Module:fi-hyphenation

local export = {}

local lang = "fi"
local sc = "Latn"
local m_str_utils = require("Module:string utilities")

local gsub = m_str_utils.gsub
local find = m_str_utils.find
local len = m_str_utils.len
local lower = m_str_utils.lower
local match = m_str_utils.match
local sub = m_str_utils.sub

local vowels = "aeiouyåäö"
local vowel = ""
local consonants = "bcdfghjklmnpqrstvwxzšžʔ*"
local consonant = ""
 
-- orthographic symbols that signify separation of syllables
local sep_symbols = "'’./ +\"-"
-- these signify that the next syllable is an "initial" syllable in a word
-- all symbols from here should also be in sep_symbols
local stressed_symbols = "/ +-"
export.sep_symbols = sep_symbols

-- diphthongs and long vowels
-- in initial syllables
local vowel_sequences_initial = {
	"i",
	"u",
	"y",
	"uo",
	"ie",
	"yö",
	"aa", "ee", "ii", "oo", "uu", "yy", "ää", "öö"
}
-- in non-initial syllables
-- further, diphthongs ending _u or _y are diphthongs only
-- in non-initial syllables if the syllable is open
local vowel_sequences_noninitial = {
	"i",
	"aa", "ee", "ii", "oo", "uu", "yy", "ää", "öö"
}
-- in non-initial *open* syllables, in addition to above
local vowel_sequences_noninitial_open = {
	"u",
	"y"
}

-- sonority by consonant (higher number is more sonorous)
local sonorities = {
	 = -1,  = -1,  = -1,  = -1,  = -1,
	 = -2,  = -2,  = -2,
	 = -3,  = -3,  = -3,
	 = -4,  = -4,  = -4,
	 = -5,  = -5,  = -5,
	 = -6,  = -6,  = -6
}

local function split_by_sonority(cluster)
	local i = len(cluster)
	-- always split two-consonant clusters evenly
	if i <= 2 then
		return i - 1
	end
	
	-- always split 'native' three-consonant clusters to AB.C
	if i == 3 then
		if match(cluster, "t$") or match(cluster, "s$") then
			native = match(cluster, "^")
		elseif match(cluster, "kk$") then
			native = match(cluster, "^")
		elseif match(cluster, "pp$") then
			native = match(cluster, "^")
		end
		if native then return 2 end
	end

	local ng = find(cluster, "ng")
	-- never split ng
	if ng then
		return ng + 1 + split_by_sonority(sub(cluster, ng + 2))
	end
	
	local max_sonority = 0
	while i > 1 do
		local sonority = sonorities
		if sonority == nil then return len(cluster) - 1 end
		if sonority >= max_sonority then break end
		max_sonority = sonority
		i = i - 1
	end
	return i
end

-- mode may be one of the following:
--		default: naive hyphenation using Finnish hyphenation rules
--		"dots": use default rules, but add dots between syllable boundaries if allowed by keep_sep_symbols
--		"sonority": split consonant clusters according to the sonority sequencing principle
--				add dots for new splits (i.e. not those incurred by symbols) followed by multiple consonants if keep_sep_symbols allows it.
--		"allow_diphthongs_everywhere": only for backwards compatibility with {{fi-hyphenation}}
function export.generate_hyphenation(word, keep_sep_symbols, mode)
	local res = {}
	local syllable = ""
	local pos = 1
	local found_vowel = false
	local initial_syllable = true
	local lower_word = lower(word)

	local dots = (mode == "dots") and (keep_sep_symbols == true or (type(keep_sep_symbols) == "string" and find(keep_sep_symbols, ".")))
	local sonority = mode == "sonority"
	local allow_diphthongs_everywhere = mode == "allow_diphthongs_everywhere"
	
	while pos <= #word do
		if match(lower_word, "^" .. consonant .. vowel, pos) then
			-- CV: end current syllable if we have found a vowel
			if found_vowel then
				if syllable and #syllable > 0 then
					table.insert(res, syllable)
					initial_syllable = false
				end
				
				found_vowel = false
				syllable = dots and "." or ""
			end
			syllable = syllable .. sub(word, pos, pos)
			pos = pos + 1
		elseif sonority and found_vowel and #syllable > 0 and match(lower_word, "^" .. consonant .. "+" .. vowel, pos) then
			-- (V)C+V: split by sonority
			local cluster = match(lower_word, "^" .. consonant .. "+", pos)
			local split_point = split_by_sonority(cluster)

			if split_point ~= nil then
				-- split at the specified position.
				syllable = syllable .. sub(cluster, 1, split_point)
				table.insert(res, syllable)
				initial_syllable = false

				local len_cluster = len(cluster)
				-- add a dot if dots are allowed, and if the split point is not before the final consonant
				local add_dot = (keep_sep_symbols == true or (type(keep_sep_symbols) == "string" and find(keep_sep_symbols, "."))) and split_point ~= len_cluster - 1
				syllable = (add_dot and "." or "") .. sub(cluster, split_point + 1)
				pos = pos + len_cluster
				found_vowel = false
			end

		elseif match(lower_word, "^" .. consonant, pos) then
			-- C: continue
			syllable = syllable .. sub(word, pos, pos)
			pos = pos + 1
		elseif match(lower_word, "^" .. vowel, pos) then
			if found_vowel then
				-- already found a vowel, end current syllable
				if syllable and #syllable > 0 then
					table.insert(res, syllable)
					initial_syllable = false
				end
				syllable = dots and "." or ""
			end	
			found_vowel = true
			
			-- check for diphthongs or long vowels
			local vowel_sequences = (allow_diphthongs_everywhere or initial_syllable) and vowel_sequences_initial or vowel_sequences_noninitial
			local seq_ok = false
			for k, v in pairs(vowel_sequences) do
				if match(lower_word, "^" .. v, pos) then
					seq_ok = true
					break
				end
			end
			
			if not seq_ok and not initial_syllable then
				for k, v in pairs(vowel_sequences_noninitial_open) do
					if match(lower_word, "^" .. v .. "", pos) or match(lower_word, "^" .. v .. "", pos) then
						seq_ok = true
						break
					end
				end
			end
			
			if seq_ok then
				syllable = syllable .. sub(word, pos, pos + 1)
				pos = pos + 2
			else
				syllable = syllable .. sub(word, pos, pos)
				pos = pos + 1
			end
		else
			local sepchar = match(lower_word, "^%(?%)?", pos)
			if sepchar then
				-- separates syllables
				if syllable and #syllable > 0 then
					table.insert(res, syllable)
				end
				
				initial_syllable = match(sepchar, "^%(?%)?")
				syllable = (keep_sep_symbols == true or (type(keep_sep_symbols) == "string" and find(keep_sep_symbols, sub(word, pos, pos)))) and sepchar or ""
				pos = pos + len(sepchar)
				found_vowel = false
			else
				-- ?: continue
				syllable = syllable .. sub(word, pos, pos)
				pos = pos + 1
			end
		end
	end
	
	if syllable and #syllable > 0 then
		table.insert(res, syllable)
	end
	
	return res
end

function export.hyphenation(frame)
	local title = mw.title.getCurrentTitle().text
	
	if type(frame) == "table" then
		local params = {
			 = {list = true, default = nil},
			
			 = {},
			 = {},
		}
		
		local args = require("Module:parameters").process(frame:getParent().args, params)
		
		hyphenation = args
		title = args or (args or title)
	end
	
	if not hyphenation or #hyphenation < 1 then
		hyphenation = export.generate_hyphenation(title, false, "diphthongs_everywhere")
	end
	
	local text = require("Module:links").full_link({lang = require("Module:languages").getByCode(lang), alt = table.concat(hyphenation, "‧"), tr = "-"})
	return "Hyphenation: " .. text
end

return export