205
edits
Module:Deranize: Difference between revisions
Only show word-initial o'aomo when necessary
m (test) |
(Only show word-initial o'aomo when necessary) |
||
| Line 76: | Line 76: | ||
["*"] = " ", | ["*"] = " ", | ||
["_"] = "", -- compatibility nbsp | ["_"] = "", -- compatibility nbsp | ||
} | |||
local vowel_lookalikes = { | |||
["s"] = true, | |||
["b"] = true, | |||
["c"] = true, | |||
["g"] = true, | |||
["f"] = true, | |||
} | |||
local diphthong_lookalikes = { | |||
["d"] = true, | |||
["p"] = true, | |||
["z"] = true, | |||
["nh"] = true, | |||
} | } | ||
| 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 106: | Line 122: | ||
if is_diphthong and reform then medial, final = v, "" end | if is_diphthong and reform 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 (reform and is_diphthong)) | |||
and coda == "" | |||
and (vowel_lookalikes[next_initial] or (reform 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] == "ꝡ" then | ||