Module:Wd: Difference between revisions
Jump to navigation
Jump to search
>Thayts (Sorting on rank upfront allows for simplified logic) |
>Thayts mNo edit summary |
||
Line 1,435: | Line 1,435: | ||
for i, v in ipairs(statements) do | for i, v in ipairs(statements) do | ||
-- rankPos will be nil for non-claim statements (e.g. qualifiers, references, etc.) | -- rankPos will be nil for non-claim statements (e.g. qualifiers, references, etc.) | ||
matches, rankPos = matchHook(self, v) | matches, rankPos = matchHook(self, v) | ||
Revision as of 04:32, 19 March 2017
Documentation for this module may be created at Module:Wd/doc
local p = {} local aliasesP = { coord = "P625", author = "P50", publisher = "P123", importedFrom = "P143", statedIn = "P248", publicationDate = "P577", startTime = "P580", endTime = "P582", retrieved = "P813", referenceURL = "P854", archiveURL = "P1065", title = "P1476", quote = "P1683", shortName = "P1813", language = "P2439", archiveDate = "P2960" } local aliasesQ = { percentage = "Q11229", prolepticJulianCalendar = "Q1985786" } local parameters = { property = "%p", qualifier = "%q", reference = "%r", separator = "%s", general = "%x" } local formats = { property = "%p[%s][%r]", qualifier = "%q[%s][%r]", reference = "%r", propertyWithQualifier = "%p[ <span style=\"font-size:smaller\">(%q)</span>][%s][%r]" } local hookNames = { -- {level_1, level_2} [parameters.property] = {"getProperty"}, [parameters.reference] = {"getReferences", "getReference"}, [parameters.qualifier] = {"getAllQualifiers"}, [parameters.qualifier.."\\d"] = {"getQualifiers", "getQualifier"} } local defaultSeparators = { ["sep"] = " " , ["sep%s"] = "," , ["sep%q"] = "; " , ["sep%q\\d"] = ", " , ["sep%r"] = "" , ["punc"] = "" } local Config = {} Config.__index = Config -- allows for recursive calls function Config.new() local cfg = {} setmetatable(cfg, Config) cfg.separators = { -- use tables so that we can pass by reference ["sep"] = {defaultSeparators["sep"]}, ["sep%s"] = {defaultSeparators["sep%s"]}, ["sep%q"] = {defaultSeparators["sep%q"]}, ["sep%r"] = {defaultSeparators["sep%r"]}, ["punc"] = {defaultSeparators["punc"]} } cfg.entity = nil cfg.propertyID = nil cfg.propertyValue = nil cfg.qualifierIDs = {} cfg.bestRank = true cfg.ranks = {true, true, false} -- preferred = true, normal = true, deprecated = false cfg.foundRank = #cfg.ranks cfg.flagBest = false cfg.flagRank = false cfg.periods = {true, true, true} -- future = true, current = true, former = true cfg.flagPeriod = false cfg.mdyDate = false cfg.pageTitle = false cfg.langCode = mw.language.getContentLanguage().code cfg.langName = mw.language.fetchLanguageName(cfg.langCode, cfg.langCode) cfg.langObj = mw.language.new(cfg.langCode) cfg.states = {} cfg.states.qualifiersCount = 0 cfg.curState = nil return cfg end local State = {} State.__index = State function State.new(cfg) local stt = {} setmetatable(stt, State) stt.conf = cfg stt.results = {} stt.parsedFormat = {} stt.separator = {} stt.movSeparator = {} stt.puncMark = {} stt.linked = false stt.rawValue = false stt.shortName = false stt.singleValue = false return stt end function unknownDataTypeError(dataType) return "Unknown or unsupported datatype '" .. dataType .. "'" end function missingRequiredParameterError() return "No required parameters defined, needing at least one" end function extraRequiredParameterError(param) return "Parameter '" .. param .. "' must be defined as optional" end -- used to make frame.args mutable and to replace #frame.args (which is always 0) with the actual amount function copyTable(tIn) local tOut = {} for i, v in pairs(tIn) do tOut[i] = v end return tOut end -- used to merge output tables together; -- note that it currently mutates the first input table function mergeTables(t1, t2) for i = 1, #t2 do t1[#t1 + 1] = t2[i] end return t1 end -- used to create the final output string when it's all done, -- so that extensionTag("ref") is only called when it's really in the final output function concatValues(valuesArray) local outString = "" for i = 1, #valuesArray do if valuesArray[i].refHash then -- add <ref> tag with the reference's hash as its name (to deduplicate references) outString = outString .. mw.getCurrentFrame():extensionTag("ref", valuesArray[i][1], {name = "wikidata-" .. valuesArray[i].refHash}) else outString = outString .. valuesArray[i][1] end end return outString end function getHookName(param, index) if hookNames[param] then return hookNames[param][index] elseif string.len(param) > 2 then return hookNames[string.sub(param, 1, 2).."\\d"][index] else return nil end end function parseWikidataURL(url) local i, j if url:match('^http[s]?://') then i, j = url:find("Q") if i then return url:sub(i) end end return nil end function parseDate(dateStr, precision) precision = precision or "d" local i, j, index, ptr local parts = {nil, nil, nil} if dateStr == nil then return parts[1], parts[2], parts[3] -- year, month, day end -- 'T' for snak values, '/' for outputs with '/Julian' attached i, j = dateStr:find("[T/]") if i then dateStr = dateStr:sub(1, i-1) end local from = 1 if dateStr:sub(1,1) == "-" then -- this is a negative number, look further ahead from = 2 end index = 1 ptr = 1 i, j = dateStr:find("-", from) if i then -- year parts[index] = tonumber(mw.ustring.gsub(dateStr:sub(ptr, i-1), "^\+(.