Module:User:Sarri.greek/grk-pronunciation

-- 2024.04.24. ] my notes with !!, todo with ??
-- this is ]
-- Now, default is med1+med2. Also view period=cla (all periods)
-- TESTS raw at ] ]
-- TEST for med at ]
--[=[!!??
STRUCTURE: This module produces at the moment consecutive lines for different language codes with pronunciations.
Each line has 3 things / 'elements':
1) the period-label (linked) e.g. period-label-cla, period-lebal-koi1... (example: 5th c. BCE Attic)
2) the IPA-label+key  (linked IPA<sup>key</sup>): e.g.  IPA-label-cla, IPA-label-koi1...
3) the actual IPA e.g. IPA-cla, IPA-koi1....
At the moment these three are unbreakable. There are no args for the editor.
One cannot call only one element. ONLY 3) or ONLY 1).
One cannot omit an element e.g. with?? ] has something like split_output "raw"
One cannot override IPA.
One cannot add a second IPA.
One cannot add inline notes before or after.

NEED for structure
* 1 Detach elements. Especially, detach the IPAs as named parameters
	Need args for editor.
	tested: cla-only, el-only etc.
* 2 OVERRIDE must be possible by editor, especially for med2, el loanwards or minor corrections
	el: if override cannot be done, the we CANNOT have modern Greek IPA
	because it is impossible to predict some pronunciations
	The same goes for some med2 (with synizesis) it may or it may not have synizesis
	depending on register.
* 3 second ipa (for optional second pronunciation)
* -- NOTES for every period line
	-- now done with tables at templates
Then, editor can make a Template with any combination of the above
Make a line for ONE period: inline (no break lines)
	period-label-cla .. IPA-label-cla .. IPA-cla in one line (no break lines)
	add parameters for all elements
Combinations: functions for one period only
	Period-titles as cla-only, el-only, from Module:User:Sarri.greek/accent qualifier/data
Combinations of many lines: if a next line follows, but <br>
	But now, we need the same width column for Period-titles
	Better do it at a template
Function: IPA only (for inflectional tables. It could be placed under the forms or under transliterations)

PROBLEMS at IPA = > > > > specific letters or clusters problems at DATA page. see /data page for /ks ps zm/ and others

* ?? if imput initial  ΆάΈέΉήΊίΌόΎύΏώ  / ΑαΕεΗηΙιΟοΥυΩω / Αί αί Εί εί Οί οί Ού ού  / Αύ αύ Εύ εύ Ηύ ηύ  then, show warning for period=el
	(not Polyt? But it works equally well for monotonic ell too)
	 or a message: Do you wish only el...
]=]--

local export = {}
local m_data = mw.loadData("Module:User:Sarri.greek/grk-pronunciation/data") --!! mw.loadData("Module:grc-pronunciation/data")

--required
local mark_implied_length = require('Module:grc-accent').mark_implied_length
local strip_accent = require('Module:grc-accent').strip_accent

local m_general_utils = require("Module:utilities")

-- ] converts sequences of diacritics to the order required by this module,
-- then replaces combining macrons and breves with spacing ones.
local m_utils = require("Module:grc-utilities")
local m_utils_data = require("Module:grc-utilities/data")
local full_link = m_utils.link
local tag_text = m_utils.tag
local diacritics = m_utils_data.diacritics
local rearrangeDiacritics = m_utils.pronunciationOrder

local m_IPA = require("Module:IPA")
local m_a = require("Module:User:Sarri.greek/accent qualifier") --!! require("Module:accent qualifier")
--!! put these under periods, according to each language
local lang = require("Module:languages").getByCode("grc") --!!this works for all but not Links to Wikipedias
local sc = require("Module:scripts").getByCode("Polyt") --!!this works fine for polytonic and monotonic

local periods = {'cla', 'koi1', 'koi2', 'med1', 'med2', 'el'} -- was {'cla', 'koi1', 'koi2', 'byz1', 'byz2'}
local inlinePeriods = {'med1', 'med2'} --!! was = {'cla', 'koi2', 'byz2'}

--!! add params for inline notes?

