wikt>Surjection at 21:10, 11 September 2025

2025-09-11T21:10:29Z

New page

local export = {}

local m_str_utils = require("Module:string utilities")

local gcodepoint = m_str_utils.gcodepoint
local rfind = m_str_utils.find
local rsubn = m_str_utils.gsub
local rmatch = m_str_utils.match
local rsplit = m_str_utils.split
local U = m_str_utils.char
local unpack = unpack or table.unpack -- Lua 5.2 compatibility

-- assigned below
local has_diacritics

-- version of rsubn() that discards all but the first return value
local function rsub(term, foo, bar)
local retval = rsubn(term, foo, bar)
return retval
end

local zwnj = U(0x200C) -- zero-width non-joiner
local alif_madda = U(0x622)
local alif_hamza_below = U(0x625)
local alif = U(0x627)
local taa_marbuuTa = U(0x629)
local laam = U(0x644)
local waaw = U(0x648)
local alif_maqSuura = "یٰ"
local yaa = "ی"
local fatHataan = U(0x64B)
local Dammataan = U(0x64C)
local kasrataan = U(0x64D)
local fatHa = U(0x64E)
local Damma = U(0x64F)
local kasra = U(0x650)
local shadda = U(0x651)
local sukuun = U(0x652)
local waaw_maj = "ۉ"
local yaa_maj = "ې"
local dagger_alif = U(0x670)
local alif_waSl = U(0x671)
--local zwj = U(0x200D) -- zero-width joiner
local lrm = U(0x200E) -- left-to-right mark
local rlm = U(0x200F) -- right-to-left mark
-- Occurs after al- in allaḏī and variants so that we can implement elision of
-- a- after a preceding vowel, after which we remove the marker.
local alladi_marker = U(0xFFF0)

local tt = {
-- consonants
["ب"]="b",
["پ"]="p",
["ت"]="t",
["ث"]="s",
["ج"]="j",
["چ"]="ch",
["ح"]="h",
["خ"]="x",
["د"]="d", ["ذ"]="z", ["ر"]="r",
["ز"]="z",
["ژ"]="zh",
["س"]="s", ["ش"]="sh",
["ص"]="s",
["ض"]="z",
["ط"]="t",
["ظ"]="z",
["ع"]="",
["غ"]="gʻ",
["ف"]="f",
["ق"]="q",
["ک"]="k",
["گ"]="g",
["ل"]="l",
["م"]="m",
["ن"]="n",
["ه"]="h",
-- toʾ marbūta (special) - always after a fátḥa (a), silent at the end of
-- an utterance, "t" in ʾizofa or with pronounced tanwīn. We catch
-- most instances of toʾ marbūta before we get to this stage.
[taa_marbuuTa]="t", -- toʾ marbūta = ة
-- control characters
[zwnj]="-", -- ZWNJ (zero-width non-joiner)
-- [zwj]="", -- ZWJ (zero-width joiner)
-- rare letters

-- semivowels or long vowels, alif, hamza, special letters
["ا"]="o", -- ʾalif
-- hamzated letters
["أ"]="ʼ", -- hamza over alif
[alif_hamza_below]="ʼ", -- hamza under alif
["ؤ"]="ʼ", -- hamza over wow
["ئ"]="ʼ", -- hamza over yo
["ء"]="ʼ", -- hamza on the line
-- long vowels
[waaw]="v", --"ū" after zamma (u) and not before diacritic
[yaa]="y", --"ī" after kasra (i) and not before diacritic
[alif_maqSuura]="o", -- ʾalif maqsūra
[alif_madda]="o", -- ʾalif madda
[alif_waSl]= "", -- hamzatu l-wasl
[dagger_alif] = "o", -- ʾalif xanjariyya = dagger ʾalif (Koranic diacritic)
-- short vowels, šádda and sukūn
[fatHataan]="an", -- fatḥatan
[Dammataan]="un", -- zammatan
[kasrataan]="in", -- kasratan
[fatHa]="a", -- fatḥa
[Damma]="u", -- zamma
[kasra]="i", -- kasra
[waaw_maj]="oʻ",
[yaa_maj]="e",
-- šadda - doubled consonant
[sukuun]="", --sukūn - no vowel
-- ligatures
["ﻻ"]="lo",
["ﷲ"]="lloh",
-- tatwīl
["ـ"]="", -- tatwīl, 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
["«"]='“', -- quotation mark
["»"]='”', -- quotation mark
["٫"]=".", -- decimal point
["٬"]=",", -- thousands separator
["٪"]="%", -- percent sign
["،"]=",", -- comma
["؛"]=";" -- semicolon
}

