Module:Lexeme

local p = {}
local wikidata = require('Module:Wikidata')
local i18nmessages = mw.loadData('Module:i18n/wikidata')

local function langSwitch(list,lang)
	lang = lang or mw.getCurrentFrame():callParserFunction("int", "lang")
	local langList = mw.language.getFallbacksFor(lang)
	table.insert(langList,1,lang)
	for i,language in ipairs(langList) do
		if list[language] then
			return list[language]
		end
	end
	return nil
end

local function i18n(str)
	local message = i18nmessages[str]
	if type(message) == 'string' then
		return message
	end
	return langSwitch(message) .. ''
end

p.getFormRepFromEntity = function(entity)
	lemmas = ''
	function valuesSortedByLanguage(t)
	    local values = {}
	    for k, v in pairs(t) do
	    	values[#values+1] = v
    	end
		table.sort(values,
			function(a,b)
				return (a[2] < b[2])
			end)
	    local i = 0
	    return function()
	        i = i + 1
	        if values[i] then
	            return values[i]
	        end
	    end
	end
	for v in valuesSortedByLanguage(entity:getRepresentations()) do
		if lemmas ~= '' then
			lemmas = lemmas .. '/' .. v[1]
		else
			lemmas = v[1]
		end
    end
	return lemmas     
end

p.getFormRep = function(frame)
	local args = frame.args
    local entity = mw.wikibase.getEntity(args[1])
    return p.getFormRepFromEntity(entity)
end

-- Note: All functions here are hacks
local function getEntity(frame)
	local args = frame.args
	local entity = args.entity
	if not entity then
		return nil
	end

	if (string.sub(entity, 1, 1) ~= 'L') or (not tonumber(string.sub(entity, 2))) then
		return nil
	end
	
	local content = mw.title.new('Lexeme:' .. entity):getContent()
	return content and mw.text.jsonDecode(content)
end

-- Returns the lemmas of the entity concatenated with '/'.
local function getLemma(entity) -- simple for simple templates like {{Q|}}}
	if not entity then
		return i18n('invalid-id')
	end

	local lemmas = ''
	local function valuesSortedByLanguage(t)
	    local values = {}
	    for k, v in pairs(t) do
	    	values[#values+1] = v
    	end
		table.sort(values,
			function(a,b)
				return (a.language < b.language)
			end)
	    local i = 0
	    return function()
	        i = i + 1
	        if values[i] then
	            return values[i]
	        end
	    end
	end
	if entity.lemmas then
		for v in valuesSortedByLanguage(entity.lemmas) do
			if lemmas ~= '' then
				lemmas = lemmas .. '/' .. '<span lang=' .. v.language .. ' dir="auto">' .. v.value .. '</span>'
			else
				lemmas = '<span lang=' .. v.language .. ' dir="auto">' .. v.value .. '</span>'
			end
		end
	end
	return lemmas 
end

-- for other Lua modules: the passed ID must include the 'L' prefix
-- debug with =p.getLemmaById('L476308')
function p.getLemmaById(id)
	return getLemma(mw.wikibase.getEntity(id))
end

-- simple for simple templates like {{Q|}}}
function p.getLemma(frame)
	return getLemma(getEntity(frame))
end

local function process_lang(langcode, lang)
	local langstring = langcode
	local nonISOstart, nonISOend = string.find(langcode, '%-x%-', 1, plain)
	if nonISOstart then
		local langlabel = wikidata._getLabel(string.sub(langcode, nonISOend+1), lang)
		langstring = mw.ustring.gsub(langstring, '%-x%-.+', ' (' .. langlabel .. ')')
		local langname = mw.language.fetchLanguageName(string.sub(langcode, 1, nonISOstart-1), lang)
		langstring = mw.ustring.gsub(langstring, '^.+ %(', langname .. ' (')
	else
		langstring = mw.language.fetchLanguageName(langstring, lang)
	end
	return langstring
end

local function getDetailLemma(entity, lang)
	if not entity then
		return i18n('invalid-id')
	end

	lemmas = ''
	for k, v in pairs(entity.lemmas) do
		local tocat = v.value .. ' <sup><i>' .. process_lang(v.language, lang) .. '</i></sup>'
		if lemmas ~= '' then
			lemmas = lemmas .. '/' .. tocat
		else
			lemmas = tocat
		end
	end
	lemmas = lemmas .. ' <sub><i>' .. wikidata._getLabel(entity.lexicalCategory, lang) .. '</i></sub>'
	return '&#32;' .. lemmas 
end

function p.getDetailLemma(frame) -- simple for simple templates like {{Q|}}}
	return getDetailLemma(getEntity(frame), frame.args.lang)
end

local function getLanguage(entity)
	return entity and entity.language or i18n('invalid-id')
end

function p.getLanguage(frame) -- simple for simple templates like {{Q|}}}
	return getLanguage(getEntity(frame))
end

local function getCategory(entity)
	return entity and entity.lexicalCategory or i18n('invalid-id')
end

