Raw BigQuery Log-data for 1st Day Installers

问题

I am fairly new to StandardSQL, and wanted to export Raw BigQuery log-data for users that installed on a given date (1st June 2017) with first 3-week's session data on and after install-date (D0-D20).

    #StandardSQL
    SELECT user_dim.first_open_timestamp_micros, user_dim.app_info.app_instance_id, event_dim.date, event_dim.name, event_dim.timestamp_micros, event_dim.previous_timestamp_micros
    FROM `your_table_id.app_events_*`
    WHERE _TABLE_SUFFIX BETWEEN '20170601' AND '20170621' AND
    (SELECT user_dim.first_open_timestamp_micros
    FROM UNNEST(user_dim) AS user
    WHERE user.first_open_timestamp_micros BETWEEN 1496275200000 AND 1496361600000);

I am only getting the following output: "Values referenced in UNNEST must be arrays. UNNEST contains expression of type STRUCT". My initial question was: How would I convert this into an array, to give me the appropriate output ?

Update after Mikhail's answer:

The following query executed, but did not give any data as output. I know there should be users that should show up, because if I preview the actual table (view screenshot), I see values that exist in the table that should be part of the output (but is not).

#standardSQL
SELECT 
  user_dim.first_open_timestamp_micros, 
  user_dim.app_info.app_instance_id, 
  event.date, event.name, 
  event.timestamp_micros, 
  event.previous_timestamp_micros
FROM `your_table_id.app_events_*`, UNNEST(event_dim) AS event
WHERE _TABLE_SUFFIX BETWEEN '20170601' AND '20170621' 
AND user_dim.first_open_timestamp_micros BETWEEN 1496275200000 AND 1496361600000
LIMIT 10;

Screenshot of the actual table:

Converting the timestamps, shows that the above session should indeed show up in the middle of the 2 user_dim.first_open_timestamp_micros-results, but does not...

1496275200000 Converted: Assuming that this timestamp is in milliseconds: GMT: Thursday, June 1, 2017 12:00:00 AM Your time zone: Thursday, June 1, 2017 2:00:00 AM GMT+02:00 DST

1496353126947000 Converted: Assuming that this timestamp is in microseconds (1/1,000,000 second): GMT: Thursday, June 1, 2017 9:38:46.947 PM Your time zone: Thursday, June 1, 2017 11:38:46.947 PM GMT+02:00 DST

1496361600000 Converted: Assuming that this timestamp is in milliseconds: GMT: Friday, June 2, 2017 12:00:00 AM Your time zone: Friday, June 2, 2017 2:00:00 AM GMT+02:00 DST

Question:

Am I missing something, why would that value not show up in the output ?

回答1:

Try below
I assumed - user_dim is record, and event_dim is record, repeated

#standardSQL
SELECT 
  user_dim.first_open_timestamp_micros, 
  user_dim.app_info.app_instance_id, 
  event.date, event.name, 
  event.timestamp_micros, 
  event.previous_timestamp_micros
FROM `your_table_id.app_events_*`, UNNEST(event_dim) AS event
WHERE _TABLE_SUFFIX BETWEEN '20170601' AND '20170621' 
AND user_dim.first_open_timestamp_micros BETWEEN 1496275200000 AND 1496361600000

来源：https://stackoverflow.com/questions/45920753/raw-bigquery-log-data-for-1st-day-installers

标签

google-bigquery

firebase-analytics