Module:User:AmazingJus/lo
- This module lacks a documentation subpage. You may create it.
- Useful links: root page • root page’s subpages • links • transclusions • testcases • user page • user talk page • userspace
This is a private module sandbox of AmazingJus, for his own experimentation. Items in this module may be added and removed at AmazingJus's discretion; do not rely on this module's stability.
local export = {}
local find = mw.ustring.find
local len = mw.ustring.len
local match = mw.ustring.match
local gmatch = mw.ustring.gmatch
local gsub = mw.ustring.gsub
local sub = mw.ustring.sub
-- Mapping of initial consonants.
local initial_conv = {
['ກ'] = 'k', ['ຂ'] = 'kh', ['ຄ'] = 'kh', ['ງ'] = 'ng',
['ຈ'] = 'ch', ['ສ'] = 's', ['ຊ'] = 's', ['ຍ'] = 'ny',
['ດ'] = 'd', ['ຕ'] = 't', ['ຖ'] = 'th', ['ທ'] = 'th', ['ນ'] = 'n',
['ບ'] = 'b', ['ປ'] = 'p', ['ຜ'] = 'ph', ['ຝ'] = 'f', ['ພ'] = 'ph', ['ຟ'] = 'f', ['ມ'] = 'm',
['ຢ'] = 'y', ['ຣ'] = 'r', ['ລ'] = 'l', ['ວ'] = 'w',
['ຫ'] = 'h', ['ອ'] = 'ʼ', ['ຮ'] = 'h', ['ຼ'] = 'r',
['ຫງ'] = 'ng',
['ຫຍ'] = 'ny',
['ຫນ'] = 'n', ['ໜ'] = 'n',
['ຫມ'] = 'm', ['ໝ'] = 'm',
['ຫຣ'] = 'r',
['ຫລ'] = 'l', ['ຫຼ'] = 'l',
['ຫວ'] = 'w',
['ກຣ'] = 'kr', ['ກລ'] = 'kl',
['ຂຣ'] = 'khr', ['ຄຣ'] = 'khr', ['ຂລ'] = 'khl', ['ຄລ'] = 'khl',
['ປຣ'] = 'pr', ['ປລ'] = 'pl',
['ພຣ'] = 'phr', ['ຟຣ'] = 'fr', ['ພລ'] = 'phl', ['ຟລ'] = 'fl',
['ດຣ'] = 'dr', ['ຕຣ'] = 'tr'
}
-- Mapping of vowel combinations.
local vowel_conv = {
['ະ'] = 'a', ['ັ'] = 'a',
['ິ'] = 'i',
['ຶ'] = 'ư', ['ຸ'] = 'u', ['ຸຍ'] = 'ui',
['ເະ'] = 'e', ['ເັ'] = 'e',
['ແະ'] = 'æ', ['ແັ'] = 'æ',
['ໂະ'] = 'o', ['ົ'] = 'o',
['ເາະ'] = 'ǫ', ['ັອ'] = 'ǫ',
['ເິ'] = 'œ',
['ເັຍ'] = 'ia', ['ັຽ'] = 'ia',
['ເຶອ'] = 'ưa',
['ົວະ'] = 'ua', ['ັວ'] = 'ua', ['ວັ'] = 'ua',
['ໄ'] = 'ai', ['ໃ'] = 'ai', ['ັຍ'] = 'ai',
['ເົາ'] = 'ao',
['ົາວ'] = 'uau',
['ຳ'] = 'am', ['ໍາ'] = 'am',
['ວຳ'] = 'uam',
['າ'] = 'ā',
['າວ'] = 'āo',
['ີ'] = 'ī',
['ື'] = 'ư̄',
['ູ'] = 'ū',
['ເ'] = 'ē',
['ແ'] = 'ǣ',
['ໂ'] = 'ō',
['ໂຍ'] = 'ōi', ['ໂຽ'] = 'ōi',
['ໍ'] = 'ǭ', ['ອ'] = 'ǭ',
['ອຍ'] = 'ǭi', ['ອຽ'] = 'ǭi',
['ເີ'] = 'œ̄',
['ເີຽ'] = 'œ̄i', ['ເີຍ'] = 'œ̄i',
['ເຍ'] = 'īa', ['ເັຽ'] = 'īa', ['ຽ'] = 'īa',
['ເືອ'] = 'ư̄a', ['ເືອຍ'] = 'ư̄ai',
['ົວ'] = 'ūa', ['ວ'] = 'ūa',
['ວຍ'] = 'uāi', ['ວຽ'] = 'uāi', ['ວຽນ'] = 'uīan',
['າຍ'] = 'āi', ['າຽ'] = 'āi',
['ວາ'] = 'uā',
['ວາຍ'] = 'uāi', ['ວາຽ'] = 'uāi',
['ແວ'] = 'ǣu', -- ແ_ວ can either be ǣu and uǣ with the first one being more common.
