使用自己的数组选择。PostgreSQL 9.5

我有以下模式：

create table reports (id bigserial primary key);
create table scens (id bigserial primary key,report_id bigint references reports(id), path_id bigint);
create table runs (id bigserial primary key,scen_id bigint references scens(id));
create table fails (
    id bigserial primary key,
    run_id bigint references runs(id),
    type varchar not null
);

create table errors (
    id bigserial primary key,
    run_id bigint references runs(id),
    type varchar not null
);
INSERT INTO reports(id)
    VALUES (1);
INSERT INTO reports(id)
    VALUES (2);

INSERT INTO scens(id, report_id)
    VALUES (555, 1);
INSERT INTO scens(id, report_id)
    VALUES (666, 2);

INSERT INTO runs(id, scen_id)
    VALUES (1, 555);
INSERT INTO runs(id, scen_id)
    VALUES (2, 666);
INSERT INTO runs(id, scen_id)
    VALUES (3, 666);
INSERT INTO runs(id, scen_id)
    VALUES (4, 666);

INSERT INTO errors(id, run_id, type)
    VALUES (DEFAULT, 2, 'ERROR2 TYPE');
INSERT INTO errors(id, run_id, type)
    VALUES (DEFAULT, 3, 'ERROR2 TYPE');

INSERT INTO fails(id, run_id, type)
    VALUES (DEFAULT, 1, 'Attachment Journal Mismatch');
INSERT INTO fails(id, run_id, type)
    VALUES (DEFAULT, 2, 'Appeared new difficulty');
INSERT INTO fails(id, run_id, type)
    VALUES (DEFAULT, 2, 'Parameters Mismatch');
INSERT INTO fails(id, run_id, type)
    VALUES (DEFAULT, 3, 'Attachment Journal Mismatch');
INSERT INTO fails(id, run_id, type)
    VALUES (DEFAULT, 3, 'Appeared new difficulty');
INSERT INTO fails(id, run_id, type)
    VALUES (DEFAULT, 3, 'Parameters Mismatch');

我需要找到一个运行（实际上更多的运行），它有错误ERROR2 TYPE并且失败“出现新的困难”和“参数不匹配”

正确的

1 -> 'ERROR2 TYPE'
1 -> 'Appeared new difficulty'
1 -> 'Parameters Mismatch'

不正确

2 -> 'Parameters Mismatch'
2 -> 'Appeared new difficulty'

因此，请求应查找请求失败且错误属于特定运行的所有运行：

1

运行不适合我，因为它没有“参数不匹配”和错误ERROR2类型。

2

运行适合我，因为它有所有错误（ERROR2类型）和所有失败（“出现新困难”和“参数不匹配”）。

3

运行适合我，因为它有所有错误（ERROR2类型）和所有失败（“出现新困难”和“参数不匹配”）。如果运行的失败次数（它有新的失败：“附件日志不匹配”）比我们请求的次数多，则可以，但它必须至少有请求的失败和错误

我怎么做

以下sql根本不起作用（它什么也找不到）：

该代码是soens；t也不起作用，它只搜索1个匹配项，但不搜索整组（'出现新困难'，'参数不匹配'）：

---

看来我需要一个十字路口。

这是我的理解：

select
    reports.id as report,
    scens.id as scen,
    scens.path_id as path,
    runs.id as run,
    array_agg(distinct fails.type) as fails,
    array_agg(distinct errors.type) as errors
from
    reports
    inner join
    scens on reports.id = scens.report_id
    inner join
    runs on runs.scen_id = scens.id
    left join
    fails on fails.run_id = runs.id
    left join
    errors on errors.run_id = runs.id
where reports.id = 2
group by 1,2,3,4
having
    array['Parameters Mismatch','Appeared new difficulty']::varchar[] <@ array_agg(fails.type) and
    array['ERROR2 TYPE']::varchar[] <@ array_agg(errors.type)
;
 report | scen | path | run |                                      fails                                      |     errors      
--------+------+------+-----+---------------------------------------------------------------------------------+-----------------
      2 |  666 |      |   2 | {"Appeared new difficulty","Parameters Mismatch"}                               | {"ERROR2 TYPE"}
      2 |  666 |      |   3 | {"Appeared new difficulty","Attachment Journal Mismatch","Parameters Mismatch"} | {"ERROR2 TYPE"}

选择
reports.id作为报告，
scens.id作为scen，
scens.path\u id作为路径，
runs.id作为run，
数组_agg（distinct fails.type）作为失败，
数组_agg（distinct errors.type）作为错误
从…起
报告
内连接
报告上的scens.id=scens.report\u id
内连接
在runs.scen_id=scens.id上运行
左连接
失败时失败。运行\u id=runs.id
左连接
errors.run_id=runs.id上的错误
其中reports.id=2
按1,2,3,4分组
有
数组['Parameters Mismatch'，'Exeated new Demobility']：：varchar[]我已经用sql解决了这个问题：

select runs.id
    from reports
    inner join scens s on reports.id=s.report_id
    inner join runs on runs.scen_id=s.id
    where exists(select * from fails where fails.path_id=s.path_id and fails.type='Rotation Mismatch')
    and exists(select * from fails where fails.path_id=s.path_id and fails.type='Disappeared inspection')
    and reports.id=2

主题可以关闭

select
    reports.id as report,
    scens.id as scen,
    scens.path_id as path,
    runs.id as run,
    array_agg(distinct fails.type) as fails,
    array_agg(distinct errors.type) as errors
from
    reports
    inner join
    scens on reports.id = scens.report_id
    inner join
    runs on runs.scen_id = scens.id
    left join
    fails on fails.run_id = runs.id
    left join
    errors on errors.run_id = runs.id
where reports.id = 2
group by 1,2,3,4
having
    array['Parameters Mismatch','Appeared new difficulty']::varchar[] <@ array_agg(fails.type) and
    array['ERROR2 TYPE']::varchar[] <@ array_agg(errors.type)
;
 report | scen | path | run |                                      fails                                      |     errors      
--------+------+------+-----+---------------------------------------------------------------------------------+-----------------
      2 |  666 |      |   2 | {"Appeared new difficulty","Parameters Mismatch"}                               | {"ERROR2 TYPE"}
      2 |  666 |      |   3 | {"Appeared new difficulty","Attachment Journal Mismatch","Parameters Mismatch"} | {"ERROR2 TYPE"}

select runs.id
    from reports
    inner join scens s on reports.id=s.report_id
    inner join runs on runs.scen_id=s.id
    where exists(select * from fails where fails.path_id=s.path_id and fails.type='Rotation Mismatch')
    and exists(select * from fails where fails.path_id=s.path_id and fails.type='Disappeared inspection')
    and reports.id=2