将查询结果附加到 PostgreSQL 中的同一结果行 - Redshift

我有一个表，有 3 列 A、B、C - 其中 A 不是主键。我们需要为每个不同的 A（按 A 分组）选择 B、C 对，并将结果附加到最终结果集的末尾。这在sql中可能吗？

A | B | C
a1| b1| c1
a1| b2| c2
a1| b3| c3
a2| b1| c2
a2| b2| c5

我需要得到

a1 | (c1,b1) ; (c2,b2);(c3;b3) 
a2 | (c2,b1) ; (c5,b2) 

作为末尾附加的行。 我通常通过 sqlalchemy 执行此操作，然后最终在 Python 中转换数据，有没有一种方法可以直接在 SQL 中执行此操作？

编辑并开放问题：红移中 string_agg() 的替代方案是什么（Postgres 8.0.2） - 有关上述用例的更多信息。

使用 string_agg 我得到ERROR: function string_agg(text, "unknown") does not exist Hint: No function matches the given name and argument types. You may need to add explicit type casts

Edit 2：使用自定义聚合函数添加错误

An error occurred when executing the SQL command:
CREATE FUNCTION cut_semicolon(text) RETURNS text AS $$
BEGIN
  RETURN SUBSTRING($1 FROM 4)

ERROR: unterminated dollar-quoted string at or near "$$
BEGIN
  RETURN SUBSTRING($1 FROM 4)"
  Position: 53

CREATE FUNCTION cut_semicolon(text) RETURNS text AS $$
                                                    ^

Execution time: 0.24s
(Statement 1 of 7 finished)

0 rows affected
END executed successfully

Execution time: 0.22s
(Statement 2 of 7 finished)

An error occurred when executing the SQL command:
$$ LANGUAGE 'plpgsql' IMMUTABLE

ERROR: unterminated dollar-quoted string at or near "$$ LANGUAGE 'plpgsql' IMMUTABLE"
  Position: 1

$$ LANGUAGE 'plpgsql' IMMUTABLE
^

Execution time: 0.22s
(Statement 3 of 7 finished)

An error occurred when executing the SQL command:
CREATE FUNCTION concat_semicolon(text, text) RETURNS text AS $$
BEGIN
  RETURN $1 || ' ; ' || $2

ERROR: unterminated dollar-quoted string at or near "$$
BEGIN
  RETURN $1 || ' ; ' || $2"
  Position: 62

CREATE FUNCTION concat_semicolon(text, text) RETURNS text AS $$
                                                             ^

Execution time: 0.22s
(Statement 4 of 7 finished)

0 rows affected
END executed successfully

Execution time: 0.22s
(Statement 5 of 7 finished)

An error occurred when executing the SQL command:
$$ LANGUAGE 'plpgsql' IMMUTABLE

ERROR: unterminated dollar-quoted string at or near "$$ LANGUAGE 'plpgsql' IMMUTABLE"
  Position: 1

$$ LANGUAGE 'plpgsql' IMMUTABLE
^

Execution time: 0.22s
(Statement 6 of 7 finished)

An error occurred when executing the SQL command:
CREATE AGGREGATE concat_semicolon(
  BASETYPE=text,
  SFUNC=concat_semicolon,
  STYPE=text,
  FINALFUNC=cut_semicolon,
  INITCOND=''
)

ERROR: SQL command "CREATE AGGREGATE concat_semicolon(
  BASETYPE=text,
  SFUNC=concat_semicolon,
  STYPE=text,
  FINALFUNC=cut_semicolon,
  INITCOND=''
)" not supported.

Execution time: 0.23s
(Statement 7 of 7 finished)


5 statements failed.
Script execution finished
Total script execution time: 1.55s

还查看了 Google 群组中的相关答案，&看起来像是替换分隔符“;”可能有帮助？- 虽然我不确定，哪个；在此函数定义中替换。 参考 ：https://groups.google.com/forum/#!topic/sql-workbench/5LHVUXTm3BI https://groups.google.com/forum/#!topic/sql-workbench/5LHVUXTm3BI

Edit 3:也许 Redshift 不支持 create 函数本身？ “错误：不支持创建函数”2013 年的帖子这样说 forums.aws.amazon.com/thread.jspa?threadID=121137

Edit 4 :

select A, concat(concat(concat(C, ',' ) , cast(B as varchar)), ',')
from  my_table
group by A,B,C


-- Is it ok to group by all A,B, C - since I can't group by A alone, which removes the related "C" columns-- 

gives -:
a1 c1b1b2b3
a2 c2b1b2

但不是 C 的所有条目（并且带有分号）

a1 c1,b1;c2,b2;c2,b3
a2 c2,b1;c5,b2

但我想要 & 之间的逗号还需要知道 A、B、C 分组是否可以？

PostgreSQL

SELECT
  a,
  STRING_AGG('(' || c || ',' || b || ')', ' ; ')
FROM
  tbl
GROUP BY
  a;

Edit: 对于 9.0 之前（引入 STRING_AGG 时）甚至 8.4 之前（添加 ARRAY_AGG 时）的 PostgreSQL 版本，您可以创建自己的自定义聚合函数 http://www.postgresql.org/docs/8.0/interactive/sql-createaggregate.html.

Edit 2：对于 8.0 之前的版本（也许 Amazon Redshift 是基于 PostgreSQL 7.4 的）不支持 $$ 语法，因此函数体需要用引号引起来，并且函数体内的引号需要转义。

CREATE FUNCTION cut_semicolon(text) RETURNS text AS '
BEGIN
  RETURN SUBSTRING($1 FROM 4);
END;
' LANGUAGE 'plpgsql' IMMUTABLE;


CREATE FUNCTION concat_semicolon(text, text) RETURNS text AS '
BEGIN
  RETURN $1 || '' ; '' || $2;
END;
' LANGUAGE 'plpgsql' IMMUTABLE;

CREATE AGGREGATE concat_semicolon(
  BASETYPE=text,
  SFUNC=concat_semicolon,
  STYPE=text,
  FINALFUNC=cut_semicolon,
  INITCOND=''
);

然后使用该聚合来代替。

SELECT
  a,
  CONCAT_SEMICOLON('(' || c || ',' || b || ')')
FROM
  tbl
GROUP BY
  a;

MySQL

SELECT
  a,
  GROUP_CONCAT(CONCAT('(', c, ',', b, ')') SEPARATOR ' ; ')
FROM
  tbl
GROUP BY
  a;