pyramid_analytics 0.2

To use pyramid_analytics, you should include the following directive in your configuration file:

pyramid.includes =
    pyramid_analytics

The following configuration options are available. Note, analytics.api_token and analytics.secret are required.:

analytics.api_token = mytoken
analytics.async = True
analytics.flush_after = 10
analytics.flush_at = 20
analytics.max_queue_size = 100000
analytics.secret = mysekret
analytics.send = True

flush_after specifies after how much time (in seconds) of no flushing that the server will flush. Used in conjunction with the flush_at size policy.

To manage the log_level and log options specified by Segment.io’s analytics-python, you should set the following configuration in your .ini file:

[logger_analytics]
level = WARN
handlers = console  # The config needs to include this handler
qualname = analytics
propagate = 0  # This "disables" logging
formatter = generic  # The config needs to include this formatter

Additionally, you’ll want to register this logger:

Under the hood, analytics-python turns logging “off” by setting the logging level to CRITICAL. Setting propagate = 0 is perhaps a more elegant solution.

All other options are detailed at Segment.io’s Python Documentation.

This library attaches an attribute, analytics, to the request object. pyramid_analytics checks for an authenticated_userid with the Pyramid api, and will identify or track data for the authenticated user.

To update the userid that this analytics package makes calls on behalf of, you should send a new UpdatedAnalyticsUserId event. Example:

from pyramid_analytics.events import UpdatedAnalyticsUserId

# ...
# User logged in
event = UpdatedAnalyticsUserId(request, userid, alias=False)
request.registry.notify(event)

This event will update the analytic. If a user registers for your service, you should set alias=True. This does not send any code to segment.io, but instead plays nicely with the Jinja2 template described below.

This analytics package sends as much data directly from your server as it can, and falls back on supplying events to the browser via the Jinja2 template. Below are the examples of identifying and tracking:

# identify the authenticated user with certain traits (timestamp and
# context are optional and described by the segment.io documentation)
request.analytics.identify(traits, timestamp=None, context=None)

# identify a (not identified) user with certain traits (timestamp and
# context are optional and described by the segment.io documentation)
request.analytics.identify_foreign(user_id, traits, timestamp=None,
                                   context=None)

# track an authenticated user with an event and properties (timestamp and
# context are optional and described by the segment.io documentation)
request.analytics.track(event, properties, timestamp=None, context=None)

# track a (not identified) user with an event and properties (timestamp and
# context are optional and described by the segment.io documentation)
request.analytics.track_foreign(user_id, event, properties, timestamp=None,
                                context=None)

When Pyramid fires the BeforeRender event, pyramid_analytics adds the following to the renderer globals:

analytics: {
   api_token: 'mytoken',
   tracking: {
        alias: userid, // or null if analytics hasn't been told to alias
        events: [{event: 'myevent',
                    properties: {property1: 'pvalue1'},
                    context: {context1: 'cvalue1'}}]
        identify: userid, // or null
    }
    tracking_json: '...' // the tracking item as json
}

An easy to use jinja2 template is provided. To use this template, you must add pyramid_analytics to your jinja2 search path. If you’re using Pyramid’s pyramid_jinja2, update your config.ini as shown below:

jinja2.directories =
    pyramid_analytics:templates
    myapp:templates

To use the template, place this in your templates after including segment.io’s javascript:

{# Segment.io's javascript precedes this... #}
{% include 'analytics.jinja2' %}