Module:vot-pronunciation
Jump to navigation
Jump to search
- This module lacks a documentation subpage. Please create it.
- Useful links: subpage list • links • transclusions • testcases • sandbox
local export = {}
local m_vot = require("Module:vot")
local m_IPA = require("Module:IPA")
local gsub_lookahead = require("Module:gsub lookahead")
local lang = m_vot.lang
local U = mw.ustring.char
--- <<< DATA START >>> ---
local LONG = "ː"
local STRESS_PRIMARY = "ˈ"
local STRESS_SECONDARY = "ˌ"
local NEVER_STRESSED = "#"
local FRONTAL = U(0x0308)
local NONSYLLABIC = U(0x032F)
local TIE = U(0x0361)
local VERYSHORT = U(0x0306)
local SCHWA_BACK = U(0xEEE0)
local SCHWA_FRONT = U(0xEEE1)
local PALATAL = "ʲ"
local IPA_VOWELS = "ɑeiouyæøɤɨ" .. SCHWA_BACK .. SCHWA_FRONT
local AUTO_STRESS = U(0xEEEE)
local IPA_CONSONANTS = m_vot.consonants .. "ɫcčCɟɕʑɲʎïx"
local IPA_CONSONANTS_GEMINATABLE = m_vot.consonants_geminatable .. "ɫcčCɕʑɲʎx"
local PALATALIZE = m_vot.palatalize
local PALATALIZE_WEAK = '"'
local UNGEMINATE = "/"
local SHIFT_STRESS = "*"
local ANY_DIACRITICS = "[" .. U(0x0300) .. "-" .. U(0x036F) .. "]*"
local SOME_DIACRITICS = "[" .. U(0x0300) .. "-" .. U(0x036F) .. "]+"
local IPA_VOWEL = "[aõäöü" .. IPA_VOWELS .. "]"
local broken_vowel_sequences = { "i" .. UNGEMINATE .. "i" }
--- <<< DATA END >>> ---
--- <<< COMMON START >>> ---
local function split_syllables(word, keep_sep_symbols)
local res = {}
local syllable = ""
local pos = 1
local found_vowel = false
-- the following consonants stick together
while pos <= mw.ustring.len(word) do
if mw.ustring.find(mw.ustring.lower(word), "^[" .. IPA_CONSONANTS .. "][" .. PALATAL .. PALATALIZE .. PALATALIZE_WEAK .. "]?" .. IPA_VOWEL, pos) then
-- CV: end current syllable if we have found a vowel
if found_vowel then
if #syllable > 0 then table.insert(res, syllable) end
found_vowel = false
syllable = ""
end
syllable = syllable .. mw.ustring.sub(word, pos, pos)
pos = pos + 1
elseif mw.ustring.find(mw.ustring.lower(word), "^[" .. IPA_CONSONANTS .. "]", pos) then
-- C: continue
syllable = syllable .. mw.ustring.sub(word, pos, pos)
pos = pos + 1
elseif mw.ustring.find(mw.ustring.lower(word), "^" .. IPA_VOWEL, pos) then
if found_vowel then
-- already found a vowel, end current syllable
if #syllable > 0 then table.insert(res, syllable) end
syllable = ""
end
found_vowel = true
-- check for diphthongs or long vowels
local seq_ok = false
local seq_ok3 = false
for k, v in pairs(broken_vowel_sequences) do
if mw.ustring.find(mw.ustring.lower(word), "^" .. v, pos) then
seq_ok3 = true
break
end
end
if not seq_ok3 then
for k, v in pairs(m_vot.vowel_sequences) do
if mw.ustring.find(mw.ustring.lower(word), "^" .. v, pos) then
seq_ok = true
break
end
end
end
if seq_ok3 then
syllable = syllable .. mw.ustring.sub(word, pos, pos + 2)
pos = pos + 3
elseif seq_ok then
syllable = syllable .. mw.ustring.sub(word, pos, pos + 1)
pos = pos + 2
else
syllable = syllable .. mw.ustring.sub(word, pos, pos)
pos = pos + 1
end
elseif mw.ustring.find(mw.ustring.lower(word), "^[" .. PALATALIZE .. PALATALIZE_WEAK .. PALATAL .. "]", pos) then
