Question

我有一个返回~97k记录的查询，因为我正在使用内连接。现在我想使用左连接来显示右表上的空记录（应该是大约300k记录）。这是实际查询（ INNER JOIN ）：

 select artnr as "Article number", ardbez1 as "Article description ", arpadrnr as "Supplier code", cd.ADPPHYSLAND as "Country of this supplier code", p.adrnr as "Parent supplier code", p.adrkname as "Name of parent", pd.ADPPHYSLAND as "Country of parent ", arppreis as "Price", arpwae as "Unit", arzwert as "COO", atatarnr as "HS Code"
    from mic_ccs_artikel, mic_ccs_artikeldetail, mic_ccs_artikelpreis, mic_dna_adressen c, mic_dna_adrdetail cd, mic_dna_adressen p, mic_dna_adrrelation, mic_dna_adrdetail pd, mic_ccs_artikelzoll, mic_ccs_artikeltarif
    where artsid = ardartsid
    and arddatstart <= sysdate
    and nvl(arddatend, '01.01.4000') >= sysdate
    and artsid = arpartsid
    and ARPPREISART = 'VP'
    and arpdatstart <= sysdate
    and nvl(arpdatend, '01.01.4000') >= sysdate
    and c.adrsid = cd.adpadrsid
    and arpadrnr = c.adrnr
    and cd.adpdatvon <= sysdate
    and nvl(cd.adpdatbis, '01.01.4000') >= sysdate
    and c.adrmandant = 'S1'
    and c.adrwerk = 'V1'
    and p.adrmandant = 'S1'
    and p.adrwerk = 'V1'
    and p.adrsid = arlrelsid1
    and c.adrsid = arlrelsid2
    and p.adrsid = pd.adpadrsid
    and pd.adpdatvon <= sysdate
    and nvl(pd.adpdatbis, '01.01.4000') >= sysdate
    and artsid = arzartsid
    and arztyp = 'URLD'
    and arzadrnr = c.adrnr
    and arzdatstart <= sysdate
    and nvl(arzdatend, '01.01.4000') >= sysdate
    and artsid = ataartsid
    and ATAREGION = 'SE' 
    and atatarart='EXPORT'
    and atadatstart <= sysdate
    and nvl(atadatend, '01.01.4000') >= sysdate
    ;

然后我试着＆＃34;转换＆＃34;它 LEFT JOIN ，但它始终显示相同的结果：

select artnr as "Article number", ardbez1 as "Article description ", arpadrnr as "Supplier code", cd.ADPPHYSLAND as "Country of this supplier code", p.adrnr as "Parent supplier code", p.adrkname as "Name of parent", pd.ADPPHYSLAND as "Country of parent ", arppreis as "Price", arpwae as "Unit", arzwert as "COO", atatarnr as "HS Code"
from mic_ccs_artikel, mic_ccs_artikeldetail, mic_ccs_artikelpreis, mic_dna_adressen c, mic_dna_adrdetail cd, mic_dna_adressen p, mic_dna_adrrelation, mic_dna_adrdetail pd, mic_ccs_artikelzoll, mic_ccs_artikeltarif
where artsid = ardartsid(+)
and arddatstart <= sysdate
and nvl(arddatend, '01.01.4000') >= sysdate
and artsid = arpartsid(+)
and ARPPREISART = 'VP'
and arpdatstart <= sysdate
and nvl(arpdatend, '01.01.4000') >= sysdate
and c.adrsid = cd.adpadrsid(+)
and arpadrnr = c.adrnr
and cd.adpdatvon <= sysdate
and nvl(cd.adpdatbis, '01.01.4000') >= sysdate
and c.adrmandant = 'S1'
and c.adrwerk = 'V1'
and p.adrmandant = 'S1'
and p.adrwerk = 'V1'
and p.adrsid = arlrelsid1(+)
and c.adrsid = arlrelsid2
and p.adrsid = pd.adpadrsid
and pd.adpdatvon <= sysdate
and nvl(pd.adpdatbis, '01.01.4000') >= sysdate
and artsid = arzartsid(+)
and arztyp = 'URLD'
and arzadrnr = c.adrnr
and arzdatstart <= sysdate
and nvl(arzdatend, '01.01.4000') >= sysdate
and artsid = ataartsid(+)
and ATAREGION = 'SE' 
and atatarart='EXPORT'
and atadatstart <= sysdate
and nvl(atadatend, '01.01.4000') >= sysdate
;

为什么它总是显示相同的结果？

Answer 1

使用左连接和where子句回应注释中的情绪可能会导致一些意外行为。有关发生这种情况的详细信息，请参阅我的回答here。基本上，您应该切换到使用ANSI语法和别名表。在你的情况下，这看起来像

select * -- select whichever fields you want here
    from mic_ccs_artikel a -- only one table in the from clause
    left join mic_ccs_artikeldetail b on b.someID = a.someID -- join conditions here
    left join mic_ccs_artikelpreis c
        on c.someID = a.someID -- join conidition
        and c.someField = 'Some Property' -- filtering here
    ...

辅助连接条件消除了在where子句中丢弃这些记录的风险。有关详细信息，请参阅my other answer。

将INNER JOIN转换为LEFT JOIN - Sql

1 个答案: