Module:ur-translit

From Wiktionary, the free dictionary
Jump to navigation Jump to search

This module will transliterate Urdu language text per WT:UR TR. It is also used to transliterate Brahui and Gojri. 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:ur-translit/testcases.

Functions

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.

Background

THIS WILL REQUIRE DIACRITICS (USED CORRECTLY), Diacritics can be found at http://udb.gov.pk (which is NOT always correct). This should work correctly for majority of the work, although is still in progress.

Read #Usage notes for tips on how to use the module correctly:

Usage notes

  1. All consonants must be paired to a vowel (a, i, u, ā, e, ī, o, ū) or a sukoon/jazm. Or else the module will return NIL (blank).
  2. Alif MUST be paired to either a consonant or a vowel, or the module will return NIL.
    if paired to a vowel, the alif will represent the vowel. If paired to a consonant, alif will represent "ā". If an initial alif is not paired to a vowel, the module will return nill.
    compare the pairings نا (), اَن (an), اُن (un), and اِن (in).
  3. Whenever there are two adjacent semivowels, the first semivowel will become the consonant and the second will be the vowel. To reverse that order, simply put a sukoon on the second semivowel.
    baṛī ye (e, ai) and choṭī ye (ī, y) are distinguished medically using diacritics. However in the final position choṭī ye is always -ī (since choṭī ye cannot be "y" in the final position), and a final baṛī ye defaults to -ē but can become -ai with a preceding zabar.
    If, for some reason, a final choṭī ye needs to be a "y", put a sukoon on it.
  4. Noon ghunna ـن٘ـ/ـں generally represents a nasal vowel. The only exceptions being when a nasal vowel isn't possible (in Urdu), such as before b, bh, j, jh, g, gh, ḍh, dh, th, q (and before kh, ṭh, ṭ and ph, unless the nasal vowel is ā, ī, or ū). Since it is impossible for a nasal vowel to appear before these sounds, the module will return an assimilated nasal.
    1. If a nasal vowel is not desired, use a sukoon to return a regular noon.
    2. If ghunna is occurring after ā, ī, or ū but needs to assimilate with ph or kh, the assimilation must be manually entered. Since unlike all other vowels, ā, ī, and ū can actually be nasalized before these letters. For all consonants not mentioned, nasal assimilation can be predicted with only a jazm/sukoon (if it even occurs).
      • Noon ghunna assimilating with "k" is not as common in Urdu as it is in Hindi, it mainly appears in english loanwords (and only at the end of a word). In which case the assimilation can be predicted with a sukoon/jazm. This assimilation should disappear if the word is inflected. Compare بَینْک (baiṅk) to بَین٘کوں (ba͠ikõ)
  5. The aspirate 'he' (i.e. do-chashme-he, ھ (h)) does NOT need ANY DIACRITICS in ANY CIRCUMSTANCES, all diacritics should either go on the previous letter or on the following letter.
    The module will work regardless but based on common practice putting a vowel on the aspirate ھ would be inappropriate.
  6. The Tashdeed/Shadda works with another diacritic, as well as alone
  7. The Sukoon/Jazm diacritic is required for consonant clusters, or else the module will return NIL

Updates

To do

  1. require an initial alif to be paired to a vowel (either a diacritic or vao/ye). needed to prevent false positives
  2. support for al- assimilation in Arabic loanwords.
    • how will we distinguish an initial al- in non arabic words? This is probably impossible and might not happen.
  3. DIFFERENCE BETWEEN ṇ, ṅ, ñ AND n .
  4. Detect when transliteration is needed and when not (i.e. if diacritics are present/needed or not)
  5. izafa/ezafe support
  6. Revert the module to transliterate initial "ای" as 'ē'
  7. اَللہ = sort out ہ + khari zabar diacritic
  8. transliteration detection often gives false positives

Example

Test Urdu:

تَرْکِ تَعَلُقّات پِہ رویا نَہ تُو نَہ مَیں لیکِن یِہ کْیا کَہ چَین سے سویا نَہ تُو نَہ مَیں

وُہ ہَمْسَفَر تھا مَگَر اُس سے ہَمْنَوَائی نَہ تھی کَہ دُھوپ چھاؤں کا عالَم رَہا جُدائی نَہ تھی

عَداوَتیں تِھیں، تَغافُل تھا رَنْجِشیں تِھیں مَگَر بِچَھڑْنے والے میں سَب کُچھ تھا بے وَفائی نَہ تھی

کاجَل ڈالو کُرْکُرا سُرْمَہ سَہا نَہ جائے جِن نَین میں پِی بَسے دُوجا کَون سَمائے؟

بِچَھڑْتے وَقْت اُن آن٘کھوں میں تھی ہَماری غَزَل غَزَل تھی وُہ جو کِسی کو کَبھی سُنائی نَہ تھی

Result:

tark-i ta'aluqqāt pe royā na tū na ma͠i lekin ye kyā ka cain se soyā na tū na ma͠i

vo hamsafar thā magar us se hamnavāī na thī ka dhūp chāõ kā 'ālam rahā judāī na thī

'adāvatẽ thī̃, taġāful thā ranjiśẽ thī̃ magar bichaṛne vāle mẽ sab kuch thā be vafāī na thī

kājal ḍālo kurkurā surma sahā na jāe jin nain mẽ pī base dūjā kaun samāe?

bichaṛte vaqt un ā̃khõ mẽ thī hamārī ġazal ġazal thī vo jo kisī ko kabhī sunāī na thī

Expected:

tark-i ta'aluqqāt pe royā na tū na ma͠i lekin ye kyā ka cain se soyā na tū na ma͠i

vo hamsafar thā magar us se hamnavāī na thī ka dhūp chāõ kā 'ālam rahā judāī na thī

'adāvatẽ thī̃, taġāful thā rañjiśẽ thī̃ magar bichaṛne vāle mẽ sab kuch thā be vafāī na thī

kājal ḍālo kurkurā surma sahā na jāe jin nain mẽ pī base dūjā kaun samāe?

bichaṛte vaqt un ā̃khõ mẽ thī hamārī ġazal ġazal thī vo jo kisī ko kabhī sunāī na thī

Test cases

3 of 69 tests failed. (refresh)