syllable = syllable .. mw.ustring.sub(word, pos, pos)
pos = pos + 1
elseif mw.ustring.find(mw.ustring.lower(word), "^[" .. UNGEMINATE .. "]", pos) then
syllable = syllable .. UNGEMINATE
pos = pos + 1
elseif mw.ustring.find(mw.ustring.lower(word), "^[" .. m_vot.sep_symbols .. AUTO_STRESS .. NEVER_STRESSED .. "%*]", pos) then
-- separates syllables
if #syllable > 0 then
table.insert(res, syllable)
end
local sepchar = mw.ustring.sub(word, pos, pos)
syllable = keep_sep_symbols and sepchar or ""
pos = pos + 1
found_vowel = false
else
-- ?: continue
syllable = syllable .. mw.ustring.sub(word, pos, pos)
pos = pos + 1
end
end
if #syllable > 0 then
table.insert(res, syllable)
end
return res
end
local function zeroth_round_of_common_replacements(text, narrow)
text = mw.ustring.gsub(text, "'", PALATALIZE)
text = mw.ustring.gsub(text, PALATALIZE_WEAK, PALATAL)
if narrow then
text = mw.ustring.gsub(text, "l([aouõ])", "ɫ%1")
text = mw.ustring.gsub(text, "lɫ", "ɫɫ")
text = mw.ustring.gsub(text, "([^" .. STRESS_PRIMARY .. STRESS_SECONDARY .. "%*-]+)", function (word)
if mw.ustring.find(word, "[aouõ]") then
return mw.ustring.gsub(word, "l(.?)", function (v) return (mw.ustring.match(v, "[ieäöü" .. PALATALIZE .. "]") and "l" or "ɫ") .. v end)
else
return word
end
end)
end
text = mw.ustring.gsub(text, "tts", "cc")
text = mw.ustring.gsub(text, "ts", "c")
text = mw.ustring.gsub(text, "ttš", "čč")
text = mw.ustring.gsub(text, "tš", "č")
text = mw.ustring.gsub(text, PALATALIZE .. "(" .. m_vot.consonants .. PALATALIZE .. ")", "%1")
return text
end
local function first_round_of_common_replacements(text)
text = mw.ustring.gsub(text, "n([-" .. AUTO_STRESS .. "]?[kg])", "ŋ%1")
text = mw.ustring.gsub(text, "n([-" .. AUTO_STRESS .. "]?[pb])", "m%1")
text = mw.ustring.gsub(text, "[aäöõü’]", {
["a"] = "ɑ",
["ä"] = "æ",
["ö"] = "ø",
["õ"] = "ɤ",
["ü"] = "y",
-- ["-"] = STRESS_SECONDARY,
})
return text
end
local function second_round_of_common_replacements(text, narrow, apical)
text = mw.ustring.gsub(text, "%" .. SHIFT_STRESS, STRESS_PRIMARY)
text = mw.ustring.gsub(text, LONG .. PALATAL, PALATAL .. LONG)
text = mw.ustring.gsub(text, PALATAL .. "+", PALATAL)
text = mw.ustring.gsub(text, "[cčCgïjšž" .. SCHWA_BACK .. SCHWA_FRONT .. "]", {
["c"] = "t͡s",
["č"] = "t͡ʃ",
["C"] = "c",
["g"] = "ɡ",
["ï"] = narrow and "ʝ" or "j",
["j"] = narrow and "ʝ" or "j",
["š"] = apical and "ʂ" or "ʃ",
["ž"] = apical and "ʐ" or "ʒ",
[SCHWA_BACK] = "ə̠",
[SCHWA_FRONT] = "ə̟",
})
text = mw.ustring.gsub(text, "[lɫ]([lɫ])", "%1" .. LONG)
return text
end
local function automatic_palatalization(text, filter) -- , regressive_filter)
text = mw.ustring.gsub(text, "(" .. filter .. "+)([äöüi])", function (c1, v1)
return c1 .. PALATAL .. v1
end)
-- text = mw.ustring.gsub(text, "(" .. regressive_filter .. "+)([" .. IPA_CONSONANTS .. "]" .. PALATAL .. ")", function (c1, c2)
-- return c1 .. PALATAL .. c2
-- end)
return text
end
local full_palatal = {
["d"] = "ɟ", ["t"] = "C", -- workaround
["z"] = "ʑ", ["n"] = "ɲ", ["l"] = "ʎ", ["s"] = "ɕ"
}
local function manual_palatalization(text)
if not mw.ustring.find(text, PALATALIZE) then return text end
text = mw.ustring.gsub(text, "([" .. IPA_CONSONANTS .. "])" .. PALATALIZE, function(c)
return full_palatal[c] or c .. PALATAL
end)
text = mw.ustring.gsub(text, PALATALIZE, "")
text = mw.ustring.gsub(text, PALATAL .. PALATAL, PALATAL)
return text
end
local IPA_diphthongs = {
"[ɑeouɤæøy]i",
"[iouɤ]ɑ",
"[iøye]æ",
"[ɑoɤei]u",
"[ɑu]ɤ",
"[æøie]y",
"[ɑiæy]e",
"[ɑi]o",
}
local function long_vowels_and_diphthongs(text)
text = mw.ustring.gsub(text, "([" .. IPA_VOWELS .. "])%1", "%1" .. LONG)
for _, diphthong in ipairs(IPA_diphthongs) do
local mod_diphthong
if mw.ustring.find(diphthong, "%]$") then
mod_diphthong = mw.ustring.gsub(diphthong, "(.)(%[[^%]]-%])", "%1" .. VERYSHORT .. "?%2")
mod_diphthong = mw.ustring.gsub(diphthong, "(%[[^%]]-%])(%[[^%]]-%])", "%1" .. VERYSHORT .. "?%2")
else
mod_diphthong = mw.ustring.sub(diphthong, 1, -2) .. VERYSHORT .. "?" .. mw.ustring.sub(diphthong, -1, -1)
end
text = mw.ustring.gsub(text, "(" .. mod_diphthong .. ")", "%1" .. NONSYLLABIC)
end
return text
end
local function long_consonants(text)
text = mw.ustring.gsub(text, "(%a)" .. PALATAL .. "%1" .. PALATAL, "%1" .. PALATAL .. LONG)
text = mw.ustring.gsub(text, "(%a)%1", "%1" .. LONG)
text = mw.ustring.gsub(text, LONG .. PALATAL, PALATAL .. LONG)
return text
end
local function add_primary_stress(text)
text = mw.ustring.gsub(text, AUTO_STRESS, "-")
text = mw.ustring.gsub(text, "-%.", "-")
text = mw.ustring.gsub(text, "-", STRESS_SECONDARY)
text = STRESS_PRIMARY .. mw.ustring.gsub(text, " ", " " .. STRESS_PRIMARY)
text = mw.ustring.gsub(text, STRESS_PRIMARY .. "([^ ]+" .. STRESS_PRIMARY .. ")", "%1")
return mw.ustring.toNFC(text)
end
local function is_stressed_syllable(syllable)
return mw.ustring.find(syllable, "^[ " .. AUTO_STRESS .. "%*-]")
end
local function add_secondary_stress(syllables, stress_last)
local distance = 0
for index, syllable in ipairs(syllables) do
if not stress_last and index == #syllables then break end
local stressed = index == 1 or is_stressed_syllable(syllable)
if stressed then
distance = 0
else
distance = distance + 1
if distance == 2 then
distance = 0
if (index == #syllables or not is_stressed_syllable(syllables[index + 1])) and not mw.ustring.find(syllable, NEVER_STRESSED) then
syllables[index] = AUTO_STRESS .. syllable
end
end
end
end
end
local function clean_ungeminate(text)
return mw.ustring.gsub(text, UNGEMINATE, "")
end
local function do_gemination(syllables, diacritic)
local try_to_geminate = false
for index, syllable in ipairs(syllables) do
local stressed = index == 1 or is_stressed_syllable(syllable)
if try_to_geminate and not stressed then
-- check if the initial consonant in this syllable is followed by two vowels
local rest = syllable .. (syllables[index + 1] or "")
if mw.ustring.find(rest, "^[" .. IPA_CONSONANTS_GEMINATABLE .. "]" .. PALATALIZE .. "?" .. m_vot.vowel .. m_vot.vowel) then
-- CVCVV -> CVC:VV
local cg = select(3, mw.ustring.find(syllable, "^([" .. IPA_CONSONANTS_GEMINATABLE .. "]" .. PALATALIZE .. "?)"))