+)$", "%1"), 10) -- remove + sign if parts[index] == -0 then parts[index] = 0 end if precision == "y" then -- we're done return parts[1], parts[2], parts[3] -- year, month, day end index = index + 1 ptr = i + 1 i, j = dateStr:find("-", ptr) if i then -- month parts[index] = tonumber(dateStr:sub(ptr, i-1), 10) if precision == "m" then -- we're done return parts[1], parts[2], parts[3] -- year, month, day end index = index + 1 ptr = i + 1 end end if dateStr:sub(ptr) ~= "" then -- day if we have month, month if we have year, or year parts[index] = tonumber(dateStr:sub(ptr), 10) end return parts[1], parts[2], parts[3] -- year, month, day end function convertUnit(unit, link) link = link or false local itemID, label, title if unit == "" or unit == "1" then return nil end itemID = parseWikidataURL(unit) if itemID then if itemID == aliasesQ.percentage then return "%" else label = mw.wikibase.label(itemID) title = nil if link or label == nil then title = mw.wikibase.sitelink(itemID) end if link then if title then return " " .. "[[" .. title .. "|" .. (label or title) .. "]]" end if not label then return " " .. "[[d:" .. itemID .. "|" .. itemID .. "]]" end end return " " .. (label or title or itemID) end end return " " .. unit end function getOrdinalSuffix(num) if tostring(num):sub(-2,-2) == '1' then return "th" -- 10th, 11th, 12th, 13th, ... 19th end num = tostring(num):sub(-1) if num == '1' then return "st" elseif num == '2' then return "nd" elseif num == '3' then return "rd" else return "th" end end function addDecimalMarks(n) local left,num,right = string.match(n,'^([^%d]*%d)(%d*)(.-)$') return left..(num:reverse():gsub('(%d%d%d)','%1,'):reverse())..right end function convertRank(rank) if rank == "preferred" then return 1 elseif rank == "normal" then return 2 elseif rank == "deprecated" then return 3 else return 4 -- default (in its literal sense) end end function datePrecedesDate(aY, aM, aD, bY, bM, bD) if aY == nil or bY == nil then return nil end aM = aM or 1 aD = aD or 1 bM = bM or 1 bD = bD or 1 if aY < bY then return true end if aY > bY then return false end if aM < bM then return true end if aM > bM then return false end if aD < bD then return true end return false end function alwaysTrue() return true end -- The following function parses a format string. -- -- The example below shows how a parsed string is structured in memory. -- Variables other than 'str' and 'child' are left out for clarity's sake. -- -- Example: -- "A %p B [%s[%q1]] C [%r] D" -- -- Structure: -- [ -- { -- str = "A " -- }, -- { -- str = "%p" -- }, -- { -- str = " B ", -- child = -- [ -- { -- str = "%s", -- child = -- [ -- { -- str = "%q1" -- } -- ] -- } -- ] -- }, -- { -- str = " C ", -- child = -- [ -- { -- str = "%r" -- } -- ] -- }, -- { -- str = " D" -- } -- ] -- function parseFormat(str) local chr, esc, param, root, cur, prev, new local params = {} local function newObject(array) local obj = {} -- new object obj.str = "" array[#array + 1] = obj -- array{object} obj.parent = array return obj end local function endParam() if param > 0 then if cur.str ~= "" then cur.str = "%"..cur.str cur.param = true params[cur.str] = true cur.parent.req[cur.str] = true prev = cur cur = newObject(cur.parent) end param = 0 end end root = {} -- array root.req = {} cur = newObject(root) prev = nil esc = false param = 0 for i = 1, #str do chr = str:sub(i,i) if not esc then if chr == '\\' then endParam() esc = true elseif chr == '%' then endParam() if cur.str ~= "" then cur = newObject(cur.parent) end param = 2 elseif chr == '[' then endParam() if prev and cur.str == "" then table.remove(cur.parent) cur = prev end cur.child = {} -- new array cur.child.req = {} cur.child.parent = cur cur = newObject(cur.child) elseif chr == ']' then endParam() if cur.parent.parent then new = newObject(cur.parent.parent.parent) if cur.str == "" then table.remove(cur.parent) end cur = new end else if param > 1 then param = param - 1 elseif param == 1 then if not string.match(chr, '%d') then endParam() end end cur.str = cur.str .. chr end else cur.str = cur.str .. chr esc = false end prev = nil end endParam() return root, params end function sortOnRank(claims) local rankPos local ranks = {{}, {}, {}, {}} -- preferred, normal, deprecated, (default) local sorted = {} for i, v in ipairs(claims) do rankPos = convertRank(v.rank) ranks[rankPos][#ranks[rankPos] + 1] = v end sorted = ranks[1] sorted = mergeTables(sorted, ranks[2]) sorted = mergeTables(sorted, ranks[3]) return sorted end function getShortName(itemID) return p._property({itemID, aliasesP.shortName}) -- "property" is single end function getLabel(ID) return p._label({ID}) end function Config:getValue(snak, raw, link, short, anyLang) raw = raw or false link = link or false short = short or false anyLang = anyLang or false if snak.snaktype == 'value' then if snak.datavalue.type == 'string' then return snak.datavalue.value elseif snak.datavalue.type == 'monolingualtext' then if anyLang then return snak.datavalue.value['text'], snak.datavalue.value['language'] elseif snak.datavalue.value['language'] == self.langCode then return snak.datavalue.value['text'] else return nil end elseif snak.datavalue.type == 'quantity' then -- strip + signs from front local value = mw.ustring.gsub(snak.datavalue.value['amount'], "^\+(.