Fix too many item lookups on items will silly amounts of authors

module_citeq.lua

@@ -80,16 +80,62 @@ local function accept_value( param_val, wd_val )
 	return val
 end
 
+-- Split string of comma-separated first names into a table of initials
+local function getInitials(str)
+	local initials = {}
+	for _, word in pairs(mw.text.split(str, ' ', true)) do
+		local initial = mw.text.trim(word)
+        if #initial > 0 then
+            table.insert(initials, initial:sub(1, 1):upper())
+        end
+	end
+	return initials
+end
+
+-- Return initials given a first name string and a name-list-style value
+-- (e.g. 'vanc' for Vancouver, or APA)
+local function getStyledNames(names, style)
+	-- Vancouver style (vanc) is handled by CS1:  Doe JQ
+	if style == 'apa' then               -- APA:  Doe, J. Q.
+		return table.concat(getInitials(names), '. ') .. '.'
+	else
+		-- Otherwise, just remove any trailing commas from names
+		return mw.text.trim(names:gsub(', ', ' '))
+	end
+end
+
 -- function to fetch a value to display
-local function makelink(v, out, link, maxpos, wdl)
+local function makelink(v, out, link, maxpos, wdl, namestyle)
 	local label
+	local max_authors_displayed = 10  -- TODO: this is probably in CS1 config somewhere
+	local item_lookup = false
 	if v.mainsnak.snaktype == "value" then
 		if v.mainsnak.datatype == "wikibase-item" then
 			local qnumber = v.mainsnak.datavalue.value.id
 			local sitelink = mw.wikibase.getSitelink(qnumber)
 			if qnumber == "Q2818964" then sitelink = nil end -- suppress link to "Various authors"
+
+			-- Last name, first name(s) from "author given names" and "author last names" qualifiers
+			local last_name
+			local first_names
+			if v.qualifiers and v.qualifiers.P9688 and v.qualifiers.P9688[1] then
+				last_name   = v.qualifiers.P9688[1].datavalue.value
+			end
+			if v.qualifiers and v.qualifiers.P9687 and v.qualifiers.P9687[1] then
+				first_names = v.qualifiers.P9687[1].datavalue.value
+			end
+			-- Otherwise, get last name, first name(s) from author item properties
+			-- but this requires a getValue WD item lookup for each author, and
+			-- lookups are limited to 500 per page, so do this after we have the
+			-- "series ordinal" order, then only do the first few since there are
+			-- article items out there with 400+ authors(!)
+			if not last_name or not first_names then
+				item_lookup = true
+			end
+
+			-- use object named as for the label
 			if v.qualifiers and v.qualifiers.P1932 then
-				label = v.qualifiers.P1932[1].datavalue.value  -- object named as
+				label = v.qualifiers.P1932[1].datavalue.value  -- "object named as"
 			else
 				label = mw.wikibase.getLabel(qnumber)
 				if label then
@@ -100,14 +146,32 @@ local function makelink(v, out, link, maxpos, wdl)
 			end
 			local position = maxpos + 1 -- Default to 'next' author.
 			-- use P1545 (series ordinal) instead of default position.
-			if v["qualifiers"] and v.qualifiers["P1545"] and v.qualifiers["P1545"][1] then
-				position = tonumber(v.qualifiers["P1545"][1].datavalue.value)
+			if v.qualifiers and v.qualifiers.P1545 and v.qualifiers.P1545[1] then
+				position = tonumber(v.qualifiers.P1545[1].datavalue.value)
 			end
 			maxpos = math.max(maxpos, position)
+
+			-- now try determining first/last names with item lookup, if needed
+			if item_lookup and position <= max_authors_displayed then
+				last_name   = last_name   or getValue({"P734", qid=qnumber, maxvals=1, ps=2})
+				first_names = first_names or getValue({"P735", qid=qnumber, maxvals=0, ps=2})
+			end
+			if first_names then
+				first_names = getStyledNames(first_names, namestyle)
+			end
+
 			if sitelink then
 				-- just the plain name,
 				-- but keep a record of the links, using the same index
+				if last_name and first_names then
+					out[position] = {
+						label = label,
+						last  = last_name or nil,
+						first = first_names or nil,
+					}
+				else
 					out[position] = label
+				end
 				link[position] = sitelink
 			else
 				if wdl then
@@ -120,24 +184,68 @@ local function makelink(v, out, link, maxpos, wdl)
 					)
 				else
 					-- no Wikidata links wanted, so just give the plain label
+					if last_name and first_names then
+						out[position] = {
+							label = label,
+							last  = last_name or nil,
+							first = first_names or nil,
+						}
+					else
 						out[position] = label
 					end
 				end
+			end
 		elseif v.mainsnak.datatype == "string" then
