Module:Lang: Difference between revisions

(One intermediate revision by one other user not shown)

Line 1,054:

local function is_latin (text, script, transl)

if mw.ustring.~~find~~ (text, 'θ', ~~1, true~~) then -- does <text> contain ~~'θ'~~?

text = mw.text.decode (text); -- convert html entities to characters

text = ~~text:~~gsub (~~'θ'~~, ''); -- ~~remove theta~~ from <text>

script = (script and script:lower()) or nil; -- ensure lowercase for comparisons

if 0 == text:len() then -- will be zero if theta was the only character in <text>

if transl then -- not nil for {{transliteration}}; assume ~~Latin theta~~ because this is 'transliteration' template

local whitelist = '[ʻʼʽʾʿΔαβγδθλσφχϑъьӾӿᾱῑ῾上入去平阳阴]'

if not transl then -- testing {{lang}} or {{langx}} <text>

if '' == mw.ustring.gsub (text, whitelist, '') then -- when <text> uses only whitelisted characters

return ('latn' == script); -- true when latn script; false when other script or no script

end

if mw.ustring.find (text, whitelist) then -- does <text> contain characters from the whitelist?

text = mw.ustring.gsub (text, whitelist, 'x'); -- replace whitelisted characters from <text> with known Latn-script char 'x'

-- if 0 == text:len() then -- will be zero if theta was the only character in <text>

if '' == text then -- will be empty string if the only character(s) in <text> are in <whitelist>

if transl then -- not nil for {{transliteration}}; assume Latn because this is 'transliteration' template

return true;

end

return (script and ('latn' == script~~:lower()~~)) or false; -- {{lang}}, {{langx}}, and {{lang-??}}; true when script is Latn; false else

return (script and ('latn' == script)) or false; -- {{lang}}, {{langx}}, and {{lang-??}}; true when script is Latn; false else

end

return unicode.is_Latin (text); -- return true when all characters in modified <text> are Latn script; false else

@@ Line 1,054: / Line 1,054: @@
 local function is_latin (text, script, transl)
-	if mw.ustring.find (text, 'θ', 1, true) then								-- does <text> contain 'θ'?
+	text = mw.text.decode (text);												-- convert html entities to characters
-		text = text:gsub ('θ', '');												-- remove theta from <text>
+	script = (script and script:lower()) or nil;								-- ensure lowercase for comparisons
-		if 0 == text:len() then													-- will be zero if theta was the only character in <text>
-			if transl then														-- not nil for {{transliteration}}; assume Latin theta because this is 'transliteration' template
+	local whitelist = '[ʻʼʽʾʿΔαβγδθλσφχϑъьӾӿᾱῑ῾上入去平阳阴]'
+	if not transl then															-- testing {{lang}} or {{langx}} <text>
+		if '' == mw.ustring.gsub (text, whitelist, '') then						-- when <text> uses only whitelisted characters
+			return ('latn' == script);											-- true when latn script; false when other script or no script
+		end
+	end
+	if mw.ustring.find (text, whitelist) then									-- does <text> contain characters from the whitelist?
+		text = mw.ustring.gsub (text, whitelist, 'x');							-- replace whitelisted characters from <text> with known Latn-script char 'x'
+--		if 0 == text:len() then													-- will be zero if theta was the only character in <text>
+		if '' == text then														-- will be empty string if the only character(s) in <text> are in <whitelist>
+			if transl then														-- not nil for {{transliteration}}; assume Latn because this is 'transliteration' template
 				return true;
 			end
-			return (script and ('latn' == script:lower())) or false;			-- {{lang}}, {{langx}}, and {{lang-??}}; true when script is Latn; false else
+			return (script and ('latn' == script)) or false;					-- {{lang}}, {{langx}}, and {{lang-??}}; true when script is Latn; false else
 		end
 		return unicode.is_Latin (text);											-- return true when all characters in modified <text> are Latn script; false else