Difference between revisions of "Module:Authority control"

Revision as of 20:35, 18 May 2015

require('Module:No globals')

local function getCatForId( id )
    local title = mw.title.getCurrentTitle()
    local namespace = title.namespace
    if namespace == 0 then
        return '[[Category:Wikipedia articles with ' .. id .. ' identifiers]]'
    elseif namespace == 2 and not title.isSubpage then
        return '[[Category:User pages with ' .. id .. ' identifiers]]'
    else
        return '[[Category:Miscellaneous pages with ' .. id .. ' identifiers]]'
    end
end

local function viafLink( id )
    if not string.match( id, '^%d+$' ) then
        return false
    end
    return '[https://viaf.org/viaf/' .. id .. ' ' .. id .. ']' .. getCatForId( 'VIAF' )
end

local function kulturnavLink( id )
    return '[http://kulturnav.org/language/en/' .. id .. ' id]' 
end

local function sikartLink( id )
    return '[http://www.sikart.ch/KuenstlerInnen.aspx?id=' .. id .. '&lng=en ' .. id .. ']' 
end

local function tlsLink( id )
	local id2 = mw.ustring.gsub(id, '%s', function(s) return mw.uri.encode(s, 'WIKI') end)
    return '[http://tls.theaterwissenschaft.ch/wiki/' .. id2 .. ' ' .. id .. ']' 
end


local function ciniiLink( id )
    return '[http://ci.nii.ac.jp/author/' .. id .. '?l=en ' .. id .. ']' 
end

local function bneLink( id )
    return '[http://catalogo.bne.es/uhtbin/authoritybrowse.cgi?action=display&authority_id=' .. id .. ' ' .. id .. ']' 
end


local function uscongressLink( id )
    return '[http://bioguide.congress.gov/scripts/biodisplay.pl?index=' .. id .. ' ' .. id .. ']' 
end

local function narapersonLink( id )
    return '[http://research.archives.gov/person/' .. id .. ' ' .. id .. ']' 
end

local function naraorganizationLink( id )
    return '[http://research.archives.gov/organization/' .. id .. ' ' .. id .. ']' 
end

local function botanistLink( id )
	local id2 = mw.ustring.gsub(id, '%s', function(s) return mw.uri.encode(s, 'PATH') end)
    return '[http://www.ipni.org/ipni/advAuthorSearch.do?find_abbreviation=' .. id2 .. ' ' .. id .. ']' 
end

local function mgpLink( id )
    -- TODO Implement some sanity checking regex
    return '[http://www.genealogy.ams.org/id.php?id=' .. id .. ' ' .. id .. ']' 
end

local function rslLink( id )
    -- TODO Implement some sanity checking regex
    return '[http://aleph.rsl.ru/F?func=find-b&find_code=SYS&adjacent=Y&local_base=RSL11&request=' .. id .. '&CON_LNG=ENG ' .. id .. ']'
end

local function leonoreLink( id )
-- Identifiants allant de LH/1/1 à LH/2794/54 (légionnaires)
-- Identifiants allant de C/0/1 à C/0/84 (84 légionnaires célèbres)
-- Identifiants allant de 19800035/1/1 à 19800035/385/51670 (légionnaires décédés entre 1954 et 1977, et quelques dossiers de légionnaires décédés avant 1954)
    if not string.match( id, '^LH/%d%d?%d?%d?/%d%d?%d?$' ) and
       not string.match( id, '^C/0/%d%d?$' ) and
	   not string.match( id, '^19800035/%d%d?%d?%d?/%d%d?%d?%d?%d?$' ) then
        return false
    end
    return '[//www.culture.gouv.fr/public/mistral/leonore_fr?ACTION=CHERCHER&FIELD_1=COTE&VALUE_1=' .. id .. ' ' .. id .. ']' 
end

local function sbnLink( id )
    if not string.match( id, '^IT\\ICCU\\%d%d%d%d%d%d%d%d%d%d$' ) and not string.match( id, '^IT\\ICCU\\%u%u[%d%u]%u\\%d%d%d%d%d%d$' ) then
        return false
    end
    return '[http://opac.sbn.it/opacsbn/opac/iccu/scheda_authority.jsp?bid=' .. id .. ' ' .. id .. ']'
end

local function nkcLink( id )
	return '[http://aleph.nkp.cz/F/?func=find-c&local_base=aut&ccl_term=ica=' .. id .. '&CON_LNG=ENG ' .. id .. ']' 
end