 			local position = maxpos + 1 -- Default to 'next' author.
 			-- use P1545 (series ordinal) instead of default position.
-			if v["qualifiers"] and v.qualifiers["P1545"] and v.qualifiers["P1545"][1] then
-				position = tonumber(v.qualifiers["P1545"][1].datavalue.value)
+			if v.qualifiers and v.qualifiers.P1545 and v.qualifiers.P1545[1] then
+				position = tonumber(v.qualifiers.P1545[1].datavalue.value)
 			end
 			maxpos = math.max(maxpos, position)
+
+			-- Names from "author given names" and "author last names" qualifiers
+			local last_name
+			local first_names
+			if v.qualifiers and v.qualifiers.P9688 and v.qualifiers.P9688[1] then
+				last_name   = v.qualifiers.P9688[1].datavalue.value
+			end
+			if v.qualifiers and v.qualifiers.P9687 and v.qualifiers.P9687[1] then
+				first_names = getStyledNames(v.qualifiers.P9687[1].datavalue.value, namestyle)
+			end
+
+			if last_name and first_names then
+				out[position] = {
+					label = v.mainsnak.datavalue.value,
+					last  = last_name or nil,
+					first = first_names or nil,
+				}
+			else
 				out[position] = v.mainsnak.datavalue.value
+			end
 		else
 			-- not a wikibase-item or a string!
 		end
 	else
-		-- code here if we want to return something when author is "unknown"
+		-- try the "object named as" qualifier, otherwise use the unknown author item
 		if v.qualifiers and v.qualifiers.P1932 then
-			label = v.qualifiers.P1932[1].datavalue.value
+			label = v.qualifiers.P1932[1].datavalue.value  -- "object named as"
+
+			-- Names from "author given names" and "author last names" qualifiers
+			local last_name
+			local first_names
+			if v.qualifiers and v.qualifiers.P9688 and v.qualifiers.P9688[1] then
+				last_name = v.qualifiers.P9688[1].datavalue.value
+			end
+			if v.qualifiers and v.qualifiers.P9687 and v.qualifiers.P9687[1] then
+				first_names = getStyledNames(v.qualifiers.P9687[1].datavalue.value, namestyle)
+			end
+			if last_name and first_names then
+				label = {
+					label = label,
+					last  = last_name or nil,
+					first = first_names or nil,
+				}
+			end
 		else
 			label = mw.wikibase.getLabel("Q4233718"):gsub("^%l", mw.ustring.upper) .. (cfg.i18n_t.trackingcats["unknown-author"] or "")
 		end
@@ -213,7 +321,7 @@ local function get_name_list (nl_type, args, qid, wdl)
 	if props and props[1] then
 		for k, v in pairs(props) do
 			if 'deprecated' ~= v.rank then										-- ignore deprecated names
-				maxpos = makelink(v, out, link, maxpos, wdl)
+				maxpos = makelink(v, out, link, maxpos, wdl, args['name-list-style'])
 			end
 		end
 	end
@@ -221,7 +329,7 @@ local function get_name_list (nl_type, args, qid, wdl)
 		-- second properties
 		for k, v in pairs(fallback) do
 			if 'deprecated' ~= v.rank then										-- ignore deprecated names
-				maxpos = makelink(v, out, link, maxpos, wdl)
+				maxpos = makelink(v, out, link, maxpos, wdl, args['name-list-style'])
 			end
 		end
 	end
@@ -235,15 +343,27 @@ local function get_name_list (nl_type, args, qid, wdl)
 	end
 	table.sort(keys) -- as they might be out of order
 	for i, k in ipairs(keys) do
-		out[k] = out[k]:gsub ('&#39;', '\'');									-- prevent cs1|2 multiple names categorization; replace html entity with the actual character
-		mw.log(i .. " " .. k .. " " .. (out[k]))
+		local label
+		if out[k].label then
+			label = out[k].label
+		else
+			label = out[k]
+		end
+		label = label:gsub('&#39;', '\'');									-- prevent cs1|2 multiple names categorization; replace html entity with the actual character
+		mw.log(i .. " " .. k .. " " .. label)
 		if args[nl_type .. i] then -- name gets overwritten
 			-- pull corresponding -link only if overwritten name is same as WD name
-			if link[k] and (args[nl_type .. i] == out[k]) then
+			if link[k] and (args[nl_type .. i] == label) then
 				args[nl_type .. '-link' .. i] = args[nl_type .. '-link' .. i] or link[k] -- author-linkn or editor-linkn
 			end
 		else -- name does not get overwritten, so pull name from WD
-			args[nl_type .. i] = out[k]
+			if out[k].first and out[k].last then
+				-- First/last names were found, so add them
+				args[nl_type .. '-last'  .. i] = args[nl_type .. '-last'  .. i] or out[k].last
+				args[nl_type .. '-first' .. i] = args[nl_type .. '-first' .. i] or out[k].first
+			else
+				args[nl_type .. i] = label
+			end
 			if link[k] then
 				args[nl_type .. '-link' .. i] = args[nl_type .. '-link' .. i] or link[k] -- author-linkn or editor-linkn
 			end
@@ -273,11 +393,12 @@ end
 
