Module:Guru-translit

Definition from Wiktionary, the free dictionary
Jump to navigation Jump to search

This module will transliterate text in the Gurmukhi script. It is used to transliterate Punjabi.

The module should preferably not be called directly from templates or other modules. To use it from a template, use {{xlit}}. Within a module, use Module:languages#Language:transliterate.

For testcases, see Module:Guru-translit/testcases.

Functions[edit]

tr(text, lang, sc)
Transliterates a given piece of text written in the script specified by the code sc, and language specified by the code lang. When the transliteration fails, returns nil.

local export = {}

local conv = {
	--symbol
	["ੴ"] = "ek oaṅkār",


	--consonants without nukta
	["ਸ"] = "s",
	["ਹ"] = "h",
	["ਕ"] = "k", ["ਖ"] = "kh", ["ਗ"] = "g", ["ਘ"] = "gh", ["ਙ"] = "ṅ",
	["ਚ"] = "ch", ["ਛ"] = "cḥ", ["ਜ"] = "j", ["ਝ"] = "jh", ["ਞ"] = "ñ",
	["ਟ"] = "ṭ", ["ਠ"] = "ṭh", ["ਡ"] = "ḍ", ["ਢ"] = "ḍh", ["ਣ"] = "ṇ",
	["ਤ"] = "t", ["ਥ"] = "th", ["ਦ"] = "d", ["ਧ"] = "dh", ["ਨ"] = "n",
	["ਪ"] = "p", ["ਫ"] = "ph", ["ਬ"] = "b", ["ਭ"] = "bh", ["ਮ"] = "m",
	["ਯ"] = "y", ["ਰ"] = "r", ["ਲ"] = "l", 	["ਵ"] = "v", ["ੜ"] = "ṛ",
	
	
	--consonants with nukta
	["ਸ਼"] = "sh",
	["ਖ਼"] = "kḥ",
	["ਗ਼"] = "ġ",
	["ਜ਼"] = "z",
	["ਫ਼"] = "f",
	["ਲ਼"] = "ḷ",
    ["ਕ਼"] = "q",
	
	-- vowels
	["ਾ"] = "ā",
	["ਿ"] = "i", ["ੀ"] = "ī",
	["ੁ"] = "u", ["ੂ"] = "ū",
	["ੇ"] = "e", ["ੈ"] = "ae",
	["ੋ"] = "o", ["ੌ"] = "au",
	
	-- other diacritics
	["ੰ"] = "N", --ṭippi: nasalize
	["ਂ"] = "N", --bindi: nasalize
	["ੱ"] = "", --addak: germinate
	["੍"] = "", --halant, supresses the inherent vowel "a"
	["ਃ"] = "h", --voiceless "h" sound (tone raiser)

	-- independent vowels
	["ਅ"] = "a", ["ਆ"] = "ā",
	["ਇ"] = "e", ["ਈ"] = "ē",
	["ਉ"] = "u", ["ਊ"] = "ū",
	["ਏ"] = "e", ["ਐ"] = "ae",
	["ਓ"] = "o", ["ਔ"] = "au",
	
	-- digits
	["੦"] = "0", ["੧"] = "1", ["੨"] = "2", ["੩"] = "3", ["੪"] = "4",
	["੫"] = "5", ["੬"] = "6", ["੭"] = "7", ["੮"] = "8", ["੯"] = "9",
}

local nasal_assim = {
	["[kg]h?"] = "ṅ",
	["[cj]h?"] = "ñ",
	["[ṭḍ]h?"] = "ṇ",
	["[td]h?"] = "n",
	["[pb]h?"] = "m",
	["n"] = "n",
	["m"] = "m",
    ["s"] = "n",
}

-- translit any words or phrases
function export.tr(text, lang, sc) 
	local c = "([ਸਹਕਖਗਘਙਚਛਜਝਞਟਠਡਢਣਤਥਦਧਨਪਫਬਭਮਯਰਲਵੜː]਼?)"
	local uae = "([ੳਅੲਉਊਆਇਈਏਐਓਔ])"
	local y = "ਯ"
	local v = "([a ਾ ਿ ੀ ੁ ੂ ੇ ੈ ੋ ੌ])"
	local virama = "([੍]?)"
	local addak = "([ੱ])"
	local n = "([ੰਂ]?)"
	local tippi = "([ੰ])"
	local nukta = "([ਸਖਗਜਫਲਕ]਼)"
	
	can_drop = mw.ustring.gsub(c,y,"")
	no_virama = mw.ustring.gsub(v,virama,"")
	
	text = text .. " "
	
