Question

目前，我将一些模拟的结果存储在具有以下结构的SQLite3表中

Table: radiation
id, timestamp,           surface_total_shortwave, cell_id
1 , 2010-01-01 09:00:00, 3.5,                     1 
2 , 2010-01-01 10:00:00, 2.5,                     1 
3 , 2010-01-01 11:00:00, 10.0,                    1 
4 , 2010-01-01 12:00:00, 6.5,                     1 
5 , 2010-01-01 09:00:00, 2.5,                     2 
6 , 2010-01-01 10:00:00, 1.5,                     2 
7 , 2010-01-01 11:00:00, 10.0,                    2 
8 , 2010-01-01 12:00:00, 5.5,                     2 
.., ..................., .....,    .
100 , 2010-01-01 09:00:00, 1.5,                   34 
101 , 2010-01-01 10:00:00, 1.5,                   34 
102 , 2010-01-01 11:00:00, 4.0,                   34 
103 , 2010-01-01 12:00:00, 3.5,                   34 
104 , 2010-01-01 09:00:00, 1.5,                   45 
105 , 2010-01-01 10:00:00, 2.5,                   45 
106 , 2010-01-01 11:00:00, 7.0,                   45 
107 , 2010-01-01 12:00:00, 2.5,                   45 
.., ..................., .....,    .

实际创建声明：

CREATE TABLE cfd(id INTEGER PRIMARY KEY, time DATETIME, u, cell_id integer)

对于每个cell_id，我都有相同数量的时间戳。

我需要减去（cell1 - cell2）组合列表的时间序列，将其分配给cell1并创建一个用户可以查询的view。

例如：

cell1 = 1 and cell2 = 34
cell1 = 2 and cell2 = 45 (duplicates are possible for cell2)

对于一个单细胞组合，我做

create view ds as
select time1, (sts1 - sts2) as sts, cell_id from
    (select time as time1, cell_id, surface_total_shortwave as sts1 from radiation where cell_id = 1)
    inner join
    (select time as time2, surface_total_shortwave as sts2 from radiation where cell_id = 34)
    on time1 = time2

当我有像

这样的映射时，如何扩展此查询

(cell1, cell2)
(1, 34)
(2, 45)
(..., ...)

结果表，假设映射仅针对上面的2个单元格（1 - > 34）和（2 - > 45）将是以下

Table: radiation
id, timestamp,           sts, cell_id
1 , 2010-01-01 09:00:00, 2.0,                     1 
2 , 2010-01-01 10:00:00, 1.0,                     1 
3 , 2010-01-01 11:00:00, 6.0,                     1 
4 , 2010-01-01 12:00:00, 3.0,                     1 
5 , 2010-01-01 09:00:00, 1.0,                     2 
6 , 2010-01-01 10:00:00, -1.0,                    2 
7 , 2010-01-01 11:00:00, 3.0,                     2 
8 , 2010-01-01 12:00:00, 3.0,                     2

EDIT 似乎可能的解决方案是创建一个临时表来存储映射

Table: mapping    
idx, cell1, cell2
1,   1    , 34
2,   2    , 45
.., ..., ...

现在我可以用这种方式重写查询

select time1, (sts1 - sts2) as sts, cell1_id, cell2_id from
    (select time as time1, cell_id as cell1_id, surface_total_shortwave as sts1 from radiation where cell_id in (1, 2))
    inner join
    (select time as time2, cell_id as cell2_id, surface_total_shortwave as sts2 from radiation where cell_id in (34, 45))
    on time1 = time2 and cell1_id = (select mapping.cell1 from mapping where mapping.cell2 = cell2_id)

Answer 1

只需连接多个查询：

CREATE VIEW ds AS

SELECT time,
       r1.surface_total_shortwave - r2.surface_total_shortwave AS sts,
       r1.cell_id
FROM radiation AS r1
JOIN radiation AS r2 USING (time)
WHERE (r1.cell_id, r2.cell_id) = (1, 34)

UNION ALL

SELECT ...
...
WHERE (r1.cell_id, r2.cell_id) = (2, 45);

或者，对所有单元格比较使用单个查询：

CREATE VIEW ds AS
SELECT time,
       r1.surface_total_shortwave - r2.surface_total_shortwave AS sts,
       r1.cell_id
FROM radiation AS r1
JOIN radiation AS r2 USING (time)
WHERE (r1.cell_id, r2.cell_id) = (1, 34)
   OR (r1.cell_id, r2.cell_id) = (2, 45);

（两个查询都需要适当的索引才能有效。哪一个更快取决于数据;你必须自己测量它。）

可以使用临时表进行映射，但只有在有更多映射的情况下才值得。

SQLite从具有多个条件的内部联接创建视图

1 个答案: