Module:User:Kiril kovachev/tok-hyph

The following documentation is located at Module:User:Kiril kovachev/tok-hyph/documentation.

Useful links: root page • root page’s subpages • links • transclusions • testcases • sandbox

Syllabifies a Toki Pona word.

E.g. for sitelen: Syllabification: si‧te‧len.

Usage

Basic usage: {{tok-hyph}}.

Optionally, you can pass a word to be syllabified instead of the page title: {{tok-hyph|sitelen}}.

Tests

All tests passed. (refresh)

test_syllabification:
Text	Expected	Actual
a	a	a
akesi	a‧ke‧si	a‧ke‧si
ala	a‧la	a‧la
alasa	a‧la‧sa	a‧la‧sa
ale	a‧le	a‧le
ali	a‧li	a‧li
anpa	an‧pa	an‧pa
ante	an‧te	an‧te
anu	a‧nu	a‧nu
awen	a‧wen	a‧wen
e	e	e
en	en	en
epiku	e‧pi‧ku	e‧pi‧ku
esun	e‧sun	e‧sun
ijo	i‧jo	i‧jo
ike	i‧ke	i‧ke
ilo	i‧lo	i‧lo
insa	in‧sa	in‧sa
jaki	ja‧ki	ja‧ki
jan	jan	jan
jasima	ja‧si‧ma	ja‧si‧ma
jelo	je‧lo	je‧lo
jo	jo	jo
kala	ka‧la	ka‧la
kalama	ka‧la‧ma	ka‧la‧ma
kama	ka‧ma	ka‧ma
kasi	ka‧si	ka‧si
ken	ken	ken
kepeken	ke‧pe‧ken	ke‧pe‧ken
kijetesantakalu	ki‧je‧te‧san‧ta‧ka‧lu	ki‧je‧te‧san‧ta‧ka‧lu
kili	ki‧li	ki‧li
kin	kin	kin
kipisi	ki‧pi‧si	ki‧pi‧si
kiwen	ki‧wen	ki‧wen
ko	ko	ko
kokosila	ko‧ko‧si‧la	ko‧ko‧si‧la
kon	kon	kon
ku	ku	ku
kule	ku‧le	ku‧le
kulupu	ku‧lu‧pu	ku‧lu‧pu
kute	ku‧te	ku‧te
la	la	la
lanpan	lan‧pan	lan‧pan
lape	la‧pe	la‧pe
laso	la‧so	la‧so
lawa	la‧wa	la‧wa
leko	le‧ko	le‧ko
len	len	len
lete	le‧te	le‧te
li	li	li
lili	li‧li	li‧li
linja	lin‧ja	lin‧ja
lipu	li‧pu	li‧pu
loje	lo‧je	lo‧je
lon	lon	lon
luka	lu‧ka	lu‧ka
lukin	lu‧kin	lu‧kin
lupa	lu‧pa	lu‧pa
ma	ma	ma
mama	ma‧ma	ma‧ma
mani	ma‧ni	ma‧ni
meli	me‧li	me‧li
meso	me‧so	me‧so
mi	mi	mi
mije	mi‧je	mi‧je
misikeke	mi‧si‧ke‧ke	mi‧si‧ke‧ke
moku	mo‧ku	mo‧ku
moli	mo‧li	mo‧li
monsi	mon‧si	mon‧si
monsuta	mon‧su‧ta	mon‧su‧ta
mu	mu	mu
mun	mun	mun
musi	mu‧si	mu‧si
mute	mu‧te	mu‧te
n	n	n
namako	na‧ma‧ko	na‧ma‧ko
nanpa	nan‧pa	nan‧pa
nasa	na‧sa	na‧sa
nasin	na‧sin	na‧sin
nena	ne‧na	ne‧na
ni	ni	ni
nimi	ni‧mi	ni‧mi
noka	no‧ka	no‧ka
o	o	o
oko	o‧ko	o‧ko
olin	o‧lin	o‧lin
ona	o‧na	o‧na
open	o‧pen	o‧pen
pakala	pa‧ka‧la	pa‧ka‧la
pali	pa‧li	pa‧li
palisa	pa‧li‧sa	pa‧li‧sa
pan	pan	pan
pana	pa‧na	pa‧na
pi	pi	pi
pilin	pi‧lin	pi‧lin
pimeja	pi‧me‧ja	pi‧me‧ja
pini	pi‧ni	pi‧ni
pipi	pi‧pi	pi‧pi
poka	po‧ka	po‧ka
poki	po‧ki	po‧ki
pona	po‧na	po‧na
pu	pu	pu
sama	sa‧ma	sa‧ma
seli	se‧li	se‧li
selo	se‧lo	se‧lo
seme	se‧me	se‧me
sewi	se‧wi	se‧wi
sijelo	si‧je‧lo	si‧je‧lo
sike	si‧ke	si‧ke
sin	sin	sin
sina	si‧na	si‧na
sinpin	sin‧pin	sin‧pin
sitelen	si‧te‧len	si‧te‧len
soko	so‧ko	so‧ko
sona	so‧na	so‧na
soweli	so‧we‧li	so‧we‧li
suli	su‧li	su‧li
suno	su‧no	su‧no
supa	su‧pa	su‧pa
suwi	su‧wi	su‧wi
tan	tan	tan
taso	ta‧so	ta‧so
tawa	ta‧wa	ta‧wa
telo	te‧lo	te‧lo
tenpo	ten‧po	ten‧po
toki	to‧ki	to‧ki
tomo	to‧mo	to‧mo
tonsi	ton‧si	ton‧si
tu	tu	tu
unpa	un‧pa	un‧pa
uta	u‧ta	u‧ta
utala	u‧ta‧la	u‧ta‧la
walo	wa‧lo	wa‧lo
wan	wan	wan
waso	wa‧so	wa‧so
wawa	wa‧wa	wa‧wa
weka	we‧ka	we‧ka
wile	wi‧le	wi‧le

-- Primary module authorship: Chernorizets (original Bulgarian syllabification code)
-- Port to Lua: Kiril Kovachev
-- Adaptation to Toki Pona: Kiril Kovachev
-- 17 April 2024.

local export = {}

local substring = mw.ustring.sub
local rsubn = mw.ustring.gsub
local rsplit = mw.text.split
local U = mw.ustring.char
local lang = require("Module:languages").getByCode("tok")
local script = require("Module:scripts").getByCode("Latn")

local hvowels_c = ""

local HYPH = U(0x2027)
local BREAK_MARKER = "."

-- 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 function char_at(str, index)
	return substring(str, index, index)
end

local function count_vowels(word)
	local _, vowel_count = mw.ustring.gsub(word, hvowels_c, "")
	return vowel_count
end

local function is_vowel(ch)
	for _, v in pairs{"a", "e", "i", "o", "u"} do
		if v == ch then
			return true
		end
	end
	return false
end

---- Main syllabification code
-- word: the word being scanned
-- left/right vowels: integers
local function find_next_syllable_onset(word, left_vowel, right_vowel)
    local n_cons = right_vowel - left_vowel - 1

    -- No consonants - syllable starts on rightVowel
    if n_cons == 0 then return right_vowel end

    -- Single consonant between two vowels - starts a syllable
    if n_cons == 1 then return left_vowel + 1 end

    -- Two ("or more") consonants between the vowels.
    -- In Toki Pona, the phonotactics only allow this if the first syllable
    -- ends in a nasal and the second begins in a consonant, so there can only
    -- ever be two consonants, and the sonority break occurs between the two
    -- consonants (so just add 2 to the left vowel).
	local sonority_break = left_vowel + 2

    return sonority_break
end

-- Returns a table of strings (list)
local function syllabify_poly(word)
    local syllables = {}

    local prev_vowel = -1
    local prev_onset = 1;
    
    for i = 1, mw.ustring.len(word) do
	    if is_vowel(mw.ustring.lower(char_at(word, i))) then
	        -- A vowel, yay!
	        local should_skip = false
	        if prev_vowel == -1 then
	            prev_vowel = i
	            should_skip = true;
	        end

	        -- This is not the first vowel we've seen. In-between
	        -- the previous vowel and this one, there is a syllable
	        -- break, and the first character after the break starts
	        -- a new syllable.
	        if not should_skip then
		        local next_onset = find_next_syllable_onset(word, prev_vowel, i)
		        table.insert(syllables, substring(word, prev_onset, next_onset - 1))
		        prev_vowel = i
		        prev_onset = next_onset
			end
	    end
    	
    end

    -- Add the last syllable
    table.insert(syllables, substring(word, prev_onset))

    return syllables
end

function export.syllabify_word(word)
    if mw.ustring.len(word) == 0 then return {} end;

    local n_vowels = count_vowels(word)
    local syllables = n_vowels <= 1 and {word} or syllabify_poly(word)

    return table.concat(syllables, HYPH)
end

function export.syllabify(term)
	local words = rsplit(term, " ")

	local out = {}
	for _, word in pairs(words) do
		table.insert(out, export.syllabify_word(word))	
	end
	return table.concat(out, " ")
end

function export.show_syllabification(frame)
	local params = {
		 = {},
	}
	
	local title = mw.title.getCurrentTitle()
	local args = require("Module:parameters").process(frame:getParent().args, params)
	local term = args or title.nsText == "Template" and "sitelen" or title.text

	local syllabification = export.syllabify(term)
	local syllables = rsplit(syllabification, HYPH)
	return require("Module:hyphenation").format_hyphenations(
		{ 
			lang = lang,
			hyphs = { { hyph = syllables } },
			sc = script,
			caption = "Syllabification",
		}
	)
end

return export

Module:User:Kiril kovachev/tok-hyph

Usage

Tests

Wikious

Boobota

Sagapedia