How to get several columns from BigQuery?

Question

I am querying the github public dataset on BigQuery. Currently, my best query for what I need looks like the following.

SELECT type, created_at, repository_name FROM [githubarchive:github.timeline]
WHERE
    (created_at CONTAINS '2012-')
AND repository_owner="twitter"
ORDER BY created_at, repository_name;

This gives me all the events ("type") from the repository_owner twitter (or any other user) for all the repositories ("repository_name") that this user owns, but in a single column.

However, what I really want is to have all the events ("type") in columns, one column for each repository ("repository_name"), more or less like this:

bootstrap     commons    twui
WatchEvent    PushEvent  PushEvent
WatchEvent    WatchEvent PushEvent

The timestamp ("created_at") is only relevant as an ordering mechanism. The columns does not need have to be equally long, and the events on a single row does not need to happening at the same time.

I will use this to put the events into the R package TraMineR to do sequence analysis.

How can I accomplish this?

Answer 1

does this work?

SELECT type, created_at, repository_name 
FROM [githubarchive:github.timeline]
WHERE
    (created_at CONTAINS '2012-')
AND repository_owner="twitter"
GROUP BY type,created_at, repository_name;

Answer 2

I'm not sure I understand exactly what you're hoping to accomplish, but it is possible to get columns via something like this:

SELECT type, bootstrap, commons, twui
FROM   (
       SELECT type,
              SUM(IF(repository_name = 'bootstrap', 1, 0)) AS bootstrap,
              SUM(IF(repository_name = 'commons', 1, 0)) AS commons,
              SUM(IF(repository_name = 'twui', 1, 0)) AS twui
       FROM   [githubarchive:github.timeline]
       WHERE  created_at CONTAINS '2012-'
       AND    repository_owner = "twitter"
       GROUP BY type
       )
ORDER BY type