['ີວ'] = 'īu', ['ິວ'] = 'iu',
['ຽວ'] = 'iāu',
['ວີວ'] = 'uīu',
}
-- Mapping of coda consonants.
local coda_conv = {
['ກ'] = 'k', ['ຂ'] = 'k', ['ຄ'] = 'k',
['ງ'] = 'ng',
['ຈ'] = 't', ['ຊ'] = 't',
['ດ'] = 't', ['ຕ'] = 't', ['ຖ'] = 't', ['ທ'] = 't',
['ສ'] = 's',
['ນ'] = 'n',
['ບ'] = 'p', ['ປ'] = 'p', ['ພ'] = 'p', ['ຟ'] = 'p',
['ມ'] = 'm',
['ຢ'] = 'y',
['ຣ'] = 'n', ['ລ'] = 'n',
['ວ'] = 'w',
[''] = '',
}
-- Special symbols.
local sp_symbols = {
['ຯ'] = '〃', ['ໆ'] = '〃',
['໌'] = '',
['໐'] = '0', ['໑'] = '1', ['໒'] = '2', ['໓'] = '3', ['໔'] = '4',
['໕'] = '5', ['໖'] = '6', ['໗'] = '7', ['໘'] = '8', ['໙'] = '9'
}
-- List of character types.
local char_type = {
['ກ'] = 'coda', ['ຂ'] = 'coda', ['ຄ'] = 'coda', ['ງ'] = 'coda',
['ຈ'] = 'coda', ['ຊ'] = 'coda', ['ຍ'] = 'ambig',
['ດ'] = 'coda', ['ຕ'] = 'coda', ['ຖ'] = 'coda', ['ທ'] = 'coda', ['ນ'] = 'coda',
['ບ'] = 'coda', ['ປ'] = 'coda', ['ຜ'] = 'cons', ['ຝ'] = 'cons', ['ພ'] = 'coda', ['ຟ'] = 'coda', ['ມ'] = 'coda',
['ຢ'] = 'coda', ['ຣ'] = 'coda', ['ລ'] = 'coda', ['ວ'] = 'ambig',
['ສ'] = 'coda', ['ຫ'] = 'cons', ['ອ'] = 'ambig', ['ຮ'] = 'cons',
['ໜ'] = 'cons', ['ໝ'] = 'cons',
['ຯ'] = 'iter_symbol',
['ະ'] = 'vowel_let', ['ັ'] = 'suf_vowel', ['າ'] = 'vowel_let', ['ຳ'] = 'suf_vowel',
['ິ'] = 'suf_vowel', ['ີ'] = 'suf_vowel', ['ຶ'] = 'suf_vowel', ['ື'] = 'suf_vowel',
['ຸ'] = 'suf_vowel', ['ູ'] = 'suf_vowel', ['ົ'] = 'suf_vowel',
['ຼ'] = 'glide',
['ຽ'] = 'vowel_let',
['ເ'] = 'pref_vowel', ['ແ'] = 'pref_vowel',
['ໂ'] = 'pref_vowel', ['ໃ'] = 'pref_vowel', ['ໄ'] = 'pref_vowel',
['ໆ'] = 'iter_symbol',
['່'] = 'tone', ['້'] = 'tone', ['໊'] = 'tone', ['໋'] = 'tone',
['໌'] = 'canc_symbol', ['ໍ'] = 'suf_vowel',
['໐'] = 'number', ['໑'] = 'number', ['໒'] = 'number', ['໓'] = 'number', ['໔'] = 'number',
['໕'] = 'number', ['໖'] = 'number', ['໗'] = 'number', ['໘'] = 'number', ['໙'] = 'number'
}
-- List of consonant classes
local cons_class = {
['ກ'] = 'mid', ['ຂ'] = 'high', ['ຄ'] = 'low', ['ງ'] = 'low',
['ຈ'] = 'mid', ['ສ'] = 'high', ['ຊ'] = 'low', ['ຍ'] = 'low',
['ດ'] = 'mid', ['ຕ'] = 'mid', ['ຖ'] = 'high', ['ທ'] = 'low', ['ນ'] = 'low',
['ບ'] = 'mid', ['ປ'] = 'mid', ['ຜ'] = 'high', ['ຝ'] = 'high', ['ພ'] = 'low', ['ຟ'] = 'low', ['ມ'] = 'low',
['ຢ'] = 'mid', ['ຣ'] = 'low', ['ລ'] = 'low', ['ວ'] = 'low',
['ຫ'] = 'high', ['ອ'] = 'mid', ['ຮ'] = 'low'
}
-- Reset the syllable table.
