模組:Utilities
該模塊導出各種通用功能，可供其他模塊使用。
pattern_escape
pattern_escape(text)
跳脫匹配模式（Lua的正則表達式版本）中使用的魔法字符。比如，"^$()%.[]*+-?" 變成 "<strong class="error"><span class="scribunto-error mw-scribunto-error-0e5118e6">脚本错误：函数“pattern_escape”不存在。</span></strong>"。
format_categories
format_categories(categories, lang, sort_key, sort_base, force_output)
形成一個分類名稱的列表（表格）。輸出是一個由所有類別組成的字符串，每個類別都使用了[[Category:...]]，並添加了給定的排序鍵。如果命名空間不是主命名空間、附錄命名空間或重構命名空間，輸出將是一個空字符串，除非給出force_output。如果沒有給定排序鍵：
- 默認的排序鍵是通過使用
sort_base（如果有給定的話）或當前的子頁面名稱，以及去除開頭的連字符（以便後綴可以在沒有鍵的情況下進行排序）而生成的。 - 如果給定的語言有排序鍵，則其將會被用來創建一個遵循該語言規則的排序鍵。
template_categorize
{{#invoke:utilities|template_categorize}}
這個函數使用於 {{categorize}}、{{catlangname}} 和 {{catlangcode}} 等模板。
catfix
這個函數添加了一個「catfix」，它被用於特定語言的分類頁面，為所有條目名稱添加語言屬性，通常是文字類（script classes）。語言屬性和文字類的添加使條目名稱顯示得更好（使用MediaWiki:Common.css中指定的特定語言或腳本樣式），這對於在瀏覽器中沒有一致字體支持的非英語語言尤其重要。
語言屬性是為所有語言添加的，但腳本類只為在其數據文件中列出文字的語言添加，或者為在Module:utilities/data的catfix_script列表中列出一個默認腳本的語言添加。有些語言顯然有一個默認文字，但在他們的數據文件中仍有其他文字，因此需要指定他們的默認文字。其他語言則沒有默認文字。
- 塞爾維亞-克羅地亞語經常以拉丁字母和西里爾字母兩種文字書寫。因為使用兩種文字，所以塞爾維亞-克羅地亞語不能在其分類頁面的條目中使用文字類別，因為一次只能指定一種文字類別。
- 俄語通常用西里爾文字（
Cyrl）書寫，但盲文（Brai）也列於其數據文件中。因此，俄語需要在catfix_script列表中加入一個條目，這樣Cyrl（西里爾字母）文字類就會被應用於其分類頁面中的條目。
要查看一種語言所列出的文字，請見Module:languages，並使用搜索框找到該語言的數據文件。要想知道一個文字代碼的含義，請在Module:scripts/data中搜索該文字代碼。
local mw = mw
local mw_text = mw.text
local package = package
local table = table
local require = require
local concat = table.concat
local decode_entities = require("Module:string utilities").decode_entities
local get_current_frame = mw.getCurrentFrame
local insert = table.insert
local ipairs = ipairs
local maxn = table.maxn
local tonumber = tonumber
local trim = mw_text.trim
local type = type
local unstrip = mw_text.unstrip
local unstripNoWiki = mw_text.unstripNoWiki
local export = {}
do
local loaded = package.loaded
local loader = package.loaders[2]
--[==[
Like require, but return false if a module does not exist instead of throwing an error.
Outputs are cached in {package.loaded}, which is faster for all module types, but much faster for nonexistent modules since require will attempt to use the full loader each time (since they don't get cached in {package.loaded}).
Note: although nonexistent modules are cached as {false} in {package.loaded}, they still won't work with conventional require, since it uses a falsy check instead of checking the return value is not {nil}.
]==]
function export.safe_require(modname)
local module = loaded[modname]
if module ~= nil then
return module
end
-- The loader returns a function if the module exists, or nil if it doesn't, and checking this is faster than using pcall with require. If found, we still use require instead of loading and caching directly, because require contains safety checks against infinite loading loops (and we do want those to throw an error).
module = loader(modname)
if module then
return require(modname)
end
loaded[modname] = false
return false
end
end
--[==[
Convert decimal to hexadecimal.
Note: About three times as fast as the hex library.
]==]
function export.dec_to_hex(dec)
dec = tonumber(dec)
if not dec or dec % 1 ~= 0 then
error("Input should be a decimal integer.")
end
return ("%x"):format(dec):upper()
end
do
local function check_level(lvl)
if type(lvl) ~= "number" then
error("Heading levels must be numbers.")
elseif lvl < 1 or lvl > 6 or lvl % 1 ~= 0 then
error("Heading levels must be integers between 1 and 6.")
end
return lvl
end
--[==[
A helper function which iterates over the headings in `text`, which should be the content of a page or (main) section.
Each iteration returns three values: `sec` (the section title), `lvl` (the section level) and `loc` (the index of the section in the given text, from the first equals sign). The section title will be automatically trimmed, and any HTML entities will be resolved.
The optional parameter `a` (which should be an integer between 1 and 6) can be used to ensure that only headings of the specified level are iterated over. If `b` is also given, then they are treated as a range.
The optional parameters `a` and `b` can be used to specify a range, so that only headings with levels in that range are returned. If only `a` is given ...
]==]
function export.find_headings(text, a, b)
a = a and check_level(a) or nil
b = b and check_level(b) or a or nil
local start, loc, lvl, sec = 1
return function()
repeat
loc, lvl, sec, start = text:match("()%f[^%z\n](==?=?=?=?=?)([^\n]+)%2[\t ]*%f[%z\n]()", start)
lvl = lvl and #lvl
until not (sec and a) or (lvl >= a and lvl <= b)
return sec and trim(decode_entities(sec)) or nil, lvl, loc
end
end
local function get_section(content, name, level, skip_st_equal) -- zh l10n
if not (content and name) then
return nil
elseif name:find("\n", 1, true) then
error("Heading name cannot contain a newline.")
end
level = level and check_level(level) or nil
name = trim(decode_entities(name))
local start
for sec, lvl, loc in export.find_headings(content, level and 1 or nil, level) do
if start and lvl <= level then
return content:sub(start, loc - 1)
elseif not start and (not level or lvl == level) and sec == name then
start, level = loc, lvl
end
end
if not start and not skip_st_equal then -- zh l10n
m_zh = require("Module:zh")
lang = require("Module:languages").getByCode("zh")
script = lang:findBestScript(name):getCode()
return script == "Hant" and get_section(content, m_zh.ts(name), orig_level, true) or script == "Hans" and get_section(content, m_zh.st(name), orig_level, true) or start
end
return start and content:sub(start)
end
--[==[
A helper function to return the content of a page section.
`content` is raw wikitext, `name` is the requested section, and `level` is an optional parameter that specifies
the required section heading level. If `level` is not supplied, then the first section called `name` is returned.
`name` can either be a string or table of section names. If a table, each name represents a section that has the
next as a subsection. For example, { {"Spanish", "Noun"}} will return the first matching section called "Noun"
under a section called "Spanish". These do not have to be at adjacent levels ("Noun" might be L4, while "Spanish"
is L2). If `level` is given, it refers to the last name in the table (i.e. the name of the section to be returned).
The returned section includes all of its subsections. If no matching section is found, return {nil}.
]==]
function export.get_section(content, names, level)
if type(names) == "string" then
return get_section(content, names, level)
end
local names_len = maxn(names)
if names_len > 6 then
error("Not possible specify more than 5 subsections: headings only go up to level 6.")
end
for i, name in ipairs(names) do
content = get_section(content, name, i == names_len and level or nil)
end
return content
end
end
--[==[
A function which returns the number of the page section which contains the current {#invoke}.
]==]
function export.get_current_section()
local frame = get_current_frame()
-- We determine the section via the heading strip marker count, since they're numbered sequentially, but the only way to do this is to generate a fake heading via frame:preprocess(). The native parser assigns each heading a unique marker, but frame:preprocess() will return copies of older markers if the heading is identical to one further up the page, so the fake heading has to be unique to the page. The best way to do this is to feed it a heading containing a nowiki marker (which we will need later), since those are always unique.
local nowiki_marker = frame:extensionTag("nowiki")
-- Note: heading strip markers have a different syntax to the ones used for tags.
local h = tonumber(frame:preprocess("=" .. nowiki_marker .. "=")
:match("\127'\"`UNIQ%-%-h%-(%d+)%-%-QINU`\"'\127"))
-- For some reason, [[Special:ExpandTemplates]] doesn't generate a heading strip marker, so if that happens we simply abort early.