function p.getCategory(frame) -- simple for simple templates like {{Q|}}}
	return getCategory(getEntity(frame))
end

function p.lexemerows(frame)
	local args = frame.args
	local entityid = args.entity
	local entity=getEntity(frame)
    if entity then
    	return frame:preprocess("|-\n| [[Lexeme:" .. entityid .. "|" .. entityid .. "]] || {{Q|" .. getLanguage(entity) .. "}} || {{Q|" .. getCategory(entity) .. "}} || " .. getLemma(entity) )
    else
    	return "|-\n| [[Lexeme:" .. entityid .. "|" .. entityid .. "]]||colspan='3'|Does not exist"
    end
end

function p.lexemerowsfr(frame)
	local args = frame.args
	local entityid = args.entity
	local entity=getEntity(frame)
    if entity then
		local lang = getLanguage(entity)
		local cat = getCategory(entity)
		local color = "white"
		if cat == "Q24905" then color = "#ff9900"
			elseif cat == "Q1084" then color = "#99e699"
		end
    	if lang=="Q150" then
    		return frame:preprocess('|- valign="top" style="background-color: '.. color .. '"\n| [[Lexeme:' .. entityid .. "|" .. entityid .. "]] ||style='white-space:nowrap'| " .. getLemma(entity) .. " || {{LinkedLabel|" .. cat .. "|fr}} (" .. cat .. ")"  )
	    else
    		return nil
    	end
    else
    	return nil
	end
end

function p.lexemerowsfrraw(frame)
	local args = frame.args
	local entityid = args.entity
	local entity=getEntity(frame)
    if entity then
		local lang = getLanguage(entity)
    	if lang=="Q150" then
    		return "{{TR lexeme fr|" .. entityid .. "}}" .. frame:preprocess("\n")
	    else
    		return nil
    	end
    else
    	return nil
	end
end


function p.LanguageStatistics(frame)
	local args = frame.args
	local s = args[1]
	local e = args[2]
	local r = {}
	for i = s,e do
		entity = mw.title.new('Lexeme:L' .. i):getContent()
		if entity then
			lang=mw.text.jsonDecode(entity).language
			if lang ~= nil then
				if r[lang] == nil then
					r[lang] = 1
				else
					r[lang] = r[lang] + 1
				end
			end
		end	
	end
	return mw.dumpObject(r):gsub("[ \n]", "")
end

function p.CategoryStatistics(frame)
	local args = frame.args
	local s = args[1]
	local e = args[2]
	local r = {}
	for i = s,e do
		entity = mw.title.new('Lexeme:L' .. i):getContent()
		if entity then
			cat=mw.text.jsonDecode(entity).lexicalCategory
			if cat ~= nil then
				if r[cat] == nil then
					r[cat] = 1
				else
					r[cat] = r[cat] + 1
				end
			end
		end	
	end
	return mw.dumpObject(r):gsub("[ \n]", "")
end

local function MergeStatistics(s)
	local r = {}
	for i, v in ipairs(s) do
		for k, t in pairs(v) do
			if r[k] == nil then
				r[k] = t
			else
				r[k] = r[k] + t
			end
		end
	end
	return r
end

local function FullLanguageStatistics()
	return MergeStatistics(require("Module:Lexeme/data").language)
end

local function FullCategoryStatistics()
	return MergeStatistics(require("Module:Lexeme/data").lexicalCategory)
end

local function itemcomp(t1,t2)
	return t1[2]==t2[2] and tonumber(string.sub(t1[1],2))<tonumber(string.sub(t2[1],2)) or t1[2]>t2[2]

end

local function FullLanguageStatisticsTable()
	local r = {}
	for k, t in pairs(FullLanguageStatistics()) do
		table.insert(r,{k,t})
	end
	table.sort(r,itemcomp)
	return r
end

local function FullCategoryStatisticsTable()
	local r = {}
	for k, t in pairs(FullCategoryStatistics()) do
		table.insert(r,{k,t})
	end
	table.sort(r,itemcomp)
	return r
end

function p.LanguageStatisticsTable(frame)
	t=""
	for i, v in ipairs(FullLanguageStatisticsTable()) do
		t = t .. frame:preprocess("|-\n| {{Q|" .. v[1] .. "}} || " .. v[2] .. "\n")
	end
    return t
end

function p.CategoryStatisticsTable(frame)
	t=""
	ct = 0
	for i, v in ipairs(FullCategoryStatisticsTable()) do
		ct = ct + v[2]
	end
	for i, v in ipairs(FullCategoryStatisticsTable()) do
		t = t .. frame:preprocess("|-\n| {{Q|" .. v[1] .. "}} || " .. v[2] .. " || " .. math.floor(v[2]/ct*10000)/100 .. "\n")
	end
    return t
end

function p.NumLanguage(frame)
    return #FullLanguageStatisticsTable()
end

function p.NumCategory(frame)
    return #FullCategoryStatisticsTable()
end

return p