local export = {}
local find = mw.ustring.find
local gsub = mw.ustring.gsub
local len = mw.ustring.len
local match = mw.ustring.match
local sub = mw.ustring.sub
local split = mw.text.split
local gsplit = mw.text.gsplit
local m_data = mw.loadData("Module:cmn-pron/data")
local m_zh = require("Module:zh")
local _m_zh_data = nil
local hom_data = nil
local function hom_data_part(pinyin)
local x = mw.ustring.toNFD(pinyin):sub(1,1)
if "a" <= x and x <= "g" then
return 1
elseif "h" <= x and x <= "m" then
return 2
elseif "n" <= x and x <= "w" then
return 3
end
return 4
end
local function get_hom_data(py)
if hom_data == nil then
hom_data = mw.loadData("Module:zh/data/cmn-hom/" .. hom_data_part(py))
end
return hom_data
end
-- if not empty
local function ine(var)
if var == "" then
return nil
else
return var
end
end
local breve, hacek, circumflex = mw.ustring.char(0x306), mw.ustring.char(0x30C), mw.ustring.char(0x302)
local decompose = mw.ustring.toNFD
local function breve_error(text)
if type(text) ~= "string" then
return
end
text = decompose(text)
if text:find(breve) then
error('The pinyin text "' .. text .. '" contains a breve. Replace it with "' .. text:gsub(breve, hacek) .. '".', 2)
end
if find(text, '') or find(text, '') then
error('The pinyin text "' .. text .. '" contains an uncommon shorthand. Replace it with "' .. gsub(text, '', { = 'zh', = 'ch', = 'sh', = 'ng', = 'Zh', = 'Ch', = 'Sh', = 'h'}) .. '".', 2)
end
end
local function m_zh_data()
if _m_zh_data == nil then _m_zh_data = mw.loadData("Module:zh/data/cmn-tag") end;
return _m_zh_data;
end
function export.py_detone(f)
local text = type(f) == 'table' and f.args or f
return mw.ustring.toNFC(gsub(mw.ustring.toNFD(text), m_data.tones, ''))
end
function export.py_transf(f)
local text = type(f) == 'table' and f.args or f
return export.py_detone(text) .. export.tone_determ(text)
end
function export.tone_determ(f)
local text = type(f) == 'table' and f.args or f
text = mw.ustring.toNFD(text)
return m_data.py_tone or '5'
end
function export.py_transform(text, detone, not_spaced)
if type(text) == 'table' then text, detone, not_spaced = text.args, text.args, text.args end
if find(text, '') then
error("Pinyin contains the hidden character: (U+200B). Please remove that character from the text.")
end
detone = ine(detone)
not_spaced = ine(not_spaced)
text = gsub(gsub(mw.ustring.toNFD(text), mw.ustring.toNFD('ê'), 'ê'), mw.ustring.toNFD('ü'), 'ü')
if find(mw.ustring.lower(text), '' .. m_data.tones .. '?' .. m_data.tones .. '') and not not_spaced then
error(("Missing apostrophe before null-initial syllable - should be \"%s\" instead."):format(gsub(text, '(' .. m_data.tones .. '?)(' .. m_data.tones .. ')', "%1'%2"))) end
original_text = text
text = gsub(text,'()()(' .. m_data.tones .. ')', '%1%3%2')
text = gsub(text,'()(' .. m_data.tones .. ')()', '%1%3%2')
if text ~= original_text then
error("Incorrect diacritic placement in Pinyin - should be \"".. text .. "\" instead.") end
text = mw.ustring.lower(text)
if not mw.ustring.find(text, m_data.tones) and text:find('') then
return gsub(text, '(%d)(%l)', '%1 %2')
end
text = gsub(text, "#", " #")
if find(text, '') then
text = gsub(text, '()$', { = ' yī', = ' bù'})
text = gsub(text, '()', ' %1 ')
text = gsub(text, '()', ' %1 ')
text = gsub(text, ' +', ' ')
text = gsub(text, '^ ', '')
text = gsub(text, ' $', '')
text = gsub(text, '%. %. %.', '...')
end
text = gsub(text, "", ' ')
text = gsub(text, '(' .. m_data.tones .. '?n?g?r?)(h?)', '%1 %2')
text = gsub(text, ' ()$', '%1')
text = gsub(text, ' () ', '%1 ')
if detone then
text = mw.ustring.gsub(text, m_data.tones, m_data.py_tone)
text = gsub(text, '()(*)', '%2%1')
text = mw.ustring.gsub(text, '() ', '%15 ')
text = mw.ustring.gsub(text, '()$', '%15')
end
if not_spaced then
text = gsub(text, ' ', '')
end
return mw.ustring.toNFC(text)
end
function export.py_ipa(text)
local ipa_initial = m_data.py_ipa_initial
local ipa_initial_tl = m_data.py_ipa_initial_tl
local ipa_final = m_data.py_ipa_final
local ipa_null = m_data.py_ipa_null
local ipa_tl_ts = m_data.py_ipa_tl_ts
local ipa_third_t_ts = m_data.py_ipa_third_t_ts
local ipa_t_values = m_data.py_ipa_t_values
local tone = {}
local tone_cat = {}
text = gsub(export.py_transform(text), '', '')
text = gsub(text, ' +', ' ')
local p = split(text, " ")
for i = 1, #p do
tone_cat = export.tone_determ(p)
p = gsub(p, '.?', m_data.py_detone)
if p == '一' then
tone_cat = (export.tone_determ(p) == '4' or p == 'ge') and '1-2' or '1-4'
p = 'yi'
elseif p == '不' then
tone_cat = (export.tone_determ(p) == '4') and '4-2' or '4'
p = 'bu'
end
end
tone_cat.length = #tone_cat
local function get_initial_and_final_IPA(a, b, c)
return a ..
