通过传递动态表名和列名来计算重复记录

Question

感谢您提前提供的所有帮助。

我需要查找所有表（超过 100 个）中是否存在重复的退出。

我知道按表查找重复项的 SQL，但无论如何我们可以传递 colu 和 table 作为动态运行 sql。

表名和key col需要动态传递

step1 : SQL 获取所有表名和关联的键 col step2：运行重复查询 选择 col1,col2,....col5 计数(1) 从 schema.tablename 按 col1,col2...col 5 分组 有 count(1) > 1

请帮助我避免手动通过 cols 和 tablenames 手动运行的手动操作。

谢谢

Answer 1

您可以使用*_TAB_COLUMNS 视图和LISTAGG 构建您的SQL 脚本。

例如：

select 
  'select '||listagg(column_name, ',') within group (order by column_name)||
  ' from '||table_name||' 
   group by '||listagg(column_name, ',') within group (order by column_name)||' 
   having count(*) >1 ' str
from user_tab_columns where table_name = 'EMPLOYEES'
group by table_name;

HR 模式中EMPLOYEES 表的上述查询的输出（当然格式化了 SQL）：

SELECT
    commission_pct,
    department_id,
    email,
    employee_id,
    first_name,
    hire_date,
    job_id,
    last_name,
    manager_id,
    phone_number,
    salary
FROM
    employees
GROUP BY
    commission_pct,
    department_id,
    email,
    employee_id,
    first_name,
    hire_date,
    job_id,
    last_name,
    manager_id,
    phone_number,
    salary
HAVING
    COUNT(*) > 1;

要遍历架构中的所有表并为所有表生成它，您可以执行以下操作：

DECLARE
  TABLE_COLUMNS VARCHAR2(10000);
  DELETE_STATEMENT VARCHAR2(10000);
  CURSOR ALL_MY_TABLES IS (SELECT TABLE_NAME FROM USER_TABLES);
BEGIN
  FOR MY_TABLE IN ALL_MY_TABLES 
  LOOP
    -- prepare the list of columns
    SELECT LISTAGG(COLUMN_NAME, ',') WITHIN GROUP (ORDER BY 1) INTO TABLE_COLUMNS FROM 
        USER_TAB_COLUMNS WHERE TABLE_NAME = 'YOUR_TABLE_NAME';
    -- prepare the delete statements
    DELETE_STATEMENT := 'DELETE FROM ' || YOUR_TABLE_NAME || ' WHERE ROWID NOT IN 
        (SELECT MIN(ROWID) FROM ' || YOUR_TABLE_NAME 
         || ' GROUP BY ' || TABLE_COLUMNS ||');';
    EXECUTE IMMEDIATE DELETE_STATEMENT;
    -- output of all delete statements in loop
    DBMS_OUTPUT.PUT_LINE(DELETE_STATEMENT);
  END LOOP;
END;
/