local export = {}
local find = mw.ustring.find
local gsub = mw.ustring.gsub
local match = mw.ustring.match
local len = mw.ustring.len
function export.extract_gloss(content, useetc)
local senses = {}
local len = mw.ustring.len
local sense_id = 0
local etc = false
local pos = 1
local zh_section
-- Find the Chinese section
while true do
local _, j, language_name, section = content:find("%f[=]==%s*([^=]+)%s*==(\n.-)\n==%f[^=]", pos)
if j == nil then
_, j, language_name, section = content:find("%f[=]==%s*([^=]+)%s*==(\n.+)", pos)
end
if j == nil then
break
else
pos = j - 1
end
if language_name:match("^%s*중국어%s*$") then
zh_section = section
break
end
end
if not zh_section then
return nil -- Return nil if the Chinese section is not found
end
-- Extract gloss from the Chinese section
for sense in zh_section:gmatch("# ([^#\n]+)") do
sense_id = sense_id + 1
if sense_id > 2 then
etc = true
break
end
table.insert(senses, sense)
end
local gloss_text = senses[1] or ""
local gloss_text_extend = gloss_text .. (senses[2] and "; " .. senses[2] or "")
gloss_text = (len(gloss_text) < 80 and len(gloss_text_extend) < 160) and gloss_text_extend or gloss_text
if gloss_text ~= gloss_text_extend then etc = true end
gloss_text = gloss_text:gsub("%.$", "")
return gloss_text
end
function export.make(frame)
local params = {
[1] = { list = true, allow_holes = true, allow_empty = true },
["s"] = { list = true },
["t"] = { list = true },
["alt"] = {}, ["type"] = {}, ["delink"] = {}, ["lit"] = {}, ["note"] = {}, ["gloss"] = {}
}
local args = require("Module:parameters").process(frame:getParent().args, params)
local comp_type = args["type"]
local s, t = {}, {}
local annotation = {}
if not frame:getParent().args["t"] then
table.insert(t, 1, mw.title.getCurrentTitle().subpageText)
end
local function insert_st(set, text)
if find(text, "^ ") or find(text, " $") then
error("Please remove the leading and / or trailing space(s) in the 's' and 't' parameters.")
end
table.insert(set, text)
end
for i = 1, #args.s do insert_st(s, args.s[i]) end
for i = 1, #args.t do insert_st(t, args.t[i]) end
s.name = "simp"
t.name = "trad"
if #t ~= 1 and #s == 0 then
table.insert(s, t[1])
end
local function asterisk(term, iscomp)
if iscomp and len(t[1]) > 1 then return "" end
if term == mw.title.getCurrentTitle().subpageText or not (mw.title.new(term) or {}).exists then return "" end
local content = mw.title.new(term):getContent()
content = gsub(content, "zh%-pron", "Ꙁ")
content = gsub(content, "zh%-see", "Ꙁ")
content = gsub(content, "[^Ꙁ]", "")
return mw.ustring.len(content) > 1 and '<sup><span class="explain" title="This form has one or more other meanings.">*</span></sup>' or ''
end
local function var_fmt(length, color)
return '\n|-\n! style="padding: 0.' .. (length > 8 and '3' or '5') ..
'em;border: 1px solid #aaa;background: #' .. (color or 'E0FFFF') ..
';font-weight: normal;font-size: smaller;" colspan="2" |'
end
local function char_gap(length)
return '\n| style="padding: 0.' .. (length > 8 and '3' or '5') ..
'em; background-color:white;border-bottom: 1px solid #aaa; font-size:x-large" lang="zh" class="Hani" | '
end
local function char_last(length)
return '\n| style="padding: 0.' .. (length > 8 and '3' or '5') ..
'em; background-color:white;border-right: 1px solid #aaa;border' ..
(length ~= 1 and '-bottom' or '') .. ': 1px solid #aaa; font-size:x-large" lang="zh" class="Hani" | '
end
local function header(length, var_count)
return ((length > 3 or var_count * length > 5) and ':{|' or '{| align=right') ..
' style="clear: right;margin: 1em;border-collapse: collapse;text-align: center"' ..
(length ~= 1 and '\n|-\n! colspan=2|' or '')
end
local function gloss_fmt(word, colspan, length)
return '\n! style="padding: 0.' .. (length > 8 and '3' or '5') ..
'em;border: 1px solid #aaa;background:#F5F5DC;font-weight: normal;font-size: 80%; width:' ..
(length <= 8 and (30 * word + 30) or (25 * word + 25)) .. 'px" colspan=' .. (colspan or 1) ..'|'
end
local function form_fmt(text, length)
for i,value in ipairs(text) do
text[i] = '[[' .. value .. ']]' .. asterisk(value, false)
end
return length ~= 1 and ((length > 8 and '' or '<span style="font-size:140%">') ..
'(<span lang="zh" class="Hani">' .. table.concat(text, "/") .. '</span>)' ..
(length > 8 and '' or '</span>')) or ''
end
local function char_fmt(text, length)
return (#text ~= 1 and (char_gap(length) .. table.concat(text, char_gap(length), 1, #text-1)) or '') .. char_last(length) .. text[#text]
end
local test_word = t[1]
local length = len(test_word)
local word_division = {}
local i = 1
local decomposable
if comp_type then
for index in mw.text.gsplit(comp_type, "", true) do
if find(mw.ustring.sub(test_word, i, i), '[，%-]') then
table.insert(word_division, { i, i } )
i = i + 1
elseif mw.ustring.sub(test_word, i, i) == '…' then
table.insert(word_division, { i, i + 1 } )
i = i + 2
end
table.insert(word_division, { i, i + index - 1 } )
i = i + index
end
if i - 1 ~= len(gsub(test_word, '…+$', '')) and not find(table.concat(t) .. table.concat(s), "[⿰⿱⿲⿳⿴⿵⿶⿷⿸⿹⿺⿻]") then
error("'type' parameter does not match word length.")
end
else
for i = 1, length do
table.insert(word_division, { i, i } )
end
decomposable = len(gsub(test_word, '…+$', '')) > 2 and true or false
end
local delink = {}
if args["delink"] then
if args["delink"] == "y" then
for del_index, _ in ipairs(word_division) do
delink[del_index] = "yes"
end
else
for position in mw.text.gsplit(args["delink"], ",") do
delink[tonumber(position)] = "yes"
end
end
end
local char_set = { ['simp'] = {}, ['trad'] = {} }
local identity = #s == 0 and {t} or {s,t}
local uncreated = {}
for _, id in ipairs(identity) do
for i, position in ipairs(word_division) do
local char_string = ""
for j = 1, #id do
local word_form = mw.ustring.sub(id[j], position[1], position[2])
if not find(char_string, word_form) then
char_string = (char_string ~= "" and (char_string .. '/') or "") .. word_form
end
end
if not find(char_string, '[，%-]') then
local hash = {}
for thing in mw.text.gsplit(char_string, (delink[i] and "" or "/")) do
table.insert(hash, '[[' .. thing .. ']]' .. asterisk(thing, true))
end
char_string = table.concat(hash, (delink[i] and "" or "/"))
end
table.insert(char_set[id.name], char_string)
end
for _, item in ipairs(id) do
if not (mw.title.new(item) or {}).exists and item ~= mw.title.getCurrentTitle().subpageText then
table.insert(uncreated, '"[[' .. item .. ']]"')
end
end
end
local scripts = { ['一-龯㐀-䶵'] = 'Hani' , ['a-zA-ZāēīōūĀĒĪŌŪａ-ｚＡ-Ｚ'] = 'Latn', ['0-9０-９'] = 'Numb', ['Ͱ-Ͽ'] = 'Grek' }
local script = {}
for range, script_name in pairs(scripts) do
if find(test_word, '[' .. range .. ']') then
table.insert(script, script_name)
end
end
local altforms = ""
if args["alt"] then
local altform_list = {}
for altform in mw.text.gsplit(args["alt"], ",") do
local altdecomp = mw.text.split(altform, ":")
local altdecomp2 = mw.text.split(altdecomp[1], "-")
table.insert(altform_list, '<span style="white-space:nowrap;">[[' .. altdecomp2[1] .. ']]</span>')
end
if #altform_list > 5 then
altforms = '<div class="vsSwitcher vsToggleCategory-glosses"><span class="vsToggleElement"> </span>' ..
'<div class="vsShow" style="display:none">' .. table.concat(altform_list, "<br>", 1, 5) ..