(ipa_initial or error(("Unrecognised initial: \"%s\""):format(b))) ..
(ipa_final or error(("Unrecognised final: \"%s\". Are you missing an apostrophe before the null-initial syllable, or using an invalid Pinyin final?"):format(c)))
end
for i, item in ipairs(p) do
if ipa_null then item = 'ˀ' .. item end
item = gsub(item, '()u', '%1ü')
if item == 'ng' then
item = ipa_final
else
item = gsub(item, '^(#?ˀ?)(?h?)(.+)$',
get_initial_and_final_IPA)
end
item = gsub(item, '(ʈ?͡?ʰ?)ir', '%1ʐ̩ɻ')
item = gsub(item, '(ʈ?͡?ʰ?)i', '%1ʐ̩')
item = gsub(item, '(t?͡?sʰ?)ir', '%1z̩ɻ')
item = gsub(item, '(t?͡?sʰ?)i', '%1z̩')
item = gsub(item, 'ʐʐ̩', 'ʐ̩')
item = gsub(item, 'ˀwɔ', 'ˀɔ')
local curr_tone_cat, next_tone_cat = tone_cat, tone_cat
if curr_tone_cat == '5' then
item = gsub(item, '^()()', function(a, b) return ipa_initial_tl .. b end)
item = gsub(item, '^(͡)()', function(a, b) return ipa_initial_tl .. b end)
item = gsub(item, 'ɤ$', 'ə')
tone = ipa_tl_ts] or ""
elseif curr_tone_cat == '3' then
if p and match(p, "#") then next_tone_cat = "#3" end
if i == tone_cat.length then
if i == 1 then tone = '²¹⁴' else tone = '²¹⁴⁻²¹⁽⁴⁾' end
else
tone = ipa_third_t_ts
end
elseif curr_tone_cat == '4' and next_tone_cat == '4' then
tone = '⁵¹⁻⁵³'
elseif curr_tone_cat == '4' and next_tone_cat == '1-4' then
tone = '⁵¹⁻⁵³'
elseif curr_tone_cat == '1-4' and next_tone_cat == '4' then
tone = '⁵⁵⁻⁵³'
else
tone = ipa_t_values
end
p = item .. tone
p = gsub(p, "#", "")
end
return table.concat(p, " ")
end
function export.py_number_to_mark(text)
local priority = { "a", "o", "e", "ê", "i", "u", "ü" }
local toneMark = { = "̄", = "́", = "̌", = "̀", = "", = "", = "" }
local mark = toneMark?$")]
local toneChars = ""
text = gsub(text, "?$", "")
for _, letter in ipairs(priority) do
text = gsub(text, letter, letter .. mark)
if find(text, toneChars) then break end
end
return mw.ustring.toNFC(gsub(text, "i("..toneChars..")u", "iu%1"))
end
function export.py_zhuyin(text)
local zhuyin_initial = m_data.py_zhuyin_initial
local zhuyin_final = m_data.py_zhuyin_final
local zhuyin_er = m_data.py_zhuyin_er
local zhuyin_tone = m_data.py_zhuyin_tone
if type(text) == 'table' then
if text.args == '' then
text = mw.title.getCurrentTitle().text
else
text = text.args
end
end
breve_error(text)
text = gsub(text, "#", "")
text = export.py_transform(text, true)
text = gsub(text, '()u', '%1ü')
text = gsub(text, '(h?)i', '%1')
text = gsub(text, '()i', '%1')
local function add_tone(syllable, tone)
if tone == '5' then
return zhuyin_tone .. syllable
else
return syllable .. zhuyin_tone
end
end
local function fun1(a, b) return add_tone((({ = 'ㄫ', = 'ㄏㄇ'}) or a), b) end
local function fun2(number) return add_tone('ㄏㄫ', number) end
local function fun3(number) return add_tone('ㄦ', number) end
local function fun4(a, b, c, d) return add_tone(zhuyin_initial .. zhuyin_final, d) .. zhuyin_er end
local word = split(text, " ", true)
for i, syllable in ipairs(word) do
if find(syllable, '^$') then
syllable = gsub(syllable, '^()()$', fun1)
elseif find(syllable, '^hng$') then
syllable = gsub(syllable, '^hng()$', fun2)
elseif find(syllable, '^er$') then
syllable = gsub(syllable, '^er()$', fun3)
else
syllable = gsub(syllable, '^(?h?)(????)(r?)()$',
fun4)
end
if find(syllable, '') then
error(("Zhuyin conversion unsuccessful: \"%s\". Are you using a valid Pinyin syllable? Is the text using a breve letter instead of a caron one?"):format(syllable))
end
word = syllable
end
text = gsub(table.concat(word, " "), ' , ', ', ')
return text
end
function export.zhuyin_py(text)
local zhuyin_py_initial = m_data.zhuyin_py_initial
local zhuyin_py_final = m_data.zhuyin_py_final
local zhuyin_py_tone = m_data.zhuyin_py_tone
if type(text) == "table" then text = text.args end
local word = split(text, " ", true)
local function process_syllable(syllable)
syllable = gsub(syllable, '^()(?)$', '%1ㄧ%2')
return gsub(syllable, '(?)(?)(?)(ㄦ?)', function(initial, final, tone, erhua)
initial = zhuyin_py_initial
final = zhuyin_py_final
if erhua ~= '' then
final = final .. 'r'
end
if initial == '' then
final = final:gsub('^()(n?g?)$', function(a, b) return a:gsub('', { = 'yi', = 'wu'}) .. b end)
final = final:gsub('^(w?u)()$', 'ue%2')
final = final:gsub('^iu$', 'iou')
final = final:gsub('^()', { = 'y', = 'w'})
final = final:gsub('^ong', 'weng')
final = gsub(final, '^ü', 'yu')
end
if initial:find('') then
final = gsub(final, '^ü', 'u')
end
local tone = zhuyin_py_tone
