Модул:category tree/poscatboiler/data/characters
Документацију овог модула можете да направите на страници Модул:category tree/poscatboiler/data/characters/док
local raw_categories = {}
local handlers = {}
local raw_handlers = {}
local m_str_utils = require("Модул:string utilities")
local codepoint = m_str_utils.codepoint
local insert = table.insert
local ulen = m_str_utils.len
local ulower = m_str_utils.lower
local umatch = m_str_utils.match
local toNFC = mw.ustring.toNFC
local toNFD = mw.ustring.toNFD
local function track(page)
require("Модул:debug/track")("poscatboiler-characters/" .. page)
return true
end
-----------------------------------------------------------------------------
-- --
-- RAW CATEGORIES --
-- --
-----------------------------------------------------------------------------
raw_categories["Термини са јединственим карактерима подкатегорије по језику"] = {
description = "Umbrella categories covering terms categorized by unusual characters contained in them.",
additional = "{{{umbrella_meta_msg}}}",
parents = {
"Кишобран метакатегорија",
{name = "изрази по својим појединачним карактерима", is_label = true, sort = " "},
},
}
-- FIXME! This should probably be deleted.
raw_categories["Слова"] = {
description = "Categories specifying individual letters, containing the languages that use those letters.",
additional = "{{{umbrella_meta}}}",
parents = {
"Почетак",
},
}
-----------------------------------------------------------------------------
-- --
-- HANDLERS --
-- --
-----------------------------------------------------------------------------
-- If char is a combining character, returns a version with a dotted circle before it.
function add_dotted_circle(char, combining)
return combining and "◌" .. char or char
end
insert(handlers, function(data)
-- NOTE: The "character" in the title may actually be a description such as
-- "gershayim". In that case, char= is specified as a parameter indicating the
-- actual character.
local titlechar = data.label:match("^који се пишу са (.+)$")
if not titlechar then
return nil
end
local params = {
["char"] = {},
["sort"] = {},
-- Not sure what used to be done with the following parameters.
["context"] = {},
["context2"] = {},
}
local args = require("Module:parameters").process(data.args, params)
if args.context or args.context2 then
track("terms-spelled-with-context")
end
local special_cases = {
numbers = {
sort = "#",
desc = "numeric digits",
},
emoji = {
sort = "⌚", -- the first emoji in our list in [[Module:headword/data]]
},
parentheses = {
sort = "(",
},
["square brackets"] = {
sort = "[",
},
["angle brackets"] = {
sort = "<",
},
braces = {
sort = "{",
},
}
if special_cases[titlechar] then
local sortkey = args.sort or special_cases[titlechar].sort
return {
description = "{{{langname}}} terms spelled with one or more " .. (special_cases[titlechar].desc or titlechar) .. ".",
parents = {{name = "изрази по својим појединачним карактерима", sort = sortkey }},
breadcrumb = titlechar,
umbrella = {
breadcrumb = titlechar,
parents = {{name = "Термини са јединственим карактерима подкатегорије по језику", sort = " " .. sortkey }}
},
}, true
end
local char = args.char or titlechar
local titlechar_is_desc = args.char and args.char ~= titlechar
if titlechar_is_desc then
track("titlechar_is_desc")
end
local lang = data.lang or require("Модул:languages").getByCode("mul")
local combining = ulen(char) == 1 and require("Модул:Unicode_data").is_combining(codepoint(char))
local specials = {["ß"] = "ẞ", ["ͅ"] = "ͅ"}
local upper = toNFD(char)
:gsub("[%z\1-\127\194-\244][\128-\191]*", function(m)
return specials[m] or m:uupper()
end)
upper = toNFC(upper)
if char ~= upper and ulen(char) == 1 then
-- We want uppercase characters; but unless we're careful, we run into an issue with
-- [[Category:English terms spelled with ı]] due to the weird behavior of this character,
-- which has standard I as its uppercase equivalent.
local standard_chars = lang:getStandardCharacters()
if lang and standard_chars then
local function err()
error("Category titles should use uppercase characters: '" .. data.label .. "'", 2)
end
if lang:getCode() ~= "hi" and lang:getCode() ~= "lo" then
if not umatch(standard_chars, upper) then
err()
end
elseif not umatch(upper, "[" .. standard_chars .. "]") then
err()
end
end
end
-- Compute description.
