Module:family tree/etymology languages: Difference between revisions

Retrieved from "https://linguifex.com/wiki/Module:family_tree/etymology_languages"

@@ Line 1: / Line 1: @@
-local language_codes = require "Module:languages/code to canonical name"
+local language_codes = require("Module:languages/code to canonical name")
+-- Prefer nrf-grn and nrf-jer over roa-grn and roa-jer (Guernsey and Jersey).
+-- Adds 2 to the weighting.
+local function isLangCode(code)
+	if language_codes[code:match("^%l+")] then return 1 else return 0 end
+end
+-- Order of preference:
+-- xx, xxx, xx-xxx, xxx-xxx, xx-xxx-xxx, xxx-xxx-xxx, xx-XX, xxx-XX, xx-XX-xxx, xxx-XX-xxx
+-- Language codes are preferred over family codes of the same format.
 local function determine_preferred_etymology_language_code(code1, code2)
-	if code2:find "^[%a-]+$" then
+	local function weighting(code)
-		if code1:find "^[%a-]+$" then
+		if code:find("^%l%l$") then
-			if not code2:find "%u%l" then
+			return 14
-				if not code1:find "%u%l" then
+		elseif code:find("^%l%l%l$") then
-					if #code2 < #code1 then
+			return 13
-						return code2
+		elseif code:find("^%l%l%-%l%l%l$") then
-					else
+			return 12
-						-- Prefer nrf-grn and nrf-jer over roa-grn and roa-jer
+		elseif code:find("^%l%l%l%-%l%l%l$") then
-						-- (Guernsey and Jersey).
+			return 10 + isLangCode(code)
-						local first_word1, first_word2 =
+		elseif code:find("^%l%l%-%l%l%l%-%l%l%l$") then
-							code1:match "^[a-z]+", code2:match "^[a-z]+"
+			return 9
-						if first_word1 and first_word2
+		elseif code:find("^%l%l%l%-%l%l%l%-%l%l%l$") then
-						and language_codes[first_word1] then
+			return 7 + isLangCode(code)
-							return code1
+		elseif code:find("^%l%l%-%u%u$") then
-						else
+			return 6
-							return code2
+		elseif code:find("^%l%l%l%-%u%u$") then
-						end
+			return 4 + isLangCode(code)
-					end
+		elseif code:find("^%l%l%-%u%u%-%l%l%l$") then
-				else
+			return 3
-					return code2
+		elseif code:find("^%l%l%l%-%u%u%-%l%l%l$") then
-				end
+			return 1 + isLangCode(code)
-			else
-				return code1
-			end
 		else
-			return code2
+			return 0
 		end
+	end
+	local weighting1, weighting2 = weighting(code1), weighting(code2)
+	if weighting1 > weighting2 then
+		return code1
+	elseif weighting1 < weighting2 then
+		return code2
+	elseif #code1 < #code2 then
+		return code1
+	elseif #code1 > #code2 then
+		return code2
+	-- If all else fails, use alphabetical order.
+	elseif code1 > code2 then
+		return code2
 	else
 		return code1
@@ Line 41: / Line 61: @@
 end
-local function invert(t)
+return require("Module:table").invert(fold(
-	local inverted = {}
+	require("Module:etymology languages/data"),
-	for k, v in pairs(t) do
-		inverted[v] = k
-	end
-	return inverted
-end
-return invert(fold(
-	require "Module:etymology languages/data",
 	{},
 	function (code, data, data_to_code)