216
edits
Module:Deranize: Difference between revisions
Update Neirani (replace vocalic pipoq with titieq)
m (test) |
(Update Neirani (replace vocalic pipoq with titieq)) |
||
| (3 intermediate revisions by the same user not shown) | |||
| Line 58: | Line 58: | ||
["diphthong"] = "", | ["diphthong"] = "", | ||
[""] = "", | [""] = "", | ||
-- | -- neirani | ||
["aı"] = "", -- DERANI LETTER DUDEO | ["aı"] = "", -- DERANI LETTER DUDEO | ||
["ao"] = " | ["ao"] = "", -- DERANI LETTER TITIEQ | ||
["eı"] = "", -- DERANI LETTER ZOZEO | ["eı"] = "", -- DERANI LETTER ZOZEO | ||
["oı"] = "", -- DERANI LETTER NHANHOQ | ["oı"] = "", -- DERANI LETTER NHANHOQ | ||
| Line 78: | Line 78: | ||
} | } | ||
local | local vowel_lookalikes = { | ||
["s"] = true, | |||
["b"] = true, | |||
["c"] = true, | |||
["g"] = true, | |||
["f"] = true, | |||
} | |||
local diphthong_lookalikes = { | |||
["d"] = true, | |||
["p"] = true, | |||
["z"] = true, | |||
["nh"] = true, | |||
} | |||
local neirani = false | |||
function deranize_word(word) | function deranize_word(word) | ||
| Line 86: | Line 101: | ||
local len = mw.ustring.len(word) | local len = mw.ustring.len(word) | ||
-- NB. this is not PCRE regex, see https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#Ustring_patterns | -- NB. this is not PCRE regex, see https://www.mediawiki.org/wiki/Extension:Scribunto/Lua_reference_manual#Ustring_patterns | ||
local toaqre = "()([mbpfndtzcsrljꝡgk']?h?)([auıoe])([̣]?)([́̈̂̀]?)([auıoe]?[auıoe]?)()([qm]?)([-·]?)()([auıoe]?)" | local toaqre = "()([mbpfndtzcsrljꝡgk']?h?)([auıoe])([̣]?)([́̈̂̀]?)([auıoe]?[auıoe]?)()([qm]?)([-·]?)()([mbpfndtzcsrljꝡgk']?h?)([auıoe]?)" | ||
local is_first_syllable = true | local is_first_syllable = true | ||
while ix <= len do | while ix <= len do | ||
local shouldbreak = false | local shouldbreak = false | ||
local pos_init, initial, medial, underdot, diacritic, final, pos_precoda, coda, hyphen, pos_postcoda, | local pos_init, initial, medial, underdot, diacritic, final, pos_precoda, coda, hyphen, pos_postcoda, next_initial, next_medial = mw.ustring.match(word, toaqre, ix) | ||
if pos_init == nil then | if pos_init == nil then | ||
break | break | ||
end | end | ||
res[#res+1] = u.toNFC(u.sub(word, ix, pos_init - 1)) | res[#res+1] = u.toNFC(u.sub(word, ix, pos_init - 1)) | ||
if coda == "m" and | if coda == "m" and next_initial == "" and next_medial ~= "" then | ||
coda = "" | coda = "" | ||
next_initial = "m" | |||
ix = pos_precoda | ix = pos_precoda | ||
else | else | ||
| Line 104: | Line 120: | ||
local v = medial..final | local v = medial..final | ||
local is_diphthong = v == "aı" or v == "ao" or v == "eı" or v == "oı" | local is_diphthong = v == "aı" or v == "ao" or v == "eı" or v == "oı" | ||
if is_diphthong and | if is_diphthong and neirani then medial, final = v, "" end | ||
local glyphs = {initial, tone, medial} | local glyphs = {initial, tone, medial} | ||
if initial == "" and is_first_syllable then glyphs[1] = "'" end | if initial == "" | ||
and is_first_syllable | |||
and (final == "" or (neirani and is_diphthong)) | |||
and coda == "" | |||
and (vowel_lookalikes[next_initial] or (neirani and diphthong_lookalikes[next_initial])) then | |||
glyphs[1] = "'" | |||
end | |||
if not is_first_syllable then glyphs[2] = "" end | if not is_first_syllable then glyphs[2] = "" end | ||
if glyphs[1] == "'" or glyphs[1] == "ꝡ" then | if glyphs[1] == "" or glyphs[1] == "'" or glyphs[1] == "ꝡ" then | ||
-- Move tone onto the first vowel | -- Move tone onto the first vowel | ||
glyphs[2], glyphs[3] = glyphs[3], glyphs[2] | glyphs[2], glyphs[3] = glyphs[3], glyphs[2] | ||
end | end | ||
if is_diphthong and not | if is_diphthong and not neirani then | ||
glyphs[#glyphs+1] = "diphthong" | glyphs[#glyphs+1] = "diphthong" | ||
elseif final ~= "" then | elseif final ~= "" then | ||
glyphs[#glyphs+1] = "hiatus" | glyphs[#glyphs+1] = "hiatus" | ||
end | end | ||
if | if neirani then | ||
glyphs[#glyphs+1] = final | glyphs[#glyphs+1] = final | ||
else | else | ||
| Line 142: | Line 164: | ||
function deranize(frame) | function deranize(frame) | ||
assert(frame.args[1] ~= nil, "This function requires at least one argument") | assert(frame.args[1] ~= nil, "This function requires at least one argument") | ||
neirani = frame.args[2] ~= nil | |||
local text = u.gsub(u.lower(u.toNFD(frame.args[1])), "i", "ı") | local text = u.gsub(u.lower(u.toNFD(frame.args[1])), "i", "ı") | ||
local converted = u.gsub(text, "(%S+)", deranize_word) | local converted = u.gsub(text, "(%S+)", deranize_word) | ||