Facetting endpoint

[cutoffthetop]

When enabling facet_field on the content endpoint, the resulting output is quite obscure.
A new /facet endpoint, that only returns facets could provide more user value:

• pagination options
• proper dictionary instead of alternating array
• meta information, like uri or display value

Proposed structure:

/facet
/facet/author
/facet/department
/facet/keyword
/facet/product
/facet/release_date
/facet/series

Available parameters and their default values would be:

q = *:*
limit = 10
offset = 0
fields = id, uri, value, […]
sort = id asc

Additional parameters for /facet/release_date:

gap = 1day
start = [NOW-100YEAR]
end = [NOW]

[cutoffthetop]

A separate facetting endpoint would require one additional request, if articles and facets are needed.
(Simply using the same query string.)
However, the number of requests needed to get a displayable text value (or other data) for each facet would drop to zero.

[ctriroli]

I am very much in support of a this feature ...
in my app I would be happy to do something like ...
http://api.zeit.de/facet/keyword?id=angela-merkel&start=2000-01-01T00:00:00Z000&gap=1YEAR
... to get the yearly keywordcount for angela-merkel ... I could also live with ...
http://api.zeit.de/facet/keyword?start=2000-01-01T00:00:00Z000&q:keyword:angela-merkel&gap=1YEAR
BTW how can we contribute to the code - or where can we find it - the repository is empty (besides of a MD5 hashfile). Do you have a sandbox environment for people like me that would be happy to code with you to speed up availability of this feature?

[cutoffthetop]

Since you are looking for facets of the release_date field, I would suggest a query syntax like this:
http://api.zeit.de/facet/release_date?q=keyword:angela-merkel&gap=1year&start=2000-01-01T00:00:00Z000
You could then use the exact same query string (q=keyword:angela-merkel) for the content endpoint and get matching articles.
At /facet/keyword I, personally, would expect keyword facets.

Regarding your second point: Yes, we are planning on open-sourcing the entire code base soon.

[ctriroli]

I agree - your query makes more sense.
Looking forward to your open-sourcing.