syllables[index - 1] = syllables[index - 1] .. cg
syllables[index] = mw.ustring.gsub(syllable, "^" .. cg, diacritic)
end
end
try_to_geminate = stressed and mw.ustring.find(syllable, "^[ " .. AUTO_STRESS .. "-]?[" .. IPA_CONSONANTS .. PALATALIZE .. TIE .. "]*" .. m_vot.vowel .. "$")
end
end
local function split_syllables_by_words(syllables)
local i = 1
return function()
local r = {}
local e = i
if e <= #syllables then
table.insert(r, (mw.ustring.gsub(syllables[e], "^%s+", "")))
e = e + 1
while e <= #syllables and not mw.ustring.find(syllables[e], "^%s") do
table.insert(r, syllables[e])
e = e + 1
end
i = e
return r
end
end
end
local function do_by_word_syllables(out_syllables, fn)
local old_syllables = {}
for k, v in pairs(out_syllables) do
old_syllables[k] = v
out_syllables[k] = nil
end
local next_word = false
for syllables in split_syllables_by_words(old_syllables) do
fn(syllables)
for i, syllable in ipairs(syllables) do
if next_word and i == 1 then
table.insert(out_syllables, " " .. syllable)
else
table.insert(out_syllables, syllable)
end
end
next_word = true
end
end
local function reduce_final_syllable(syl)
local allowed_finals = {
"(" .. m_vot.consonant .. ")%1",
"g[lɫnr]",
"mp",
"šk",
"lt"
}
if not mw.ustring.find(syl, m_vot.consonant .. m_vot.consonant .. PALATALIZE .. "?$") then
return syl
end
for _, allowed_final in ipairs(allowed_finals) do
if not mw.ustring.find(syl, allowed_final .. PALATALIZE .. "?$") then
return mw.ustring.gsub(syl, PALATALIZE .. "$", "")
end
end
return mw.ustring.sub(mw.ustring.gsub(syl, PALATALIZE .. "$", ""), 1, -2)
end
local function is_syllable_stressed_at(syllable, index)
return index == 1 or is_stressed_syllable(syllable)
end
local function do_reduction_word(syllables, narrow, reduce_completely)
local prev_was_stressed = false
local prev_was_long = false
local syllables_since_last_stressed = 0
local final_vowel_dropped = false
for index, syllable in ipairs(syllables) do
local stressed = is_syllable_stressed_at(syllable, index)
local final = index == #syllables
if stressed then
syllables_since_last_stressed = 0
else
syllables_since_last_stressed = syllables_since_last_stressed + 1
end
prev_was_long = prev_was_long
if not stressed and ((prev_was_stressed and prev_was_long) or (syllables_since_last_stressed > 1 or prev_was_long)) then
syllables[index] = mw.ustring.gsub(syllable, "(" .. m_vot.vowel .. "+)(.*)", function (nucleus, coda)
if mw.ustring.find(nucleus, "(" .. m_vot.vowel .. ")%1") then
return mw.ustring.sub(nucleus, 1, 1) .. coda
end
if not narrow then
local broad_reduce = { ["a"] = "õ", ["ä"] = "e" }
return (broad_reduce[nucleus] or nucleus) .. coda
end
--if mw.ustring.find(nucleus, "i[aä]") then
--return (syllable.find(PALATALIZE) and "" or PALATALIZE) .. mw.ustring.sub(nucleus, 2) .. coda
--end
if mw.ustring.find(nucleus, m_vot.vowel .. m_vot.vowel) then
return nucleus .. coda
end
local reduced = {
["a"] = SCHWA_BACK, ["ä"] = SCHWA_FRONT
}
if not reduced[nucleus] then
return nucleus .. coda
end
if final and reduce_completely and #coda < 1 and mw.ustring.match(nucleus, "[aä]") then
if mw.ustring.find(syllable, "j[aä]") then
return reduced[nucleus] .. VERYSHORT
end
final_vowel_dropped = true
return mw.ustring.find(nucleus, "ä") and PALATAL or ""
end
return (reduced[nucleus] or nucleus) .. coda
end)
end
-- reduce the next syllable only if the current syllable is stressed and heavy
prev_was_stressed = stressed
prev_was_long = mw.ustring.find(syllable, m_vot.vowel .. "[" .. IPA_CONSONANTS .. m_vot.vowels .. "]")
end
if final_vowel_dropped then
syllables[#syllables - 1] = reduce_final_syllable(syllables[#syllables - 1] .. syllables[#syllables])
syllables[#syllables] = nil
end
end
local function do_reduction(syllables, narrow, reduce_completely)
do_by_word_syllables(syllables, function(s) do_reduction_word(s, narrow, reduce_completely) end)
end
local diphthongize_broad = {
["e"] = "ie", ["o"] = "uo", ["ø"] = "yø"
}
local diphthongize_narrow = {
["e"] = "ɪ̆e", ["o"] = "ʊ̆o", ["ø"] = "ʏ̆ø"
}
local function do_diphthongization_word(syllables, narrow, reduce_completely)
for index, syllable in ipairs(syllables) do
local stressed = is_syllable_stressed_at(syllable, index)
syllables[index] = mw.ustring.gsub(syllable, "([eoø])%1", function (v)
return ((narrow and not stressed) and diphthongize_narrow or diphthongize_broad)[v]
end)
end
end
local function do_diphthongization(syllables, narrow)
do_by_word_syllables(syllables, function(s) do_diphthongization_word(s, narrow) end)
end
local function pass_diacritics_through(map, consonant)
local consonant, diacritics = mw.ustring.match(consonant, "([" .. IPA_CONSONANTS .. "])([" .. PALATAL .. "]?)")
