Wikidata:Property proposal/Search formatter URL

Search formatter URL

Originally proposed at Wikidata:Property proposal/Generic

   Done: search formatter URL (P4354) (Talk and documentation)

Description	Web page search URL; URI template from which "$1" can be automatically replaced with the sting to be searched for
Data type	String
Domain	websites & web services; Wikidata external ID properties
Allowed values	Search URLs, with "$1" replacing the search string
Example	ORCID iD (Q51044) → https://orcid.org/orcid-search/quick-search/?searchQuery=$1 ORCID iD (P496) → https://orcid.org/orcid-search/quick-search/?searchQuery=$1 e.g. https://orcid.org/orcid-search/quick-search/?searchQuery=David%20Levy
See also	formatter URL (P1630), formatter URI for RDF resource (P1921), third-party formatter URL (P3303)

Motivation

We should be able to programmatically indicate how to use a website's search facility; not least for external IDs. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:04, 15 June 2017 (UTC)

Discussion

•  Comment What is suppose to run into $1? The label? If, yes. Which language? — Finn Årup Nielsen (fnielsen) (talk) 20:17, 15 June 2017 (UTC)
  • @Fnielsen: Whatever anyone wants to search the target site for. In the specific Scholia search which we discussed elsewhere, yes, you could use ?authorLabel. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:06, 16 June 2017 (UTC)
•  Support, though what happens for those search facilities that expect more than a simple search string? Mahir256 (talk) 21:17, 19 June 2017 (UTC)
  • For example..? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:59, 20 June 2017 (UTC)

@Pigsonthewing: I will admit I don't have a ready example of this, but a lot of search pages for catalogues require a specific class of result to be returned (whether authors, books, publishers, or the like) that is typically specified within the URL. Should a separate URL be needed under this proposal for each type of search result? Mahir256 (talk) 18:51, 21 June 2017 (UTC)

•  Weak support perhaps browsers will maintain this better d1g (talk) 09:22, 1 August 2017 (UTC)
•  Support ديفيد عادل وهبة خليل 2 (talk) 19:55, 1 October 2017 (UTC)
•  Comment @Pigsonthewing: Something like this is needed, but I think it can take some more thinking/refinement.

The first important question: does this apply to class of entity / property or to website? Your example is entity/property, but shouldn't we also think about websites? Websites tie up to entities: eg to find things about Ontotext (Q7095072), I'll want to use the search box on http://www.ontotext.com.

The http://www.opensearch.org/ spec exposes the "search box" of any site using an "OpenSearch description document". There's a lot of detail in that document (encoding, language, adult rating, result type (HTML, Atom, RSS), etc). Don't we want to capture that info, or at least the location of the document?

OpenSearch has many extensions: Referrer Relevance Parameter Suggestions Geo Time Advertisement Commerce Mobile Spelling SRU (maybe some of them are relevant?)

VOID (the eminent LOD dataset description ontology) also ties to that: https://www.w3.org/TR/void/#opensearch

You assume the results will be HTML but as OpenSearch shows, Atom and RSS are also often used. Does this mean we also need "Search formatter URL for Atom" and "Search formatter URL for RSS"? We need a more generic approach!

--Vladimir Alexiev (talk) 10:09, 3 October 2017 (UTC)

My examples are for both. You can refer to RSS or Atom with qualifiers. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:14, 3 October 2017 (UTC)

• @Pigsonthewing, Vladimir Alexiev, ديفيد عادل وهبة خليل 2, D1gggg, Mahir256, Fnielsen:  Done Created as search formatter URL (P4354). ChristianKl (talk) 11:54, 12 October 2017 (UTC)