TextExpectedActualDiffers at
test_translit_urdu:
Passedاِیرانِیīrānīīrānī
Passedماشاءاَللّٰہmāśā'allāhmāśā'allāh
Passedپَیدائِشpaidāiśpaidāiś
Passedبَرْقِیاتbarqiyātbarqiyāt
Passedعَقْل'aql'aql
Passedعِزَّت'izzat'izzat
Passedعَین'ain'ain
Passedعالَم'ālam'ālam
Passedعَورَت'aurat'aurat
Passedشُرُوعśurū'śurū'
Passedاِشْعاعiś'ā'iś'ā'
Passedتَعَلُّقاتta'alluqātta'alluqāt
Passedتَعَلُّقta'alluqta'alluq
Passedمُتَعَلِّقmuta'alliqmuta'alliq
Passedمتعلق(nil)(nil)N/A
Passedعُمْر'umr'umr
Passedدَفْعَہdaf'adaf'a
Passedبَچَّہbaccabacca
Passedقُوَّتquvvatquvvat
Passedمَۓ عِشْقma-ye 'iśqma-ye 'iśq
Passedشیرِ پَن٘جابśer-i pañjābśer-i pañjāb
Passedمَلْکَۂ دُنْیاmalka-yi dunyāmalka-yi dunyā
Passedجَمُّوںjammū̃jammū̃
Passedآمāmām
Passedاِشْتِراکِیَّتiśtirākiyyatiśtirākiyyat
Passedسِسَکْناsisaknāsisaknā
Passedپُلpulpul
Passedعِیسیٰ'īsā'īsā
Passedاَعْلیٰa'lāa'lā
Passedلَفْظlafzlafz
Passedحاضِرhāzirhāzir
Passedبَہوراbahorābahorā
Passedنَہِیںnahī̃nahī̃
Passedاِشْتِمالِیَتiśtimāliyatiśtimāliyat
Passedچَوڑاcauṛācauṛā
Passedتِھیںthī̃thī̃
Passedکُتّاkuttākuttā
Passedپَہْلےpahlepahle
Passedکِھلائیkhilāīkhilāī
Passedکھلائی(nil)(nil)N/A
Passedٹَھہَرْناṭhaharnāṭhaharnā
Passedتَیمُورtaimūrtaimūr
Passedفَوراًfauranfauran
Passedکوئےkoekoe
Passedمَنَّتوںmannatõmannatõ
Passedگان٘وںgā̃õgā̃õ
Passedمَیںma͠ima͠i
Passedآئیāīāī
Passedمَکَّھنmakkhanmakkhan
Passedخُداxudāxudā
Passedکَئیkaīkaī
Passedکُئیkuīkuī
Passedچائےcāecāe
Passedکُھلْواؤkhulvāokhulvāo
Passedغَدّارġaddārġaddār
Passedبَیٹھوbaiṭhobaiṭho
Passedبَطَّخbattaxbattax
Passedمُتَّحِدَۂmuttahida-yimuttahida-yi
Passedساؤُتھ اَفْرِیقَہsāuth afrīqasāuth afrīqa
Passedکُلِّیَّہkulliyyakulliyya
Passedدائِرَۃُdāiratudāiratu
Passedسُورَۃsūrasūra
Passedبِلّاbillābillā
Failedدائِرَۃُ الْمَعارِفdāiratu l-ma'ārifdāiratu اlma'ārif9
Failedدائِرَۃْ اُلْمَعارِفdāirah ulma'ārifdāirat ulma'ārif6
Passedآیَتُ اْللّٰہāyatu llāhāyatu llāh
Passedصَیّادsaiyādsaiyād
Passedگُرْدَہgurdagurda
Failedکہاں(nil)khā̃N/A

--[=[

FIXME:

1. support for Arabic al- (copy from fa-cls-translit)

]=]
local U = require("Module:string/char")
local gsub = mw.ustring.gsub
local export = {}

local fatHataan = U(0x64B)
local zabar = U(0x64E)
local zer = U(0x650)
local pesh = U(0x64F)
local zwnj = U(0x200C) -- Is this even used in Urdu? Why was it included in the previous version?
local highhmz = U(0x654)
local tashdid = U(0x651) -- also called tashdid
local jazm = "ْ"
local he = "ہ"
local ghunna = U(0x658)
local dagger_alif = U(0x670)

local consonants = "ببپتثجچحخدذرزژسشصضطظعغفقکگلࣇمنݨؤڷہئھٹڈڑ"
local consonantS = "ببپتثجچحخدذرزژسشصضطظعغفقکگڷلࣇمنݨہھٹڈڑ"
local consonantS2 = "یببپتثجچحخدذرزژسشصضطظعغفقکگلࣇڷمنݨوؤہھئٹڈڑ" 
local semivowel = "یو"
local vowels = "āایئےۓوؤ"
local indvowels = "آایےوؤ"
local hes = "ہح"
local diacritics = "َُِّْٰ"
local ZZP = "َُِ"
local lrm = U(0x200e) -- left-to-right mark
local rlm = U(0x200f) -- right-to-left mark

local consonants_needing_vowels = "ببپتثجچحخدذرزژسشصضطظعغفقکڷگلࣇمنںݨہئٹڈڑءﷲ"
-- consonants on the right side; includes alif madda
local rconsonants = consonants_needing_vowels .. "ویآ"
-- consonants on the left side; does not include alif madda
local lconsonants = consonants_needing_vowels
local space_like = "%s'" .. '"'
local space_like_class = "[" .. space_like .. "]"