+)$", "%1") if not raw then value = addDecimalMarks(value) local unit = convertUnit(snak.datavalue.value['unit'], link) if unit then value = value .. unit end end return value elseif snak.datavalue.type == 'time' then local y, m, d, p, yDiv, yRound, yFull, value, calendarID, dateStr local yFactor = 1 local sign = 1 local suffix = "" local mayAddCalendar = false local calendar = "" local precision = snak.datavalue.value['precision'] if precision == 11 then p = "d" elseif precision == 10 then p = "m" else p = "y" yFactor = 10^(9-precision) end y, m, d = parseDate(snak.datavalue.value['time'], p) if y < 0 then sign = -1 y = y * sign end -- if precision is tens/hundreds/thousands/millions/billions of years if precision <= 8 then yDiv = y / yFactor -- if precision is tens/hundreds/thousands of years if precision >= 6 then mayAddCalendar = true if precision <= 7 then -- round centuries/millenniums up (e.g. 20th century or 3rd millennium) yRound = math.ceil(yDiv) if not raw then if precision == 6 then suffix = " millennium" else suffix = " century" end suffix = getOrdinalSuffix(yRound) .. suffix else -- if not verbose, take the first year of the century/millennium -- (e.g. 1901 for 20th century or 2001 for 3rd millennium) yRound = (yRound - 1) * yFactor + 1 end else -- precision == 8 -- round decades down (e.g. 2010s) yRound = math.floor(yDiv) * yFactor if not raw then suffix = "s" end end if raw and sign < 0 then -- if BCE then compensate for "counting backwards" -- (e.g. -2019 for 2010s BCE, -2000 for 20th century BCE or -3000 for 3rd millennium BCE) yRound = yRound + yFactor - 1 end else local yReFactor, yReDiv, yReRound -- round to nearest for tens of thousands of years or more yRound = math.floor(yDiv + 0.5) if yRound == 0 then if precision <= 2 and y ~= 0 then yReFactor = 1e6 yReDiv = y / yReFactor yReRound = math.floor(yReDiv + 0.5) if yReDiv == yReRound then -- change precision to millions of years only if we have a whole number of them precision = 3 yFactor = yReFactor yRound = yReRound end end if yRound == 0 then -- otherwise, take the unrounded (original) number of years precision = 5 yFactor = 1 yRound = y mayAddCalendar = true end end if precision >= 1 and y ~= 0 then yFull = yRound * yFactor yReFactor = 1e9 yReDiv = yFull / yReFactor yReRound = math.floor(yReDiv + 0.5) if yReDiv == yReRound then -- change precision to billions of years if we're in that range precision = 0 yFactor = yReFactor yRound = yReRound else yReFactor = 1e6 yReDiv = yFull / yReFactor yReRound = math.floor(yReDiv + 0.5) if yReDiv == yReRound then -- change precision to millions of years if we're in that range precision = 3 yFactor = yReFactor yRound = yReRound end end end if not raw then if precision == 3 then suffix = " million years" elseif precision == 0 then suffix = " billion years" else yRound = yRound * yFactor if yRound == 1 then suffix = " year" else suffix = " years" end end else yRound = yRound * yFactor end end else yRound = y mayAddCalendar = true end if mayAddCalendar then calendarID = parseWikidataURL(snak.datavalue.value['calendarmodel']) if calendarID and calendarID == aliasesQ.prolepticJulianCalendar then if not raw then if link then calendar = " ([[Julian calendar|Julian]])" else calendar = " (Julian)" end else calendar = "/Julian" end end end if not raw then local ce = nil if sign < 0 then ce = "BCE" elseif precision <= 5 then ce = "CE" end if ce then if link then ce = "[[Common Era|" .. ce .. "]]" end suffix = suffix .. " " .. ce end value = tostring(yRound) if m then dateStr = self.langObj:formatDate("F", "1-"..m.."-1") if d then if self.mdyDate then dateStr = dateStr .. " " .. d .. "," else dateStr = d .. " " .. dateStr end end value = dateStr .. " " .. value end value = value .. suffix .. calendar else value = tostring(yRound * sign) if m then value = value .. "-" .. m if d then value = value .. "-" .. d end end value = value .. calendar end return value elseif snak.datavalue.type == 'globecoordinate' then -- logic from https://github.com/DataValues/Geo local precision, numDigits, strFormat, value, globe local latValue, latitude, latDegrees, latMinutes, latSeconds local latDirection = "N" local lonValue, longitude, lonDegrees, lonMinutes, lonSeconds local lonDirection = "E" local degSymbol = "°" local minSymbol = "'" local secSymbol = '"' local partsGlue = ", " if raw then degSymbol = "/" minSymbol = "/" secSymbol = "/" partsGlue = "/" end latitude = snak.datavalue.value['latitude'] longitude = snak.datavalue.value['longitude'] if latitude < 0 then latDirection = "S" latitude = math.abs(latitude) end if longitude < 0 then lonDirection = "W" longitude = math.abs(longitude) end precision = snak.datavalue.value['precision'] latitude = math.floor(latitude / precision + 0.5) * precision longitude = math.floor(longitude / precision + 0.5) * precision numDigits = math.ceil(-math.log10(3600 * precision)) if numDigits < 0 or numDigits == -0 then numDigits = 0 end strFormat = "%." .. numDigits .. "f" -- use string.format() to strip decimal point followed by a zero (.0) for whole numbers latSeconds = tonumber(string.format(strFormat, math.floor(latitude * 3600 * 10^numDigits + 0.5) / 10^numDigits)) lonSeconds = tonumber(string.format(strFormat, math.floor(longitude * 3600 * 10^numDigits + 0.5) / 10^numDigits)) latMinutes = math.floor(latSeconds / 60) lonMinutes = math.floor(lonSeconds / 60) latSeconds = latSeconds - (latMinutes * 60) lonSeconds = lonSeconds - (lonMinutes * 60) latDegrees = math.floor(latMinutes / 60) lonDegrees = math.floor(lonMinutes / 60) latMinutes = latMinutes - (latDegrees * 60) lonMinutes = lonMinutes - (lonDegrees * 60) latValue = latDegrees .. degSymbol lonValue = lonDegrees .. degSymbol if precision < 1 then latValue = latValue .. latMinutes .. minSymbol lonValue = lonValue .. lonMinutes .. minSymbol end if precision < (1 / 60) then latSeconds = string.format(strFormat, latSeconds) lonSeconds = string.format(strFormat, lonSeconds) latValue = latValue .. latSeconds .. secSymbol lonValue = lonValue .. lonSeconds .. secSymbol end latValue = latValue .. latDirection lonValue = lonValue .. lonDirection value = latValue .. partsGlue .. lonValue if link then globe = parseWikidataURL(snak.datavalue.value['globe']) if globe then globe = mw.wikibase.label(globe):lower() else globe = "earth" end value = "[https://tools.wmflabs.org/geohack/geohack.php?language="..self.langCode.."¶ms="..latitude.."_"..latDirection.."_"..longitude.."_"..lonDirection.."_globe:"..globe.." "..value.."]" end return value elseif snak.datavalue.type == 'wikibase-entityid' then local value = "" local title = nil local itemID = "Q" .. snak.datavalue.value['numeric-id'] if raw then if link then return "[[d:" .. itemID .. "|" .. itemID .. "]]" else return itemID end end if short then value = getShortName(itemID) end if value == "" then value = mw.wikibase.label(itemID) end if link or value == nil then title = mw.wikibase.sitelink(itemID) end if link then if title then value = "[[" .. title .. "|" .. (value or title) .. "]]" elseif not value then value = "[[d:" .. itemID .. "|" .. itemID .. "]]" end elseif not value then value = (title or itemID) end return value else return '<strong class="error">' .. unknownDataTypeError(snak.datavalue.type) .. '.</strong>' end elseif snak.snaktype == 'somevalue' then if raw then return " " -- single space represents 'somevalue' else return "unknown" end elseif snak.snaktype == 'novalue' then if raw then return "" -- empty string represents 'novalue' else return "none" end else return nil end end function Config:getSingleRawQualifier(claim, qualifierID) local qualifiers if claim.qualifiers then qualifiers = claim.qualifiers[qualifierID] end if qualifiers and qualifiers[1] then return self:getValue(qualifiers[1], true) -- raw = true else return nil end end function Config:snakEqualsValue(snak, value) local snakValue = self:getValue(snak, true) -- raw = true if snakValue and snak.snaktype == 'value' and snak.datavalue.type == 'wikibase-entityid' then value = value:upper() end return snakValue == value end function Config:setRank(rank) local rankPos if rank == "best" then self.bestRank = true self.flagBest = true -- mark that 'best' flag was given return end if rank:sub(1,9) == "preferred" then rankPos = 1 elseif rank:sub(1,6) == "normal" then rankPos = 2 elseif rank:sub(1,10) == "deprecated" then rankPos = 3 else return end -- one of the rank flags was given, check if another one was given before if not self.flagRank then self.ranks = {false, false, false} -- no other rank flag given before, so unset ranks self.bestRank = self.flagBest -- unsets bestRank only if 'best' flag was not given before self.flagRank = true -- mark that a rank flag was given end if rank:sub(-1) == "+" then for i = rankPos, 1, -1 do self.ranks[i] = true end elseif rank:sub(-1) == "-" then for i = rankPos, #self.ranks do self.ranks[i] = true end else self.ranks[rankPos] = true end end function Config:setPeriod(period) local periodPos if period == "future" then periodPos = 1 elseif period == "current" then periodPos = 2 elseif period == "former" then periodPos = 3 else return end -- one of the period flags was given, check if another one was given before if not self.flagPeriod then self.periods = {false, false, false} -- no other period flag given before, so unset periods self.flagPeriod = true -- mark that a period flag was given end self.periods[periodPos] = true end function Config:processFlag(flag) if not flag then return false else flag = mw.text.trim(flag) end if flag == "linked" then self.curState.linked = true return true elseif flag == "raw" then self.curState.rawValue = true if self.curState == self.states[parameters.reference] then -- raw reference values end with periods and require a separator different from "" self.separators["sep%r"][1] = " " end return true elseif flag == "short" then self.curState.shortName = true return true elseif flag == "mdy" then self.mdyDate = true return true elseif flag == "best" or flag:match('^preferred[+-]?$') or flag:match('^normal[+-]?$') or flag:match('^deprecated[+-]?$') then self:setRank(flag) return true elseif flag == "future" or flag == "current" or flag == "former" then self:setPeriod(flag) return true elseif flag == "" then -- ignore empty flags and carry on return true else return false end end function Config:processFlagOrCommand(flag) local param = "" if not flag then return false else flag = mw.text.trim(flag) end if flag == "property" or flag == "properties" then param = parameters.property elseif flag:match('^qualifier[s]?$') then self.states.qualifiersCount = self.states.qualifiersCount + 1 param = parameters.qualifier .. self.states.qualifiersCount self.separators["sep"..param] = {defaultSeparators["sep%q\\d"]} elseif flag:match('^reference[s]?