Hadoop tool to import data from the GitHub public timeline XML format.
Run with:
$ hadoop jar hubstats.jar hubstats.HubStats input output
The following fields are output:
- event_id - The unique numeric event id
- event_type - The type of event
- at - The date and time of the event
- actor - The GitHub account that created the event
- repo_account - The GitHub account associated with the related repository
- repo_name - The name of the repository
- branch - The Git branch name
- tag - The Git tag name
- alternate_id - An alternative numeric id if the event relates to an Issue, PullRequest or Gist
- subtype - Further subtype of the event, for example an Issues event can be 'Opened' or 'Closed'
Sample output from a run over 30 gigs of the timeline requests (about 300,000 GitHub events in a bit over 11 days) is available in sample-output.gz
.
$ gzcat sample-output.gz | awk '{print $3}' | sort | uniq -c | sort -nr
173260 Push
48186 Create
34782 Watch
12256 Gist
11756 Issues
9453 Follow
8731 Fork
8202 Gollum
6672 PullRequest
5132 Delete
4027 CommitComment
2527 Member
1666 Download
1053 ForkApply
288 Public