-- not all letters here are used by urdu
local mapping = {
	["آ"] = 'ā', ["ب"] = 'b', ["پ"] = 'p', ["ت"] = 't', ["ٹ"] = 'ṭ', ["ث"] = 's',
	["ج"] = 'j', ["چ"] = 'c', ["ح"] = 'h', ["خ"] = 'x', 
	["د"] = 'd', ["ڈ"] = 'ḍ', ["ذ"] = 'z', ["ر"] = 'r', ['ڑ'] = "ṛ", ["ز"] = 'z', ["ژ"] = 'ź',
	["س"] = 's', ["ش"] = 'ś', ["ص"] = 's', ["ض"] = 'z', 
	["ط"] = 't', ["ظ"] = 'z', ["غ"] = 'ġ', ["ف"] = 'f', ["ق"] = 'q',
	["ک"] = 'k', ["گ"] = 'g', ["ݨ"] = 'ṇ', ["ࣇ"] = 'ḷ', ["ڷ"] = 'ł',
	["ل"] = 'l', ["م"] = 'm', ["ن"] = 'n', ["و"] = 'o', ["ہ"] = 'h', ["ی"] = 'e', ["ے"] = 'e', ["۔"] = ".", ["ں"] = '̃',

	["ھ"] = "h",
	

	["ع"] = '\'',
	["ء"] = '\'',
	["أ"] = '',
	
	-- diacritics
	[zabar] = "a",
	[zer] = "i",
	[pesh] = "u",
	[jazm] = "", -- also sukun - no vowel
	[zwnj] = "-", -- ZWNJ (zero-width non-joiner)
	
	-- ligatures
	["ﻻ"] = "lā",
	["ﷲ"] = "allāh",
	
	-- kashida
	["ـ"] = "-", -- kashida, no sound
	
	-- numerals
	["۱"] = "1", ["۲"] = "2", ["۳"] = "3", ["۴"] = "4", ["۵"] = "5",
	["۶"] = "6", ["۷"] = "7", ["۸"] = "8", ["۹"] = "9", ["۰"] = "0",
	
	-- punctuation (leave on separate lines)
	["؟"] = "?", -- question mark
	["۔"] = ".", -- period
	["،"] = ",", -- comma
	["؛"] = ";", -- semicolon
	["«"] = '“', -- quotation mark
	["»"] = '”', -- quotation mark
	["٪"] = "%", -- percent
	["؉"] = "‰", -- per mille
	["٫"] = ".", -- decimals
	["٬"] = ",", -- thousand
	["ۓ"] = "-ye", 
	[highhmz] = "-yi",
}

local punctuation = "%-:%(%)%[%]*&٫؛؟،ـ«\".\'!»٪؉۔"
local numbers = "۱۲۳۴۵۶۷۸۹۰"

local ain = 'ع'
local alif = 'ا'
local ye = 'ی'
local ye2 = 'ئ'
local ye3 = "ے"
local vao = "و"
local aspirate = 'ھ'
local highhmz = U(0x654)
local aiu = "āīūآ"
local n_exceptions = "[^" .. aiu .. "]" -- for nasalization exceptions

local before_diacritic_checking_subs = {
	------------ transformations prior to checking for diacritics --------------
	{U(0x06E5), "و"},
	{U(0x06E6), "ی"},
	-- ignore dagger alif placed over regular alif or alif maqṣūra
	{"([" .. alif .. ye .. "])" .. dagger_alif, alif},
	{"([^" .. alif .. ye .. "])" .. fatHataan, alif .. fatHataan},
}

