Thrift module: events

Enumeration: Platform

The platform that served this request to the reader.

R2	0
NEXT_GEN	1
IOS_NATIVE_APP	2
ANDRIOD_NATIVE_APP	3

Enumeration: SuspectStatus

VALID	0	This event is valid and should be processed normally
INVALID_APP_RESUME_BUG	1	This event is invalid, becuase we think it was only sent because of a persistent bug in the mobile apps where they erroneously sent page view notifcation after a device suspend-resume
INVALID_WEB_NOT_VISIBLE	2	This event is invalid, becuase although we tracked it, the resulting web page was not visible. In the future, ophan's tracking code will deal with this and report only when the page does subsequently become visible.
INVALID_WEB_NOT_SUCCESSFUL	3	This event does not represent a usual page view, as the user was returned a non-200 sucess code. (Which we track by putting special ophan tracking on the error page displayed.) Unless you're specifically interested in errors, you should ignore these as page views.

Enumeration: EventType

ROBOT	1	NO LONGER USED previously (when ophan processed this data) indicated that the event arose from web server logs with no evidence of actual user interaction
WEB_PAGE_VIEW	2	This represents a page view delivered to a user on a website
WEB_PAGE_VIEW_ADDITIONAL	3	This represents additional information about a page view delivered on the website NOTE: ophan (currently) makes no guarantees about order of message delivery so it's perfectly possible to receive additional events about a page view before the page view itself
NATIVE_APP_PAGE_VIEW	4	Represents data about a page view within our native mobile apps

Typedef: Path

Base type: string

The path extracted from the url (without query string or fragment), including a preceeding "/". So http://www.theguardian.com/uk/sport becomes "/uk/sport".

Note that ophan performs some procesisng on this path - especially from the native mobile apps - to try to make this match up with a valid url on www.theguardian.com

Struct: AssignedId

Key	Field	Type	Description	Requiredness	Default value
1	id	string	The actual id	required
2	isNew	bool	Whether the id was generated and set for the first time on this request. If undefined, it's not known whether the id is a new one or not.	optional

Ophan assigns various ids to things when necessary (on Web, by dropping cookies). This struct indicates the id and whether we freshly assigned an id on this request.

Struct: Geo

Key	Field	Type	Description	Requiredness	Default value
1	lat	double	Latitude	required
2	lon	double	Longitude	required

A geographical location

Struct: Location

Key	Field	Type	Description	Requiredness	Default value
1	xForwardedForHeader	string	The full value of the X-Forwarded-For header supplied. Normally you'll want to ignore this as read the other fields that are derived from that header.	optional
2	ip	string	The ip address of the client	optional
3	geo	Geo	The resolved geolocation of the ip address	optional
4	countryCode	string	The two letter country code of the ip address; Note that guardian internal traffic (within the Guardian offices) is assigned a country code of "GNM"	optional
5	countryName	string	Human readable country name	optional
6	city	string	Human readable city name	optional
7	continent	string	Human readable continent name	optional
8	subdivisions	list<string>	Administrative subdivisions, e.g. "Wales", "Berkshire"	optional

Where this request was made from, derived by ip address lookup.

Struct: Page

Key	Field	Type	Description	Requiredness	Default value
1	url	string	The full URL, potentially including query string and fragment identifier Note that for native mobile apps, we synthesise a url with an invented hostname of "native-app.theguardian.com".	required
2	host	string	The hostname extracted from the url. Note this is the *full* hostname, with no abbreviation performed	optional
3	path	Path	The path extracted from the url. See documentation for the Path typedef for more information.	optional
4	campaignId	set<string>	The values of all of the CMP and INTCMP parameters	optional
5	platform	Platform	The platform that served this page	optional

Details about the page that was served to the user

Struct: Referrer

Key	Field	Type	Description	Requiredness	Default value
1	raw	string	the raw value of document.referrer as reported by the browser, or invented for mobile apps	required
2	simplifiedHost	string	the host that this referral came from Note that in most cases we trim down the hostname included in "raw" to only include the first value before the top level domain, so m.facebook.com => facebook.com Exceptions include google (not trimmed, so it's easy to see the difference between www. and news.); outbrain (not trimmed, to identify between traffic.outbrain.com and paid.outbrain.com. Also, t.co is mapped to twitter.com	optional
3	path	Path	the path of the referrer, if the previous page was served by the guardian	optional
4	component	string	the component that was clicked on for this referral, if the previous page was served by the guardian	optional
5	platform	Platform	the platform of the referrer, if the previous page was served by the guardian	optional
6	viewId	string	the viewId of the referrer, if the previous page was served by the guardian	optional
7	email	string	if this was from a guardian email, what email it was	optional

Information about the referrer - previous page - that the reader navigated to this one from.

Struct: PageView

Key	Field	Type	Description	Requiredness	Default value
1	validity	SuspectStatus	Whether we view this event as "suspect". most consumers of this stream should ignore events that are do not have a status of VALID.	optional	UNKNOWN
2	page	Page	Details about the page displayed	optional
3	userAgent	ua.UserAgent	The user agent that made this request	optional
4	location	Location	Details about the location of the user made this request	optional
5	referrer	Referrer	Details about the referrer to this page view Will only be present on _PAGE_VIEW events where a referer was received	optional
6	httpStatus	i16	The http status returned to the user for this page view	required	200

Details about a page view - only populated for _PAGE_VIEW event types

Struct: Event

Key	Field	Type	Description	Requiredness	Default value
1	eventType	EventType	The type of this event, from the list above. Note: this is marked as optional because what thrift does when new event types are added to the enum above that your version of this definition doesn't understand, this field will become undefined. You should probably ignore events where this field is undefined.	optional
2	uniqueEventId	string	Unique id associated with this event. Ophan never makes better than at-least-once delivery promises, so you must ensure that processing two events with the same uniqueEventId has no effect	required
3	dt	i64	The date time (in millis since epoch UTC) at which this event occurred.	required
4	pageViewId	string	The page view for which this event is associated. Ophan may send multiple events relating to the same page view, which may contain updates to any previously supplied data or new data. You should treat the one with the highest timestamp (dt) as the most accurate.	optional
5	browserId	AssignedId	The unqiue id associated with this browser. Currently this is maintained by setting a cookie for web events, or otherwise determined for native apps.	optional
6	visitId	AssignedId	The unique id associated with this "visit". For web reports, the visit id is a refreshed session cookie that expires after 30 minutes of activity. Mobile apps do not currently set this value.	optional
7	userId	string	If the user is logged in, the identity user id.	optional
8	pageView	PageView	For *_PAGE_VIEW event types, details about the page view	optional

Struct: EventBatch

Key	Field	Type	Description	Requiredness	Default value
1	events	list<Event>		default

Used internally within ophan to e.g. put batches of tracker events on a queue.