local he = "ه"

local sun_letters = "تثدذرزسشصضطظلن"
-- For use in implementing sun-letter assimilation of ال (al-)
local ttsun1 = {}
local ttsun2 = {}
local ttsun3 = {}
for cp in gcodepoint(sun_letters) do
local ch = U(cp)
ttsun1[ch] = tt[ch]
ttsun2["l-" .. ch] = tt[ch] .. "-" .. ch
table.insert(ttsun3, tt[ch])
end
-- For use in implementing elision of al-
local sun_letters_tr = table.concat(ttsun3, "")

local consonants_needing_vowels = "بتثجحخدذرزسشصضطظعغفقکلمنهپچژڤگڨڧڢںأإؤئءةﷲ"
-- 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 .. "وی"
-- Arabic semicolon, comma, question mark; tatwīl; period, exclamation point,
-- single quote for bold/italic, double quotes for quoted material
local punctuation = "؟،؛" .. "ـ" .. ".!'" .. '"'
local space_like = "%s'" .. '"' .. zwnj .. "\n" .. "%p"
local space_like_class = "[" .. space_like .. "]"
local numbers = "۱۲۳۴۵۶۷۸۹۰"

local before_diacritic_checking_subs = {
------------ transformations prior to checking for diacritics --------------
-- random Koranic marks and presentation forms
{U(0x06E1), sukuun}, -- "Small High Dotless Head of Khah" (variant of sukūn)
{U(0x06DA), ""}, -- "Small High Jeem"
{U(0x06DF), ""}, -- "Small High Rounded Zero" (FIXME: correct?)
{U(0x08F0), U(0x64B)}, -- "Open Fathatan"
{U(0x08F1), U(0x64C)}, -- "Open Dammatan"
{U(0x08F2), U(0x64D)}, -- "Open Kasratan"
{U(0x06E4), ""}, -- "Small High Madda" (FIXME: correct?)
{U(0x06D6), ""}, -- "Small High Ligature Sad with Lam with Alef Maksura" (FIXME: there are others we need to do)
{U(0x06E5), "و"},
{U(0x06E6), "ی"},
-- convert llh for alloh into ll+shadda+dagger-alif+h
{"لله", "للّٰه"},
-- shadda+short-vowel (including tanwīn vowels, i.e. -an -in -un) gets
-- replaced with short-vowel+shadda during NFC normalisation, which
-- MediaWiki does for all Unicode strings; however, it makes the
-- transliteration process inconvenient, so undo it.
{"([" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. dagger_alif .. "])" .. shadda, shadda .. "%1"},
-- ignore Koranic gemination at beginning of word due to assimilation of preceding consonant
{" ([" .. lconsonants .. "])" .. shadda, " %1"},
-- ignore alif jamīla (otiose alif in 3pl verb forms)

-- alif_madda, ayn
{"(" .. sukuun .. ")" .. alif_madda , "%1" .. "ء" .. fatHa .. alif},
{"([" .. lconsonants .. sukuun .. "])" .. "ع"
, "%1" .. "ء" },
{ "ع" .. sukuun , "ء" .. sukuun },

-- final he
{"([" .. fatHa .. Damma .. kasra .. "])" .. "ه" .. "(" .. space_like_class .. ")" , "%1%2" },
{"([" .. fatHa .. Damma .. kasra .. "])" .. "ه" .. "$" , "%1" },
{"([" .. fatHa .. Damma .. kasra .. "])" .. he .. "$" , "%1%2" },