local has_diacritics_subs = {
	-- remove arabic ye (ruins conversions)
	{"لل" ..  he , ""},
	{"لل" .. tashdid ..  he , ""},
	{"لل" .. tashdid .. dagger_alif ..  he , ""},
	{"ۃ" , ""},
	-- aspirated consonants should cound as 1 consonant not two
	{"([" .. consonants .. "][".. ZZP .. diacritics .. "?])" ..  aspirate , "%1"},
	{"([" .. consonants .. "])" ..  aspirate , "%1"},
	{ aspirate , ""},
	-- remove punctuation and tashdid
	{"[" .. punctuation .. tashdid .. highhmz .. zwnj .. numbers .. "]", ""},
	-- noon gunna and silent consonants can be removed
	{ ".. [".. ZZP .. indvowels .. diacritics .. "?] .. ([" .. consonantS2 .. "])" .. "([".. ghunna .. jazm .."])" .. "([" .. consonantS2 .. "])"  , ""},
	{ "([" .. consonants .. "])" .. ghunna , ""},
	{ "([" .. consonantS2 .. "])" .. jazm , ""},
	{ "([" .. consonantS2 .. "])" .. "یٰ" , ""},
	-- must go before removing final consonants
	{"[".. ZZP .. diacritics .. "]" .. alif , alif },
	{fatHataan , "" },
	{ "([" .. consonantS2 .. "])" .. "[" .. ZZP .. diacritics .. indvowels .. "?]" .. "([ںۓۂۂ])", "" },
	{ "([ںۓۂۂ])", "" },
	{ "([" .. ye .. alif .. "])" .. dagger_alif, alif},
	{ dagger_alif .. ye , alif},
	{ alif .. "[".. ZZP .. diacritics .. "]" , ""},
	{ "[".. ZZP .. diacritics .. "]" .. alif , alif},
	{ dagger_alif .. "([" .. ye .. alif .. "])", alif},
	-- Remove consonants at end of word or utterance, so that we're OK with
	-- words lacking iʿrāb (must go before removing other consonants).
	-- If you want to catch places without iʿrāb, comment out the next two lines.
	{"[" .. lconsonants .. "]$", ""},
	-- closed consonants
	{"([" .. consonantS2 .. "])[" .. indvowels .. ZZP .. "]", ""},
	-- remove consonants (or alif) when followed by diacritics
	-- must go after removing tashdid
	-- do not remove the diacritics yet because we need them to handle
	-- long-vowel sequences of diacritic + pseudo-consonant
	{"[" .. lconsonants .. alif .. "]([" .. fatHataan .. zabar .. pesh .. zer .. jazm .. dagger_alif .. "])", "%1"},
	-- the following two must go after removing consonants w/diacritics because
	{"([" .. rconsonants .. "])([".. ZZP .. diacritics .. "?][" .. indvowels .. "?])([" .. consonantS2 .. "])", ""},
	{"[" .. indvowels .. "]([" .. rconsonants .. "])", ""},
	{"[".. ZZP .. diacritics .. "]([" .. lconsonants .. "])", ""},
	{"([" .. consonants .. "])[" .. indvowels .. ZZP .. diacritics .. "]", ""},
	{"([" .. rconsonants .. "])(" .. space_like_class .. ")", ""},
	{"[" .. lconsonants .. "]" .. zabar .. "[".. ye .. ye3 .. vao .. "]", ""},
	-- we only want to treat vocalic wāw/yā' in them (we want to have removed
	-- remove vaw
	{ "[" .. lconsonants .. "]" .. vao, ""},
	{"ؤ" .. pesh , ""},
	{"ؤ", ""},
	-- remove ye
	{ "[" .. lconsonants .. "]" .. ye, ""},
	{ye3, ""},
	{"([" .. consonants .. "][" .. ZZP .. "])" .. he,""},
	-- remove fatḥa/fatḥatan + alif/alif-maqṣūra
	{"[" .. fatHataan .. zabar .. "][" .. alif .. ye .. "]", ""},
	-- remove diacritics and independant vowels
	{"[" .. fatHataan .. zabar .. pesh .. zer .. jazm .. dagger_alif .. "]", ""},
	{ "[" .. indvowels .. "]" , ""},
	{ "[".. semivowel .."]" .. "[" .. indvowels .. "]" , ""},
	-- remove numbers, hamzatu l-waṣl, alif madda
	{"[" .. numbers .. "ٱ" .. "آ" .. "]", ""},
	{"%s", ""},
}

