sql - 如何仅在不存在时插入

Question

我想在表中插入数据，但如果它已经存在于表中，那么它不应该添加。
这是我想出的代码，但它仍然添加具有相同值的多个数据。

insert into nol_art_izm([ART_ID],[DAT])
    select distinct
        v.id_art, {fn now()}
    from
        openxml(@hDoc, '/art_kompl/nol_voac') with #vc xd
        inner join nol_voac v on xd.id_art = v.id_art
    where
        not exists(select * from nol_art_izm where nol_art_izm.art_id=xd.id_art)

我希望没有任何重复的“ART_ID”值

Answer 1

注意:此答案仅适用于 SQL Server 2008...

使用 MERGE陈述。 MERGE 语句的优点是它清楚地表达了仅在没有匹配项时才想插入的意图。对于 future 的读者，这可能会有所帮助，因为涉及 INSERT .. SELECT 的替代方案更难破译。

-- This is where you're "merging" data into
MERGE INTO nol_art_izm dst

-- This is your merge data source
USING (
  -- Use DISTINCT here, to prevent possible duplicates from the below INNER JOIN
  SELECT DISTINCT v.id_art 
  FROM openxml(@hDoc, '/art_kompl/nol_voac') with #vc xd
  INNER JOIN nol_voac v on xd.id_art = v.id_art
) src

-- This is your "join" condition, used to decide whether to perform an
-- INSERT or UPDATE
ON (dst.art_id = src.id_art)

-- When source and target don't match (see ON clause), perform an insert
WHEN NOT MATCHED THEN INSERT ([ART_ID],[DAT])
  VALUES (src.id_art, {fn now()})

此语句省略了 WHEN MATCHED THEN UPDATE 子句，因为您只对执行 INSERTs 感兴趣，而不是 UPDATEs