-- #1: handle zamma + wow + alif (final -ū)
{Damma .. waaw .. alif, Damma .. waaw},
-- #2: handle wow + sukūn + alif (final -w in -aw in defective verbs)
-- this must go before the generation of w, which removes the waw here.
{waaw .. sukuun .. alif, waaw .. sukuun},
-- ignore final alif or alif maqsūra following fatḥatan (e.g. in accusative
-- singular or words like عَصًا "stick" or هُدًى "guidance"; this is called
-- tanwin nasb)

{"(" .. fatHa .. "?)" .. yaa .. dagger_alif, "%1" .. alif},
{fatHataan .. "[" .. alif .. alif_maqSuura .. "]", fatHataan},
-- same but with the fatḥatan placed over the alif or alif maqsūra
-- instead of over the previous letter (considered a misspelling but
-- common)
{"[" .. alif .. alif_maqSuura .. "]" .. fatHataan, fatHataan},
-- toʾ marbūta should always be preceded by fatḥa, alif, alif madda or
-- dagger alif; infer fatḥa if not
{"([^" .. fatHa .. alif .. alif_madda .. dagger_alif .. "])" .. taa_marbuuTa, "%1" .. fatHa .. taa_marbuuTa},
-- similarly for alif between consonants, possibly marked with shadda
-- (does not apply to initial alif, which is silent when not marked with
-- hamza, or final alif, which might be pronounced as -an)
{"([" .. lconsonants .. "]" .. shadda .. "?)" .. alif .. "([" .. rconsonants .. "])",
"%1" .. fatHa .. alif .. "%2"},
-- infer fatḥa in case of non-fatḥa + alif/alif-maqsūra + dagger alif
{"([^" .. fatHa .. "])([" .. alif .. alif_maqSuura .. "]" .. dagger_alif .. ")", "%1" .. fatHa .. "%2"},
-- infer kasra in case of hamza-under-alif not + kasra
{alif_hamza_below .. "([^" .. kasra .. kasrataan .. "])", alif_hamza_below .. kasra .. "%1"},
-- ignore dagger alif placed over regular alif or alif maqsūra
{"([" .. alif .. alif_maqSuura .. "])" .. dagger_alif, "%1"},

----------- rest of these concern definite article alif-lom ----------
-- in kasra/zamma + alif + lam, make alif into hamzatu l-wasl, so we
-- handle cases like بِالتَّوْفِیق (bi-t-tawfīq) correctly
{"([" .. Damma .. kasra .. "])" .. alif .. laam, "%1" .. alif_waSl .. laam},
-- al + consonant + shadda (only recognize word-initially if regular alif): remove shadda
{"^(" .. alif .. fatHa .. "?" .. laam .. "[" .. lconsonants .. "])" .. shadda, "%1"},
{"(" .. space_like_class .. alif .. fatHa .. "?" .. laam .. "[" .. lconsonants .. "])" .. shadda, "%1"},
{"(" .. alif_waSl .. fatHa .. "?" .. laam .. "[" .. lconsonants .. "])" .. shadda, "%1"},
-- handle l- hamzatu l-wasl or word-initial al-
{"^" .. alif .. fatHa .. "?" .. laam, "al-"},
{"(" .. space_like_class .. ")" .. alif .. fatHa .. "?" .. laam, "%1al-"},
-- next one for bi-t-tawfīq
{"([" .. Damma .. kasra .. "])" .. alif_waSl .. fatHa .. "?" .. laam, "%1-l-"},
-- next one for remaining hamzatu l-wasl (at beginning of word)
{alif_waSl .. fatHa .. "?" .. laam, "l-"},
-- special casing if the l in al- has a shadda on it (as in الَّذِی "that"),
-- so we don't mistakenly double the dash; insert a special marker here so
-- that we know later to elide the a- after a vowel
{"l%-" .. shadda, "l" .. alladi_marker .. "l"},
-- implement assimilation of sun letters
{"l%-[" .. sun_letters .. "]", ttsun2},
}