-- declared as local above
local function has_diacritics(text)
	local count
	text, count = gsub(text, "[" .. lrm .. rlm .. "]", "")
	if count > 0 then
		require("Module:debug").track("ur-translit/lrm or rlm")
	end
	for _, sub in ipairs(has_diacritics_subs) do
		text = gsub(text, unpack(sub))
	end
	return #text == 0
end

function export.tr(text, lang, sc)
	if type(text) == "table" then
		local function f(x) return (x ~= "") and x or nil end
		text, lang, sc, omit_i3raab, force_translit =
			f(text.args[1]), f(text.args[2]), f(text.args[3]), f(text.args[4]), f(text.args[5])
	end
	for _, sub in ipairs(before_diacritic_checking_subs) do
		text = gsub(text, sub[1], sub[2])
	end

	if not force_translit and not has_diacritics(text) then
		require("Module:debug").track("ur-translit/lacking diacritics")
		return nil
	end
	
	--define the "end" of a word
	text = gsub(text, "#", "HASHTAG")
	text = gsub(text, " | ", "# | #")
	text = gsub(text, "\n" , "#".."\n" .. "#")
	text = gsub(text, "(["..punctuation.."])" , "#".."%1" .. "#")
	text = "##" .. gsub(text, " ", "# #") .. "##"
	text = gsub(text, zwnj, "#"..zwnj.."#")
	-- hastags now mark the beginning and end of a word
	
	--exceptions
	text = gsub(text, "#" .. vao .. he .. "#", "#vo#")
	text = gsub(text, "#" .. vao .. pesh .. he .. "#", "#vo#")
	text = gsub(text, "#" .. "پ" .. he .. "#", "#pe#")
	text = gsub(text, "#" .. "پ" .. zer .. he .. "#", "#pe#")
	text = gsub(text, "#" .. ye .. he .. "#", "#ye#")
	text = gsub(text, "#" .. ye .. zer .. he .. "#", "#ye#")
	
	--character reformatting
	--to make an exceptions for a word, put hashtags on both sides
	text = gsub(text, "ۂ", he .. highhmz)
	text = gsub(text, highhmz, "#"..highhmz.."#")
	--text = gsub(text, 'ىٰ', "ā") -- the first letter is U+0649 (Arabic alif maqṣūra), it doesn't belong here
	text = gsub(text, 'یٰ', "ā") -- the first letter is U+06CC
	text = gsub(text, 'ٰ', "ā")
	text = gsub(text, 'ا' .. fatHataan, "an")
	text = gsub(text, 'لا', "ﻻ")
	text = gsub(text, "ة" 	, "ۃ")
	text = gsub(text, "ۃ" .. "([" .. ZZP .. jazm .. "])", "ت%1")
	text = gsub(text, "ۃ" , he)
	
	-- Tashdeed
	text = gsub(text, '([' .. consonantS2 .. '])' .. tashdid, "%1%1")
	text = gsub(text, '([' .. consonantS2 .. '])' .. tashdid .. '([' .. ZZP .. '])', "%1%1%2")
	-- For some reason the tashdeed gets pushed after the other diacritics, so this line is necessary for tashdeed to work with other diacritics
	text = gsub(text, '([' .. consonants .. '])' .. '([' .. ZZP .. '])' .. tashdid, "%1%1%2")
	text = gsub(text, '([' .. ZZP .. '])' .. aspirate, aspirate.."%1") 
	text = gsub(text, dagger_alif .. aspirate, aspirate.."%1")
	text = gsub(text, ye .. '([' .. ZZP .. '])' .. tashdid, "yy%1")
	text = gsub(text,  vao .. '([' .. ZZP .. '])' .. tashdid, "vv%1")
	text = gsub(text, ye .. tashdid .. '([' .. ZZP .. '])', "yy%1")
	text = gsub(text, vao .. tashdid .. '([' .. ZZP .. '])', "vv%1")
	