$') then param = parameters.reference else return self:processFlag(flag) end if self.states[param] then return false end -- create a new State for each command self.states[param] = State.new(self) -- use "%x" as the general parameter name self.states[param].parsedFormat = parseFormat(parameters.general) -- will be overwritten for param=="%p" -- set the separator self.states[param].separator = self.separators["sep"..param] -- will be nil for param=="%p", which will be set separately if string.sub(flag, -1) ~= 's' then self.states[param].singleValue = true end self.curState = self.states[param] return true end function Config:rankMatches(rankPos) if self.bestRank then return (self.ranks[rankPos] and self.foundRank >= rankPos) else return self.ranks[rankPos] end end function Config:timeMatches(claim) local startTime = nil local startTimeY = nil local startTimeM = nil local startTimeD = nil local endTime = nil local endTimeY = nil local endTimeM = nil local endTimeD = nil if self.periods[1] and self.periods[2] and self.periods[3] then -- any time return true end local now = os.date('!*t') startTime = self:getSingleRawQualifier(claim, aliasesP.startTime) if startTime and startTime ~= "" and startTime ~= " " then startTimeY, startTimeM, startTimeD = parseDate(startTime) end endTime = self:getSingleRawQualifier(claim, aliasesP.endTime) if endTime and endTime ~= "" and endTime ~= " " then endTimeY, endTimeM, endTimeD = parseDate(endTime) elseif endTime == " " then -- end time is 'unknown', assume it is somewhere in the past; -- we can do this by taking the current date as a placeholder for the end time endTimeY = now['year'] endTimeM = now['month'] endTimeD = now['day'] end if startTimeY ~= nil and endTimeY ~= nil and datePrecedesDate(endTimeY, endTimeM, endTimeD, startTimeY, startTimeM, startTimeD) then -- invalidate end time if it precedes start time endTimeY = nil endTimeM = nil endTimeD = nil end if self.periods[1] then -- future if startTimeY and datePrecedesDate(now['year'], now['month'], now['day'], startTimeY, startTimeM, startTimeD) then return true end end if self.periods[2] then -- current if (startTimeY == nil or not datePrecedesDate(now['year'], now['month'], now['day'], startTimeY, startTimeM, startTimeD)) and (endTimeY == nil or datePrecedesDate(now['year'], now['month'], now['day'], endTimeY, endTimeM, endTimeD)) then return true end end if self.periods[3] then -- former if endTimeY and not datePrecedesDate(now['year'], now['month'], now['day'], endTimeY, endTimeM, endTimeD) then return true end end return false end function State:claimMatches(claim) local matches, rankPos -- if a property value was given, check if it matches the claim's property value if self.conf.propertyValue then matches = self.conf:snakEqualsValue(claim.mainsnak, self.conf.propertyValue) else matches = true end -- check if the claim's rank and time period match rankPos = convertRank(claim.rank) matches = (matches and self.conf:rankMatches(rankPos) and self.conf:timeMatches(claim)) return matches, rankPos end function State:out() local result -- collection of arrays with value objects local valuesArray -- array with value objects local sep = "" local out = {} -- array with value objects local function walk(formatTable, result) local valuesArray = {} -- array with value objects for i, v in pairs(formatTable.req) do if not result[i] or not result[i][1] then -- we've got no result for a parameter that is required on this level, -- so skip this level (and its children) by returning an empty result return {} end end for i, v in ipairs(formatTable) do if v.param then valuesArray = mergeTables(valuesArray, result[v.str]) else valuesArray[#valuesArray + 1] = {v.str} end if v.child then valuesArray = mergeTables(valuesArray, walk(v.child, result)) end end return valuesArray end -- iterate through the results from back to front, so that we know when to add separators for i = #self.results, 1, -1 do result = self.results[i] -- if there is already some output, then add the separators if #out > 0 then sep = self.separator[1] -- fixed separator result[parameters.separator] = { {self.movSeparator[1]} } -- movable separator else sep = "" result[parameters.separator] = { {self.puncMark[1]} } -- optional punctuation mark end valuesArray = walk(self.parsedFormat, result) if #valuesArray > 0 then valuesArray[#valuesArray + 1] = {sep} out = mergeTables(valuesArray, out) end end -- reset state before next iteration self.results = {} return out end -- level 1 hook function State:getProperty(claim) local value = {self.conf:getValue(claim.mainsnak, self.rawValue, self.linked, self.shortName)} -- create one value object if #value > 0 then return {value} -- wrap the value object in an array and return it else return {} -- return empty array if there was no value end end -- level 1 hook function State:getQualifiers(claim, param) local qualifiers if claim.qualifiers then qualifiers = claim.qualifiers[self.conf.qualifierIDs[param]] end if qualifiers then -- iterate through claim's qualifier statements to collect their values; -- return array with multiple value objects return self.conf.states[param]:iterate(qualifiers, {[parameters.general] = hookNames[parameters.qualifier.."