--!! there are commands, not some made up names
local title = mw.title.getCurrentTitle()
local pagename = title.text
local namespace = title.nsText

local rsplit = mw.text.split
local rfind = mw.ustring.find
local usub = mw.ustring.sub
local rmatch = mw.ustring.match
local rsubn = mw.ustring.gsub
local ulen = mw.ustring.len --
local ulower = mw.ustring.lower
local U = mw.ustring.char

local function fetch(s, i)
	--[==[
	because we fetch a single character at a time so often
	out of bounds fetch gives ''
	]==]
	i = tonumber(i)
	
	if type(i) ~= "number" then
		error("fetch requires a number or a string equivalent to a number as its second argument.")
	end
	
	if i == 0 then
		return ""
	end
	
	local n = 0
	for character in string.gmatch(s, "*") do
		n = n + 1
		if n == i then
			return character
		end
	end
	
	return ""
end

--!! these are IPA things.
--Combining diacritics are tricky.
local tie = U(0x35C)				-- tie bar
local nonsyllabic = U(0x32F)		-- combining inverted breve below
local high = U(0x341)				-- combining acute tone mark
local low = U(0x340)				-- combining grave tone mark
local rising = U(0x30C)				-- combining caron
local falling = diacritics.Latin_circum	-- combining circumflex
local midHigh = U(0x1DC4)			-- mid–high pitch
local midLow = U(0x1DC6)			-- mid–low pitch
local highMid = U(0x1DC7)			-- high–mid pitch
local voiceless = U(0x325)			-- combining ring below
local aspirated = 'ʰ'
local macron = '¯'
local breve = '˘'

--?? what does this do?
--!!  = "ι",		 = m_utils_data.all,
local function is(text, X)
	if not text or not X then
		return false
	end
	pattern = m_data.chars or error('No data for "' .. X .. '".', 2)
	if X == "frontDiphth" or X == "Greekdiacritic" then
		pattern = "^" .. pattern .. "$"
	else
		pattern = "^$"
	end
	return mw.ustring.find(text, pattern)
end

--!! env = environment MAKE diphthongs αι ει?? οι, αυ ευ ηυ but not when iota has dialytics (diaeresis)
--!!  = "ιηευ",	 = "ι",	 = "ϊΐῒῗ",
local env_functions = {
	preFront = function(term, index)
		local letter1, letter2 = fetch(term, index + 1), fetch(term, index + 2)
		return is(strip_accent(letter1), "frontVowel") or (is(strip_accent(letter1 .. letter2), "frontDiphth") and not is(letter2, "iDiaer"))
	end,
	isIDiphth = function(term, index)
		local letter = fetch(term, index + 1)
		return strip_accent(letter) == 'ι' and not m_data.diaer
	end,
	isUDiphth = function(term, index)
		local letter = fetch(term, index + 1)
		return strip_accent(letter) == 'υ' and not m_data.diaer
	end,
	hasMacronBreve = function(term, index)
		return fetch(term, index + 1) == macron or fetch(term, index + 1) == breve
	end,
}

local function decode(condition, x, term)
	--[==[
		"If" and "and" statements.
		Note that we're finding the last operator first, 
		which means that the first will get ultimately get decided first.
		If + ("and") or / ("or") is found, the function is called again,
		until if-statements are found.
		In if-statements:
		* A number represents the character under consideration:
			 -1 is the previous character, 0 is the current, and 1 is the next.
		* Equals sign (=) checks to see if the character under consideration
			is equal to a character.
		* Period (.) plus a word sends the module to the corresponding entry
			in the letter's data table.
		* Tilde (~) calls a function on the character under consideration,
			if the function exists.
	]==]
	if mw.ustring.find(condition, '') then
		-- Find slash or plus sign preceded by something else, and followed by anything
		-- (including another sequence of slash or plus sign and something else).
		local subcondition1, sep, subcondition2 = mw.ustring.match(condition, "^(-)()(.*)$")
		if not (subcondition1 or subcondition2) then
			error('Condition "' .. tostring(condition) .. '" is improperly formed')
		end
		
		if sep == '/' then		-- logical operator: or
			return decode(subcondition1, x, term) or decode(subcondition2, x, term)
		elseif sep == '+' then	-- logical operator: and
			return decode(subcondition1, x, term) and decode(subcondition2, x, term)
		end
	elseif mw.ustring.find(condition, '=') then				-- check character identity
		local offset, char = unpack(mw.text.split(condition, "="))
		if namespace == "Module" or namespace == "Template" then
			mw.log(term, offset, char, x + offset, fetch(term, x + offset), char == fetch(term, x + offset) )
		end
		return char == fetch(term, x + offset) -- out of bounds fetch gives ''
	elseif mw.ustring.find(condition, '%.') then				-- check character quality
		local offset, quality = unpack(mw.text.split(condition, "%."))
		local character = fetch(term, x + offset)
		return m_data and m_data or false
	elseif mw.ustring.find(condition, '~') then				-- check character(s) using function
		local offset, func = unpack(mw.text.split(condition, "~"))
		return env_functions and env_functions(term, x + offset) or false
	end
end

local function check(p, x, term)
	if type(p) == 'string' or type(p) == 'number' then
		return p
	elseif type(p) == 'table' then   --This table is sequential, with a variable number of entries.
		for _, possP in ipairs(p) do
			if type(possP) == 'string' or type(possP) == 'number' then
				return possP
			elseif type(possP) == 'table' then    --This table is paired, with two values: a condition and a result.
				rawCondition, rawResult = possP, possP
				if decode(rawCondition, x, term) then
					return (type(rawResult) == 'string') and rawResult or check(rawResult, x, term)
				end	
			end
		end
	else
		error('"p" is of unrecongized type ' .. type(p))
	end
end