if final:find('') then
final = final:gsub("()", "%1" .. tone)
elseif final:find('i') then
final = final:gsub("(i)", "%1" .. tone)
elseif final:find('') then
final = final:gsub("()", "%1" .. tone)
else
final = gsub(final, "^(?)(.)", "%1" .. "%2" .. tone)
end
return initial .. final
end)
end
for i, syllable in ipairs(word) do
word = process_syllable(syllable)
end
return mw.ustring.toNFC(table.concat(word, " "))
end
function export.py_wg(text)
local py_wg_initial = m_data.py_wg_initial
local py_wg_final = m_data.py_wg_final
local py_wg_syl = m_data.py_wg_syl
if type(text) == 'table' then text = text.args end
local function process_syllable(initial, final, tone)
for text, replace in pairs(py_wg_final) do
final = gsub(final, text, replace)
end
if (initial == "zh" or initial == "ch") and final == "i" then
final = "ih"
end
local untoned = (py_wg_initial or initial) .. final
for text, replace in pairs(py_wg_syl) do
untoned = gsub(untoned, text, replace)
end
return untoned .. '<sup>' .. tone .. '</sup>'
end
local function process_word(word)
local uppercase
if mw.ustring.find(word, '^%u') then uppercase = true else uppercase = false end
local text = gsub(export.py_transform(word, true), '', '')
text = gsub(gsub(text, ' +', ' '), '', { = 'yi1', = 'bu4'})
text = gsub(text, '()u', '%1ü')
local p = split(text, " ", true)
for i = 1, #p do
p = gsub(p, '^(?h?)(.+)()$', process_syllable)
end
text = table.concat(p, "-")
if uppercase then
text = gsub(text, '^%l', mw.ustring.upper)
end
return text
end
local words = split(text, " ")
for i, word in ipairs(words) do
words = process_word(word)
end
return table.concat(words, " ")
end
local function temp_bg(text, bg)
if bg == 'y' then
return '<' .. text .. '>'
end
return text
end
local function make_bg(text, bg)
if bg == 'y' then
return '<span style="background-color:#F5DEB3">' .. text .. '</span>'
else
return text
end
end
function export.py_gwoyeu(text, original_text)
local initials = m_data.py_gwoyeu_initials
local finals = m_data.py_gwoyeu_finals
if type(text) == 'table' then text = text.args end
if text:find('^%s') or text:find('%s$') then error('invalid spacing') end
local words = split(text, " ")
local count = 0
for i, word in ipairs(words) do
local uppercase
if mw.ustring.find(word, '^%u') then uppercase = true else uppercase = false end
word = export.py_transform(word, true, true)
word = gsub(word, "()", "%1 ")
word = gsub(word, " $", "")
word = gsub(word, '()', ' %1 ')
word = gsub(word, ' +', ' ')
word = gsub(word, ' $', '')
word = gsub(word, '^ ', '')
local syllables = split(word, " ")
for j, syllable in ipairs(syllables) do
count = count + 1
if not find(syllable, '^+$') then
local current = sub(mw.title.getCurrentTitle().text, count, count)
if find(current, '^$') then
local exceptions = { = 'i', = 'chi', = 'ba', = 'bu'}
syllables = exceptions
else
local initial, final, tone = '', '', ''
syllable = gsub(syllable, '()u', '%1ü')
syllable = gsub(syllable, '^(h?)i(r?)$', '%1ɨ%2')
if mw.ustring.find(syllable, '(?h?)(??g?r?)()') then
syllable = gsub(syllable, '(?h?)(??g?r?)()', function(a, b, c)
initial = initials or error('Unrecognised initial:' .. a); final = finals or error('Unrecognised final:' .. b); tone = c
return (initial .. final .. tone) end)
elseif not find(mw.title.getCurrentTitle().text, "") then
error('Unrecognised syllable:' .. syllable)
end
local original = initial..final..tone
if initial:find('^$') then
final = initial .. final
initial = ''
end
if initial .. final == "e'l" then
final = "el"
end
final = gsub(final, '()%1', '%1')
local len = len(initial) + len(final)
local detone = initial..final
local replace = detone
local fullstop = false
if tone == 5 or tone == '5' then
fullstop = true
if original_text then
tone = split(export.py_transform(original_text, true), ' '):match('')
elseif initial:find('^$') then
tone = 2
else tone = 1 end
if tone == 5 or tone == '5' then
tone = export.tone_determ(m_zh.py(current))
end
end
if tone == 1 or tone == '1' then
if initial:find('^$') then
replace = initial .. 'h' .. sub(detone, 2, len)
else
replace = detone
end
elseif tone == 2 or tone == '2' then
if not initial:find('^$') then
if final:find('^') then
replace = gsub(detone, '', { = 'y', = 'w'}, 1)
replace = gsub(replace, 'l?$', { = 'yi', = 'wu', = 'wul',})
else
replace = gsub(detone, '(+)', '%1r')
end
else
replace = detone
end
elseif tone == 3 or tone == '3' then
if final:find("^iu?e'l$") then
detone = detone:gsub("'", '')
