Module:collation: Difference between revisions

Jump to navigation Jump to search

Module:collation (view source)

Revision as of 18:22, 5 August 2024

1,320 bytes added ,  5 August 2024
no edit summary
(Created page with "local export = {} -- Custom functions for generating a sortkey that will achieve the desired sort -- order. -- name of module and name of exported function local custom_sort_...")
 
No edit summary
Line 1: Line 1:
local export = {}
local export = {}
local require = require
local byte = string.byte
local concat = table.concat
local find = string.find
local get_plaintext = require("Module:utilities").get_plaintext
local match = string.match
local memoize = require("Module:fun").memoize
local remove = table.remove
local sort = table.sort
local string_sort -- defined below as export.string_sort
local sub = string.sub
local trim = mw.text.trim
local type = type


-- Custom functions for generating a sortkey that will achieve the desired sort
-- Custom functions for generating a sortkey that will achieve the desired sort
-- order.
-- order.
-- name of module and name of exported function
-- name of module and name of exported function
local custom_sort_functions = {
local custom_funcs = {
ahk = { "Mymr-sortkey", "makeSortKey" },
aio = { "Mymr-sortkey", "makeSortKey" },
blk = { "Mymr-sortkey", "makeSortKey" },
egy = { "egy-utilities", "make_sortkey" },
egy = { "egy-utilities", "make_sortkey" },
kac = { "Mymr-sortkey", "makeSortKey" },
kht = { "Mymr-sortkey", "makeSortKey" },
ksw = { "Mymr-sortkey", "makeSortKey" },
kyu = { "Mymr-sortkey", "makeSortKey" },
["mkh-mmn"] = { "Mymr-sortkey", "makeSortKey" },
mnw = { "Mymr-sortkey", "makeSortKey" },
my  = { "Mymr-sortkey", "makeSortKey" },
phk = { "Mymr-sortkey", "makeSortKey" },
pwo = { "Mymr-sortkey", "makeSortKey" },
omx = { "Mymr-sortkey", "makeSortKey" },
shn = { "Mymr-sortkey", "makeSortKey" },
tjl = { "Mymr-sortkey", "makeSortKey" },
}
}


local function is_lang_object(lang)
local function is_lang_object(lang)
return type(lang) == "table" and type(lang.getCanonicalName) == "function"
return type(lang) == "table" and type(lang.getCanonicalName) == "function"
end
local function check_lang_object(funcName, argIdx, lang)
if not is_lang_object(lang) then
error("bad argument #" .. argIdx .. " to " .. funcName
.. ": expected language object, got " .. type(lang) .. ".", 2)
end
end
end


Line 24: Line 46:
.. ": expected function object, got " .. type(func) .. ".", 2)
.. ": expected function object, got " .. type(func) .. ".", 2)
end
end
return true
end
end


-- UTF-8-encoded characters that do not belong to the Basic Multilingual Plane
local function make_sortkey_func(lang, make_sortbase)
-- (that is, with code points greater than U+FFFF) have byte sequences that
local langcode = lang:getCode()
-- begin with the bytes 240 to 244.
local makeDisplayText = lang.makeDisplayText
local function contains_non_BMP(str)
local custom_func = custom_funcs[langcode]
return str:find '[\240-\244]'
end
local makeSortKey
 
if custom_func then
do
local _makeSortKey = require("Module:" .. custom_func[1])[custom_func[2]]
local byte, min = string.byte, math.min
function makeSortKey(_, text)
function export.laborious_comp(item1, item2)
return _makeSortKey(text, langcode)
local l1, l2 = #item1, #item2
for i = 1, min(l1, l2) do
local char1, char2 = byte(item1, i, i), byte(item2, i, i)
if char1 ~= char2 then
return char1 < char2
end
end
end
return l1 < l2
else
makeSortKey = lang.makeSortKey
end
end
end
function export.make_sortkey_func(lang, keyfunc)
check_lang_object("make_sortkey_func", 1, lang)
keyfunc = keyfunc or function(elem) return elem end
check_function("make_sortkey_func", 2, keyfunc)
local cache = {}
return make_sortbase and check_function("make_sortkey_func", 2, make_sortbase) and function(element)
return (makeSortKey(
local custom_sort_function = custom_sort_functions[lang:getCode()]
lang,
local makeSortKey =
(makeDisplayText(
custom_sort_function and require("Module:" .. custom_sort_function[1])[custom_sort_function[2]]
lang,
or function(text)
get_plaintext(make_sortbase(element))
return lang:makeSortKey(text)
))
end
))
end or function(element)
local m_links = require("Module:links")
return (makeSortKey(
lang,
return function (element)
(makeDisplayText(
element = keyfunc(element)
lang,
local result = cache[element]
get_plaintext(element)
))
if result then
))
return result
end
result = m_links.remove_links(element)
result = mw.ustring.gsub(result, "[%p ]", "")
result = makeSortKey(lang:makeEntryName(result))
cache[element] = result
return result
end
end
end
end