local function nclLink( id )
    if not string.match( id, '^%d+$' ) then
        return false
    end
    return '[http://aleweb.ncl.edu.tw/F/?func=accref&acc_sequence=' .. id .. '&CON_LNG=ENG ' .. id .. ']' 
end

local function ndlLink( id )
	return '[http://id.ndl.go.jp/auth/ndlna/' .. id .. ' ' .. id .. ']' 
end

local function sudocLink( id )
    if not string.match( id, '^%d%d%d%d%d%d%d%d[%dxX]$' ) then
        return false
    end
    return '[http://www.idref.fr/' .. id .. ' ' .. id .. ']' 
end

local function hlsLink( id )
    if not string.match( id, '^%d+$' ) then
        return false
    end
    return '[http://www.hls-dhs-dss.ch/textes/f/F' .. id .. '.php ' .. id .. ']'
end

local function lirLink( id )
    if not string.match( id, '^%d+$' ) then
        return false
    end
    return '[http://www.e-lir.ch/e-LIR___Lexicon.' .. id .. '.450.0.html ' .. id .. ']'
end

local function splitLccn( id )
    if id:match( '^%l%l?%l?%d%d%d%d%d%d%d%d%d?%d?$' ) then
        id = id:gsub( '^(%l+)(%d+)(%d%d%d%d%d%d)$', '%1/%2/%3' )
    end
    if id:match( '^%l%l?%l?/%d%d%d?%d?/%d+$' ) then
         return mw.text.split( id, '/' )
    end
    return false
end

local function append(str, c, length)
    while str:len() < length do
        str = c .. str
    end
    return str
end

local function lccnLink( id )
    local parts = splitLccn( id )
    if not parts then
        return false
    end
    local lccnType = parts[1] ~= 'sh' and 'names' or 'subjects'
    id = parts[1] .. parts[2] .. append( parts[3], '0', 6 )
    return '[http://id.loc.gov/authorities/' .. lccnType .. '/' .. id .. ' ' .. id .. ']' .. getCatForId( 'LCCN' )
end

local function mbLink( id )
    -- TODO Implement some sanity checking regex
    return '[//musicbrainz.org/artist/' .. id .. ' ' .. id .. ']' .. getCatForId( 'MusicBrainz' )
end

--Returns the ISNI check digit isni must be a string where the 15 first elements are digits
local function getIsniCheckDigit( isni )
    local total = 0
    for i = 1, 15 do
        local digit = isni:byte( i ) - 48 --Get integer value
        total = (total + digit) * 2
    end
    local remainder = total % 11
    local result = (12 - remainder) % 11
    if result == 10 then
        return "X"
    end
    return tostring( result )
end

--Validate ISNI (and ORCID) and retuns it as a 16 characters string or returns false if it's invalid
--See http://support.orcid.org/knowledgebase/articles/116780-structure-of-the-orcid-identifier
local function validateIsni( id )
    id = id:gsub( '[ %-]', '' ):upper()
    if not id:match( '^%d%d%d%d%d%d%d%d%d%d%d%d%d%d%d[%dX]$' ) then
        return false
    end
    if getIsniCheckDigit( id ) ~= string.char( id:byte( 16 ) ) then
        return false
    end
    return id
end

local function isniLink( id )
    id = validateIsni( id )
    if not id then
        return false
    end
    return '[http://isni-url.oclc.nl/isni/' .. id .. ' ' .. id:sub( 1, 4 ) .. ' ' .. id:sub( 5, 8 ) .. ' '  .. id:sub( 9, 12 ) .. ' '  .. id:sub( 13, 16 ) .. ']' .. getCatForId( 'ISNI' )
end

local function orcidLink( id )
    id = validateIsni( id )
    if not id then
        return false
    end
    id = id:sub( 1, 4 ) .. '-' .. id:sub( 5, 8 ) .. '-'  .. id:sub( 9, 12 ) .. '-'  .. id:sub( 13, 16 )
    return '[http://orcid.org/' .. id .. ' ' .. id .. ']' .. getCatForId( 'ORCID' )
end

local function gndLink( id )
    return '[http://d-nb.info/gnd/' .. id .. ' ' .. id .. ']' .. getCatForId( 'GND' )
end

local function selibrLink( id )
	if not string.match( id, '^%d+$' ) then
        return false
    end
    return '[//libris.kb.se/auth/' .. id .. ' ' .. id .. ']' .. getCatForId( 'SELIBR' )
end