return map[consonant] .. diacritics
end
local voiceless_sounds = "kptcčfsšh"
local function do_voicing(text, always_devoiced)
local devoice = { ["g"] = "k", ["b"] = "p", ["d"] = "t", ["z"] = "s", ["ž"] = "š", ["ʑ"] = "ɕ" }
local semivoice = { ["g"] = "g̊", ["b"] = "b̥", ["d"] = "d̥", ["z"] = "z̥", ["ž"] = "ž̥", ["ʑ"] = "ɕ̊" }
if always_devoiced then semivoice = devoice end
local consonants_to_devoice = "[bdgzž][" .. PALATAL .. "]?"
local vowel = "[" .. IPA_VOWELS .. "]"
-- b/d/g/z/ž is semivoiced if it is not followed by anything
text = mw.ustring.gsub(text, "(" .. consonants_to_devoice .. ")$",
function (consonant)
return pass_diacritics_through(semivoice, consonant)
end)
-- b/d/g/z/ž is devoiced if it is followed by a voiceless sound
text = gsub_lookahead(text, "(" .. consonants_to_devoice .. ")([%s" .. AUTO_STRESS .. "-]+)([" .. voiceless_sounds .. "])",
function (consonant, space, after)
return pass_diacritics_through(devoice, consonant) .. space, after
end)
return text
end
local palatalize_filter = "[dfghkmnprstvz]"
local kattila_palatalize_filter = "[dlnrstz]"
-- local regressive_palatalize_filter = "[dnrstz]"
--- <<< COMMON END >>> ---
--- <<< DIALECTS START >>> ---
-- narrow_level 0 = broad, 1 = rhyme, 2 = narrow
-- Luutsa, Liivtšülä
local function IPA_luutsa_liivtsula(text, narrow_level)
text = zeroth_round_of_common_replacements(text, narrow_level > 1)
if narrow_level > 0 then
local syllables = split_syllables(text, true)
add_secondary_stress(syllables)
text = table.concat(syllables)
end
text = mw.ustring.gsub(text, NEVER_STRESSED, "")
local syllables = split_syllables(text, true)
if narrow_level > 1 then
do_gemination(syllables, LONG)
do_reduction(syllables, true, false)
end
text = table.concat(syllables)
if narrow_level > 0 then text = do_voicing(text) end
if narrow_level > 1 then
text = automatic_palatalization(text, palatalize_filter) -- , regressive_palatalize_filter) -- palatalization
text = mw.ustring.gsub(text, "h([kg])", "x%1")
end
text = clean_ungeminate(text)
text = mw.ustring.gsub(text, "j" .. PALATALIZE, PALATALIZE)
text = manual_palatalization(text)
text = first_round_of_common_replacements(text)
text = long_vowels_and_diphthongs(text)
text = long_consonants(text)
text = second_round_of_common_replacements(text, narrow_level > 1)
return add_primary_stress(text)
end
-- Jõgõperä
local function IPA_jogopera(text, narrow_level)
text = zeroth_round_of_common_replacements(text, narrow_level > 1)
if narrow_level > 0 then
local syllables = split_syllables(text, true)
add_secondary_stress(syllables)
text = table.concat(syllables)
end
text = mw.ustring.gsub(text, NEVER_STRESSED, "")
local syllables = split_syllables(text, true)
if narrow_level > 1 then
do_gemination(syllables, LONG)
do_reduction(syllables, true, true)
end
text = table.concat(syllables)
if narrow_level > 0 then text = do_voicing(text) end
if narrow_level > 1 then
text = automatic_palatalization(text, palatalize_filter) -- , regressive_palatalize_filter) -- palatalization
text = mw.ustring.gsub(text, "h([kg])", "x%1")
end
text = clean_ungeminate(text)
text = mw.ustring.gsub(text, "j" .. PALATALIZE, PALATALIZE)
text = manual_palatalization(text)
text = first_round_of_common_replacements(text)
text = long_vowels_and_diphthongs(text)
text = long_consonants(text)
text = second_round_of_common_replacements(text, narrow_level > 1)
return add_primary_stress(text)
end
-- Kattila
local function IPA_kattila(text, narrow_level)
text = zeroth_round_of_common_replacements(text, narrow_level > 1)
if narrow_level > 0 then
local syllables = split_syllables(text, true)
add_secondary_stress(syllables, true)
text = table.concat(syllables)
end
text = mw.ustring.gsub(text, NEVER_STRESSED, "")
local syllables = split_syllables(text, true)
if narrow_level > 1 then
do_gemination(syllables, LONG)
end
text = table.concat(syllables)
if narrow_level > 0 then text = do_voicing(text, true) end
if narrow_level > 1 then
text = mw.ustring.gsub(text, "h([kpt])", "H%1")
text = mw.ustring.gsub(text, "[hH]", {["h"] = "ɦ", ["H"] = "h"})
end
text = clean_ungeminate(text)
text = mw.ustring.gsub(text, "j" .. PALATALIZE, PALATALIZE)
text = manual_palatalization(text)
text = first_round_of_common_replacements(text)
if narrow_level > 0 then
local syllables = split_syllables(text, true)
do_diphthongization(syllables, narrow_level > 1)
text = table.concat(syllables)
end
text = long_vowels_and_diphthongs(text)
text = long_consonants(text)
text = second_round_of_common_replacements(text, narrow_level > 1, true)
return add_primary_stress(text)
end
--- <<< DIALECTS END >>> ---
--- <<< INTERFACE START >>> ---
local function cleanup_for_hyphenate(text)
local no_hyph_symbols = "[" .. UNGEMINATE .. "%-]"
return mw.ustring.gsub(mw.ustring.gsub(text, no_hyph_symbols, ""), "%*", ".")
end
local function run_reductions(text)
local syllables = split_syllables(text, true)
add_secondary_stress(syllables)
local prev_was_stressed = false
local prev_was_long = false
local syllables_since_last_stressed = 0
for index, syllable in ipairs(syllables) do
local stressed = is_syllable_stressed_at(syllable, index)
local final = index == #syllables
if stressed then
syllables_since_last_stressed = 0
else
syllables_since_last_stressed = syllables_since_last_stressed + 1
end
prev_was_long = prev_was_long
if not stressed and ((prev_was_stressed and prev_was_long) or (syllables_since_last_stressed > 1 or prev_was_long)) then
syllables[index] = mw.ustring.gsub(syllable, "(" .. m_vot.vowel .. "+)(.*)", function (nucleus, coda)
if mw.ustring.find(nucleus, "(" .. m_vot.vowel .. ")%1") then
return mw.ustring.sub(nucleus, 1, 1) .. coda
end
local broad_reduce = { ["a"] = "õ", ["ä"] = "e" }
return (broad_reduce[nucleus] or nucleus) .. coda
end)
end
-- reduce the next syllable only if the current syllable is stressed and heavy
prev_was_stressed = stressed
prev_was_long = mw.ustring.find(syllable, m_vot.vowel .. "[" .. IPA_CONSONANTS .. m_vot.vowels .. "]")
end
return mw.ustring.gsub(table.concat(syllables, ""), "[" .. NEVER_STRESSED .. AUTO_STRESS .. "]", "")
end
local function match_spelling_with_title_for_hyphenation(sp, title)
return title
end
local function hyphenate_matches(sp, title)
local resp = run_reductions(mw.ustring.lower(mw.ustring.gsub(sp, "%*", ".")))