\\d"][2], count = 1}) -- pass qualifier State with level 2 hook else return {} -- return empty array end end -- level 2 hook function State:getQualifier(snak) local value = {self.conf:getValue(snak, self.rawValue, self.linked, self.shortName)} -- create one value object if #value > 0 then return {value} -- wrap the value object in an array and return it else return {} -- return empty array if there was no value end end -- level 1 hook function State:getAllQualifiers(claim, param, result, hooks) local out = {} -- array with value objects local sep = self.conf.separators["sep"..parameters.qualifier][1] -- iterate through the output of the separate "qualifier(s)" commands for i = 1, self.conf.states.qualifiersCount do -- if a hook has not been called yet, call it now if not result[parameters.qualifier..i] then self:callHook(parameters.qualifier..i, hooks, claim, result) end -- if there is output for this particular "qualifier(s)" command, then add it if result[parameters.qualifier..i] and result[parameters.qualifier..i][1] then -- if there is already some output, then add the separator if #out > 0 then out[#out + 1] = {sep} end out = mergeTables(out, result[parameters.qualifier..i]) end end return out end -- level 1 hook function State:getReferences(claim) if claim.references then -- iterate through claim's reference statements to collect their values; -- return array with multiple value objects return self.conf.states[parameters.reference]:iterate(claim.references, {[parameters.general] = hookNames[parameters.reference][2], count = 1}) -- pass reference State with level 2 hook else return {} -- return empty array end end -- level 2 hook -- logic determined based on https://www.wikidata.org/wiki/Help:Sources function State:getReference(statement) local snakValue, lang, property local value = "" local ref = {} local snaks = {} local params = {} local leadParams = {} if statement.snaks then for i, v in pairs(statement.snaks) do if v[1] then snaks[i] = v[1] end end if snaks[aliasesP.importedFrom] then snaks[aliasesP.importedFrom] = nil end if snaks[aliasesP.referenceURL] and snaks[aliasesP.title] then params["url"] = self.conf:getValue(snaks[aliasesP.referenceURL]) params["title"] = self.conf:getValue(snaks[aliasesP.title], false, false, false, true) -- anyLang = true if snaks[aliasesP.publicationDate] then params["date"] = self.conf:getValue(snaks[aliasesP.publicationDate]) end if snaks[aliasesP.retrieved] then params["access-date"] = self.conf:getValue(snaks[aliasesP.retrieved]) end if snaks[aliasesP.archiveURL] then params["archive-url"] = self.conf:getValue(snaks[aliasesP.archiveURL]) end if snaks[aliasesP.archiveDate] then params["archive-date"] = self.conf:getValue(snaks[aliasesP.archiveDate]) end if snaks[aliasesP.author] then params["author"] = self.conf:getValue(snaks[aliasesP.author]) end if snaks[aliasesP.publisher] then params["publisher"] = self.conf:getValue(snaks[aliasesP.publisher]) end if snaks[aliasesP.quote] then params["quote"] = self.conf:getValue(snaks[aliasesP.quote], false, false, false, true) end -- anyLang = true if snaks[aliasesP.language] then snakValue = self.conf:getValue(snaks[aliasesP.language]) if self.conf.langName ~= snakValue then params["language"] = snakValue end end value = mw.getCurrentFrame():expandTemplate{title="cite_web", args=params} else for i, v in pairs(snaks) do property = getLabel(i) if property ~= "" then snakValue, lang = self.conf:getValue(v, false, (i == aliasesP.statedIn), false, true) -- link = true/false, anyLang = true if lang and lang ~= self.conf.langCode then snakValue = "''" .. snakValue .. "'' (" .. mw.language.fetchLanguageName(lang, self.conf.langCode) .. ")" end if i == aliasesP.referenceURL or i == aliasesP.statedIn then leadParams[#leadParams + 1] = snakValue elseif i ~= aliasesP.language or self.conf.langName ~= snakValue then params[#params + 1] = property .. ": " .. snakValue end end end value = table.concat(leadParams, "; ") params = table.concat(params, "; ") if params ~= "" then if value ~= "" then value = value .. "; " end value = value .. params end if value ~= "" then value = value .. "." end end if value ~= "" then ref = {value} -- create one value object if not self.rawValue then -- this should become a <ref> tag, so safe the reference's hash for later ref.refHash = statement.hash end ref = {ref} -- wrap the value object in an array end end return ref end function State:callHook(param, hooks, statement, result) local valuesArray, refHash -- call a parameter's hook if it has been defined and if it has not been called before if not result[param] and hooks[param] then valuesArray = self[hooks[param]](self, statement, param, result, hooks) -- array with value objects -- add to the result if #valuesArray > 0 then result[param] = valuesArray result.count = result.count + 1 else result[param] = {} -- an empty array to indicate that we've tried this hook already return true -- miss == true end end return false end -- iterate through claims, claim's qualifiers or claim's references to collect values function State:iterate(statements, hooks, matchHook) matchHook = matchHook or alwaysTrue local matches = false local rankPos = nil local result, gotRequired for i, v in ipairs(statements) do -- rankPos will be nil for non-claim statements (e.g. qualifiers, references, etc.) matches, rankPos = matchHook(self, v) if matches then result = {count = 0} -- collection of arrays with value objects local function walk(formatTable) local miss for i2, v2 in pairs(formatTable.req) do -- call a hook, adding its return value to the result miss = self:callHook(i2, hooks, v, result) if miss then -- we miss a required value for this level, so return false return false end if result.count == hooks.count then -- we're done if all hooks have been called; -- returning at this point breaks the loop return true end end for i2, v2 in ipairs(formatTable) do if result.count == hooks.count then -- we're done if all hooks have been called; -- returning at this point prevents further childs from being processed return true end if v2.child then walk(v2.child) end end return true end gotRequired = walk(self.parsedFormat) -- only append the result if we got values for all required parameters on the root level if gotRequired then -- if we have a rankPos (only with matchHook() for complete claims), then update the foundRank if rankPos and self.conf.foundRank > rankPos then self.conf.foundRank = rankPos end -- append the result self.results[#self.results + 1] = result -- break if we only need a single value if self.singleValue then break end end end end return self:out() end function p.property(frame) return p._property(copyTable(frame.args)) end function p._property(args) return execCommand(args, "property") end function p.properties(frame) return p._properties(copyTable(frame.args)) end function p._properties(args) return execCommand(args, "properties") end function p.qualifier(frame) return p._qualifier(copyTable(frame.args)) end function p._qualifier(args) return execCommand(args, "qualifier") end function p.qualifiers(frame) return p._qualifiers(copyTable(frame.args)) end function p._qualifiers(args) return execCommand(args, "qualifiers") end function p.reference(frame) return p._reference(copyTable(frame.args)) end function p._reference(args) return execCommand(args, "reference") end function p.references(frame) return p._references(copyTable(frame.args)) end function p._references(args) return execCommand(args, "references") end function execCommand(args, funcName) _ = Config.new() _:processFlagOrCommand(funcName) -- process first command (== function name) local parsedFormat, formatParams, claims local hooks = {count = 0} local nextArg = args[1] local nextIndex = 2 -- process flags and commands while _:processFlagOrCommand(nextArg) do nextArg = args[nextIndex] nextIndex = nextIndex + 1 end if nextArg then nextArg = mw.text.trim(nextArg) else nextArg = "" end -- check for optional item ID if nextArg:sub(1,1):upper() == "Q" then _.entity = mw.wikibase.getEntity(nextArg) -- item ID given _.propertyID = mw.text.trim(args[nextIndex] or "") -- property ID nextIndex = nextIndex + 1 else _.entity = mw.wikibase.getEntity() -- no item ID given, use item connected to current page _.propertyID = nextArg -- property ID end -- check if given property ID is an alias if aliasesP[_.propertyID] then _.propertyID = aliasesP[_.propertyID] end _.propertyID = _.propertyID:upper() if _.states.qualifiersCount > 0 then -- do further processing if "qualifier(s)" command was given if #args - nextIndex + 1 > _.states.qualifiersCount then -- claim ID or literal value has been given nextArg = args[nextIndex] -- don't trim because might be single space representing 'somevalue' nextIndex = nextIndex + 1 _.propertyValue = nextArg end for i = 1, _.states.qualifiersCount do nextArg = mw.text.trim(args[nextIndex] or "") -- is a qualifierID nextIndex = nextIndex + 1 -- check if given qualifier ID is an alias if aliasesP[nextArg] then nextArg = aliasesP[nextArg] end _.qualifierIDs[parameters.qualifier..i] = nextArg:upper() end elseif _.states[parameters.reference] then -- do further processing if "reference(s)" command was given nextArg = args[nextIndex] nextIndex = nextIndex + 1 _.propertyValue = nextArg -- claim ID or literal value (possibly nil) end -- check for special property value 'somevalue' or 'novalue' if _.propertyValue then if _.propertyValue ~= "" and mw.text.trim(_.propertyValue) == "" then _.propertyValue = " " -- single space represents 'somevalue', whereas empty string represents 'novalue' else _.propertyValue = mw.text.trim(_.propertyValue) end end -- parse the desired format, or choose an appropriate format if args["format"] then parsedFormat, formatParams = parseFormat(mw.text.trim(args["format"])) elseif _.states.qualifiersCount > 0 then if _.states[parameters.property] then parsedFormat, formatParams = parseFormat(formats.propertyWithQualifier) else parsedFormat, formatParams = parseFormat(formats.qualifier) end elseif _.states[parameters.property] then parsedFormat, formatParams = parseFormat(formats.property) else parsedFormat, formatParams = parseFormat(formats.reference) -- if only "reference(s)" has been given, make the emtpy string the default separator (except when raw) if not _.states[parameters.reference].rawValue then _.separators["sep"][1] = "" end end -- if exactly one "qualifier(s)" command has been given, make "sep%q" point to "sep%q1" to make them equivalent -- must come BEFORE overriding the separator values if _.states.qualifiersCount == 1 then _.separators["sep"..parameters.qualifier] = _.separators["sep"..parameters.qualifier.."1"] end -- process overridden separator values; -- must come AFTER parsing the formats for i, v in pairs(_.separators) do if args[i] then _.separators[i][1] = args[i] end end -- make sure that at least one required parameter has been defined if not next(parsedFormat.req) then error(missingRequiredParameterError()) end -- make sure that the separator parameter "%s" is not amongst the required parameters if parsedFormat.req[parameters.separator] then error(extraRequiredParameterError(parameters.separator)) end -- define the hooks that should be called (getProperty, getQualifiers, getReferences); -- only define a hook if both its command ("propert(y|ies)", "reference(s)", "qualifier(s)") and its parameter ("%p", "%r", "%q1", "%q2", "%q3") have been given for i, v in pairs(_.states) do -- e.g. 