--?? handle lines/periods separately too?
--!! add notes NEED somewhere here
local function convert_term(term, periodstart)
	if not term then error('The variable "term" in the function "convert_term" is nil.') end
	local IPAs = {}
	local start
	local outPeriods = {}
	if periodstart and periodstart ~= "" then
		start = false
	else
		start = true
	end
	for _, period in ipairs(periods) do 
		if period == periodstart then
			start = true
		end
		if start then
			IPAs = {}
			table.insert(outPeriods, period)
		end
	end
	local length, x, advance, letter, p = mw.ustring.len(term), 1, 0, '', nil
	while x <= length do
		letter = fetch(term, x)
		local data = m_data
		if not data then		-- no data found
			-- explicit pass
		else
			-- check to see if a multicharacter search is warranted
			advance = data.pre and check(data.pre, x, term) or 0
			p = (advance ~= 0) and m_data.p or data.p
			for _, period in ipairs(outPeriods) do
				table.insert(IPAs, check(p, x, term))
			end
			x = x + advance
		end
		x = x + 1
	end

	--Concatenate the IPAs
	for _, period in ipairs(outPeriods) do
		IPAs = { IPA = table.concat(IPAs, '')}
	end
	
	return IPAs, outPeriods
end


local function find_syllable_break(word, nVowel, wordEnd)
	if not word then error('The variable "word" in the function "find_syllable_break" is nil.') end
	if wordEnd then
		return mw.ustring.len(word)
		
--!! unbreakable or special consonants for el, ...
--!! ks ps & zm at el, med2, (med1?), I don't know about koi and grc
--!! check example λοκσι λοξι
	elseif period == 'el' or period == 'med2' or period == 'med1'
	then
		if mw.ustring.match(word, nVowel - 1, "z") then
			if mw.ustring.match(word, nVowel - 2, "m") then
			return nVowel - 4
			end
		end
		--!! consDoule (ks ps if they represent ξ ψ but not κσ)
		--!! if el τσ τζ = with ties t͡s d͡z
		--!! if med, el αυφ αυβ do not repeat aff avv Same for ευ ηυ


	elseif is(fetch(word, nVowel - 1), "liquid") then
		if is(fetch(word, nVowel - 2), "obst") then
			return nVowel - 3
		elseif fetch(word, nVowel - 2) == aspirated and is(fetch(word, nVowel - 3), "obst") then
			return nVowel - 4
		else
			return nVowel - 2
		end
	elseif is(fetch(word, nVowel - 1), "cons") then
		return nVowel - 2
	elseif fetch(word, nVowel - 1) == aspirated and is(fetch(word, nVowel - 2), "obst") then
		return nVowel - 3
	elseif fetch(word, nVowel - 1) == voiceless and fetch(word, nVowel - 2) == 'r' then
		return nVowel - 3
	else
		return nVowel - 1
	end
end

local function syllabify_word(word)
	local syllables = {}
	--[[	cVowel means "current vowel", nVowel "next vowel",
			sBreak "syllable break".							]]--
	local cVowel, nVowel, sBreak, stress, wordEnd, searching
	while word ~= '' do
		cVowel, nVowel, sBreak, stress = false, false, false, false
		
		--First thing is to find the first vowel.
		searching = 1
		cVowelFound = false
		while not cVowel do
			letter = fetch(word, searching)
			local nextLetter = fetch(word, searching + 1)
			if cVowelFound then
				if (is(letter, "vowel") and nextLetter ~= nonsyllabic) or is(letter, "cons") or letter == '' or letter == 'ˈ' then
					cVowel = searching - 1
				elseif is(letter, "diacritic") then
					searching = searching + 1
				elseif letter == tie then
					cVowelFound = false
					searching = searching + 1
				else
					searching = searching + 1
				end
			else
				if is(letter, "vowel") then
					cVowelFound = true
				elseif letter == 'ˈ' then
					stress = true
				end
				searching = searching + 1
			end
		end
	
		--Next we try and find the next vowel or the end.
		searching = cVowel + 1
		while (not nVowel) and (not wordEnd) do
			letter = fetch(word, searching)
			if is(letter, "vowel") or letter == 'ˈ' then
				nVowel = searching
			elseif letter == '' then
				wordEnd = true
			else
				searching = searching + 1
			end
		end
		
--?? keep ψ = ps and ξ = ks toghther .ps. .ks. at med1 med2 el?		
		--Finally we find the syllable break point.
		sBreak = find_syllable_break(word, nVowel, wordEnd)
		
		--Pull everything up to and including the syllable Break.
		local syllable = usub(word, 1, sBreak)
		
		--If there is a stress accent, then we need to move it to the 
		--beginning of the syllable, unless it is a monosyllabic word,
		--in which case we remove it altogether.
		if stress then
			if next(syllables) or syllable ~= word then
				syllable = 'ˈ' .. rsubn(syllable, 'ˈ', '')
			else 
				syllable = rsubn(syllable, 'ˈ', '')
			end
			stress = false
		end
		table.insert(syllables, syllable)
		word = usub(word, sBreak + 1)
	end
	
	local out = nil
	
	if #syllables > 0 then
		out = table.concat(syllables, '.')
		out = rsubn(out, '%.ˈ', 'ˈ')
	end
	return out
end

local function syllabify(IPAs, periods)
	--Syllabify
	local word_ipa = ''
	local ipa = {}
	for _, period in ipairs(periods) do
		ipa = {}
		for _, word in ipairs(rsplit(IPAs.IPA, ' ')) do
			word_ipa = syllabify_word(word)
			if word_ipa then
				table.insert(ipa, word_ipa)
			end
		end
		IPAs.IPA = table.concat(ipa, ' ')
	end
	return IPAs
end

--??TODO is everything automatically brachy for  koi1 koi2 med1 med2? 
local function make_ambig_note(ambig, ambig_letter_list)
	-- The table ambig is filled with all the ambiguous vowels that have been found in the term.
	local ambig_note = ''
	if ambig and #ambig > 0 then
		local agr = (#ambig > 1) and { 's ', 'each one' } or { ' ', 'it' }
			
		ambig_note = '\n<p class="previewonly">Mark the vowel length for 5th century Attic of the ambiguous vowel' .. agr
			.. mw.text.listToText(ambig) .. ' by adding a macron after ' .. agr
			.. ' if it is long, or a breve if it is short. By default, ] assumes it is short if unmarked.'
			.. '<br/><small></small>'
--??TODO Take this Category off? no, but it should only apply to cla.
			.. m_general_utils.format_categories(
				{ 'Ancient Greek terms with incomplete pronunciation' }, lang)
			..'</p>\n'
	end
	return ambig_note
end


local function make_table(IPAs, ambig, periods, ambig_letter_list)
	--Final format
	local inlineProns = {}
	local listOfProns = {}
	local fullProns = {}
	local periods2 = {}
	
--[=[

--!! add notes	
--?? need periodnotes td next to eadh periodline	 local inline_period_notes = {}
	local listOfNotes = {}
	
	for _, period in ipairs(periods) do
--!! I change grc to grk at -- m_a.show({'grc-' .. period})
		table.insert(fullProns, '* ' .. m_a.show({'grk-' .. period}) .. ' ' ..  m_IPA.format_IPA_full { lang = lang, items = {{pron = '/' .. IPAs.IPA .. '/'}}, } .. notes_full({{inline_notes = ' || ' .. {period .. '-note'} }}))
		periods2 = true
	end
	
	for _, period in ipairs(inlinePeriods) do
		if periods2 then
			local pron = '/' .. IPAs.IPA .. '/'
			table.insert(inlineProns, {pron = pron})
			table.insert(listOfProns, pron)
		end
	end
	
	for _, period in ipairs(periods) do
		if periods2 then
			local inline_notes = ' || ' .. {period .. '-note'}
		table.insert(inlineNotes, {inline_notes = inline_notes})
		table.insert(listOfNotes, inline_notes)
		end
	end
	
]=]--

	for _, period in ipairs(periods) do
--!! I change grc to grk at -- m_a.show({'grc-' .. period})
		table.insert(fullProns, '* ' .. m_a.show({'grk-' .. period}) .. ' ' ..  m_IPA.format_IPA_full { lang = lang, items = {{pron = '/' .. IPAs.IPA .. '/'}} })
		periods2 = true
	end
	
	for _, period in ipairs(inlinePeriods) do
		if periods2 then
			local pron = '/' .. IPAs.IPA .. '/'
			table.insert(inlineProns, {pron = pron})
			table.insert(listOfProns, pron)
		end
	end
--??  is this the length of IPA or the titles or both? I fixed the titles, small and balanced. 
--?? THE IPA length it is too big. Need autofit? and after it a note for eachline? 
--?? and manual like med1=xyz
	local inlineIPAlength = math.floor( math.max( mw.ustring.len("IPA(key): " .. table.concat(listOfProns, ' → ') or "") * 0.68, mw.ustring.len("(15th c. Medieval of Constantinople) IPA(key): /" .. IPAs.med2.IPA .. "/") * 0.68 ) )
	
	local inline = '\n<div class="vsShow" style="display:none">\n* ' .. m_IPA.format_IPA_full { lang = lang, items = inlineProns, separator = ' → ' } .. '</div>'
--!! ambiguous for cal
	local full = '\n<div class="vsHide">\n' .. table.concat(fullProns, '\n') .. make_ambig_note(ambig, ambig_letter_list) .. '</div>'
--!! I do not want More/Less hide/show
--!! take off switcher  -- <div class="vsSwitcher" 
--?? is the IPA length here? It is too big. And I need a note/per line. And a manual med2=xyz What is toggle... float right?? 
	return '<div data-toggle-category="pronunciations" style="width: ' .. inlineIPAlength .. 'em; max-width:100%;"><span class="vsToggleElement" style="float: right;">&nbsp;</span>' .. inline .. full .. '</div>'
end

--!! make period default = med1 at ], not here
function export.create(frame)
--?? if  == 'med1' then default = "med1" else {default = "med1"} end},
	local params = {
		 = {default = pagename},
		 = {default = "cla"},
	}
--!! change "grc-pronunciation", "create")
	local args = require("Module:parameters").process(frame.getParent and frame:getParent().args or frame, params, nil, "User:Sarri.greek/gkm-pronunciation", "create")
	
	local term = ulower(args)
	local old = term
	term = m_utils.standardDiacritics(term)
	term = mark_implied_length(term)
	--[[
	if mw.ustring.toNFD(old) ~= term then
		mw.log(old .. " > " .. term)
	end
	]]
	
	local decomposed = mw.ustring.toNFD(term)
	if rfind(decomposed, "" .. m_utils_data.diacritic .. "*") then
		error("Macrons and breves cannot be placed after the letters ε, ο, η, or ω.")
	end
	
	local ambig, ambig_letter_list
	if args.period == "cla" then
		ambig, ambig_letter_list = m_utils.findAmbig(term)
	end
	term = rsubn(term, 'ς', 'σ')
	term = rsubn(term, 'ῤ', 'ρ')
	term = rearrangeDiacritics(term)
	
	local IPAs, periods = convert_term(term, args.period)
	
	IPAs = syllabify(IPAs, periods)
	
	return make_table(IPAs, ambig, periods, ambig_letter_list)
end

function export.example(frame)
--!! adding little title without bullet
	local little_title = '<div style="width:500px; padding-top:0px; padding-bottom:0px; background:#fafafa; font-size:11px; line-height:105%;"><i>An approximation of non-dialectal pronunciation.</i> (? = uncertain or debated)</div>\n'

	local output = { '{| class="wikitable"' } --?? why cannot i put here '|' .. little_title?
	
	local params = {
		 = {}
	}
--!! changed  "grc-pronunciation", "example")	--!! it works WHATEVER i write
	local args = require("Module:parameters").process(frame:getParent().args, params, nil, "User:Sarri.greek/grk-pronunciation", "example")
	
	local terms = mw.text.split(args, ",%s+")
	
	for _, term in pairs(terms) do
--?? cla? I am not changing this, as everything works fine with it.		
		local period = rmatch(term, "%(period ?= ?(+)%)") or "cla" 
		local entry = rmatch(term, "(+) %(") or term or error('No term found in "' .. term .. '".') 
		local link = full_link(entry)
		local IPA = export.create{ entry,  = period }
--??	local periodnotes = ''
		table.insert(output, "\n|-\n| " .. link .. " || " .. IPA) --??  " || " .. periodnotes
	end
	
	table.insert(output, "\n|}")
	
	return table.concat(output)
end

return export
--Things we still need:
--Voicing of sigma around (after?) voiced stops. 
--Proper alerts for editors, especially on ambiguous vowels.

--?? that IPA is too long. Put notes at end like qual
--?? present lines/periods independently too
--?? el (perhaps med1 too) must have manual option for override (for loadwords)
--?? allow manuals e.g. med2=xxxx med2-note=as in Cretan dialect