end
detone = detone:gsub('^', { = 'yi', = 'wu'})
if final:find('') and (not final:find('^') or initial..final=="ie") and (not final:find('^') or initial..final=="uo") then
replace = detone:gsub('', { = 'e', = 'o'}, 1)
elseif final:find('') then
replace = detone:gsub('(?)(\168?)', '%1%2%2', 1) --\195\168 is the utf8 for è
else
error('Unrecognised final:'..final)
end
elseif tone == 4 or tone == '4' then
if final:find("^iu?e'l$") then
detone = detone:gsub("'", '')
end
detone = detone:gsub('^', { = 'yi', = 'wu'})
if detone:find('l?$') then
replace = detone:gsub("l?$", { = 'y', = 'w', = 'wl'})
elseif detone:find('$') then
replace = detone:gsub("l?$", { = 'nn', = 'q', = 'll', = 'ql', = 'hl'})
else
replace = detone .. 'h'
end
replace = replace:gsub('yi()', 'y%1')
replace = replace:gsub('wu()', 'w%1')
end
if fullstop then replace = '.' .. replace end
syllables = syllable:gsub(original, replace)
end
end
end
words = table.concat(syllables, "")
if uppercase then
words = gsub(words, '^%l', mw.ustring.upper)
end
end
return table.concat(words, " ")
end
-- Converts Hanyu Pinyin into Tongyong Pinyin.
function export.py_tongyong(text)
if type(text) == 'table' then text = text.args end
local ty_tone = {
= "", = "\204\129", = "\204\140", = "\204\128", = "\204\138"
}
local function num_to_mark(syllable, tone)
tone = ty_tone
if tone ~= "" then
if syllable:find('') then
syllable = syllable:gsub("()", "%1" .. tone)
elseif syllable:find('o') then
syllable = syllable:gsub("(o)", "%1" .. tone)
elseif syllable:find('') then
syllable = syllable:gsub("()", "%1" .. tone)
end
end
return syllable
end
local words = {}
for word in gsplit(text, " ") do
local cap = mw.ustring.find(word, "^%u")
word = export.py_transform(word, true)
local syllables = {}
for syllable in gsplit(word, " ") do
syllable = syllable:gsub("(h?)i", "%1ih")
syllable = syllable:gsub("ü", "yu")
syllable = syllable:gsub("()u", "%1yu")
syllable = syllable:gsub("iu", "iou")
syllable = syllable:gsub("ui", "uei")
syllable = syllable:gsub("()eng", "%1ong")
syllable = syllable:gsub("wen", "wun")
syllable = syllable:gsub("iong", "yong")
syllable = syllable:gsub("^zh", "jh")
syllable = syllable:gsub("^q", "c")
syllable = syllable:gsub("^x", "s")
syllable = #syllables ~= 0 and syllable:gsub("^()", "'%1") or syllable
syllable = syllable:gsub("^(+)()$", num_to_mark)
table.insert(syllables, syllable)
end
word = table.concat(syllables, "")
word = cap and word:gsub("^.", string.upper) or word
table.insert(words, word)
end
return mw.ustring.toNFC(table.concat(words, " "))
end
function export.py_format(text, cap, bg, simple, nolink)
if cap == false then cap = nil end
if bg == false then bg = 'n' else bg = 'y' end
if simple == false then simple = nil end
if nolink == false then nolink = nil end
text = mw.ustring.toNFD(text)
local phon = text
local title = mw.title.getCurrentTitle().text
local cat = ''
local spaced = mw.ustring.toNFD(export.py_transform(text))
local space_count
spaced, space_count = gsub(spaced, ' ', '@')
local consec_third_count
for _ = 1, space_count do
spaced, consec_third_count = gsub(spaced, "(+)̌(*)@(+̌)", function(a, b, c)
return temp_bg(a..'́'..b, bg)..'@'..c end, 1)
if consec_third_count > 0 then
phon = gsub(spaced, '@', '')
end
end
text = gsub(text, "#", "")
phon = gsub(phon, "#", "")
if title:find('一') and not text:find('一') and not simple then
cat = cat .. ']'
end
if text:find('') and not simple then
text = gsub(text, '$', { = 'yī', = 'bù'})
phon = gsub(phon, '$', { = 'yī', = 'bù'})
if find(text, '一') then
if find(text, '一*') then
cat = cat .. ']'
phon = gsub(phon, '一(*)', function(a) return temp_bg('yì', bg) .. a end)
text = gsub(text, '一(*)', 'yī%1')
end
if find(text, '一*̀') or find(text, '一ge$') or find(text, '一ge') then
cat = cat .. ']'
phon = gsub(phon, '一(*̀)', function(a) return temp_bg('yí', bg) .. a end)
phon = gsub(phon, '一ge', temp_bg('yí', bg) .. 'ge')
text = gsub(text, '一(*)', 'yī%1')
end
end
if find(text, '不 ??h?*̀') then
cat = cat .. ']'
phon = gsub(phon, '不( ??h?*̀)', function(a) return temp_bg('bú', bg) .. a end)
end
end
text = gsub(text, '', { = 'yī', = 'bù'})
text = gsub(text, '兒', function() return make_bg('r', bg) end) -- character is deleted
phon = gsub(phon, '<(+)>', '<span style="background-color:#F5DEB3">%1</span>')
if not simple then
if cap then
text = gsub(text, '^%l', string.upper)
phon = gsub(phon, '^%l', string.upper)
end
if not nolink then
text = ']'