'formatParams["%q1"] or formatParams["%q"]' to define hook even if "%q1" was not defined to be able to build a complete value for "%q" if formatParams[i] or formatParams[string.sub(i, 1, 2)] then hooks[i] = getHookName(i, 1) hooks.count = hooks.count + 1 end end -- the "%q" parameter is not attached to a state, but is a collection of the results of multiple states (attached to "%q1", "%q2", "%q3", ...); -- so if this parameter is given then this hook must be defined separately, but only if at least one "qualifier(s)" command has been given if formatParams[parameters.qualifier] and _.states.qualifiersCount > 0 then hooks[parameters.qualifier] = getHookName(parameters.qualifier, 1) hooks.count = hooks.count + 1 end -- create a state for "properties" if it doesn't exist yet, which will be used as a base configuration for each claim iteration; -- must come AFTER defining the hooks if not _.states[parameters.property] then _.states[parameters.property] = State.new(_) end -- set the parsed format and the separators (and optional punctuation mark) _.states[parameters.property].parsedFormat = parsedFormat _.states[parameters.property].separator = _.separators["sep"] _.states[parameters.property].movSeparator = _.separators["sep"..parameters.separator] _.states[parameters.property].puncMark = _.separators["punc"] if _.entity and _.entity.claims then claims = _.entity.claims[_.propertyID] end if claims then -- first sort the claims on rank to pre-define the order of output (preferred first, then normal, then deprecated) claims = sortOnRank(claims) -- then iterate through the claims to collect values return concatValues(_.states[parameters.property]:iterate(claims, hooks, State.claimMatches)) -- pass property State with level 1 hooks and matchHook else return "" end end function p.label(frame) return p._label(copyTable(frame.args)) end function p._label(args, _) _ = _ or Config.new() _.curState = State.new(_) local ID local label = "" local title = nil local nextArg = args[1] local nextIndex = 2 while _:processFlag(nextArg) do nextArg = args[nextIndex] nextIndex = nextIndex + 1 end if nextArg then nextArg = mw.text.trim(nextArg) else nextArg = "" end ID = nextArg if ID ~= "" then if aliasesP[ID] then ID = aliasesP[ID] end ID = ID:upper() -- check if this is a valid ID, and if the number is not larger than max int (to prevent error) if not string.match(ID, '^[QP]%d+$') or tonumber(string.match(ID, '%d+')) > 2147483647 then return "" end if _.curState.rawValue then if mw.wikibase.getEntity(ID) or mw.wikibase.resolvePropertyId(ID) then if _.curState.linked then if ID:sub(1,1) == "P" then label = "[[d:Property:" .. ID .. "|" .. ID .. "]]" else label = "[[d:" .. ID .. "|" .. ID .. "]]" end else label = ID end end else if ID:sub(1,1) == "P" then if not _.pageTitle then label = mw.wikibase.label(ID) or "" if _.curState.linked and label ~= "" then label = "[[d:Property:" .. ID .. "|" .. label .. "]]" end end else if not _.pageTitle then if _.curState.shortName then label = getShortName(ID) end -- at this point, 'label' will be a string and not nil if label == "" then label = mw.wikibase.label(ID) end else -- set 'label' to nil so 'title' will always prevail label = nil end -- at this point, 'label' will be nil or a non-empty string if _.curState.linked or label == nil then title = mw.wikibase.sitelink(ID) end if _.curState.linked and title then label = "[[" .. title .. "|" .. (label or title) .. "]]" else label = label or title or "" end end end else if _.curState.rawValue then label = mw.wikibase.getEntityIdForCurrentPage() or "" if _.curState.linked and label ~= "" then label = "[[d:" .. label .. "|" .. label .. "]]" end else if not _.pageTitle then label = mw.wikibase.label() else -- set 'label' to nil so 'title' will always prevail label = nil end if _.curState.linked or label == nil then title = mw.title.getCurrentTitle().prefixedText end if _.curState.linked then label = "[[" .. title .. "|" .. (label or title) .. "]]" -- not much use since it links to the current page, but does add wiki mark-up else label = label or title end end end return label end function p.title(frame) return p._title(copyTable(frame.args)) end function p._title(args, _) _ = _ or Config.new() _.pageTitle = true return p._label(args, _) end -- main function that is supposed to be used by wrapper templates function p.main(frame) local f, args, i, v -- get the parent frame to take the arguments that were passed to the wrapper template frame = frame:getParent() or frame f = mw.text.trim(frame.args[1] or "") if f == "main" then error('The function "main" cannot be called twice') end assert(p["_"..f], 'The function "' .. f .. '" does not exist') -- copy arguments from immutable to mutable table args = copyTable(frame.args) -- remove the function name from the list table.remove(args, 1) return p["_"..f](args) end return p