    --initial alif
    text = gsub(text, pesh .. vao .. alif, "uā")
    text = gsub(text, "(["..consonantS2.."])" .. alif, "%1ā") 
    --alifs paired to a consonant are a vowel
    text = gsub(text, jazm .. alif, "-") -- invisible ZWNJ
    text = gsub(text, jazm .. "آ", "-ā") -- invisible ZWNJ
    text = gsub(text, "(["..consonantS2.."])" .. "آ", "%1'ā") 
    	text = gsub(text, pesh .. vao .. zabar .. alif , "ūā" )
    text = gsub(text, zabar .. alif, "ā")
    text = gsub(text, "(" .. ghunna .. ")" .. alif, "%1ā")
    text = gsub(text, "(["..diacritics.."])" .. alif, "%1")
    text = gsub(text, "(["..ZZP.."])" .. alif, "%1")
    --alifs not paired to a consonant are a glottal stop (not shown currently)
    text = gsub(text, alif.."(["..diacritics.."])".. "(["..consonantS2.."])", "%1%2")
    text = gsub(text, alif..ye.."#", "ī")
    text = gsub(text, alif..ye, "e")
    text = gsub(text, alif..ye3, "e")
    text = gsub(text, alif..zabar..ye3, "ai")
    text = gsub(text, alif..vao, "o")
    text = gsub(text, alif..zer..ye, "ī")
    text = gsub(text, alif..pesh..vao, "ū")
    text = gsub(text, alif.."(["..diacritics.."])", "%1")
    
    
    -- convert semi vowels
    text = gsub(text, vao.. "(["..diacritics..ZZP.."])", "v%1")
    text = gsub(text, ye.. "(["..diacritics..ZZP.."])", "y%1")
    text = gsub(text, ye .. "ā", "yā")
    text = gsub(text, vao.. "ā", "vā")
    text = gsub(text, ye .. "(["..zabar.."]?)" .. ye3, "y%1"..ye3.."")
    text = gsub(text, vao .. "(["..zabar.."]?)" .. ye3, "v%1"..ye3.."")
    text = gsub(text, ye .. "(["..semivowel.."])(["..semivowel.."])", "e%1%2")
    text = gsub(text, vao .. "(["..semivowel.."])(["..semivowel.."])", "o%1%2")
    text = gsub(text, ye .. "(["..semivowel.."])", "y%1")
    text = gsub(text, vao .. "(["..semivowel.."])", "v%1")
    
    -- conversions for vaav/vaw/vao
    text = gsub(text, pesh.. vao, "ū")
    text = gsub(text, zabar .. vao, "au")
    text = gsub(text, vao.. "(["..diacritics..ZZP.."])", "v%1")
    text = gsub(text, "(["..diacritics..ZZP.."])" .. vao, "%1v")
    -- conversions for ye
    text = gsub(text, zer.. ye, "ī")
    text = gsub(text, ye .. "#", "ī#")
    text = gsub(text, zabar.. ye, "ai")
    text = gsub(text, zabar.. ye3, "ai")
    text = gsub(text, ye .. "(["..diacritics..ZZP.."])", "y%1")
    text = gsub(text, "(["..diacritics..ZZP.."])" .. ye , "%1y")
    
