Question

sh-3.2$ bq shell
Welcome to BigQuery! (Type help for more information.)
bqaccount> select 1;
Waiting on bqjob_r42d56b0f0c2a3eb9_00000164b439b383_1 ... (0s) Current status: DONE
+-----+
| f0_ |
+-----+
|   1 |
+-----+
bqaccount> select f0_ from (select 1);
Error in query string: Error processing job 'bqaccount:bqjob_r7d8d3d2925de7c0c_00000164b439def3_1': Unrecognized name: f0_ at [1:8]

是否可以从没有别名的查询中选择特定列？

Answer 1

使用BigQuery旧版SQL可以引用fn_

#legacySQL
SELECT f2_, f0_, f1_
FROM (SELECT 1, 2, 3)

结果是

Row f2_ f0_ f1_  
1   3   1   2

尽管

似乎在BigQuery Standard SQL中不受支持

因此，以下是BigQuery标准SQL的一种解决方法

#standardSQL
SELECT 
  cols[OFFSET(0)],
  cols[OFFSET(1)],
  cols[OFFSET(2)],
  cols[OFFSET(3)],
  cols[OFFSET(4)]
FROM (SELECT 1, 2, 3, 'abc', CURRENT_DATE()) t, 
UNNEST([
  STRUCT<cols ARRAY<STRING>>(REGEXP_EXTRACT_ALL(TO_JSON_STRING(t), r'"[^"]?":"?(\w*)"?'))
])

或另一种完全相同的方法（无论您会发现哪种方式更好）

#standardSQL
SELECT 
  cols[OFFSET(0)],
  cols[OFFSET(1)],
  cols[OFFSET(2)],
  cols[OFFSET(3)],
  cols[OFFSET(4)]
FROM (
  SELECT REGEXP_EXTRACT_ALL(TO_JSON_STRING(t), r'"[^"]?":"?(\w*)"?') cols
  FROM (SELECT 1, 2, 3, 'abc', CURRENT_DATE()) t
)

有结果

Row f0_ f1_ f2_ f3_ f4_  
1   1   2   3   abc 2018

Answer 2

您可以将其强制转换为具有列名称的结构：

SELECT CAST(t AS STRUCT<f0_ INT64, f1_ INT64, f2_ INT64>).f0_
FROM (SELECT 1, 2, 3) AS t

Answer 3

我认为第一个？在...

SELECT REGEXP_EXTRACT_ALL（TO_JSON_STRING（t），r'“ [^”]？“：”？（\ w *）“？'）cols

...应为*或+

如何选择没有别名的bigquery表的第n列？

3 个答案: