Question

表DDL：

CREATE TABLE entities_patents (
    entity_id INTEGER REFERENCES entities(entity_id) NOT NULL,
    patent_id INTEGER REFERENCES patents (patent_id) NOT NULL,
    doc_number text NOT NULL,
    created_at timestamp with time zone DEFAULT current_timestamp,
    updated_at timestamp with time zone DEFAULT current_timestamp,
    deleted boolean,
    PRIMARY KEY (entity_id, patent_id)
);

不幸的是，表中有重复项，我之前无法添加pk约束。尝试这样做现在导致：

DETAIL:  Key (entity_id, patent_id)=(123123, 811231333) is duplicated.

有没有办法删除这些副本？

Answer 1

你可以使用ROW_NUMBER（），所以试试这个:(未经测试可能会有一些语法问题）

DELETE FROM entities_patents t
WHERE EXISTS(SELECT * FROM (
                    SELECT p.entity_id,p.patent_id,p.created_at,p.updated_at,
                    ROW_NUMBER() OVER(PARTITION BY p.entity_id,p.patent_id ORDER BY p.created_at DESC,p.updated_at DESC) as rnk
                    FROM entities_patents p) s
              WHERE s.rnk > 1 and s.entity_id = t.entity_id and s.patent_id = t.patent_id
                  AND s.created_at = t.created_at and s.updated_at = t.updated_at)

Answer 2

WITH CTE AS
  (SELECT *,ROW_NUMBER() OVER(PARTITION by entity_id, patent_id
                              ORDER BY created_at) AS rn
   FROM entities_patents)
DELETE
FROM entities_patents
WHERE (entity_id,patent_id) IN (SELECT entity_id,patent_id FROM CTE WHERE rn >1)

Answer 3

使用group by和count

的另一种方法

DELETE 
FROM entities_patents B  
USING (
      select entity_id, patent_id
      from entities_patents
      group by entity_id, patent_id
      having count(*) > 1
) C 
WHERE B.entity_id = C.entity_id AND
      B.patent_id = C.patent_id

如何使用PostgreSQL删除2列的重复行？

3 个答案: