Module:Sandbox/Erutuon: Difference between revisions

Revision as of 07:42, 2 July 2018

Revision as of 02:07, 3 July 2018

Line 36:

local Latn_pattern = table.concat {

'[',

'[', -- this is a set so include opening bracket

'\n\32-\127',

'\n\32-\127', -- C0 Controls and Basic Latin U+0020–U+007E (20 - 7E) + (U+0010 and U+007F <poem>...</poem> support)

'\194\160-\194\172', -- C1 Controls and Latin-1 Supplement U+00A0-U+00AC (C2 A0 - C2 AC)

'\195\128-\195\191', -- (skip shy) U+00C0–U+00FF (C3 80 - C3 BF)

'\196\128-\197\191', -- Latin Extended-A U+0100–U+017F (C4 80 - C5 BF)

'\198\128-\201\143', -- Latin Extended-B U+0180–U+024F (C6 80 - C9 8F)

'\225\184\128-\225\187\191', -- Latin Extended Additional U+1E00-U+1EFF (E1 B8 80 - E1 BB BF)

'\226\177\160-\226\177\191', -- Latin Extended-C U+2C60–U+2C7F (E2 B1 A0 - E2 B1 BF)

'\234\156\160-\234\159\191', -- Latin Extended-D U+A720-U+A7FF (EA 9C A0 - EA 9F BF)

'\234\172\176-\234\173\175', -- Latin Extended-E U+AB30-U+AB6F (EA AC B0 - EA AD AF)

'\239\172\128-\239\172\134', -- Alphabetic Presentaion Forms U+FB00-U+FB06 (EF AC 80 - EF AC 86)

'\239\188\129-\239\188\188', -- Halfwidth and Fullwidth Forms U+FF01-U+FF3C (EF BC 81 - EF BC BC)

'–', -- ndash

'—', -- mdash

'«', '»',

'«', '»', -- guillemets commonly used in several 'Latn' languages

']',

']', -- close the set

};

Line 188:

local singles = script_data.singles

local ranges = script_data.ranges

local function clear (self)

for _, key in ipairs(m_table.keysToList(self, false)) do

self[key] = nil

end

end

local counts = {}

Line 195 ⟶ 201:

self[script_code] = (self[script_code] or 0) + (amount or 1)

end,

clear = function (self)clear,

}

for _, key in ipairs(m_table.keysToList(self, false)) do

})

self[key] = nil

local codepoints_per_script = {}

setmetatable(codepoints_per_script, {

__index = {

add = function(self, script_code, codepoint)

self[script_code] = self[script_code] or { n = 0 }

if self[script_code].n <= 0x20

and not (codepoint <= 0x9F and (codepoint +>= 10x80

or codepoint <= 0x1F)) then

if self[script_code].n == 0x20 then

local period = ('.'):byte()

for _ = 1, 3 do

self[script_code].n = self[script_code].n + 1

self[script_code][self[script_code].n] = period

end

else

if script_code == "Zinh" then -- probably combining character

self[script_code].n = self[script_code].n + 1

self[script_code][self[script_code].n] = 0x25CC

end

self[script_code].n = self[script_code].n + 1

self[script_code][self[script_code].n] = codepoint

end

end

end,

clear = clear,

}

})

Line 217 ⟶ 246:

if singles[codepoint] then

counts:increment(singles[codepoint])

codepoints_per_script:add(singles[codepoint], codepoint)

codepoint = codepoint + 1

else

Line 223 ⟶ 253:

local count = 0

while codepoint <= range[2] and codepoint <= block[2] do

codepoint = codepoint + 1

count = count + 1

codepoints_per_script:add(range[3], codepoint)

codepoint = codepoint + 1

end

counts:increment(range[3], count)

Line 254 ⟶ 285:

fun.map(

function (count, script)

return ('<abbr title="%s">%s</abbr> (<span title="%s">%d</span>)"')

:format(

script_data.aliases[script], script,

('<abbr title="%s">%s</abbr>')

:format(script_data.aliasescodepoints_per_script[script], script),

and mw.text.nowiki(mw.ustring.char(

unpack(codepoints_per_script[script])))

or "",

count)

end,

Line 267 ⟶ 301:

", "))

end

-- mw.logObject(codepoints_per_script, block[3])

counts:clear()

codepoints_per_script:clear()

end

output:insert "|}"