE 'source_variant' END,
      2025,CASE WHEN c.correction_type='former' THEN 2025 ELSE NULL END,@source_id
    FROM tmp_school_name_corrections c JOIN schools s ON s.official_name=c.target_school_name;

    SET @groups_before=(SELECT COUNT(*) FROM admission_groups WHERE school_id IS NULL);
    UPDATE admission_groups target JOIN tmp_school_name_corrections c ON c.raw_name=target.school_name_raw
      JOIN schools s ON s.official_name=c.target_school_name
    SET target.school_id=s.school_id WHERE target.school_id IS NULL;
    SET @matched_groups=@groups_before-(SELECT COUNT(*) FROM admission_groups WHERE school_id IS NULL);

    SET @major_before=(SELECT COUNT(*) FROM admission_major_lines WHERE school_id IS NULL);
    UPDATE admission_major_lines target JOIN tmp_school_name_corrections c ON c.raw_name=target.school_name_raw
      JOIN schools s ON s.official_name=c.target_school_name
    SET target.school_id=s.school_id WHERE target.school_id IS NULL;
    SET @matched_major=@major_before-(SELECT COUNT(*) FROM admission_major_lines WHERE school_id IS NULL);

    SET @cutoff_before=(SELECT COUNT(*) FROM admission_cutoff_lines WHERE school_id IS NULL);
    UPDATE admission_cutoff_lines target JOIN tmp_school_name_corrections c ON c.raw_name=target.school_name_raw
      JOIN schools s ON s.official_name=c.target_school_name
    SET target.school_id=s.school_id WHERE target.school_id IS NULL;
    SET @matched_cutoff=@cutoff_before-(SELECT COUNT(*) FROM admission_cutoff_lines WHERE school_id IS NULL);

    CREATE TEMPORARY TABLE tmp_invalid_school_rows (
      raw_name VARCHAR(255) CHARACTER SET utf8mb4 COLLATE utf8mb4_bin NOT NULL COMMENT '误入学校名称字段的原始文本',
      reason VARCHAR(512) NOT NULL COMMENT '判定无效数据行的原因',
      PRIMARY KEY(raw_name)
    ) ENGINE=InnoDB COMMENT='确认无效的招生来源行清理规则临时表';
    INSERT INTO tmp_invalid_school_rows VALUES
    {invalid_values};

    SET @deleted_groups=(SELECT COUNT(*) FROM admission_groups g JOIN tmp_invalid_school_rows i ON i.raw_name=g.school_name_raw);
    DELETE g FROM admission_groups g JOIN tmp_invalid_school_rows i ON i.raw_name=g.school_name_raw;
    SET @deleted_major=(SELECT COUNT(*) FROM admission_major_lines m JOIN tmp_invalid_school_rows i ON i.raw_name=m.school_name_raw);
    DELETE m FROM admission_major_lines m JOIN tmp_invalid_school_rows i ON i.raw_name=m.school_name_raw;
    SET @deleted_cutoff=(SELECT COUNT(*) FROM admission_cutoff_lines c JOIN tmp_invalid_school_rows i ON i.raw_name=c.school_name_raw);
    DELETE c FROM admission_cutoff_lines c JOIN tmp_invalid_school_rows i ON i.raw_name=c.school_name_raw;

    INSERT INTO data_quality_issues (
      entity_type,entity_key,issue_code,severity,field_name,issue_message,
      issue_fingerprint,resolution_status,detected_at,resolved_at,resolved_by,resolution_note
    )
    SELECT 'admission_group',CAST(g.admission_group_id AS CHAR),'COMPOSITE_ADMISSION_UNIT','info','school_id',
      '该招生单元下的专业分别由多所民航院校承办，不应绑定单一学校ID。',
      SHA2(CONCAT('admission_group|',g.admission_group_id,'|COMPOSITE_ADMISSION_UNIT'),256),
      'false_positive',NOW(),NOW(),'system_import',
      '保留school_id为空；7个专业计划的承办院校记录在专业名称与备注中。'
    FROM admission_groups g
    WHERE g.admission_year=2025 AND g.province_name='上海市'
      AND g.school_name_raw='民航飞行技术类'
    ON DUPLICATE KEY UPDATE resolution_status='false_positive',resolved_at=NOW(),
      resolved_by='system_import',resolution_note=VALUES(resolution_note);

    UPDATE dataset_coverage SET record_count=(
      SELECT COUNT(*) FROM admission_cutoff_lines l
      WHERE l.admission_year=2025 AND l.province_name='山西省'
    ),notes='GaokaoCompass 2025院校投档线CSV原始1097条；259条缺院校名称未入业务表，31条误入表头已清理，保留807条有效记录且全部关联稳定学校ID。'
    WHERE dataset_name='院校专业组投档线' AND admission_year=2025
      AND province_name='山西省';

    UPDATE ingestion_runs SET finished_at=NOW(),status='success',
      rows_updated=@matched_groups+@matched_major+@matched_cutoff,
      rows_rejected=@deleted_groups+@deleted_major+@deleted_cutoff,
      message=CONCAT('补匹配专业组',@matched_groups,'条、专业线',@matched_major,
        '条、投档线',@matched_cutoff,'条；删除误入表头或类别行',
        @deleted_groups+@deleted_major+@deleted_cutoff,'条。')
    WHERE ingestion_run_id={sql_string(run_id)};
    COMMIT;
    """
    mysql_run(args, sql)
    print(mysql_run(args, "SELECT message FROM ingestion_runs WHERE ingestion_run_id=" + sql_string(run_id) + ";", True).strip())


if __name__ == "__main__":
    main()