local function bnfLink( id )
    --Add cb prefix if it has been removed
    if not string.match( id, '^cb.+$' ) then
        id = 'cb' .. id
    end

    return '[http://catalogue.bnf.fr/ark:/12148/' .. id .. ' ' .. id .. '] [http://data.bnf.fr/ark:/12148/' .. id .. ' (data)]' .. getCatForId( 'BNF' )
end

local function bpnLink( id )
    if not string.match( id, '^%d+$' ) then
        return false
    end
    return '[http://www.biografischportaal.nl/en/persoon/' .. id .. ' ' .. id .. ']' .. getCatForId( 'BPN' )
end

local function ridLink( id )
    return '[http://www.researcherid.com/rid/' .. id .. ' ' .. id .. ']' .. getCatForId( 'RID' )
end

local function bibsysLink( id )
    return '[http://ask.bibsys.no/ask/action/result?cmd=&kilde=biblio&cql=bs.autid+%3D+' .. id .. '&feltselect=bs.autid ' .. id .. ']' .. getCatForId( 'BIBSYS' )
end

local function ulanLink( id )
    return '[//www.getty.edu/vow/ULANFullDisplay?find=&role=&nation=&subjectid=' .. id .. ' ' .. id .. ']' .. getCatForId( 'ULAN' )
end

local function nlaLink( id )
	return '[//nla.gov.au/anbd.aut-an' .. id .. ' ' .. id .. ']' .. getCatForId( 'NLA' )
end

local function rkdartistsLink( id )
	return '[https://rkd.nl/en/explore/artists/' .. id .. ' ' .. id .. ']' .. getCatForId( 'RKDartists' )
end

local function getIdsFromWikidata( item, property )
    local ids = {}
    if not item.claims[property] then
        return ids
    end
    for _, statement in pairs( item.claims[property] ) do
		if statement.mainsnak.datavalue then
			table.insert( ids, statement.mainsnak.datavalue.value )
		end
    end
    return ids
end

local function matchesWikidataRequirements( item, reqs )
    for _, group in pairs( reqs ) do
        local property = 'p' .. group[1]
        local qid = group[2]
        if item.claims[property] ~= nil then
            for _, statement in pairs ( item.claims[property] ) do
            	if statement.mainsnak.datavalue ~= nil then
	                if statement.mainsnak.datavalue.value['numeric-id'] == qid then
    	                return true
        	        end
        	    end
            end
        end
    end
    return false
end

local function createRow( id, label, rawValue, link, withUid )
    if link then
        if withUid then
            return '* ' .. label .. ' <span class="uid">' .. link .. '</span>\n'
        else
            return '* ' .. label .. ' ' .. link .. '\n'
        end
    else
        return '* <span class="error">The ' .. id .. ' id ' .. rawValue .. ' is not valid.</span>[[Category:Wikipedia articles with faulty authority control identifiers (' .. id .. ')]]\n'
    end
end

