Record download stats for resources (extension)

[rgrp]

As a User (especially as a Package Owner/Maintainer?) I want to know how many times a resource has been downloaded (and when).

So let's record download stats (as in clicks on the link for a resource).

Implementation

• Use the existing support for this feature in google analytics or piwiki
• For google analytics see:
  • Integrate to record: ​http://www.google.com/support/analytics/bin/answer.py?answer=55529
  • Accessing analytics data via API (for showing download counts): ​http://code.google.com/apis/analytics/docs/gdata/gdataDeveloperGuide.html
• Limitations only record downloads by browsers with js turned on.
  • This is OK I think (in any case machine download via e.g. datapkg does not get downloaded)

Old Spec (do it in CKAN)

• Record info of form: resource id (or url?), timestamp
• Do this via javascript capturing of onclick event talking to an api
• API: /api/resource/{id}/download
  • POST to increment (how do we stop spamming -- could use a nonce setup with a random string set on each page load for the js)
  • GET to get data back { total: X, day_count: [ [yyyy-mm-dd, count], ... ] }

Questions

• Do we record ip addresses (to handle de-botting etc)?
• Do we count preview clicks as well?

[sebbacon]

Could consider using third-party analytics tracking here, which will also give referrer etc data for free?

Would probably be bes provided in the form of optional piwik or google analytics integration. Being able to say in the UI how many downloads there have been would need piwik.

[sebbacon]

I did a very quick hacky thing at the end of last week on top of the "insert google analytics code" extension we discussed, to work out "most popular packages" based off data harvested from the Google Analytics API. Needs making generic, tests etc but could be a starting point: ​https://bitbucket.org/sebbacon/ckanext-googleanalytics/src

[sebbacon]

(and it would also need some proper caching as the GA API is very slow)

[sebbacon]

The current implementation I referenced above will be a good starting point.

Work that remains:

• Add download click tracking to individual download links (currently we just record page views for packages, not downloads)
• Somehow cache the download stats against each package (the Google API is very slow); package reddis or sqlite or similar as a local storage for the extension
• Expose download information in the relevant places in the UI (all users? package owners? where?)

This is about 2 days' work. Unlikely to get it done in this sprint.

[sebbacon]

Completed; software at ​https://bitbucket.org/okfn/ckanext-googleanalytics/src