end
if '', { = 'yī', = 'bù'}) .. ']]' ~= text then
phon = gsub(phon, '', { = 'yī', = 'bù'})
text = text .. ' '
end
if mw.title.getCurrentTitle().nsText ~= 'Template' and not nolink then
text = text .. cat
end
end
return mw.ustring.toNFC(text)
end
function export.make_tl(original_text, tl_pos, bg, cap)
if bg == false then bg = 'n' else bg = 'y' end
local _, countoriginal = gsub(original_text, " ", " ")
local spaced = export.py_transform(original_text)
if sub(spaced, -1, -1) == ' ' then spaced = sub(spaced, 1, -2) end
local _, count = gsub(spaced, " ", " ")
local index = {}
local start, finish
local pos = 1
for i = 1, count, 1 do
if i ~= 1 then pos = (index + 1) end
index = mw.ustring.find(spaced, ' ', pos)
end
if tl_pos == 2 then
start = index - count + countoriginal + 2
finish = index - count + countoriginal
elseif tl_pos == 3 then
start = index - count + countoriginal + 3
finish = index - count + countoriginal + 1
else
start = count == 0 and 1 or (index - count + countoriginal + 1)
finish = -1
end
local text = (sub(original_text, 1, start-1) .. make_bg(gsub(sub(original_text, start, finish), '.', m_data.py_detone), bg))
if finish ~= -1 then text = (text .. sub(original_text, finish+1, -1)) end
if cap == true then text = gsub(text, '^%l', string.upper) end
return text
end
function export.tag(first, second, third, fourth, fifth)
local text = "(''"
local tag = {}
local tagg = first or "Standard Chinese"
tag = (second ~= '') and second or "Standard Chinese"
tag = (third ~= '') and third or nil
tag = (fourth ~= '') and fourth or nil
tag = (fifth ~= '') and fifth or nil
text = text .. ((tagg == '') and table.concat(tag, ", ") or tagg) .. "'')"
text = gsub(text, 'Standard Chinese', "]")
text = gsub(text, 'Mainland', "]")
text = gsub(text, 'Taiwan', "]")
text = gsub(text, 'Beijing', "]")
text = gsub(text, 'erhua', "]")
text = gsub(text, 'Min Nan', "]")
text = gsub(text, 'shangkouzi', "'']''")
return text
end
function export.straitdiff(text, pron_ind, tag)
local conv_text = text
for i = 1, #text do
if m_zh_data().MT then conv_text = 'y' end
end
if tag == 'tag' then
conv_text = (conv_text == 'y') and m_zh_data().MT_tag')] or ''
elseif pron_ind == 1 or pron_ind == 2 or pron_ind == 3 or pron_ind == 4 or pron_ind == 5 then
local reading = {}
for a, b in pairs(m_zh_data().MT) do
reading = b
if reading then reading = gsub(reading, "^()", "'%1") end
end
conv_text = gsub(text, '.', reading)
text = gsub(text, "^'", "")
text = gsub(text, " '", " ")
if conv_text == text and tag == 'exist' then return nil end
end
conv_text = gsub(conv_text, "^'", "")
return conv_text
end
function export.str_analysis(text, conv_type, other_m_vars)
if type(text) == 'table' then text, conv_type = text.args, (text.args or "") end
local MT = m_zh_data().MT
text = gsub(text, '=', '—')
text = gsub(text, ',', '隔')
text = gsub(text, '隔 ', ', ')
if conv_type == 'head' or conv_type == 'link' then
if find(text, '隔cap—') then
text = gsub(text, '', { = 'Yī', = 'Bù'})
end
text = gsub(text, '', { = 'yī', = 'bù'})
end
local comp = split(text, '隔', true)
local reading = {}
local alternative_reading = {}
local zhuyin = {}
--[[
-- not used
local param = {
'1n', '1na', '1nb', '1nc', '1nd', 'py', 'cap', 'tl', 'tl2', 'tl3', 'a', 'audio', 'er', 'ertl', 'ertl2', 'ertl3', 'era', 'eraudio',
'2n', '2na', '2nb', '2nc', '2nd', '2py', '2cap', '2tl', '2tl2', '2tl3', '2a', '2audio', '2er', '2ertl', '2ertl2', '2ertl3', '2era', '2eraudio',
'3n', '3na', '3nb', '3nc', '3nd', '3py', '3cap', '3tl', '3tl2', '3tl3', '3a', '3audio', '3er', '3ertl', '3ertl2', '3ertl3', '3era', '3eraudio',
'4n', '4na', '4nb', '4nc', '4nd', '4py', '4cap', '4tl', '4tl2', '4tl3', '4a', '4audio', '4er', '4ertl', '4ertl2', '4ertl3', '4era', '4eraudio',
'5n', '5na', '5nb', '5nc', '5nd', '5py', '5cap', '5tl', '5tl2', '5tl3', '5a', '5audio', '5er', '5ertl', '5ertl2', '5ertl3', '5era', '5eraudio'
}
--]]
if conv_type == '' then
return comp
elseif conv_type == 'head' or conv_type == 'link' then
for i, item in ipairs(comp) do
if not find(item, '—') then
if find(item, '') then
local M, T, t = {}, {}, {}
for a, b in pairs(MT) do
M = b; T = b; t = b;
M = gsub(M, "^()", "'%1")
T = gsub(T, "^()", "'%1")
if t then t = gsub(t, "^()", "'%1") end
end
local mandarin = gsub(item, '.', M)
local taiwan = gsub(item, '.', T)
mandarin = gsub(mandarin, "^'", "")
mandarin = gsub(mandarin, " '", " ")
if conv_type == 'link' then return mandarin end
taiwan = gsub(taiwan, "^'", "")
