Seems like a problem in english too:

https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/en.wikinews.org/all-access/user/Men%27s_Health/daily/2016030213/2016030913

I'm not sure if this is analytics it wiki escaping issue. The HTML escaping
happens before URL encoding

Hi,
This requests works: https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/ru.wikipedia/all-access/all-agents/Men's%20Health/daily/2016010100/2016010123
We don't have knowledge of russian characters, can you help us precise the issue?
Thanks

{{Graph:PageViews|100|Men's Health}} on en wiki works fine, encoding the value as https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/en.wikipedia.org/all-access/user/Men%27s%20Health/daily/2015120517/2016031417
or another words -- Men%27s%20Health

{{Graph:PageViews}} on the Men's Health article tries to get the data from https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/en.wikipedia.org/all-access/user/Men%26%2339%3Bs%20Health/daily/2015120517/2016031417
which is incorrect: Men%26%2339%3Bs%20Health

The problem is in the template itself - {{urlencode:{{{2|{{ARTICLEPAGENAME}} }}}|PATH}} produces Men%26%2339%3Bs%20Health. Suggestions?

This problem can easily be solved with Lua, or decoding html entries, e.g. {{#htmldecode | {{ARTICLEPAGENAME}} }} (doesn't exist). The {{ARTICLEPAGENAMEE}} is not good because it uses URL query encoding, not URL path encoding (first does not encode the '/' symbol). The problem is that having Lua reduces portability of the template - lua code has to be kept in sync as well.

Sorry, no suggestions from me, I think the pageview API is doing the right thing, just blindly decoding.

Change 289131 had a related patch set uploaded (by Yurik):
Add {{urlencode:...|REPATH}} magic keyword

https://gerrit.wikimedia.org/r/289131

Change 289131 abandoned by Yurik:
Add {{urlencode:...|REPATH}} magic keyword

https://gerrit.wikimedia.org/r/289131