BigQuery 连接两个表并聚合结果答案

【问题标题】：BigQuery joining two tables and aggregating resultsBigQuery 连接两个表并聚合结果
【发布时间】：2019-02-16 20:53:34
【问题描述】：

我试图从本质上连接两个表，找到两者之间的共同数字，找出 call_counterparty 的名称并创建一个结果表。我希望它显示 call_counterparty 的而不是正在查询的号码，在我们的例子中是（from_number）。例如，如果您打电话给我，您将是我的 in_call 交易对手，而我将是您的 out_call 交易对手。如果我给您打电话，您将是我的 out_call 交易对手，而我将是您的 in_call 交易对手。我不确定如何区分“call_counterparty”。

简而言之，被询问的人打电话或接听电话的人是谁，他们的关联姓名是什么？如果要连接的表中不存在它们的名称，则显示 null。假设我的电话号码是 2228881111...假设您甚至可以给自己打电话。

我正在使用的两个 BigQuery 数据库表是：

bigquery.call_log_table

from_number,to_number
2228881111,9997772222
2228881111,8883337777
8883337777,2228881111
2228881111,1112229999
6663332222,2228881111
2228881111,2228881111

bigquery.contact_list_table

number,name
9997772222,John
8883337777,Jill
2228881111,Me

这是我当前的代码，没有左连接或其他：

SELECT CASE WHEN call_log.from_number = 
'2228881111' THEN 
call_log.to_number ELSE call_log.from_number END AS 
call_counterparty,
SUM(call_log.to_number = '2228881111') AS in_call,
SUM(call_log.from_number = '2228881111') AS out_call

FROM `bigquery.call_log_table` AS call_log

WHERE '2228881111' in (call_log.from_number, call_log.to_number)
GROUP BY call_counterparty

当前结果：

call_counterparty,in_call,out_call
9997772222,0,1
8883337777,1,1
1112229999,0,1
6663332222,1,0
2228881111,1,1

我想要的结果：

call_counterparty,counterparty_name,in_call,out_call
9997772222,John,0,1
8883337777,Jill,1,1
1112229999,<null>,0,1
6663332222,<null>1,0
2228881111,Me,1,1

我尝试了各种方法，花了几个小时，用谷歌搜索了很多，但我根本无法弄清楚。任何帮助将不胜感激！

【问题讨论】：

标签： google-bigquery

【解决方案1】：

尝试这样的查询：

WITH call_records AS (
  SELECT 
    CASE WHEN call_log.from_number = '2228881111' THEN call_log.to_number ELSE call_log.from_number END AS call_counterparty
    , SUM(IF(call_log.to_number = '2228881111',1,0)) AS in_call
    , SUM(IF(call_log.from_number = '2228881111',1,0)) AS out_call
  FROM `dataset.call_log_table` AS call_log
  WHERE '2228881111' in (call_log.from_number, call_log.to_number)
  GROUP BY call_counterparty
)
SELECT call_counterparty, contacts.name, in_call, out_call
FROM call_records 
LEFT JOIN `dataset.contact_list_table` as contacts
ON call_records.call_counterparty = contacts.number
GROUP BY call_counterparty, contacts.name, in_call, out_call

【讨论】：

那行得通@vinoaj！我确实必须将contacts.name 定义为contact_name 并按contact_name 代替contacts.name 进行分组，并且效果很好！你太棒了，我非常感谢像你这样愿意提供帮助的开发人员。从字面上看，从第二天 22:00 -> 07:00 开始工作，试图解决这个问题。再次感谢您！

【解决方案2】：

以下是 BigQuery 标准 SQL

#standardSQL
SELECT 
  CASE WHEN call_log.from_number = '2228881111' THEN call_log.to_number ELSE call_log.from_number END AS call_counterparty
  , contacts.name
  , SUM(IF(call_log.to_number = '2228881111',1,0)) AS in_call
  , SUM(IF(call_log.from_number = '2228881111',1,0)) AS out_call
FROM `project.dataset.call_log_table` AS call_log
LEFT JOIN `project.dataset.contact_list_table` AS contacts
ON contacts.number = CASE WHEN call_log.from_number = '2228881111' THEN call_log.to_number ELSE call_log.from_number END
WHERE '2228881111' IN (call_log.from_number, call_log.to_number)
GROUP BY call_counterparty, contacts.name

【讨论】：