-- If the letter has a lowercase form, show it.
local character = require("Модул:links").full_link(
{
term = char,
alt = combining and add_dotted_circle(char, true) or nil,
lang = lang,
tr = combining and "-" or nil,
},
"term"
)
if ulower(char) ~= char then
character = "upper case " .. character .. " or lower case " ..
require("Модул:links").full_link(
{
term = ulower(char),
lang = lang
},
"term"
)
end
if titlechar_is_desc then
character = character .. " (" .. titlechar .. ")"
end
local description = "{{{langname}}} термини који се пишу са " .. character .. "."
-- Set tagged character for displaytitle and breadcrumb.
local tagged_titlechar = not titlechar_is_desc and
require("Модул:script utilities").tag_text(titlechar, lang, nil, "term") or nil
local tagged_char = titlechar_is_desc and titlechar or
require("Модул:script utilities").tag_text(add_dotted_circle(char, combining), lang, nil, "term")
local han = umatch(char, "^[" .. require("Модул:scripts").getByCode("Hani").characters .. "]+$")
-- Make the sortkey. Always use Hani-sortkey for Han characters, as this circumvents any reading-based sortkey methods.
local sortkey = args.sort or han and require("Module:Hani-sortkey").makeSortKey(char) or lang:makeSortKey(char)
-- Use the char as a fallback.
if sortkey == "" then
sortkey = char
end
return {
description = description,
additional = "Note that categories of the form '''''LANG terms spelled with CHAR''''' are intended for characters not "
.. "part of the standard repertoire of a language (e.g. Cyrillic characters in English or Latin characters in Russian).",
displaytitle = not titlechar_is_desc and "{{{langname}}} термини који се пишу са " .. tagged_titlechar or nil,
parents = {{name = "изрази по својим појединачним карактерима", sort = sortkey }},
breadcrumb = tagged_char,
umbrella = not han and {
displaytitle = not titlechar_is_desc and "Термини који се пишу са " .. tagged_titlechar .. " по језику" or nil,
breadcrumb = tagged_char,
parents = {{name = "Термини са јединственим карактерима подкатегорије по језику", sort = " " .. sortkey }}
} or false,
}, true
end)
-----------------------------------------------------------------------------
-- --
-- RAW HANDLERS --
-- --
-----------------------------------------------------------------------------
-- Special-cased categories that we allow, for Turkish letters.
local letter_cat_allow_list = require("Модул:table/listToSet") {
"İi",
}
insert(raw_handlers, function(data)
-- Only recognize cases consisting of an uppercase letter followed by the
-- corresponding lowercase letter, either as the entire category name or
-- followed by a colon (for cases like [[Category:Gg: ⠛]]). Cases that
-- don't fit this profile (e.g. for Turkish [[Category:İi]] and
-- [[Category:Iı]]) need to call {{letter cat}} directly. Formerly this
-- handler was much less restrictive and would fire on categories named
-- [[Category:zh:]], [[Category:RFQ]], etc.
local upper, lower = umatch(data.category, "^(%u)(%l)%f[:%z]")
if not upper or not letter_cat_allow_list[data.category] and lower:uupper() ~= upper then
return nil
end
return {
description = ('Languages that use the uppercase letter "%s" (lowercase equivalent "%s").'):format(upper, lower),
parents = {"Слова"},
}
end)
return {RAW_CATEGORIES = raw_categories, HANDLERS = handlers, RAW_HANDLERS = raw_handlers}