	--two consonants
	
	text = mw.ustring.gsub(text,
		c .. c,
		"%1a%2") --example ਕਰ
	
	--[[
	-- Has no effect.
	text = mw.ustring.gsub(text,
		c .. v .. c,
		"%1%2%3") --example ਸ਼ੇਰ
	--]]
	
	text = mw.ustring.gsub(text,
		c .. c .. v ,
		"%1a%2%3") --example ਦਰੀ
	
	text = mw.ustring.gsub(text,
		c .. v .. c .. v ,
		"%1%2%3%4") --example ਖੀਰਾ 

		-- addak
	text = mw.ustring.gsub(text,
		c .. addak .. c,
		"%1a%2%3") --example ਰੱਬ
	
	text = mw.ustring.gsub(text,
		c .. v .. addak .. c,
		"%1%2%3%4") --example ਸਿੱਖ 
	
	text = mw.ustring.gsub(text,
		c .. addak .. c .. v ,
		"%1a%2%3%4") --example ਗੱਡੀ 
	
	text = mw.ustring.gsub(text,
		c .. v .. addak .. c .. v ,
		"%1%2%3%4%5") --example ਤਿੱਖਾ 
	
		-- paer vich haha
	
	text = mw.ustring.gsub(text,
		c .. c .. virama .. c,
		"%1a%2%3%4") --example ਹੜ੍ਹ
	
	text = mw.ustring.gsub(text,
		c .. c .. virama .. c .. v,
		"%1a%2%3%4%5") --example ਘੜ੍ਹੀ
	
	text = mw.ustring.gsub(text,
		c .. v .. c .. virama .. c .. v,
		"%1%2%3%4%5%6") --example ਮਾੜ੍ਹੀ
	
		-- tippi
	text = mw.ustring.gsub(text,
		c .. tippi .. c,
		"%1a%2%3") --example ਬੰਦ
	
	text = mw.ustring.gsub(text,
		c .. v .. tippi .. c,
		"%1%2%3%4") --example ਜਿੰਦ 
	
	text = mw.ustring.gsub(text,
		c .. tippi .. c .. v,
		"%1a%2%3%4") --example ਬੰਦੀ
	
	text = mw.ustring.gsub(text,
		c .. v .. tippi .. c .. v,
		"%1%2%3%4%5") --example ਜ਼ਿੰਦਾ
	
	
	--three consonants
	
	text = mw.ustring.gsub(text,
		c .."a" .. c .. c,
		"%1a%2a%3") --example ਕਦਰ 
	
	text = mw.ustring.gsub(text,
		c .. v .. c .. c,
		"%1%2%3a%4") --example ਜਿਸਮ
	
	text = mw.ustring.gsub(text,
		c .. c .. v .. c,
		"%1a%2%3%4") --example ਤਸੀਰ
	
	text = mw.ustring.gsub(text,
		c .. c .. c .. v,
		"%1a%2%3%4") --example ਤਸਬੀ
	
	text = mw.ustring.gsub(text,
		c .. v .. c  .. "a" .. c .. v,
		"%1%2%3%4%5") --example ਖ਼ਾਲਸਾ 
	
	text = mw.ustring.gsub(text,
		c .. v .. c .. v .. c .. v,
		"%1%2%3%4%5%6") --example ਖ਼ੁਲਾਸਾ
	
		-- addak
		
	text = mw.ustring.gsub(text,
		c .. addak .. c .. "a" .. c .. v,
		"%1%2%3%4%5%6") --example ਨੱਠਣਾ
		
	text = mw.ustring.gsub(text,
		c .. v .. addak .. c .. "a" .. c .. v,
		"%1%2%3%4%5%6") --example ਸਿੱਖਣਾ
	
		-- tippi
		
	text = mw.ustring.gsub(text,
		c .. tippi .. c .. c,
		"%1a%2%3a%4") --example ਜੰਗਲ
	
	text = mw.ustring.gsub(text,
		c .. tippi .. c .. v,
		"%1a%2%3%4") --example ਰੰਗਲਾ
	
	text = mw.ustring.gsub(text,
		c .. v .. tippi .. c .. c .. v,
		"%1%2%3%4a%5%6") --example ਜ਼ਿੰਦਗੀ
	
		-- paer vich haha
		
	text = mw.ustring.gsub(text,
		c .. "a" .. c .. virama .. c .. "a" .. c .. v,
		"%1a%2%3%4%5%6") --example ਪੜ੍ਹਣਾ
	
	--four consonants
		
	text = mw.ustring.gsub(text,
		c .. c .. c .. c,
		"%1a%2%3a%4") --example ਸਰਬਤ
	
	text = mw.ustring.gsub(text,
		c .. v .. c .. c .. c,
		"%1%2%3%4a%5") --example ਕੁਦਰਤ
	
	text = mw.ustring.gsub(text,
		c .. "a" .. c .. v .. c .. c,
		"%1a%2%3%4a%5") --example ਕਬੂਤਰ
	
	text = mw.ustring.gsub(text,
		c .. c .. c .. v .. c,
		"%1a%2%3%4%5") --example 
	
	text = mw.ustring.gsub(text,
		c .. c .. c .. c .. v,
		"%1a%2%3%4%5") --example 
	
	text = mw.ustring.gsub(text, 
		c .. c .. c .. v .. c .. v,
		"%1a%2%3%4%5%6") --example ਦਰਵਾਜ਼ਾ
	
