来自 BigQuery Standard SQL 的漂亮打印值 table 在 CLI 上使用 bq 加入了 tables？

Question

有没有一种方法可以使用标准 SQL 使用 bq 命令行工具来漂亮地格式化 Big Query "value table" 结果集？这在使用 Standard SQL 的 BigQuery 控制台中按预期工作，但在 bq 中不起作用。我找不到方便的解决方法。它也可以使用 Legacy SQL.

按预期工作

我的用例：我正在探索 BigQuery 中包含大量列的表。 BigQuery 中的匿名查询要求 selected 列的名称是唯一的，否则您会收到消息 "Duplicate column names in the result are not supported. Found duplicate(s):..."。当我只是探索数据时，这是一个非常严格的要求。幸运的是，通过在 select 表达式...

中使用值表，BigQuery Web 控制台中有一种解决方法

with
left_side as
(
select 1 as id, 'hello' as col1 union all 
select 2 as id, 'there' as col1
)
,right_side as
(
select 1 as id, 'hello' as col1 union all 
select 2 as id, 'world' as col1
)
select a, b
  from left_side a
       join
       right_side b
          on a.id = b.id

...这会产生以下表格结果集，其中自动枚举的列非常好...

然而，当在命令行上传递给 bq 时，完全相同的查询如下（为方便起见，通过此处的文档）。通常，我在文件中有查询并将其重定向到 bq，例如 bq query --use_legacy_sql=false < bq_test4.sql

bq query --use_legacy_sql=false <<BQ_TEST
with
left_side as
(
select 1 as id, 'hello' as col1 union all 
select 2 as id, 'there' as col1
)
,right_side as
(
select 1 as id, 'hello' as col1 union all 
select 2 as id, 'world' as col1
)
select a, b
  from left_side a
       join
       right_side b
          on a.id = b.id
BQ_TEST

...无论传递给 bq...

的任何 --format 选项如何，都会产生相同的基本结果集

+---------------------------+---------------------------+
|             a             |             b             |
+---------------------------+---------------------------+
| {"id":"1","col1":"hello"} | {"id":"1","col1":"hello"} |
| {"id":"2","col1":"there"} | {"id":"2","col1":"world"} |
+---------------------------+---------------------------+

这样的结果集对我没有帮助。

好的，我可以将 json 转换为表格结果集...

with
left_side as
(
select 1 as id, 'hello' as col1 union all 
select 2 as id, 'there' as col1
)
,right_side as
(
select 1 as id, 'hello' as col1 union all 
select 2 as id, 'world' as col1
)
select json_extract_scalar(to_json_string(a), "$['id']") as `a_id`
      ,json_extract_scalar(to_json_string(a), "$['col1']") as `a_col1`
      ,json_extract_scalar(to_json_string(b), "$['id']") as `b_id`
      ,json_extract_scalar(to_json_string(b), "$['col1']") as `b_col1`
  from left_side a
       join
       right_side b
          on a.id = b.id

...这会产生什么，而这正是 Legacy SQL 会产生什么...

+------+--------+------+--------+
| a_id | a_col1 | b_id | b_col1 |
+------+--------+------+--------+
| 1    | hello  | 1    | hello  |
| 2    | there  | 2    | world  |
+------+--------+------+--------+

必须枚举列违背了我的用例的意图。

在使用标准 SQL 时，有什么方法可以避免在加入之前或加入之后枚举列？

Answer 1

仍然需要一些输入 - 但版本不那么冗长

#standardSQL
WITH left_side AS (
  SELECT 1 AS id, 'hello' AS col1 UNION ALL 
  SELECT 2 AS id, 'there' AS col1
), right_side AS (
  SELECT 1 AS id, 'hello' AS col1 UNION ALL 
  SELECT 2 AS id, 'world' AS col1
)
SELECT *
FROM (SELECT NULL a_id, NULL a_col1 UNION ALL SELECT * FROM left_side) a
JOIN (SELECT NULL b_id, NULL b_col1 UNION ALL SELECT * FROM right_side) b
ON a_id = b_id

结果

Row a_id    a_col1  b_id    b_col1   
1   1       hello   1       hello    
2   2       there   2       world

所以在 bq 中将如下所示...

+------+--------+------+--------+
| a_id | a_col1 | b_id | b_col1 |
+------+--------+------+--------+
| 1    | hello  | 1    | hello  |
| 2    | there  | 2    | world  |
+------+--------+------+--------+

来自 BigQuery Standard SQL 的漂亮打印值 table 在 CLI 上使用 bq 加入了 tables？

Pretty print value table from BigQuery Standard SQL joined tables using bq on CLI?

join

google-bigquery

bq