Module:Babel/data/check

local export = {}

--[=[
Prints out a wikitable with warnings for the following:
- All-uppercased language codes (as GEZ and TI were, before the switch to module)
- Whitespace in language codes (matches stuff like "User ojp" where there should be "ojp")
- Invalid language codes (also checks whether defined in custom_codes)
- Gender switches missing delimiter (like {masculine+feminine} instead of {masculine+feminine+})
- Messages missing proper markup: bolding, category links
]=]--

local m_data = mw.loadData("Module:Babel/data")
local get_by_code = require("Module:languages").getByCode
local sc_get_by_code = require("Module:scripts").getByCode
local char = string.char
local byte = string.byte
local sub = string.sub

local warnings = {}

local n = 1
local function ins(...)
	warnings = ...
	n = n + 1
end

for code, glyph in pairs(m_data.script_glyphs) do
	if not sc_get_by_code(code) then
		ins({ = code .. ": invalid script code", is_root = true})
	end
	if glyph == "A" then
		ins({ = glyph .. ": redundant " .. code .. " glyph", is_root = true})	
	end
end

for code, _ in pairs(m_data.custom_codes) do
	if get_by_code(code, false, true) then
		ins({ = code .. ": redundant custom code", is_root = true})
	end
	local has_message = false
	local subpage = mw.loadData("Module:Babel/data/" .. sub(code, 1, 1))
	for _, i in ipairs{"0", "1", "2", "3", "4", "5", "N"} do
		if subpage then
			has_message = true
			break
		end
	end
	if not has_message then
		ins({ = code .. ": custom code does not have any corresponding messages and should be removed", is_root = true})
	end
end

for _, content in pairs(m_data.proglangs) do
	if not content.link then
		ins({ = code .. ": missing link", is_root = true})
	end
end

local function get_subpage_warnings(letter)
	for lang, message in pairs(mw.loadData("Module:Babel/data/" .. letter)) do
		local lang_n = lang
		local proficiency = lang:match("$")
		lang = lang:gsub("-$", "")
		local truncated = lang:gsub("-$", ""):gsub("-%u%l%l%l", "")
		if not get_by_code(truncated, false, true) and not m_data.custom_codes then
			if mw.language.isKnownLanguageTag(lang:lower()) then
				ins({lang, lang .. ": wikimedia language code not in WT:LOL"})
			else
				ins({lang, lang .. ": invalid language code"})
			end
		end
		if truncated:find("^+$") then
			ins({lang, lang .. ": uppercase language code"})
		end
		if truncated:find("%s") then
			ins({lang, lang .. ": space in language code"})
		end
		if message:find("{+}") and not message:find("%++%+*}") then
			ins({lang, lang_n .. ": gender switch missing second delimiter"})
		end
		if proficiency ~= "0" and not message:find("%]+%]%]") then
			ins({lang, lang_n .. ": missing $1 cat link"})
		elseif proficiency == "0" and message:find("%]+%]%]") then
			ins({lang, lang_n .. ": has extraneous $1 cat link"})
		end
		if not message:find("%]+%]%]") then
			ins({lang, lang_n .. ": missing $2 cat link"})
		end
		if not message:find("'''") or message:find("%w%|]+%]%]%w") then
			ins({lang, lang_n .. ": missing bolding"})
		end
		if message:find("''' ?'''") then
			ins({lang, lang_n .. ": has adjacent bold tags"})
		end
		if message:find("$3") then
			ins({lang, lang_n .. ": missing lang name"})
		end
		if message:lower():find(":cat")then
			ins({lang, lang_n .. ": has raw category link"})
		end
	end
end

local letter = "a"
local warnings_hash, warnings_res = {}, {}

for i = 1, 26 do
	get_subpage_warnings(letter)
	letter = char(byte(letter) + 1)
end

for n, i in ipairs(warnings) do
	-- remove duplicates from table
	if not warnings_hash] then
		warnings_res = i
		warnings_hash] = true
	end
end

function export.show()
	local purge = '<span style="color: #0022BB;border:' ..
	'1px solid #AAAAAA; background-color:' ..
	'#F9F9F9; padding-left: 10px;padding-right:10px;' ..
	'font-size:90%;text-weight: none; text-decoration: none;' ..
	'font-style:normal;">' ..
	'[[Special:ApiSandbox#action=purge&format=json&generator=prefixsearch' ..
	'&formatversion=2&gpssearch=Module%3ABabel%2Fdata|Purge all data pages]]' ..
	'</span>'
	if not warnings then return ";Number of warnings: 0\n" .. purge end
	wikitable = '{| class="wikitable"\n!subpage\n!warnings\n|-\n'
	for _, i in ipairs(warnings_res) do
		local subpage = "/data"
		if not i.is_root then subpage = subpage .. "/" .. sub(i, 1, 1) end
		wikitable = wikitable .. "|]\n|" .. i .. "\n|-\n"
	end
	return ";Number of warnings:" .. #warnings_res .. "\n" .. purge .. "\n" .. wikitable .. "|}"
end

return export