Module:Sandbox/Hike395/Wikidata utils

require('strict')

-- Is the string a valid QID?
local function validQid(qid)
    return qid and mw.ustring.find(qid,"^[Qq]%d+$")
end

-- lookup nestedIndex in outer table
-- Example:  fetchNested(foo,{'bar','baz','quux'}) will return
--     foo.bar and foo.bar.baz and foo.bar.baz.quux
local function fetchNested(outer,nestedIndex)
    if not outer or type(outer) ~= 'table' or not nestedIndex or type(nestedIndex) ~= 'table' then
        return nil
    end
    local current = outer
    for _, indx in ipairs(nestedIndex) do
        current = current[indx]
        if not current then
            return current
        end
    end
    return current
end

-- Mapping from wikidata properties (of a reference or "stated in" entity) to citation parameters
local pid2param = {
    P50 = "author",
    P123 = "publisher",
    P407 = "language",
    P577 = "date",
    P813 = "access-date",
    P854 = "url",
    P856 = "url",
    P953 = "url",
    P1476 = "title",
    P2699 = "url",
}

-- scan through snaks, loading citation args with correct values
local function scanSnaks(args, snaks)
    for pid, data in pairs(snaks) do
        data = fetchNested(data,{1,'mainsnak'}) or data[1] -- handle the case where the snak is buried in the claims
        local param = pid2param[pid]
        if pid == "P248" then  --- "stated in": must look up different entity for data
            local statedId = fetchNested(data,{'datavalue','value','id'})
            local statedEntity = statedId and mw.wikibase.getEntity(statedId)
            local statedClaims = statedEntity and statedEntity.claims
            if statedClaims then
                scanSnaks(args, statedClaims)
            end
            local statedLabel = fetchNested(statedEntity,{'labels','en','value'})
            args.title = args.title or statedLabel
        elseif param then
            local renderedData = data and mw.wikibase.renderSnak(data)
            if param == "title" then
                -- set language from title if not otherwise specified
                args.language = args.language or fetchNested(data,{'datavalue','value','language'})
            end
            if pid == "P856" then --- use official URL only if URL is missing
                args[param] = args[param] or renderedData
            else
                args[param] = renderedData
            end
        end
    end
end

-- function to assemble wikicode to create citation, given reference property
local function citeSource(snaks)
    local args = {}
    args.mode = "cs1"
    scanSnaks(args, snaks)
    -- citation templates are unhappy without a title, so return a wikilink
    if not args.title then  
        return args.url and "["..args.url.."]"
    end
    if not args.url then
        args["access-date"] = nil  -- don't allow access date without url
    end
    local result = '{{'..(args.url and 'cite web' or 'citation')
    for k,v in pairs(args) do
        result = result..'|'..k..'='..v
    end
    result = result..'}}'
    return result
end

-- Core function that looks up value, reference, and data for property
-- Arguments:
--   qid = entity id in Wikidata
--   pid = property id on that entity
--   unit = if non-nil, desired unit for value (expressed as a qid for that unit)
--   asOf = date or partial date string to prefer. If not given, find the latest version
-- Returns table only if Wikidata property is source externally
--   amount = value that is looked up (or nil)
--   source = wiki markup for reference
--   asOf = date that value is valid (if given)
local function getValueAndRef(qid,pid,unit,asOf)
    local results = nil
    local latest = nil
    local statements = mw.wikibase.getAllStatements(qid, pid ) 
    for _, statement in pairs(statements) do
        local amount
        local wrongDate = true
        local value = fetchNested(statement,{'mainsnak','datavalue','value'})
        if value then
            amount = value.amount
        end
        -- check if unit is correct, if given
        if unit then
            local storedUnit = value.unit
            if not storedUnit or not mw.ustring.find(value.unit or '',unit,1,true) then
                amount = nil
            end
        end
        -- check if asOf matches stored date
        -- if asOf not specified, remember stored date
        local storedAsOf = fetchNested(statement,{'qualifiers','P585',1,'datavalue','value','time'})
        local date
        if storedAsOf then
            date = mw.ustring.match(storedAsOf,'^%+(%d+%-%d+%-%d+)')
            local justYear = mw.ustring.match(date,'^(%d+)%-0+%-0+$')
            date = justYear or date
            if asOf then
                wrongDate = date and not mw.ustring.find(date,asOf,1,true)
            end
        end
        -- check to see if this statement is adequately sourced (filter out "imported from" refs)
        local source
        local refs = statement['references']
        if refs then
            for _, ref in pairs(refs) do
                local renderedRef = ref.snaks and mw.wikibase.renderSnaks(ref.snaks)
                if renderedRef and not mw.ustring.find(renderedRef,'Wiki') then
                    source = citeSource(ref.snaks)
                end
            end
        end
        if amount and source then -- yes, it is adequately sourced,
            local theseResults = {}
            theseResults.amount = tonumber(amount)
            theseResults.source = source
            theseResults.asOf = date
            -- if we haven't found the preferred date
            if wrongDate then
                if date and (not latest or date > latest) then -- store latest if dated
                    results = theseResults
                    latest = date
                elseif not date then -- store first if undated
                    results = results or theseResults
                end
            -- here, we found the correct date, so return it
            else
                results = theseResults
                break
            end
        end
    end
    return results
end

return {validQid=validQid, fetchNested=fetchNested, citeSource=citeSource, getValueAndRef=getValueAndRef}