resp = mw.ustring.gsub(resp, "'", PALATALIZE)
resp = mw.ustring.gsub(resp, "([bdgzž])([kpsšt])", function(c1, c2) return ({b = "p", d = "t", g = "k", z = "s", ["ž"] = "š"})[c1] .. c2 end)
resp = mw.ustring.gsub(cleanup_for_hyphenate(resp), "%.", "")
return resp == mw.ustring.lower(title)
end
local function hyphenate(text)
return m_vot.split_syllables(cleanup_for_hyphenate(text))
end
local function spell_long_consonants(text)
text = mw.ustring.gsub(text, "(t[sš])" .. "(" .. PALATALIZE .. "?)" .. LONG,
function (c, p) return "t" .. c .. p end)
text = mw.ustring.gsub(text, "([" .. m_vot.consonants .. "])" .. "(" .. PALATALIZE .. "?)" .. LONG,
function (c, p) return c .. c .. p end)
text = mw.ustring.gsub(text, "iï", "i")
return text
end
local function generate_rhyme(tuple)
local text = tuple.rhyme
local index = mw.ustring.find(text, "[" .. STRESS_PRIMARY .. STRESS_SECONDARY .. "][^" .. STRESS_PRIMARY .. STRESS_SECONDARY .. "]*$")
if index ~= nil then text = mw.ustring.sub(text, index + 1) end
index = mw.ustring.find(text, "[" .. IPA_VOWELS .. "]")
if index == nil then return nil end
return mw.ustring.sub(text, index)
end
local function make_IPAs(fn, forms, variety)
local p = {}
for _, form in ipairs(forms) do
form = mw.ustring.lower(form)
local suffix = mw.ustring.find(form, "^%-")
local prefix = mw.ustring.find(form, "%-$")
if suffix then form = mw.ustring.gsub(form, "^%-", "") end
if prefix then form = mw.ustring.gsub(form, "%-$", "") end
local broad = fn(form, 0)
local rhyme = fn(form, 1)
local narrow = fn(form, 2)
if prefix then
broad = broad .. "-"
rhyme = nil
narrow = narrow .. "-"
end
if suffix then
broad = "-" .. mw.ustring.gsub(broad, "^" .. STRESS_PRIMARY, "")
rhyme = nil
narrow = "-" .. mw.ustring.gsub(narrow, "^" .. STRESS_PRIMARY, "")
end
table.insert(p, { broad = broad, rhyme = rhyme, narrow = narrow })
end
local result = {
forms = p,
varieties = { variety }
}
return result
end
local function format_IPAs(tuple, title, has_spaces)
local dialects = require("Module:accent qualifier").format_qualifiers(tuple.varieties)
local p = {}
for _, form in ipairs(tuple.forms) do
table.insert(p, {pron = "/" .. form.broad .. "/"})
table.insert(p, {pron = "[" .. form.narrow .. "]"})
end
return "* " .. dialects .. " " .. m_IPA.format_IPA_full(lang, p, nil, nil, nil, has_spaces)
end
local function get_arg_list(param, fallback, allow_dash, required)
if not param or #param == 0 then return required and fallback or nil end
if not allow_dash and #param == 1 and param[1] == "-" then return nil end
if #param == 1 and param[1] == "+" then return fallback end
return param
end
local varieties = {
["Lu"] = { "Luutsa", IPA_luutsa_liivtsula },
["Li"] = { "Liivtšülä", IPA_luutsa_liivtsula },
["J"] = { "Jõgõperä", IPA_jogopera },
["K"] = { "Kattila", IPA_kattila },
}
local variety_groups = {
{ "LL", {"Lu", "Li"}, true },
{ nil, "J", false },
{ nil, "K", false },
}
local varieties_merged = {}
for _, group in ipairs(variety_groups) do
if group[1] then
varieties_merged[group[1]] = group[2]
end
end
local function get_variety(variety_code)
if varieties[variety_code] then
return varieties[variety_code]
end
if varieties_merged[variety_code] then
local subvarieties = varieties_merged[variety_code]
local names = {}
local fn = nil
for _, subvariety_code in ipairs(subvarieties) do
local subvariety = varieties[subvariety_code]
fn = subvariety[2]
table.insert(names, subvariety[1])
end
return {table.concat(names, ", "), fn}
end
error("Unrecognized variety code: " .. variety_code)
end
function export.get_variety(variety_code)
return get_variety(variety_code)[1]
end
function export.generate_one(form, variety_code, transcription)
local name, fn = unpack(get_variety(variety_code))
local result = make_IPAs(fn, {form}, name).forms[1]