			-- tippi

	text = mw.ustring.gsub(text, 
		c .. "a" .. tippi .. c .. "a" .. c .. v .. c,
		"%1a%2%3%4%5%6") --example ਧੰਨਵਾਦ 
	
	text = mw.ustring.gsub(text, 
		c .. tippi .. c .. c .. v .. c,
		"%1a%2%3%4%5") --example ਫ਼ਰਜ਼ੰਦ
	
		--addak
		
	text = mw.ustring.gsub(text, 
		c .. "a" .. addak .. c .. "a" .. c .. c,
		"%1a%2%3%4a%5") --example ਯੱਕਦਮ
	
	-- five consonants

	text = mw.ustring.gsub(text, 
		c .. v .. c .. c .. "a" .. c .. v .. c,
		"%1%2%3a%4%5%6%7") --example ਵੇਪਰਵਾਹ
	
	text = mw.ustring.gsub(text, 
		c .. v .. c .. "a" .. c .. c .. "a" .. c .. v,
		"%1%2%3%4a%5%6%7") --example ਪਾਰਦਰਸ਼ੀ
	
	--addak
	
	text = mw.ustring.gsub(text, 
		c  .. v .. addak  ..  c ..  c .. c .. v .. c,
		"%1%2%3%4a%5%6%7%8") --example ਸ਼ੁੱਕਰਵਾਰ
	
	text = mw.ustring.gsub(text, 
		c .. "a" .. c .. c .. "a" .. c .. "a" .. addak .. c,
		"%1a%2a%3%4a%5%6") --example ਮਗਰਮੱਛ
	
	text = mw.ustring.gsub(text, 
		c .. v .. c .. v .. c .. c .. "a" .. c .. c .. v,
		"%1%2%3%4%5a%6a%7%8%9") --example ਯੂਨੀਵਰਸਟੀ
	
	-- uae
	
	text = mw.ustring.gsub(text, 
		uae .. c .. c ,
		"%1%2a%3") --example ਅਦਬ
	
	text = mw.ustring.gsub(text, 
		uae .. c .. "a".. c .. v ,
		"%1%2%3%4") --example ਆਖਣਾ
	
	text = mw.ustring.gsub(text, 
		uae .. tippi .. c .. "a".. c .. v ,
		"%1%2%3%4%5") --example ਅੰਤੜੀ
	
	text = mw.ustring.gsub(text, 
		uae .. c .. "a".. c .. c ,
		"%1%2%3a%4") --example ਅਫ਼ਸਰ

	text = mw.ustring.gsub(text, 
		uae .. c .. "a" .. c .. v .. c .. v ,
		"%1%2%3%4%5%6") --example ਅਲਮਾਰੀ
	
	text = mw.ustring.gsub(text, 
		uae .. c .. "a" .. c .. v .. c ,
		"%1%2%3%4%5") --example ਇਕਰਾਰ
	
	text = mw.ustring.gsub(text, 
		uae .. "ਸ" .. "ਤ" .. "a" .. c .. v ,
		"%1st%2%3") --example ਉਸਤਰਾ
	
	text = mw.ustring.gsub(text, 
		c .. v .. uae .. c .. c ,
		"%1%2%3%4a%5") --example ਸਾਈਕਲ
	
	text = mw.ustring.gsub(text, 
		c .. v .. c .. v .. uae .. c .. "a" .. c .. v,
		"%1%2%3%4%5%6%7%8") --example ਵਿਦਿਆਰਥੀ
	
	text = mw.ustring.gsub(text,can_drop .. "ਈ", "%1aī") -- example ਕਈ
	text = mw.ustring.gsub(text,v .. "ਸ" .. "a" .. "ਤ" , "%1st") --example ਦੋਸਤ
	text = mw.ustring.gsub(text,can_drop .. "ਸ" .. "a" .. "ਤ" , "%1st") --example ਹਸਤੀ
	text = mw.ustring.gsub(text,v .. "ā" .. "ਸ" .. "ਤ", "%1āsat" ) -- example ਰਿਆਸਤ, ਸਿਆਸਤ
	text = mw.ustring.gsub(text,"ਸ" .. "a" .. "ਟ", "sṭ")
	text = mw.ustring.gsub(text,"ਸ" .. "a" .. "ਥ", "sth")
	text = mw.ustring.gsub(text,"ਸ਼" .. "a" .. "ਤ", "sht")
	text = mw.ustring.gsub(text,"ਸ਼" .. "a" .. "ਥ", "shth")
	text = mw.ustring.gsub(text,"ਸ਼" .. "a" .. "ਟ", "shṭ")
	
	text = mw.ustring.gsub(text,nukta,conv)
	text = mw.ustring.gsub(text,".",conv)
	
	for key,val in pairs(nasal_assim) do
		text = mw.ustring.gsub(text,"N("..key..")",val.."%1")
	end
	text = mw.ustring.gsub(text,"([aiueēoāīū])N ", "%1̃ ")
	text = mw.ustring.gsub(text,"(.?)N", "%1̃")
	
	text = mw.ustring.gsub(text,"ː(.)","%1%1")
	
	text = mw.ustring.gsub(text," ?।",".")
	
	text = mw.ustring.gsub(text," $","")
	
	return mw.ustring.toNFC(text)
end
 
return export