taiwan = gsub(taiwan, " '", " ")
local tt = gsub(item, '.', t)
if find(text, 'cap—') then
mandarin = gsub(mandarin, '^%l', mw.ustring.upper)
taiwan = gsub(taiwan, '^%l', mw.ustring.upper)
tt = gsub(tt, '^%l', mw.ustring.upper)
end
if tt == item then
zhuyin = export.py_zhuyin(mandarin, true) .. ', ' .. export.py_zhuyin(taiwan, true)
reading = mandarin .. ']], [[' .. taiwan
else
tt = gsub(tt, "^'", "")
tt = gsub(tt, " '", " ")
zhuyin = export.py_zhuyin(mandarin, true) .. ', ' .. export.py_zhuyin(taiwan, true) .. ', ' .. export.py_zhuyin(tt, true)
reading = mandarin .. ']], ], [[' .. tt
end
else
if conv_type == 'link' then return item end
zhuyin = export.py_zhuyin(item, true)
reading = item
if len(mw.title.getCurrentTitle().text) == 1 and #mw.text.split(export.py_transform(item), " ") == 1 then
alternative_reading = ") .. "|" .. mw.ustring.gsub(export.py_transf(reading), '()', '<sup>%1</sup>') .. "]]"
end
end
if reading ~= '' then reading = ' .. ']]' end
reading = gsub(reading, "#", "")
end
comp = item
if conv_type == 'link' then return comp end
end
local id = m_zh.ts_determ(mw.title.getCurrentTitle().text)
local accel
if id == 'trad' then
accel = '<span class="form-of pinyin-t-form-of transliteration-' .. m_zh.ts(mw.title.getCurrentTitle().text)
elseif id == 'simp' then
accel = '<span class="form-of pinyin-s-form-of transliteration-' .. m_zh.st(mw.title.getCurrentTitle().text)
elseif id == 'both' then
accel = '<span class="form-of pinyin-ts-form-of'
end
accel = accel .. '" lang="cmn" style="font-family: Consolas, monospace;">'
local result = other_m_vars and "*: <small>('']'')</small>\n*::" or "*:"
result = result .. "<small>('']'')</small>: " .. accel .. gsub(table.concat(reading, ", "), ", ,", ",")
if alternative_reading then
result = result .. " (" .. table.concat(alternative_reading, ", ") .. ")"
end
result = result .. (other_m_vars and "</span>\n*::" or "</span>\n*:")
result = result .. "<small>('']'')</small>: " .. '<span lang="zh-Bopo" class="Bopo">' .. gsub(table.concat(zhuyin, ", "), ", ,", ",") .. "</span>"
return result
elseif conv_type == '2' or conv_type == '3' or conv_type == '4' or conv_type == '5' then
if not find(text, '隔') or (comp and find(comp, '—')) then
return ''
else
return comp
end
else
for i = 1, #comp, 1 do
local target = '^' .. conv_type .. '—'
if find(comp, target) then
text = gsub(comp, target, '')
return text
end
end
text = ''
end
return text
end
function export.homophones(pinyin)
local text = ''
if mw.title.getCurrentTitle().nsText == '' then
local args = get_hom_data(pinyin).list
text = '<div style="visibility:hidden; float:left"><sup><span style="color:#FFF">edit</span></sup></div>'
for i, term in ipairs(args) do
if i > 1 then
text = text .. "<br>"
end
if mw.title.new(term).exists and term ~= mw.title.getCurrentTitle().text then
local forms = { term }
local content = mw.title.new(term):getContent()
local template = match(content, "{{zh%-forms*}}")
if template then
local simp = match(template, "|s=()+")
if simp then
table.insert(forms, simp)
end
for tradVar in mw.ustring.gmatch(template, "|t=()+") do
table.insert(forms, tradVar)
end
for simpVar in mw.ustring.gmatch(template, "|s=()+") do
table.insert(forms, simpVar)
end
term = table.concat(forms, "／")
end
end
text = text .. mw.getCurrentFrame():expandTemplate{ title = "Template:zh-l", args = { term, tr = "-" } }
end
text = text .. ']'
end
return text
end
local function erhua(word, erhua_pos, pagename)
local title = split(pagename, '')
local linked_title = ''
local syllables = split(export.py_transform(word), ' ')
local count = #syllables
erhua_pos = find(erhua_pos, '') and split(erhua_pos, ';') or { count }
for _, pos in ipairs(erhua_pos) do
pos = tonumber(pos)
title = title .. '兒'
syllables = syllables .. 'r'
end
local title = table.concat(title)
if mw.title.new(title).exists then
linked_title = ' (' .. require("Module:zh/link").link(nil, nil, {title, tr='-'}) .. ')'
end
for i, syllable in pairs(syllables) do
if i ~= 1 and mw.ustring.toNFD(syllable):find('^') then
syllables = "'" .. syllable
end
end
word = table.concat(syllables, '')
return (export.tag('', '', 'erhua-ed') .. linked_title), word
end
export.erhua = erhua
function export.make(frame)
local args = frame:getParent().args
return export.make_args(args)
end
function export.make_args(args)
local pagename = mw.title.getCurrentTitle().text
local text = {}
local reading = {args or '', args or '', args or '', args or '', args or ''}
