Page MenuHomePhabricator

No results for Special:BlankPage or Special:BlankPage/RTRC
Closed, DeclinedPublic

Description

The page views API (or at least the UI) does not recognize either Special:BlankPage or Special:BlankPage/RTRC .

I'm guessing maybe you blacklisted it to not distort the rankings?

It might be useful to exclude it from the rankings, but still allow you to get metrics. In particular, we would like metrics for Special:BlankPage/RTRC . See https://meta.wikimedia.org/wiki/User:Krinkle/Tools/Real-Time_Recent_Changes .

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript

Currently BlankPage is one of the special pages we don't count as pageviews explicitly:

https://github.com/wikimedia/analytics-refinery-source/blob/master/refinery-core/src/main/java/org/wikimedia/analytics/refinery/core/PageviewDefinition.java#L79

The owners of this definition are the research team, so I'd issue this request with them, but we in general we only count pageviews that we think are content our users are consuming. For us, this goes on the radar, pending any change of mind from research.

Please see: https://meta.wikimedia.org/wiki/Research:Page_view#Are_Special:_pages_considered_pageviews.3F

Best way (to date) to get aproximate pageviews for Special:BlankPage/RTRC would be to do a query on webrequest table on the available data for the last 60 days. You could request data for the 2 or 3 days and extrapolate pageviews for that.

This only matches against the url path, which means means it only applies to English. As such, https://nl.wikipedia.org/wiki/Speciaal:LegePagina, for example, is counted at https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/nl.wikipedia.org/all-access/all-agents/Speciaal:LegePagina/daily/20180101/20180801.

Seems like this is a good use case for the X-Analytics information we emit, where the canonical name is exposed. For the above url, as example:

x-analytics: ns=-1;special=Blankpage

That would also take care of any variations because MediaWiki will process it accordingly (e.g. via 'title' query string, or alternate casing given that special page titles are case-insensitive). /cc T196558.

My gadget is terribly abusing how special pages happen to work in MediaWiki, and is not something I think the Analytics team should try to accommodate in their infrastructure.

I'm pretty sure the original use case is no longer current (declining because of that).

Having said that, this question can be answered by querying the webrequests data directly with a query like path LIKE "/wiki/%:%/RTRC" (the variables are for the case-insensitive and localised nature of "Special:Blankpage").

To satisfy my own curiosity, here are my approximated RTRC stats for February 2019.

rtrc-201902.sql
SELECT COUNT(*), uri_host FROM webrequest WHERE year=2019 AND month=2 AND webrequest_source="text" AND uri_path LIKE "/wiki/%:%/RTRC" GROUP BY uri_host;
Total number of sessions openings for RTRC (2019-02)6530

Top 15:

RTRC session startsuri_host
1182nl.wikipedia.org
1062ar.wikipedia.org
850fa.wikipedia.org
348pt.wikipedia.org
275en.wikipedia.org
248mwl.wikipedia.org
230he.wikipedia.org
225commons.wikimedia.org
222it.wikipedia.org
203fr.wiktionary.org
172gl.wikipedia.org
156de.wikipedia.org
148kk.wikipedia.org
128hi.wikipedia.org
106zh.wikipedia.org
103simple.wikipedia.org

This doesn't account for:

  • session starts that involve a browser cache-hit to the RTRC page.
  • sessions starts via index.php?title=.
  • session starts where the url has url-escaped the slash Special:Blankpage/RTRC.