--In this order: name of the parameter, label, propertyId in Wikidata, formatting function
local conf = {
    { 'VIAF', '[[Virtual International Authority File|VIAF]]', 214, viafLink },
    { 'LCCN', '[[Library of Congress Control Number|LCCN]]', 244, lccnLink },
    { 'ISNI', '[[International Standard Name Identifier|ISNI]]', 213, isniLink },
    { 'ORCID', '[[ORCID]]', 496, orcidLink },
    { 'GND', '[[Integrated Authority File|GND]]', 227, gndLink },
    { 'SELIBR', '[[LIBRIS|SELIBR]]', 906, selibrLink },
    { 'SUDOC', '[[Système universitaire de documentation|SUDOC]]', 269, sudocLink },    
    { 'BNF', '[[Bibliothèque nationale de France|BNF]]', 268, bnfLink },
    { 'BPN', '[[Biografisch Portaal|BPN]]', 651, bpnLink },
    { 'RID', '[[ResearcherID]]', 1053, ridLink },
    { 'BIBSYS', '[[BIBSYS]]', 1015, bibsysLink },
    { 'ULAN', '[[Union List of Artist Names|ULAN]]', 245, ulanLink },
    { 'HDS', '[[Historical Dictionary of Switzerland|HDS]]', 902, hlsLink },
    { 'LIR', '[[Historical Dictionary of Switzerland#Lexicon_Istoric_Retic|LIR]]', 886, lirLink },
    { 'MBA', '[[MusicBrainz]]', 434, mbLink },
    { 'MGP', '[[Mathematics Genealogy Project|MGP]]', 549, mgpLink },    
    { 'NLA', '[[National Library of Australia|NLA]]', 409, nlaLink },
    { 'NDL', '[[National Diet Library|NDL]]', 349, ndlLink },
    { 'NCL', '[[National Central Library|NCL]]', 1048, nclLink },
    { 'NKC', '[[National Library of the Czech Republic|NKC]]', 691, nkcLink },
    { 'Léonore', '[[:fr:Base Léonore|Léonore]]', 640, leonoreLink }, 
    { 'SBN', '[[Istituto Centrale per il Catalogo Unico|ICCU]]', 396, sbnLink },     
    { 'RLS', '[[Russian State Library|RLS]]', 947, rslLink },
    { 'Botanist', '[[Author citation (botany)|Botanist]]', 428, botanistLink },
    { 'NARA-person', '[[National Archives and Records Administration|NARA]]', 1222, narapersonLink },
    { 'NARA-organization', '[[National Archives and Records Administration|NARA]]', 1223, naraorganizationLink },
    { 'USCongress', '[[Biographical Directory of the United States Congress|US Congress]]', 1157, uscongressLink },
    { 'BNE', '[[Biblioteca Nacional de España|BNE]]', 950, bneLink },
    { 'CINII', '[[CiNii]]', 271, ciniiLink },
    { 'TLS', '[[Theaterlexikon der Schweiz|TLS]]', 1362, tlsLink },
    { 'SIKART', '[[SIKART]]', 781, sikartLink },
    { 'KULTURNAV', '[[KulturNav]]', 1248, kulturnavLink },
    { 'RKDartists', '[[Netherlands Institute for Art History#Online artist pages|RKD]]', 650, rkdartistsLink },
}

-- Check that the Wikidata item has this property-->value before adding it
local reqs = {}

local p = {}

function p.authorityControl( frame )
    local parentArgs = frame:getParent().args
    --Create rows
    local elements = {}

    --redirect PND to GND
    if (parentArgs.GND == nil or parentArgs.GND == '') and parentArgs.PND ~= nil and parentArgs.PND ~= '' then
        parentArgs.GND = parentArgs.PND
    end

    --Wikidata fallback if requested
    local item = mw.wikibase.getEntityObject()
    if item ~= nil and item.claims ~= nil then
        for _, params in pairs( conf ) do
            if params[3] ~= 0 then
                local val = parentArgs[params[1]]
                if not val or val == '' then
                	local canUseWikidata = nil
                    if reqs[params[1]] ~= nil then
                        canUseWikidata = matchesWikidataRequirements( item, reqs[params[1]] )
                    else
                        canUseWikidata = true
                    end
                    if canUseWikidata then
                        local wikidataIds = getIdsFromWikidata( item, 'P' .. params[3] )
                        if wikidataIds[1] then
                            parentArgs[params[1]] = wikidataIds[1]
                        end
                    end
                end
            end
        end
    end

    --Worldcat
    if parentArgs['WORLDCATID'] and parentArgs['WORLDCATID'] ~= '' then
        table.insert( elements, createRow( 'WORLDCATID', '', parentArgs['WORLDCATID'], '[//www.worldcat.org/identities/' .. parentArgs['WORLDCATID'] .. ' WorldCat]', false ) ) --Validation?
    elseif parentArgs['LCCN'] and parentArgs['LCCN'] ~= '' then
        local lccnParts = splitLccn( parentArgs['LCCN'] )
        if lccnParts then
            table.insert( elements, createRow( 'LCCN', '', parentArgs['LCCN'], '[//www.worldcat.org/identities/lccn-' .. lccnParts[1] .. lccnParts[2] .. '-' .. lccnParts[3] .. ' WorldCat]', false ) )
        end
    end

    --Configured rows
    local rct = 0
    for k, params in pairs( conf ) do
        local val = parentArgs[params[1]]
        if val and val ~= '' then
            table.insert( elements, createRow( params[1], params[2] .. ':', val, params[4]( val ), true ) )
            rct = rct + 1
        end
    end
    local Navbox = require('Module:Navbox')
    local elementscats = ''
    if rct > 13 then
    	elementscats  = '[[Category:AC with ' .. rct .. ' elements]]'
	end
	
	if #elements ~= 0 then
		return Navbox._navbox( {
			name  = 'Authority control',
			bodyclass = 'hlist',
			group1 = '[[Authority control]]' .. elementscats,
			list1 = table.concat( elements )
			} )
	else
		return ""
	end
end

return p