local p = {}
local namedRefTable = mw.loadData( 'Modul:Unicode2NamedRef' )
local needsNamedRefs, needsUTF16
local function getArgNums(pargs, prefix)
-- Returns an ordered table containing the numbers (>0) of the arguments
-- that exist for the specified prefix. Includes 0 if prefix without number
-- is used as argument. For example, if the prefix was 'seznam', and
-- 'seznam1', 'seznam', and 'seznam5' exist, it would return {0, 1, 5}.
local nums = { }
if pargs[prefix] then table.insert(nums, 0) end
for k, v in pairs(pargs) do
local num = tostring(k):match('^' .. prefix .. '([1-9]%d*)$')
if num then table.insert(nums, tonumber(num)) end
end
table.sort(nums)
return nums
end
local function unicode2Named(unicode)
-- Returns named character reference for given Unicode code point or nil.
local named = replacementTable[tonumber(unicode,16)]
return named
end
local function unicode2UTF8(unicode)
local utf8 = {}
if tonumber(unicode,16)<0x110000 then
if tonumber(unicode,16)>0xffff then -- 4B
utf8[4] = 0xf0 + floor(tonumber(unicode,16)/0x40000)
utf8[3] = 0x80 + floor(tonumber(unicode,16)%0x40000/0x1000)
utf8[2] = 0x80 + floor(tonumber(unicode,16)%0x1000/0x40)
utf8[1] = 0x80 + tonumber(unicode,16)%0x40
elseif tonumber(unicode,16)>0x7ff then -- 3B
utf8[3] = 0x80 + floor(tonumber(unicode,16)%0x40000/0x1000)
utf8[2] = 0x80 + floor(tonumber(unicode,16)%0x1000/0x40)
utf8[1] = 0x80 + tonumber(unicode,16)%0x40
elseif tonumber(unicode,16)>0x7f then -- 2B
utf8[2] = 0x80 + floor(tonumber(unicode,16)%0x1000/0x40)
utf8[1] = 0x80 + tonumber(unicode,16)%0x40
else -- 1B
utf8[1] = tonumber(unicode,16)
end
end
return utf8
end
local function unicode2UTF16(unicode)
local utf16 = {}
if tonumber(unicode,16)<0x110000 then
if tonumber(unicode,16)>0xffff then -- 2B
utf16[2] = 0xd800 + floor(tonumber(unicode,16)/0x400)
utf16[1] = 0xdc00 + tonumber(unicode,16)%0x400
else -- 1B
utf16[1] = tonumber(unicode,16)
end
end
return utf16
end
function p.charmap(frame)
pargs = frame:getParent().args
local tableClass = pargs['tableClass'] or 'wikitable'
local tableStyle = pargs['tableStyle'] or 'text-align:right'
local maps = getArgNums(pargs,"map")
local name,csname,image,encoding,unicode,utf8,utf16,numref,namedref,a,s = {},{},{},{},{},{},{},{},{}
for i, _ in pairs(pargs) do
table.insert (image,pargs["image" .. tostring(i)] or pargs["obrázek " .. tostring(i)] or ("&#x" .. pargs[i] ..";"))
table.insert (name,pargs["name" .. tostring(i)] or pargs["Unicode název " .. tostring(i)] or "")
table.insert (csname,pargs["csname" .. tostring(i)] or pargs["Unicode česky " .. tostring(i)] or "")
table.insert (encoding,"dec")
table.insert (encoding,"hex")
table.insert (unicode,tonumber(pargs[i],16))
if pargs["info" .. tostring(i)] or (pargs[info] and (pargs["info" .. tostring(i)]==nil)) then
table.insert (unicode,'[http://www.fileformat.info/info/unicode/char/' .. pargs[i] .. " U+" .. pargs[i] .. "]")
else
table.insert (unicode,"U+" .. pargs[i])
end
a = unicode2UTF8(pargs[i])
s = {}
for j, _ in a do s[5-j]=string.format('%d',a[j]) end
table.insert (utf8,table.concat(s," "))
s = {}
for j, _ in a do s[5-j]=string.format('%xx',a[j]) end
table.insert (utf8,table.concat(s," "))
a = unicode2UTF16(pargs[i])
if tonumber(pargs[i],16)>0xffff then
table.insert (utf16,a[2] .. ' ' .. a[1])
table.insert (utf16,string.format('%xxxx',a[2]) .. ' ' .. string.format('%xxxx',a[1]))
needsUTF16 = true
else
table.insert (utf16,a[1])
table.insert (utf16,string.format('%xxxx',a[1]))
end
table.insert (utf16,unicode2UTF16(pargs[i]))
table.insert (numref,'&#' .. tonumber(pargs[i],16))
table.insert (numref,'&#' .. pargs[i])
table.insert (namedref,unicode2Named(pargs[i]) or "")
if unicode2Named(parg[i]) then needsNamedRefs = true end
-- ... =============TODO maps=================
end
result = '<table class="' .. tableClass .. '" style="' .. tableStyle .. '"><thead><tr><th>Znak</th><th colspan="2">' .. table.concat (image, '</th><th colspan="2">') .. '</th></tr><tr><th>Název v Unicodu</th><td colspan="2">' .. table.concat (name, '</td><td colspan="2">') .. '</td></tr><tr><th>Český název</th><td colspan="2">' .. table.concat (csname, '</td><td colspan="2">') .. '</td></tr><tr><th>Kódování</th><th>' .. table.concat (encoding, '</th><th>') .. '</th></tr></thead><tbody><tr><th>[[Unicode]]</th><td>' .. table.concat (unicode, '</td><td>') .. '</td></tr><tr><th>[[UTF-8]]</th><td>' .. table.concat (utf8, '</td><td>') .. '</td></tr>'
if needsUTF16 then
result = result .. '<tr><th>[[UTF-16]]</th><td>' .. table.concat (utf16, '</td><td>') .. '</td></tr>'
end
result = result .. '<tr><th>[[HTML entita|Číselná entita]]</th><td>' .. table.concat (numref, '</td><td>') .. '</td></tr>'
if needsNamedRefs then
result = result .. '<tr><th>[[HTML entita|Názvová entita]]</th><td colspan="2">' .. table.concat (namedref, '</td><td colspan="2">') .. '</td></tr>'
end
-- ... =============TODO maps=================
result = result .. '</tbody></table>'
return result
end
return p