args = ine(args)
if reading ~= '' then
local title = export.tag((args or ''), (args or ''), (args or export.straitdiff(args, 1, 'tag')), (args or ''), (args or ''))
local pinyin = export.straitdiff(reading, 1, '')
table.insert(text, export.make_table(title, pinyin, (args or ''), (args or ''), (args or ''), (args or ''), (args or ''), (args or args or '')))
if args and args ~= '' then
title, pinyin = erhua(pinyin, args, pagename)
table.insert(text, export.make_table(title, pinyin, '', (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
end
end
if reading ~= '' or export.straitdiff(reading, 2, 'exist') then
if args and args ~= '' then tagb = args else tagb = export.straitdiff(args, 2, 'tag') end
title = export.tag((args or ''), (args or ''), tagb, (args or ''), (args or ''))
pinyin = (reading ~= '') and reading or export.straitdiff(reading, 2, '')
table.insert(text, export.make_table(title, pinyin, (args or ''), (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
table.insert(text, ', 1, '') or args) .. ']]')
if args and args ~= '' then
title, pinyin = erhua(pinyin, args, pagename)
table.insert(text, export.make_table(title, pinyin, '', (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
end
if reading ~= '' or export.straitdiff(reading, 3, 'exist') then
if args and args ~= '' then tagb = args else tagb = export.straitdiff(args, 3, 'tag') end
title = export.tag((args or ''), (args or ''), tagb, (args or ''), (args or ''))
if reading ~= '' then pinyin = reading else pinyin = export.straitdiff(reading, 3, '') end
table.insert(text, export.make_table(title, pinyin, (args or ''), (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
if args and args ~= '' then
title, pinyin = erhua(pinyin, args, pagename)
table.insert(text, export.make_table(title, pinyin, '', (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
end
if reading ~= '' or export.straitdiff(reading, 4, 'exist') then
if args and args ~= '' then tagb = args else tagb = export.straitdiff(args, 4, 'tag') end
title = export.tag((args or ''), (args or ''), tagb, (args or ''), (args or ''))
if reading ~= '' then pinyin = reading else pinyin = export.straitdiff(reading, 4, '') end
table.insert(text, export.make_table(title, pinyin, (args or ''), (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
if args and args ~= '' then
title, pinyin = erhua(pinyin, args, pagename)
table.insert(text, export.make_table(title, pinyin, '', (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
end
if reading ~= '' or export.straitdiff(reading, 5, 'exist') then
if args and args ~= '' then tagb = args else tagb = export.straitdiff(args, 5, 'tag') end
title = export.tag((args or ''), (args or ''), tagb, (args or ''), (args or ''))
if reading ~= '' then pinyin = reading else pinyin = export.straitdiff(reading, 5, '') end
table.insert(text, export.make_table(title, pinyin, (args or ''), (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
if args and args ~= '' then
title, pinyin = erhua(pinyin, args, pagename)
table.insert(text, export.make_table(title, pinyin, '', (args or ''), (args or ''), (args or ''), (args or ''), (args or args or ''), true))
end
end
end
end
end
if (args or '') .. (args or '') .. (args or '') .. (args or '') .. (args or '') .. (args or '') ~= '' then
table.insert(text, ', 1, '') .. ']]')
end
return table.concat(text)
end
function export.make_audio(args)
local text, reading, pinyin = {}, {}, ""
local audio = {
args or args or '',
args or args or '',
args or args or '',
args or args or '',
args or args or '',
}
for i=1,5 do
reading = args or ''
if i == 1 then
pinyin = export.straitdiff(reading, 1, '')
else
pinyin = (reading ~= '') and reading or export.straitdiff(reading, i, '')
end
pinyin = export.py_format(pinyin, false, false, true)
add_audio(text, audio, pinyin)
end
return table.concat(text)
end
function add_audio(text, audio, pinyin)
if audio and audio ~= "" then
if audio == "y" then audio = string.format('zh-%s.ogg', pinyin) end
table.insert(text, '\n*:: [[File:')
table.insert(text, audio)
table.insert(text, ']]')
table.insert(text, ']')
end
end
function export.make_table(title, pinyin, py, cap, tl, tl2, tl3, a, novariety)
py = ine(py);cap = ine(cap);tl = ine(tl);tl2 = ine(tl2);tl3 = ine(tl3);a = ine(a);novariety = ine(novariety)
local text = {}
local pinyin_simple_fmt = export.py_format(pinyin, false, false, true)