-- Transliterate the word(s) in TEXT. LANG (the language) and SC (the script)
-- are ignored. OMIT_I3RAAB means leave out final short vowels (ʾiʿrob).
-- GRAY_I3RAAB means render transliterate short vowels (ʾiʿrob) in gray.
-- FORCE_TRANSLIT causes even non-vocalized text to be transliterated
-- (normally the function checks for non-vocalized text and returns nil,
-- since such text is ambiguous in transliteration).
function export.tr(text, lang, sc, omit_i3raab, gray_i3raab, force_translit)
-- make it possible to call this function from a template
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 = rsub(text, sub[1], sub[2])
end

if not force_translit and not has_diacritics(text) then
require("Module:debug").track("ar-translit/lacking diacritics")
return nil
end

------------ transformations after checking for diacritics --------------
-- Replace plain alif with hamzatu l-wasl when followed by fatḥa/zamma/kasra.
-- Must go after handling of initial al-, which distinguishes alif-fatḥa
-- from alif w/hamzatu l-wasl. Must go before generation of ū and ī, which
-- eliminate the zamma/kasra.
text = rsub(text, alif .. "([" .. fatHa .. Damma .. kasra .. waaw_maj .. yaa_maj .. "])", alif_waSl .. "%1")
-- zamma + waw not followed by a diacritic is ū, otherwise w
text = rsub(text, Damma .. waaw .. "([^" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. shadda .. sukuun .. dagger_alif .. waaw_maj .. yaa_maj .. "])", "ū%1")
text = rsub(text, Damma .. waaw .. "$", "ū")
-- kasra + yaa not followed by a diacritic (or ū from prev step) is ī, otherwise y
text = rsub(text, kasra .. yaa .. "([^" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. shadda .. sukuun .. dagger_alif .. waaw_maj .. yaa_maj .. "ū])", "ī%1")
text = rsub(text, kasra .. yaa .. "$", "ī")
-- final he
text = rsub(text, "([" .. fatHa .. Damma .. kasra .. "])" .. he .. "$", "%1")
text = rsub(text, "([" .. fatHa .. Damma .. kasra .. "])" .. he .. "(" .. space_like_class .. zwnj .. ")", "%1%2")
text = rsub(text, zwnj, "")
-- convert shadda to double letter.
text = rsub(text, "(.)" .. shadda, "%1%1")
if not omit_i3raab and gray_i3raab then -- show ʾiʿrob grayed in transliteration
-- decide whether to gray out the t in ﺓ. If word begins with al- or l-, yes.
-- Otherwise, no if word ends in a/i/u, yes if ends in an/in/un.
text = rsub(text, "^(a?l%-[^%s]+)" .. taa_marbuuTa .. "([" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. "])",
'%1t%2')
text = rsub(text, "(" .. space_like_class .. "a?l%-[^%s]+)" .. taa_marbuuTa .. "([" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. "])",
'%1t%2')
text = rsub(text, taa_marbuuTa .. "([" .. fatHa .. Damma .. kasra .. "])", "t%1")
text = rsub(text, taa_marbuuTa .. "([" .. fatHataan .. Dammataan .. kasrataan .. "])",
't%1')
text = rsub(text, ".", {
[fatHataan] = 'an',
[kasrataan] = 'in',
[Dammataan] = 'un'
})
text = rsub(text, "([" .. fatHa .. Damma .. kasra .. "])(" .. space_like_class .. ")",
function(vowel, space)
vowel_repl = {
[fatHa] = 'a ',
[kasra] = 'i ',
[Damma] = 'u '
}
return vowel_repl[vowel] .. space
end
)
text = rsub(text, "[" .. fatHa .. Damma .. kasra .. "]$", {
