PostgreSQL SELECT 結果具有不同的 ID，更喜歡來自另一個表的特定值

Question

我有一個 some_dictionary 表和一個 some_dictionary_language 表，其中包含多種語言的數據（以及對 some_dictionary 的引用）。

我需要在 some_dictionary.id 上獲得唯一值，更喜歡給定語言的結果，如果不是默認值。

給定 sql：

create table some_dictionary(
    id bigserial primary key,
    some_text varchar(5),
    some_array integer[]
);

create table some_dictionary_language(
    id bigserial primary key,
    some_dictionary_id bigint not null REFERENCES some_dictionary(id),
    language varchar(64) not null,
    name varchar(128) not null
);

insert into some_dictionary (some_text, some_array)
values
    ('text2', '{1, 32, 2}'),
    ('text1', '{5, 9, 1}'),
    ('text4', '{1, 97, 4}'),
    ('text3', '{616, 1, 55}'),
    ('text5', '{8, 1}'),
    ('text6', '{1}');

insert into some_dictionary_language (some_dictionary_id, language, name)
values
    (2, 'POLISH', 'nazwa2'),
    (1, 'ENGLISH', 'name1'),
    (3, 'ENGLISH', 'name3'),
    (2, 'ENGLISH', 'name2'),
    (1, 'POLISH', 'nazwa1'),
    (1, 'SPANISH', 'nombre1'),
    (4, 'SPANISH', 'nombre1'),
    (5, 'ENGLISH', '5name'),
    (6, 'ENGLISH', '6name'),
    (6, 'POLISH', 'nazwa5'),
    (5, 'POLISH', 'nazwa6');

給定條件參數：

langugage = 'POLISH' or if not, default = 'ENGLISH'
phrase in some_text or name = 'na'
element in some_array = 1
page = 1 size = 10

我的 select 聲明沒有區別：

select d.id, d.some_text, d.some_array, dl.name, dl.language  
from some_dictionary d 
join some_dictionary_language dl on d.id = dl.some_dictionary_id
where dl."language" in ('POLISH', 'ENGLISH')
and (d.some_text ilike '%na%' or dl.name ilike '%na%')
and 1 = ANY(d.some_array)

select 結果：

d.id    d.some_text d.some_array    dl.name     dl.lanugage
2   text1       {5,9,1}     nazwa2          POLISH
1   text2       {1,32,2}    name1       ENGLISH
3   text4       {1,97,4}    name3       ENGLISH
2   text1       {5,9,1}     name2       ENGLISH
1   text2       {1,32,2}    nazwa1          POLISH
5   text5       {8,1}       5name       ENGLISH
6   text6       {1}     6name       ENGLISH
6   text6       {1}     nazwa5          POLISH
5   text5       {8,1}       nazwa6          POLISH

預期 select 結果在 d.id 和首選語言 POLISH 上不同，否則默認為 ENGLISH：

d.id    d.some_text d.some_array    dl.name     dl.lanugage
1   text2       {1,32,2}    nazwa1          POLISH
2   text1       {5,9,1}     name2       POLISH
3   text4       {1,97,4}    name3       ENGLISH (default!)
5   text5       {8,1}       nazwa6          POLISH
6   text6       {1}     6name       POLISH

我試圖做這樣的事情：

select distinct on (id) * from (
    select d.id, d.some_text, d.some_array, dl.name, dl.language  
    from some_dictionary d 
    join some_dictionary_language dl on d.id = dl.some_dictionary_id
    where dl."language" in ('POLISH', 'ENGLISH')
    and (d.some_text ilike '%na%' or dl.name ilike '%na%')
    and 1 = ANY(d.some_array)
    order by case when dl."language" = 'POLISH' then 1 end
) sub offset 0 row fetch next 10 rows only;

但它沒有正常工作：

d.id    d.some_text d.some_array    dl.name     dl.lanugage
1   text2       {1,32,2}    nazwa1          POLISH
2   text1       {5,9,1}     name2       ENGLISH
3   text4       {1,97,4}    name3       ENGLISH
5   text5       {8,1}       nazwa6          POLISH
6   text6       {1}     6name       ENGLISH

Answer 1

確定 DISTINCT ON 如何選擇其記錄的 ORDER BY 應該與 DISTINCT ON 本身處於同一級別，而不是在子查詢中。

如果這樣做，錯誤消息會告訴您問題所在，您必須先按 DISTINCT ON 列排序，然后再按決勝列排序。 像這樣：

select distinct on (id) d.id, d.some_text, d.some_array, dl.name, dl.language  
from some_dictionary d 
join some_dictionary_language dl on d.id = dl.some_dictionary_id
where dl."language" in ('POLISH', 'ENGLISH')
and (d.some_text ilike '%na%' or dl.name ilike '%na%')
and 1 = ANY(d.some_array)
order by id, case when dl."language" = 'POLISH' then 1 end;

當 DISTINCT ON 查詢沒有 ORDER BY 時，它只是組成一個僅由 DISTINCT ON 列組成的查詢，而每個組中保留的行可以任意選擇。