local pinyin_simple_fmt_nolink = export.py_format(pinyin, false, false, true, true)
if not novariety then
table.insert(text, '* ]')
else
table.insert(text, '<br>')
end
table.insert(text, '\n** <small>' .. title .. '</small>')
local hom_found
if get_hom_data(pinyin_simple_fmt).list then
hom_found = true
else
hom_found = false
table.insert(text, '<sup><small><abbr title="Add Mandarin homophones"><span class="plainlinks">="edit"})) .. ' +]</span></abbr></small></sup>')
end
table.insert(text, "\n*** <small>'']''</small>: ")
local id = m_zh.ts_determ(mw.title.getCurrentTitle().text)
if id == 'trad' then
table.insert(text, '<span class="form-of pinyin-t-form-of transliteration-')
table.insert(text, m_zh.ts(mw.title.getCurrentTitle().text))
elseif id == 'simp' then
table.insert(text, '<span class="form-of pinyin-s-form-of transliteration-')
table.insert(text, m_zh.st(mw.title.getCurrentTitle().text))
else -- both
table.insert(text, '<span class="form-of pinyin-ts-form-of')
end
table.insert(text, '" lang="cmn" style="font-family: Consolas, monospace;">')
if py then
table.insert(text, py)
else
if cap then
table.insert(text, export.py_format(pinyin, true, true))
else
table.insert(text, export.py_format(pinyin, false, true))
end
if tl or tl2 or tl3 then
table.insert(text, ' → ')
if tl then tl_pos = 1 elseif tl2 then tl_pos = 2 elseif tl3 then tl_pos = 3 end
if cap then
table.insert(text, export.make_tl(export.py_format(pinyin, true, false, true, true), tl_pos, true, true))
else
table.insert(text, export.make_tl(pinyin_simple_fmt_nolink, tl_pos, true))
end
end
if tl then table.insert(text, ' <small>(toneless final syllable variant)</small>')
elseif tl2 or tl3 then table.insert(text, ' <small>(toneless variant)</small>') end
end
table.insert(text, "</span>\n*** <small>'']''</small>: ")
table.insert(text, '<span lang="zh-Bopo" class="Bopo">')
table.insert(text, export.py_zhuyin(pinyin_simple_fmt, true))
if tl or tl2 or tl3 then
table.insert(text, ' → ')
table.insert(text, export.py_zhuyin(export.make_tl(pinyin_simple_fmt_nolink, tl_pos, false), true))
end
table.insert(text, '</span>')
if tl then table.insert(text, ' <small>(toneless final syllable variant)</small>')
elseif tl2 or tl3 then table.insert(text, ' <small>(toneless variant)</small>') end
table.insert(text, "\n*** <small>'']''</small>: <code>")
if tl or tl2 or tl3 then
table.insert(text, export.py_gwoyeu(export.make_tl(pinyin_simple_fmt_nolink, tl_pos, false), pinyin_simple_fmt))
else
table.insert(text, export.py_gwoyeu(pinyin_simple_fmt))
end
table.insert(text, '</code>')
table.insert(text, "\n*** <small>'']''</small>: <code>")
if tl or tl2 or tl3 then
table.insert(text, export.py_tongyong(export.make_tl(pinyin_simple_fmt_nolink, tl_pos, false), pinyin_simple_fmt))
else
table.insert(text, export.py_tongyong(pinyin_simple_fmt))
end
table.insert(text, '</code>')
table.insert(text, "\n*** <small>'']''</small>: <code>")
if tl or tl2 or tl3 then
table.insert(text, export.py_wg(export.make_tl(pinyin_simple_fmt_nolink, tl_pos, false), pinyin_simple_fmt))
else
table.insert(text, export.py_wg(pinyin_simple_fmt))
end
table.insert(text, '</code>')
table.insert(text, '\n*** <small>Sinological ] <sup>(])</sup></small>: <span class="IPA">/')
table.insert(text, export.py_ipa(pinyin))
if tl or tl2 or tl3 then
table.insert(text, '/ → /')
table.insert(text, export.py_ipa(export.make_tl(pinyin_simple_fmt_nolink, tl_pos, false)))
end
table.insert(text, '/</span>')
-- if a then
-- if a == 'y' then a = 'zh-' .. pinyin_simple_fmt .. '.ogg' end
-- table.insert(text, '\n*** <div style="display:inline-block; position:relative; top:0.5em;">[[File:')
-- table.insert(text, a)
-- table.insert(text, ']]</div>]')
-- end
if hom_found then
table.insert(text, "\n*** <small>Homophones</small>: " ..
'<table class="wikitable" style="width:15em;margin:0; position:left; text-align:center">' ..
'<tr><th class="mw-customtoggle-cmnhom" style="color:#3366bb"></th></tr>' ..
'<tr class="mw-collapsible mw-collapsed" id="mw-customcollapsible-cmnhom">' ..
'<td><sup><div style="float: right; clear: right;"><span class="plainlinks">[')
table.insert(text, tostring(mw.uri.fullUrl("Module:zh/data/cmn-hom/" .. hom_data_part(pinyin_simple_fmt), {="edit"})))
table.insert(text, ' edit]</span></div></sup>')
table.insert(text, export.homophones(mw.ustring.lower(pinyin_simple_fmt)))
table.insert(text, '</td></tr></table>')
end
return table.concat(text)
end
return export