[fatHa] = 'a',
[kasra] = 'i',
[Damma] = 'u'
})
text = rsub(text, '', "")
elseif omit_i3raab then -- omit ʾiʿrob in transliteration
text = rsub(text, "[" .. fatHataan .. Dammataan .. kasrataan .. "]", "")
text = rsub(text, "[" .. fatHa .. Damma .. kasra .. "](" .. space_like_class .. ")", "%1")
text = rsub(text, "[" .. fatHa .. Damma .. kasra .. "]$", "")
end
-- toʾ marbūta should not be rendered by -t if word-final even when
-- ʾiʿrob (desinential inflection) is shown; instead, use (t) before
-- whitespace, nothing when final; but render final -ﺍﺓ and -ﺁﺓ as -oh,
-- consistent with Wehr's dictionary
-- Left-to-right or right-to-left mark at end of text will prevent toʾ marbūta
-- from being transliterated correctly.
text = string.gsub(text, lrm, "")
text = string.gsub(text, rlm, "")
text = rsub(text, "([" .. alif .. alif_madda .. "])" .. taa_marbuuTa .. "$", "%1h")
-- Ignore final toʾ marbūta (it appears as "a" due to the preceding
-- short vowel). Need to do this after graying or omitting word-final
-- ʾiʿrob.
text = rsub(text, taa_marbuuTa .. "$", "")
text = rsub(text, taa_marbuuTa .. "(%p)", "%1")
if not omit_i3raab then -- show ʾiʿrob in transliteration
text = rsub(text, taa_marbuuTa .. "(" .. space_like_class .. ")", "(t)%1")
else
-- When omitting ʾiʿrob, show all non-absolutely-final instances of
-- toʾ marbūta as (t), with trailing ʾiʿrob omitted.
text = rsub(text, taa_marbuuTa, "(t)")
end
-- tatwīl should be rendered as - at beginning or end of word. It will
-- be rendered as nothing in the middle of a word (FIXME, do we want
-- this?)
text = rsub(text, "^ـ", "-")
text = rsub(text, "(" .. space_like_class .. ")ـ",
"%1-")
text = rsub(text, "ـ$", "-")
text = rsub(text, "ـ(" .. space_like_class .. ")", "-%1")
-- Now convert remaining Arabic chars according to table.
text = rsub(text, ".", tt)
text = rsub(text, "ao", "o")
-- Implement elision of al- after a final vowel. We do this
-- conservatively, only handling elision of the definite article and related
-- terms (specifically, relative pronoun الَّذِی (allaḏī) and variants) rather
-- than elision in other cases of hamzat al-wasl (e.g. form-I imperatives
-- or form-VII and above verbal nouns) partly because elision in
-- these cases isn't so common in MSA and partly to avoid excessive
-- elision in case of words written with initial bare alif instead of
-- properly with hamzated alif. Possibly we should reconsider.
text = rsub(text, "([aiuoīū]'* +'*)a([" .. sun_letters_tr .. "][%-" .. alladi_marker .. "])",
"%1%2")
if gray_i3raab then
text = rsub(text, "([aiuoīū]'*'* +'*)a([" .. sun_letters_tr .. "][%-" .. alladi_marker .. "])",
"%1%2")
end
-- remove indicator of allaḏī, which has served its purpose
text = rsub(text, alladi_marker, "")
-- Special-case the transliteration of alloh, without the hyphen.
text = rsub(text, "^(a?)l%-loh", "%1lloh")
text = rsub(text, "(" .. space_like_class .. "a?)l%-loh", "%1lloh")
-- Compress multiple spaces, which may occur e.g. when removing Koranic diacritics.
text = rsub(text, "(%s)%s+", "%1")
-- Remove length (not native in uzbek)
text = rsub(text, "ī", "i")
text = rsub(text, "ū", "u")

