Module:hi-translit

-- Transliteration for Hindi (possibly other languages using Devanagari script, except for Sanskrit)
local export = {}

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

local gmatch = m_str_utils.gmatch
local gsub = m_str_utils.gsub
local match = m_str_utils.match
local plain_gsub = m_str_utils.plain_gsub
local reverse = m_str_utils.reverse
local toNFC = mw.ustring.toNFC

local conv = {
	-- consonants
	 = 'k',  = 'kh',  = 'g',  = 'gh',  = 'ṅ', 
	 = 'c',  = 'ch',  = 'j',  = 'jh',  = 'ñ', 
	 = 'ṭ',  = 'ṭh',  = 'ḍ',  = 'ḍh',  = 'ṇ',
	 = 't',  = 'th',  = 'd',  = 'dh',  = 'n',
	 = 'p',  = 'ph',  = 'b',  = 'bh',  = 'm', 
	 = 'y',  = 'r',  = 'l',  = 'v',  = 'ḷ',
	 = 'ś',  = 'ṣ',  = 's',  = 'h',
	 = 'q',  = 'x',  = 'ġ',  = 'ḻ',
	 = 'z',  = 'ẓ',  = 'ź',  = 'ṛ',  = 'ṛh',
	 = 'f',  = 'ṉ',  = 'ṟ',  = 'ẏ',  = 'w',
	--  = 'jñ',

	-- vowel diacritics
	 = 'i',  = 'u',  = 'e',  = 'o',
     = 'ǒ',  = 'ě',
	 = 'ā',  = 'ī',  = 'ū', 
	 = 'ŕ',
	 = 'ai',  = 'au',
	 = 'ŏ',
	 = 'ĕ',

	-- vowel signs
	 = 'a',  = 'i',  = 'u',  = 'e',  = 'o',
	 = 'ā',  = 'ī',  = 'ū',  = 'ě',  = 'ǒ',
	 = 'ŕ', 
	 = 'ai',  = 'au', 
	 = 'ŏ',
	 = 'ĕ',
	
	 = 'om',
	
	-- chandrabindu
	 = '̃',
	
	-- anusvara
	 = '̃',
	
	-- visarga
	 = 'ḥ',
	
	-- virama
	 = '',
	
	-- numerals
	 = '0',  = '1',  = '2',  = '3',  = '4',
	 = '5',  = '6',  = '7',  = '8',  = '9',
	
	-- punctuation
	 = '.', -- danda
	 = '.', -- double danda
	 = '', -- compound separator
	
	-- abbreviation sign
	 = '.',
}

local nasal_assim_short = {
	 = 'ङ',  = 'ङ',  = 'ङ',  = 'ङ',  = 'ङ',
	 = 'ञ',  = 'ञ',  = 'ञ',  = 'ञ',  = 'ञ',  
	 = 'ण',  = 'ण',  = 'ण',  = 'ण',  = 'ण',
	 = 'न',  = 'न',  = 'न',  = 'न',  = 'न',
	 = 'म',  = 'म',  = 'म',  = 'म',  = 'म',
	 = 'ँ',  = 'न',  = 'न',   = 'म',
	 = 'ञ',  = 'ण',  = 'न',  = 'ँ',
	 = 'न',  = 'म',   = 'ङ',  = 'ङ',  = 'ङ',
	 = 'ँ',  = 'ँ'
}

local nasal_assim_long = {
	 = 'ँ',  = 'ँ',  = 'ङ',  = 'ङ',  = 'ँ',
	 = 'ँ',  = 'ँ',  = 'ञ',  = 'ञ',  = 'ँ',  
	 = 'ँ',  = 'ँ',  = 'ण',  = 'ण',  = 'ँ',
	 = 'ँ',  = 'ँ',  = 'न',  = 'न',  = 'ँ',
	 = 'ँ',  = 'ँ',  = 'म',  = 'म',  = 'ँ',
	 = 'ँ',
	 = 'न',  = 'म',   = 'ङ',  = 'ङ',  = 'ङ',
	 = 'ँ',  = 'ँ'
}

-- These clusters when occurring word-finally will not trigger a schwa added
-- after them even though the second consonant is in special_cons, which normally
-- causes the extra schwa to be added. NOTE: The clusters are reversed from their
-- ultimate effect, e.g. the first cluster is written 'ml' but actually applies
-- to words ending in 'lm'. The clusters below overall refer to the six clusters
-- describable by , i.e. rm, rn, rv, lm, ln, lv (plus rṇ).
-- Plus some word-final geminates.
local perm_cl = {
	 = true,  = true,  = true,
	 = true,  = true,  = true,  = true,
	 = true,  = true,  = true,  = true,  = true
}

local all_cons, special_cons = 'कखगघङचछजझञटठडढणतथदधनपफबभमयरलवषशसह', 'यरलवहनमञण'
local vowel, vowel_sign = '*aिुृेोाीूैौॉॅॆॊ\'', 'अइउएओआईऊऋऐऔऑऍ\''
local long_vowel, short_vowel = 'ाीूेैोौआईऊएऐओऔ', '*aिुृॆॊॅॉअइउऋऍऑऎऒ\''
local syncope_pattern = '()(़?)a(़?)(?)'

function export.tr(text, lang, sc)
	-- treat anusvara + nasal as geminate nasal after short vowels
	text = gsub(text, '()ं()', '%1%2्%2')
	-- word-final apostrophe (e.g. from bold formatting) does not delete schwa
	text = gsub(text, '(़?)(\'%A)', '%1a%2')
	text = gsub(text, '(़?)(\')$', '%1a%2')
	text = gsub(text, '(़?)(?)', function(c, d)
		return c .. (d == "" and 'a' or d) end)
	-- to detect words, include Unicode 0900-0963 and 0971-097F (plus 'a' and '*'),
	-- excluding e.g. danda and abbreviation dot ॰
	for word in gmatch(text, "+") do
		local orig_word = word
		word = reverse(word)
		word = gsub(word, '^a(़?)()(.)(.?)', function(opt, first, second, third)
			return (((match(first, '') and match(second, '्') and not perm_cl)
				or match(first .. second, 'य'))
				and 'a' or "") .. opt .. first .. second .. third end)
		while match(word, syncope_pattern) do
			word = gsub(word, syncope_pattern, '%1%2%3%4')
		end
		word = reverse(word)
		-- The special_vowel category consists of ī/e/o/ai/au, these vowels are often 
		-- written with anusvara for 'aesthetic' reasons as the vowel diacritic 
		-- gets in the way of the candrabindu. For ī/e/o/ai/au, anusvara will trigger 
		-- nasal_assim_long and candrabindu will force a nasal vowel. To force a nasal 
		-- consonant before voiceless stops (mostly in loanwords), respell with 
		-- homorganic nasal ङ/ञ/ण/न/म + virama ्
		-- Exception: vowel 'e' when written as standalone ए
        local special_vowel, normal_vowel = 'ीेैोौईऐओऔ', '*aिुाूृॆॅॊॉअइउआऊऋऎऍऒऑए\''
        word = gsub(word, '()ँ(.़?)', function(prev, succ)
			return prev .. "̃" .. succ
		end)
		--sometimes chandrabindu != anusvara
		word = gsub(word, '()ं(़?)', function(prev, succ)
			return prev .. (nasal_assim_short or "̃") .. succ
		end)
		word = gsub(word, '()ँ(़?)', function(prev, succ)
			return prev .. "̃" .. succ
		end)
		--force chandrabindu to behave as anusvara
		word = gsub(word, 'ँ', 'ं')
		word = gsub(word, '()ं(.़?)', function(prev, succ)
			return prev .. (nasal_assim_short or "̃") .. succ
		end)
		word = gsub(word, '()ं(.़?)', function(prev, succ)
			return prev .. (nasal_assim_long or "̃") .. succ
		end)
		text = plain_gsub(text, orig_word, word)
	end
	text = gsub(text, '.़?', conv)
	text = gsub(text, 'a()̃', 'a͠%1')
    text = gsub(text, 'ñz', 'nz')
    text = gsub(text, '%*', 'a')
	return toNFC(text)
end

return export