'</div><div class="vsHide">' .. table.concat(altform_list, "<br>") .. '</div></div>'
else
altforms = table.concat(altform_list, "<br>")
end
altforms = var_fmt(length, 'F0FFE0') .. '이체자' .. mw.ustring.sub(char_last(length), 1, -45)
.. 'font-size:90%; padding-top: 10px; padding-bottom: 10px; border-right: 1px solid #aaa; border-bottom: 1px solid #aaa"'
.. ' colspan="' .. #word_division .. '"|' .. altforms
end
local anagram = ""
if len(t[1]) == 2 and not find(t[1], "(.)%1") then
local anagram_link = mw.ustring.sub(t[1], 2, 2) .. mw.ustring.sub(t[1], 1, 1)
local anagram_content = mw.title.new(anagram_link):getContent() or false
if anagram_content and find(anagram_content, "== 중국어 ==") then
anagram = var_fmt(length, 'F0FFE0') .. '어구전철' .. mw.ustring.sub(char_last(length), 1, -45)
.. 'font-size:90%; padding-top: 10px; padding-bottom: 10px; border-right: 1px solid #aaa; border-bottom: 1px solid #aaa"'
.. ' colspan="' .. #word_division .. '"|[[' .. anagram_link .. "]]"
end
end
local literal = (args["lit"] or args["note"]) and '\n|-' ..
gloss_fmt(length, #word_division + 2, length) ..
(args["lit"] and '글자 그대로: “' .. args["lit"] ..
(find(args["lit"], "%.$") and "”" or "”.") or args["note"]) or ""
local gloss = {}
if args['gloss'] == '-' then
gloss = { gloss_fmt(length * 1.6, #word_division, length) .. '(음역)' }
elseif length == 1 then
gloss = {}
else
for i, position in ipairs(word_division) do
local character = mw.ustring.sub(t[1], position[1], position[2])
local gloss_text = args[1][i]
if gloss_text == "-" then gloss_text = "(음역)" end
if not gloss_text then
gloss_text = ""
end
if find(gloss_text, "-", nil, true) then
local nonbreaking_hyphen = mw.ustring.char(0x2011)
gloss_text = mw.ustring.gsub(gloss_text, "^%-", nonbreaking_hyphen)
gloss_text = mw.ustring.gsub(gloss_text, "(%s)%-", "%1" .. nonbreaking_hyphen)
end
if gloss_text == "" and position[2] > position[1] then
local senses = {}
local content = mw.title.new(character):getContent() or false
if content then
gloss_text = export.extract_gloss(content, false)
end
end
gloss_text = gsub(gloss_text, ";[^a-zA-Z0-9]+;", ";")
gloss_text = gsub(gloss_text, ";[^a-zA-Z0-9]*$", "")
gloss_text = gsub(gloss_text, ";+", ";")
if len(gsub(gloss_text, '[^;]', '')) > 2 then
gloss_text = '<div class="vsSwitcher vsToggleCategory-glosses"><span class="vsToggleElement"> </span><div class="vsShow" style="display:none">' ..
match(gloss_text, '^[^;]+;[^;]+;[^;]+') .. '</div><div class="vsHide">' .. gloss_text .. '</div></div>'
end
local word_length = match(character, '[，…%-]') and 0 or
(delink[i] and len(character) or len(character) * (len(gsub(char_set['trad'][i], '[^/]', '')) + 1))
table.insert(gloss, gloss_fmt(word_length, 1, length) .. gloss_text)
end
end
local PAGENAME = mw.title.getCurrentTitle().text
local content = mw.title.new(PAGENAME):getContent()
if content then
local applicable_pos = { ["Noun"] = 1, ["Verb"] = 1, ["Adjective"] = 1, ["Adverb"] = 1,
["Definitions"] = 1, ["Pronunciation"] = 1 }
local previous_level = 2
local subheading_wanted
for equals, heading_text in string.gmatch(content, "\n(%=%=+)([^%=]+)%=%=+") do
local current_level = #equals
if subheading_wanted and current_level ~= previous_level + 1 then
if applicable_pos[heading_text] then
error("The heading \"===어원 ''n''===\" or \"===발음 ''n''===\" must be followed by a subheading one level lower.")
end
end
previous_level = current_level
subheading_wanted = string.find(heading_text, " [0-9]+") and true or false
end
end
return table.concat{
header(length, math.max(#t, #s)), table.concat(gloss, ""),
(#identity == 1 and
(var_fmt(length) .. '[[정체]]/[[간체]]<br>' ..
form_fmt(t, length) .. char_fmt(char_set['trad'], length))
or
var_fmt(length) .. '[[정체]] ' ..
form_fmt(t, length) .. char_fmt(char_set['trad'], length) ..
(var_fmt(length) .. '[[간체]] ' ..
form_fmt(s, length) .. char_fmt(char_set['simp'], length))
), altforms, anagram, literal, '\n|}', table.concat(annotation)
}
end
return export