local function reset_syllable()
return { curr = {}, initial = {}, vowel = {}, tone = {}, coda = {} }
end
-- Split the entry into individual syllables.
function export.split_syll(text)
-- (temp function)
if type(text) == 'table' then
text = text.args[1]
end
-- Store split syllables.
local syllables = {}
local curr_syll = reset_syllable()
-- Iterate through Lao characters.
for lao_text in gmatch(text, '[ກ-ໝ]+') do
local c, c_types = {}, {}
-- Classify each character in the syllable.
for i = 1, len(lao_text) do
c[i] = sub(lao_text, i, i)
c_types[i] = char_type[c[i]]
end
-- Parse the entry by identifying each character's type.
for i = 1, #c + 1 do
local type_curr, type_next = c_types[i], c_types[i + 1]
local curr_vowel_full = table.concat(curr_syll.vowel)
-- Prefix vowels are always the start of a new syllable.
if type_curr == 'pref_vowel' or i == #c + 1 then
if #curr_syll.curr ~= 0 then
table.insert(syllables, table.concat(curr_syll.curr))
curr_syll = reset_syllable()
end
table.insert(curr_syll.vowel, c[i])
table.insert(curr_syll.curr, c[i])
-- Glide consonants always follow the initial consonant.
elseif type_curr == 'glide' then
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
-- Suffix vowels and vowel letters are always part of the same syllable.
elseif type_curr == 'suf_vowel' or type_curr == 'vowel_let' then
table.insert(curr_syll.vowel, c[i])
table.insert(curr_syll.curr, c[i])
-- Same with tone marks.
elseif type_curr == 'tone' then
table.insert(curr_syll.tone, c[i])
table.insert(curr_syll.curr, c[i])
-- Some consonants can be at the end of a syllable.
elseif type_curr == 'coda' then
if #curr_syll.coda == 0 and initial_conv[table.concat(curr_syll.initial)..c[i]] and (#curr_syll.vowel == 0 or char_type[curr_vowel_full] == 'pref_vowel') then
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
elseif #curr_syll.coda == 0 and #curr_syll.initial ~= 0 and (type_next ~= 'glide' and type_next ~= 'suf_vowel' and type_next ~= 'vowel_let' and type_next ~= 'tone')
and not (type_next == 'ambig' and match(c_types[i + 2], 'co')) then
table.insert(curr_syll.coda, c[i])
table.insert(curr_syll.curr, c[i])
else
table.insert(syllables, table.concat(curr_syll.curr))
curr_syll = reset_syllable()
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
end
-- However, some consonants can only start a syllable.
elseif type_curr == 'cons' then
if #curr_syll.coda == 0 and initial_conv[table.concat(curr_syll.initial)..c[i]] and (#curr_syll.vowel == 0 or char_type[curr_vowel_full] == 'pref_vowel') then
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
else
table.insert(syllables, table.concat(curr_syll.curr))
curr_syll = reset_syllable()
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
end
-- Ambiguous characters can be either the start or end of a syllable.
elseif type_curr == 'ambig' then
if #curr_syll.initial == 0 or char_type[table.concat(curr_syll.vowel)] == 'pref_vowel' then
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
elseif c[i] == 'ຍ' and #curr_syll.vowel == 0 then
table.insert(syllables, table.concat(curr_syll.curr))
curr_syll = reset_syllable()
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
elseif #curr_syll.initial ~= 0 and (#curr_vowel_full == 0 or vowel_conv[curr_vowel_full..c[i]] and
(type_next ~= 'glide' and type_next ~= 'suf_vowel' and type_next ~= 'vowel_let' and type_next ~= 'tone')) then
table.insert(curr_syll.vowel, c[i])
table.insert(curr_syll.curr, c[i])
else
table.insert(syllables, table.concat(curr_syll.curr))
curr_syll = reset_syllable()
table.insert(curr_syll.initial, c[i])
table.insert(curr_syll.curr, c[i])
end
-- The iteration symbol should be treated as its own syllable.
elseif type_curr == 'iter_symbol' then
if #curr_syll.curr ~= 0 then
table.insert(syllables, table.concat(curr_syll.curr))
end
curr_syll = reset_syllable()
table.insert(curr_syll.curr, c[i])
-- Similar with the cancel symbol.
elseif type_curr == 'canc_symbol' then
table.insert(curr_syll.curr, c[i])
-- As well as with numbers.
elseif type_curr == 'number' then
table.insert(curr_syll.curr, sp_symbols[c[i]])
end
end
-- Add syllables to the syllables table.
for _, syllable in ipairs(curr_syll.curr) do
table.insert(syllables, syllable)
end
end
return table.concat(syllables, '-')
end
return export