 -- gets the language codes of a Wikidata entry as a table
 local function _lang_code(qid)
-	local lc = _getPropOfProp( {qid = qid, prop1 = "P407", prop2 = "P424", ps = 1} )
+	-- P407 "language of work or name"
+	local lc = _getPropOfProp( {qid = qid, prop1 = "P407", prop2 = "P424", ps = 1} )  -- "Wikimedia language code"
 	if lc then return mw.text.split( lc, "[, ]+" ) end
-	lc = _getPropOfProp( {qid = qid, prop1 = "P407", prop2 = "P218", ps = 1} )
+	lc = _getPropOfProp( {qid = qid, prop1 = "P407", prop2 = "P218", ps = 1} )  -- "ISO 639-1 code"
 	if lc then return mw.text.split( lc, "[, ]+" ) end
-	return _getLangOfProp(qid, "P1476")
+	return _getLangOfProp(qid, "P1476") -- "title"
 end
 local function lang_code(frame)
 	return table.concat(_lang_code(frame.args.qid or mw.text.trim(frame.args[1] or "")), ", ")
@@ -302,8 +423,11 @@ local function yearFromDate(datestr)
 	return datestr
 end
 
--- Attempt to determine heuristically if we're citing a news article
-local function isNewsArticle(qid, args)
+-- Attempt to determine heuristically what we're citing
+local function getTemplateFromInstance(qid, args)
+	local chapteritems = {
+		'Q1980247',   -- chapter
+	}
 	local newsitems = {
 		'Q5707594',   -- news article
 		'Q30070590',  -- magazine article
@@ -322,7 +446,9 @@ local function isNewsArticle(qid, args)
 	}
 	-- most of the time a cheap lookup to a single value will work
 	if args['instance-of'] and in_array(args['instance-of'], newsitems) then
-		return true
+		return 'news'
+	elseif args['instance-of'] and in_array(args['instance-of'], chapteritems) then
+		return 'chapter'
 	end
 	-- otherwise check multiple non-deprecated values
 	local instance_of = mw.wikibase.getBestStatements(qid, 'P31')
@@ -330,12 +456,14 @@ local function isNewsArticle(qid, args)
 		for k, v in pairs(instance_of) do
 			if v.mainsnak.snaktype == "value" and v.mainsnak.datatype == "wikibase-item" then
 				if in_array(v.mainsnak.datavalue.value.id, newsitems) then
-					return true
+					return 'news'
+				elseif in_array(v.mainsnak.datavalue.value.id, chapteritems) then
+					return 'chapter'
 				end
 			end
 		end
 	end
-	return false
+	return ''
 end
 
 -- Figure out a URL access value, from the "online access status" (P6954) qualifier of an identifier
@@ -702,14 +830,20 @@ local function _cite_q (citeq_args)
 	-- code to make a guess what template to use from the supplied parameters
 	-- (first draft for proof-of-concept)
 	local pubyear = yearFromDate(citeq_args['publication-date'])
-	local is_news = isNewsArticle(qid, citeq_args)
-	if is_news then
+	local guessedType = getTemplateFromInstance(qid, citeq_args)
+	if guessedType == 'news' then
 		template = template or "news"
 		-- some items use P123 (publisher) as the newspaper instead of P1433 (published in)
 		if citeq_args.publisher and not citeq_args.journal then
 			citeq_args.journal = citeq_args.publisher
 			citeq_args.publisher = nil
 		end
+	elseif guessedType == 'chapter' then
+		template = template or "book"
+		citeq_args['publication-date'] = pubyear or citeq_args['publication-date']
+		citeq_args.chapter = citeq_args.title
+		citeq_args.title = citeq_args.journal
+		citeq_args.journal = nil
 	elseif citeq_args.journal then
 		template = template or "journal"
 		citeq_args['publication-date'] = pubyear or citeq_args['publication-date']
@@ -874,6 +1008,9 @@ local function _cite_q (citeq_args)
 
 	-- remove any remaining non-CS1 fields
 	citeq_args['instance-of'] = nil
+	if citeq_args['name-list-style'] and citeq_args['name-list-style'] == 'apa' then
+		citeq_args['name-list-style'] = nil
+	end
 
 	-- render the template
 	local lc_template = template and template:lower() or "citation"