    -- final he and izafa/ezafe
    text = gsub(text, "e" .. zer .. "#", "e-yi#")
    text = gsub(text, "ī" .. zer .. "#", "ī-yi#")
    text = gsub(text, "y" .. zer .. "#", "-yi#")
    text = gsub(text, zer .. "#", "-i#")
    text = gsub(text, "(["..ZZP.."])" .. he .. "#" .. zwnj, "%1-")
    text = gsub(text, "(["..ZZP.."])" .. he .. "#", "%1#")
    text = gsub(text, zabar .. he .. "#", "a#")
    
    -- noon ghunna assimilation/nasalization
    --remove impossible nasal vowels
    text = gsub(text, "ن" .. ghunna .. "([ب])", "m%1") -- nasal vowels are impossible before b
    text = gsub(text, "ن" .. ghunna .. "ت" .. aspirate, "nth") 
    text = gsub(text, "ن" .. ghunna .. "([قگ])",	"ṅ%1") -- impossible before q and g
    text = gsub(text, "(" .. n_exceptions .. ")" .. "ن" .. ghunna .. "ٹ"	.. aspirate	, "%1ṇṭh")
    text = gsub(text, "(" .. n_exceptions .. ")" .. "ن" .. ghunna .. "پ" .. aspirate, "%1mph")
    text = gsub(text, "(" .. n_exceptions .. ")" .. "ن" .. ghunna .. "ک" .. aspirate, "%1ṅkh")
    text = gsub(text, "ن" .. ghunna .. "([ج])", "ñ%1") -- impossible before j
    text = gsub(text, "ن".. ghunna .. "ڈ" .. aspirate, "ṇḍh") -- aspirated d/D cant be nasalized
    text = gsub(text, "ن".. ghunna .. "د" .. aspirate, "ndh") -- aspirated d/D cant be nasalized
    --other nasals
    text = gsub(text, "ن" .. jazm .. "([کگق])" .. "#",	"ṅ%1#")
    text = gsub(text, "ن" .. ghunna .. "([کگق])" .. jazm .. "#",	"ṅ%1#")
	text = gsub(text, "ن" .. jazm .. "([دتر])", "n%1") -- dental
	text = gsub(text, "ن" .. ghunna .. "([ٹڈ])" .. jazm .. "#", "ṇ%1#")
	text = gsub(text, "ن" .. ghunna .. "([چج])" .. jazm .. "#", "ñ%1#") -- postalveolar
	text = gsub(text, "ن" .. ghunna .. "([چج]".. aspirate ..")" .. jazm .. "#", "ñ%1#") 
	-- if noon ghunna cannot assimilate, it becomes a nasal vowel.
	text = gsub(text, "ن" .. ghunna, "ں")
	text = gsub(text, "ؤ" .. pesh .. "ں" .. "#", ye2 .. "ū" .. "ں" .. "#")
    
    -- get rid of hashtags (not needed)
    text = gsub(text, "#", "")
    text = gsub(text, "HASHTAG", "#")
    text = string.gsub(text, lrm, "")
	text = string.gsub(text, rlm, "")
    -- convert all characters
    text = gsub(text, '.', mapping)
    
    -- vowel fixes
    -- nasalized dipthongs
    text = gsub(text, 'a([iu])̃', 'a͠%1')
	
	-- alif
	-- Final corrections
	text = gsub(text, "lll", "ll")
	text = gsub(text, "āa", "ā")
	text = gsub(text, "aaa", "ā")
	text = gsub(text, "āā", "ā")
	text = gsub(text, "aa", "ā")
	
	--now get rid of the zero consonants
	text = gsub(text, "ئ", "")
	text = gsub(text, "u" .. "ؤ" , "u")
	text = gsub(text, "ؤ" .. "u" .. "$", "ū")  -- ؤُ is rendered 'ū' word-finally, short 'u' otherwise
	text = gsub(text, "ؤ" .. "u" .. "([ ,.;?!-])", "ū%1")
	text = gsub(text, "ؤ" .. "u" , "u")
	text = gsub(text, "ؤ", "o")
	
	text = mw.ustring.toNFC(text)
	
	return text
end

return export