return text
end

local has_diacritics_subs = {
-- FIXME! What about lam-alif ligature?
-- remove punctuation and shadda
-- must go before removing final consonants
{"[" .. punctuation .. shadda .. "]", ""},
{zwnj, " "},
-- Remove consonants at end of word or utterance, so that we're OK with
-- words lacking iʿrob (must go before removing other consonants).
-- If you want to catch places without iʿrob, comment out the next two lines.
{"[" .. lconsonants .. "]$", ""},
{"[" .. lconsonants .. "]([%)%]}]?" .. space_like_class .. ")", "%1"},
-- remove consonants (or alif) when followed by diacritics
-- must go after removing shadda
-- do not remove the diacritics yet because we need them to handle
-- long-vowel sequences of diacritic + pseudo-consonant
{"[" .. lconsonants .. alif .. "]([" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. sukuun .. dagger_alif .. waaw_maj .. yaa_maj .. "])", "%1"},
-- the following two must go after removing consonants w/diacritics because
-- we only want to treat vocalic wow/yo' in them (we want to have removed
-- wow/yo' followed by a diacritic)
-- remove zamma + wow
{Damma .. waaw, ""},
-- remove kasra + yo'
{kasra .. yaa, ""},
-- remove majhuls
{waaw_maj, ""},
{yaa_maj, ""},
-- remove fatḥa/fatḥatan + alif/alif-maqsūra
{"[" .. fatHataan .. fatHa .. "][" .. alif .. alif_maqSuura .. "]", ""},
-- remove diacritics
{"[" .. fatHataan .. Dammataan .. kasrataan .. fatHa .. Damma .. kasra .. sukuun .. dagger_alif .. "]", ""},
-- remove numbers, hamzatu l-wasl, alif madda
{"[" .. numbers .. "ٱ" .. "آ" .. "]", ""},
-- remove non-Arabic characters
{"[^" .. U(0x0600) .. "-" .. U(0x06FF) .. U(0x0750) .. "-" .. U(0x077F) ..
U(0x08A0) .. "-" .. U(0x08FF) .. U(0xFB50) .. "-" .. U(0xFDFF) ..
U(0xFE70) .. "-" .. U(0xFEFF) .. "]", ""}
}

-- declared as local above
function has_diacritics(text)
local orig_text = text
local count
text, count = rsubn(text, "[" .. lrm .. rlm .. "]", "")
if count > 0 then
require("Module:debug").track("ar-translit/lrm or rlm")
end
for _, sub in ipairs(has_diacritics_subs) do
text = rsub(text, unpack(sub))
end
if #text > 0 then
mw.log(("Check for missing diacritics failed; original text '%s', text without diacritics '%s'"):format(
orig_text, text))
end
return #text == 0
end

-- Return true if transliteration TR is an irregular transliteration of
-- ARABIC. Return false if ARABIC can't be transliterated. For purposes of
-- establishing regularity, hyphens are ignored and word-final toʾ marbūta
-- can be transliterated as "(t)", "" or "t".
function export.irregular_translit(arabic, tr)
if not arabic or arabic == "" or not tr or tr == "" then
return false
end
local regtr = export.tr(arabic)
if not regtr or regtr == tr then
return false
end
local arwords = rsplit(arabic, " ")
local regwords = rsplit(regtr, " ")
local words = rsplit(tr, " ")
if #regwords ~= #words or #regwords ~= #arwords then
return true
end
for i=1,#regwords do
local regword = regwords[i]
local word = words[i]
local arword = arwords[i]
-- Resolve final (t) in auto-translit to t, h or nothing
if rfind(regword, "%(t%)$") then
regword = rfind(word, "oh$") and rsub(regword, "%(t%)$", "h") or
rfind(word, "t$") and rsub(regword, "%(t%)$", "t") or
rsub(regword, "%(t%)$", "")
end
-- Resolve clitics + short a + alif-lom, which may get auto-transliterated
-- to contain long o, to short a if the manual translit has it; note
-- that currently in cases with assimilated l, the auto-translit will
-- fail, so we won't ever get here and don't have to worry about
-- auto-translit l against manual-translit assimilated char.
local clitic_chars = "^[وفیییل]" -- separate line to avoid L2R display weirdness
if rfind(arword, clitic_chars .. fatHa .. "?[" .. alif .. alif_waSl .. "]" .. laam) and rfind(word, "^[wfkl]a%-") then
regword = rsub(regword, "^([wfkl])o", "%1a")
end
-- Ignore hyphens when comparing
if rsub(regword, "%-", "") ~= rsub(word, "%-", "") then
return true
end
end
return false
end

return export

Module:uz-afg-translit - Revision history

Sware: 1 revision imported

wikt>Surjection at 21:10, 11 September 2025

← Older revision	Revision as of 12:01, 21 April 2026
(No difference)