function export.make_compare_func(lang, non_BMP, keyfunc)
-- When comparing two elements with code points outside the BMP, the less-than
local make_sortkey = export.make_sortkey_func(lang, keyfunc)
-- operator treats all code points above U+FFFF as equal because of a bug in
-- glibc. See [[phab:T193096#4161287]]. Instead, compares bytes, which always
-- yields the same result as comparing code points in valid UTF-8 strings.
-- UTF-8-encoded characters that do not belong to the Basic Multilingual Plane
-- (that is, with code points greater than U+FFFF) have byte sequences that
-- begin with the bytes 240 to 244.
do
-- Memoize match because we've set the `simple` flag, which means it
-- should only be used with fixed additional arguments (in this case, the
-- pattern).
local match = memoize(match, true)
-- When comparing two elements with code points outside the BMP, the
function export.string_sort(item1, item2)
-- less-than operator does not work correctly because of a bug in glibc.
if match(item1, "^[^\240-\244]*$") and match(item2, "^[^\240-\244]*$") then
-- See [[phab:T193096]].
return item1 < item2
if non_BMP then
return function (elem1, elem2)
return export.laborious_comp(make_sortkey(elem1), make_sortkey(elem2))
end
end
else
local i = 0
return function (elem1, elem2)
while true do
return make_sortkey(elem1) < make_sortkey(elem2)
i = i + 1
local b1, b2 = byte(item1, i, i), byte(item2, i, i)
if not b1 then
return b2 and true or false
elseif b1 ~= b2 then
return b2 and b1 < b2 or false
end
end
end
end
end
string_sort = export.string_sort
end
end


function export.sort(elems, lang, keyfunc)
function export.sort(elems, lang, make_sortbase)
local non_BMP
if not is_lang_object(lang) then
for _, elem in ipairs(elems) do
return sort(elems)
if keyfunc then
elem = keyfunc(elem)
end
if contains_non_BMP(elem) then
non_BMP = true
break
end
end
end
return table.sort(elems, is_lang_object(lang) and export.make_compare_func(lang, non_BMP, keyfunc) or nil)
local make_sortkey = memoize(make_sortkey_func(lang, make_sortbase), true)
return sort(elems, function(elem1, elem2)
return string_sort(make_sortkey(elem1), make_sortkey(elem2))
end)
end
end


Line 124: Line 138:
end
end
local elems = require("Module:table").shallowClone(args)
local m_table = require("Module:table")
local elems = m_table.shallowcopy(args)
local m_languages = require("Module:languages")
local m_languages = require("Module:languages")
local lang
local lang
if args.lang then
if args.lang then
lang = m_languages.getByCode(args.lang) or m_languages.err(args.lang, 'lang')
lang = m_languages.getByCode(args.lang) or m_languages.err(args.lang, "lang")
else
else
local code = table.remove(elems, 1)
local code = remove(elems, 1)
code = code and mw.text.trim(code)
code = code and trim(code)
lang = m_languages.getByCode(code) or m_languages.err(code, 1)
lang = m_languages.getByCode(code) or m_languages.err(code, 1)
end
end
local i = 1
while true do
local elem = elems[i]
while elem do
elem = trim(elem, "%s")
if elem ~= "" then
break
end
remove(elems, i)
elem = elems[i]
end
if not elem then
break
elseif not ( -- Strip redundant wikilinks.
not match(elem, "^()%[%[") or
find(elem, "[[", 3, true) or
find(elem, "]]", 3, true) ~= #elem - 1 or
find(elem, "|", 3, true)
) then
elem = sub(elem, 3, -3)
elem = trim(elem, "%s")
end
elems[i] = elem .. "\n"
i = i + 1
end
elems = m_table.removeDuplicates(elems)
export.sort(elems, lang)
export.sort(elems, lang)
return table.concat(elems, args.sep or "|")
return concat(elems, args.sep or "|")
end
end


return export
return export
Retrieved from "https://linguifex.com/wiki/Special:MobileDiff/374534"

Navigation menu