REST API backend for tracking user campaign clicks and redirecting them to appropriate links.

To run the backend locally you must first install the Google App Engine SDK for Python. See the official installation instructions here. The you can run the backend using the following command:

cd <path_to_project_folder>
/usr/bin/python2.7 <path_to_gae_sdk>/dev_appserver.py --host localhost .

Here is a list of few backend API calls using curl. For complete reference see the API reference section.

echo "create a campaign"
curl -X POST -v -H "Authorization: Basic dHJhY2tlcjp0cmFja2Vy" \
-H "Content-type: application/json" \
-d '{"name": "Campaign name", "link": "http://campaign_link.com", "platforms": ["android", "ios"]}' \
http://localhost:8080/api/admin/campaign

echo "generate a click (copy the campaign id from the output above)"
curl -X GET -v http://localhost:8080/api/campaign/6323566249246720/platform/ios

echo "get campaign details"
curl -X GET -v -H "Authorization: Basic dHJhY2tlcjp0cmFja2Vy" \
-H "Content-type: application/json" \ http://localhost:8080/api/admin/campaign/6323566249246720

Deployment to Google App Engine is performed using the appcfg.py script that is provided with the SDK. To deploy it, use the following command:

cd <path_to_project_folder>
/usr/bin/python2.7 <path_to_gae_sdk>/appcfg.py -A <YOUR_PROJECT_ID> update app.yaml
/usr/bin/python2.7 <path_to_gae_sdk>/appcfg.py -A <YOUR_PROJECT_ID> update_indexes .

You need to provide your project id that you can obtain from the Google Cloud Platform Console.

Demo backend deployment is available here: http://click-tracker-1268.appspot.com

Unit tests are provided with the project. Make sure you have a Python package webtest installed. You can install it by running sudo pip install webtest. To run the tests simply invoke the supplied script:

cd <path_to_project_folder>
/usr/bin/python2.7 test_runner.py <path_to_gae_sdk> .

Settings for the backend are applied through the use of environment variables in file app.yaml. You can set the username and password for the private API calls and counter update interval (see section Assumptions for details).

Handles incoming clicks for given campaign_id and platform_name. If click is valid then user is redirected to url defined in the campaign and statistic about this click is saved. All invalid clicks (e.g. for non existing campaigns, platforms) users are redirected to http://outfit7.com.

All access to private endpoints is restricted with HTTP Basic Authorization. All calls therefore need the Authorization: Basic <credentials> header to be set.

List all existing campaigns.

Create a new campaign. The campaign data must be JSON encoded, Content-Type set to 'application/json' and have the following form:

{
    "name": "Name of the campaign",
    "link": "Link to where the user will be redirected",
    "platforms": ["android", "ios", "wp"]
}

The platforms parameter lists on which platforms this campaign will be active. Possible values are android, ios and wp.

Get information about the existing campaign.

Update the existing campaign. Data must be provided the same way as for the creation of a new campaign, however, parameters that are not being updated can be omitted.

Delete the existing campaign.

Retrieves the number of clicks for given campaign on the given platform.

List all existing campaigns available on a given platform.

Retrieve the number of clicks on the given platform.

To circumvent the Google App Engine Datastore limits on the number of updates to entites (limit of 1 update per second) memcache was employed to temporarily store the counter delta values (number of clicks since the last update to the Datastore) and stored into the Datastore at predefined intervals (every platform counter is permanently updated on every N seconds, where N is configurable through the environment variable TRACKER_COUNTER_UPDATE_INTERVAL_LENGTH, see app.yaml). The assumption is that occasional loss (due to memcache) of N seconds worth of clicks is not critical. Memcache also incurs N seconds of delay to the click statistics. However, memcache has some benefits over implementation using sharded counters, such as the cost of read/write operations and speed.