FOR 循环声明部分中的语法错误

Question

我正在尝试在 Postgres 中使用 FOR 循环来计算我的数据库中每个地理位置的（变量）范围内的数据平均值。我正在使用 python/psycopg2。代码如下：

query ='''
DECLARE geoids RECORD;
BEGIN
    RAISE NOTICE 'Beginning average calculation';
    FOR geoids IN select*from census_blocks
        WHERE ST_contains((select geom from census_cbsa
             WHERE cbsafp10='%s'),census_blocks.geom)
        LOOP
        INSERT INTO temp_avgs VALUES
            (geoids,
            select avg(select alljobs from accessibility_results
            WHERE geoid=geoids
                AND deptime BETWEEN '%s' and '%s'
                AND threshold='%s')
        END LOOP;
END;
'''

我收到的错误是

psycopg2.ProgrammingError: syntax error at or near "RECORD"
LINE 2:  DECLARE geoids RECORD;

如果我删除此DECLARE 语句（因为我相信选择值上的LOOP 变量会自动声明为RECORD），错误变为：

psycopg2.ProgrammingError: syntax error at or near "RAISE"
LINE 4:   RAISE NOTICE 'Beginning average calculation';

我应该如何正确格式化这个查询？

Answer 1

带循环的程序解决方案

您正在使用 PL/pgSQL 代码，但正试图将其表述为 SQL 查询。这是不可能的。

使用DO 语句或（因为您想使用参数）创建plpgsql function：

CREATE OR REPLACE FUNCTION foo(_cbsafp10     ?type?     -- replace with ...
                              ,_deptime_from timestamp? -- ... actual data types
                              ,_deptime_to   timestamp?
                              ,_threshold    ?type?)
  RETURNS void AS
$func$
DECLARE
   rec RECORD;
BEGIN

FOR rec IN
   SELECT b.*
   FROM   census_blocks b
   JOIN   census_cbsa   c ON ST_contains(c.geom, b.geom)
   WHERE  c.cbsafp10 = _cbsafp10
LOOP
   INSERT INTO temp_avgs    -- you might add a target list for safety. depends ..
   SELECT rec.*, avg(alljobs)
   FROM   accessibility_results
   WHERE  geoid = rec.geoid -- assuming you join on column "geoid"?
   AND    deptime BETWEEN _deptime_from AND _deptime_to
   AND    threshold = _threshold;
END LOOP;

END
$func$ LANGUAGE plpgsql;

确保在客户端中正确转义引号！

• 选择值上的 LOOP 变量不会自动声明为任何值。
• 替换不必要的子查询。
• 第二个错误消息的直接原因：RAISE 是一个 plpgsql 命令，不是一个 SQL 命令。

卓越的基于集合的解决方案

这将演示 plpgsql 函数的基础知识。但是，正如@Gordon 已经提供的那样，只需使用单个INSERT 语句做同样的事情。进一步解开：

INSERT INTO temp_avgs   -- you might add a target list for safety. depends ..
SELECT b.*, avg(alljobs)
FROM   census_cbsa           c
JOIN   census_blocks         b ON ST_contains(c.geom, b.geom)
JOIN   accessibility_results a ON a.geoid = b.geoid     
WHERE  c.cbsafp10 = %s
AND    a.deptime BETWEEN %s AND %s
AND    a.threshold = %s
GROUP  BY b.geoid;    -- assuming b.geoid is the primary key

Answer 2

如何将其表述为单选。像这样的：

INSERT INTO temp_avgs(geoid, average)
   select cb.geoid, avg(aj.alljobs)
   from census_blocks cb join
        accessibility_results ar
        on cb.geoid = ar.geoid
   WHERE ST_contains((select geom from census_cbsa WHERE cbsafp10='%s'), cb.geom) and
         ar.deptime BETWEEN '%s' and '%s' AND ar.threshold='%s'
   group by cb.geoid;

我不确定所有列的来源，所以表别名有点猜测。