if transcription then result = result[transcription] end
return result
end
function export.generate_multiple(forms, variety_code, transcription)
local param, name, fn = unpack(get_variety(variety_code))
local result = make_IPAs(fn, forms, name).forms
if transcription then
for i, form in ipairs(result) do
result[i] = form[transcription]
end
end
return result
end
local function add_IPAs(IPAs, spellings, main_code, args, required)
local name, fn = unpack(get_variety(main_code))
local forms = get_arg_list(args, spellings, false, required)
if forms then
table.insert(IPAs, make_IPAs(fn, forms, name))
end
end
function export.show(frame)
local title = mw.title.getCurrentTitle().text
local hyphenation = nil
local rhymes = nil
local categories = {}
local params = {
[1] = { list = true },
["LL"] = { list = true }, -- Luutsa-Liivtšülä
["Lu"] = { list = true }, -- Luutsa
["Li"] = { list = true }, -- Liivtšülä
["J"] = { list = true }, -- Jõgõperä
["K"] = { list = true }, -- Kattila,
["dial"] = { type = "boolean" },
["title"] = {}, -- for debugging or demonstration only
}
local args = require("Module:parameters").process(frame:getParent().args, params)
title = args["title"] or title
local dialectal = args.dial
local spellings = get_arg_list(args[1], { mw.ustring.lower(title) }, true, true)
local IPAs = {}
for _, variety_group in ipairs(variety_groups) do
local param, codes, always = unpack(variety_group)
if param then
local split = false
for _, code in ipairs(codes) do
if #args[code] > 0 then
split = true
break
end
end
if split then
for _, code in ipairs(codes) do
add_IPAs(IPAs, spellings, code, args[code] or (param and args[param] or nil), always and not dialectal)
end
else
add_IPAs(IPAs, spellings, param, args[param], always and not dialectal)
end
else
add_IPAs(IPAs, spellings, codes, args[codes], always and not dialectal)
end
end
if #IPAs < 1 then
error("No dialects to display IPA for")
end
local results = {}
local has_spaces = mw.ustring.find(title, " ")
for _, tuple in ipairs(IPAs) do
table.insert(results, format_IPAs(tuple, title, has_spaces))
end
if not hyphenation then
hyphenation = {}
if not has_spaces then
local sp = spellings[1]
if not hyphenate_matches(sp, title) then
-- try to geminate
local syllables = m_vot.split_syllables(sp, true)
do_gemination(syllables, LONG)
sp = spell_long_consonants(clean_ungeminate(table.concat(syllables)))
end
if hyphenate_matches(sp, title) then
table.insert(hyphenation, hyphenate(match_spelling_with_title_for_hyphenation(sp, title)))
end
end
end
if not rhymes then
rhymes = {}
if not has_spaces then
local found_rhymes = {}
for _, tuple in ipairs(IPAs) do
for _, form in ipairs(tuple.forms) do
if form.rhyme then
local rhyme = generate_rhyme(form)
if not found_rhymes[rhyme] then
found_rhymes[rhyme] = true
table.insert(rhymes, rhyme)
end
end
end
end
end
end
if #rhymes > 0 then
local sylkeys = {}
local sylcounts = {}
-- get all possible syllable counts from syllabifications
for i, h in ipairs(hyphenation) do
local hl = #h
if hl > 0 and not sylkeys[hl] then
table.insert(sylcounts, hl)
sylkeys[hl] = true
end
end
local rhymeobjs = {}
for _, rhyme in ipairs(rhymes) do
table.insert(rhymeobjs, {rhyme = rhyme})
end
table.insert(results, "* " .. require("Module:rhymes").format_rhymes(
{ lang = lang, rhymes = rhymeobjs, num_syl = sylcounts }))
end
if #hyphenation > 0 then
local hyphs = {}
for i, h in ipairs(hyphenation) do
table.insert(hyphs, { ["hyph"] = h })
end
table.insert(results, "* " .. require("Module:hyphenation").format_hyphenations(
{ lang = lang, hyphs = hyphs, caption = "Hyphenation" }))
end
return table.concat(results, "\n") .. require("Module:utilities").format_categories(categories, lang)
end
--- <<< INTERFACE END >>> ---
return export