if not h then
return 0
end
-- The only way to get the section number is to increment the heading count, so we store the offset in nowiki strip markers which can be retrieved by procedurally unstripping nowiki markers, counting backwards until we find a match.
local n, offset = tonumber(nowiki_marker:match("\127'\"`UNIQ%-%-nowiki%-([%dA-F]+)%-QINU`\"'\127"), 16)
while not offset and n > 0 do
n = n - 1
offset = unstripNoWiki(("\127'\"`UNIQ--nowiki-%08X-QINU`\"'\127"):format(n))
:match("^HEADING\1(%d+)") -- Prefix "HEADING\1" prevents collisions.
end
offset = offset and (offset + 1) or 0
frame:extensionTag("nowiki", "HEADING\1" .. offset)
return h - offset
end
do
local L2_sections
--[==[
A function which returns the name of the L2 language section which contains the current {#invoke}.
]==]
function export.get_current_L2()
local section = export.get_current_section()
if section == 0 then
return
end
L2_sections = L2_sections or mw.loadData("Module:headword/data").page.L2_sections
while section > 0 do
local L2 = L2_sections[section]
if L2 then
return L2
end
section = section - 1
end
end
end
--[==[
A helper function to strip wiki markup, giving the plaintext of what is displayed on the page.
]==]
function export.get_plaintext(text)
text = text
:gsub("%[%[", "\1")
:gsub("%]%]", "\2")
-- Remove strip markers and HTML tags.
text = unstrip(text):gsub("<[^<>\1\2]+>", "")
-- Parse internal links for the display text, and remove categories.
text = require("Module:links").remove_links(text)
-- Remove files.
for _, falsePositive in ipairs({"File", "Image"}) do
text = text:gsub("\1" .. falsePositive .. ":[^\1\2]+\2", "")
end
-- Parse external links for the display text.
text = text:gsub("%[(https?://[^%[%]]+)%]",
function(capture)
return capture:match("https?://[^%s%]]+%s([^%]]+)") or ""
end)
-- Any remaining square brackets aren't involved in links, but must be escaped to avoid creating new links.
:gsub("\1", "[[")
:gsub("\2", "]]")
:gsub("%[", "[")
:gsub("]", "]")
-- Strip bold, italics and soft hyphens.
:gsub("('*)'''(.-'*)'''", "%1%2")
:gsub("('*)''(.-'*)''", "%1%2")
:gsub("", "")
-- Get any HTML entities.
-- Note: don't decode URL percent encoding, as it shouldn't be used in display text and may cause problems if % is used.
text = decode_entities(text)
return trim(text)
end
do
local title_obj, category_namespaces, page_data, pagename, pagename_defaultsort
--[==[
Format the categories with the appropriate sort key.
* `categories` is a list of categories. Each entry in the list can be either a string (the full category, minus
the {"Category:"} prefix) or an object. In the latter case, the object should have fields
** `cat`: the full category, minus the {"Category:"} prefix (required);
** `lang`: optional language object to override the overall `lang`;
** `sort_key`: optional sort key to override the overall `sort_key`;
** `sort_base`: optional sort base to override the overall `sort_base`;
** `sc`: optional script object to override the overall `sc`.
* `lang` is an object encapsulating a language; if {nil}, the object for language code {"und"} (undetermined) will
be used. `lang` is used when computing the sort key (either from the subpage name or sort base).
* `sort_key` is placed in the category invocation, and indicates how the page will sort in the respective category.
Normally '''do not use this'''. Instead, leave it {nil}, and if you need to a control the sort order, use
{sort_base}, so that language-specific normalization is applied on top of the specified sort base. If neither
{sort_key} nor {sort_base} is specified, the default is to apply language-specific normalization to the subpage
name; see below.
* `sort_base` lets you override the default sort key while still maintaining appropriate language-specific
normalization. If {nil} is specified, this defaults to the subpage name, which is the portion of the full pagename
after subtracting the namespace prefix (and, in certain namespaces such as {User:}, but notably not in the
mainspace, after subtracting anything up through the final slash). The actual sort key is derived from the sort
base approximately by lowercasing, applying language-specific normalization and then uppercasing; note that the
same process is applied in deriving the sort key when no sort base is specified. For example, for French, Spanish,
etc. the normalization process maps accented letters to their unaccented equivalents, so that e.g. in French,
{{m|fr|ça}} sorts after {{m|fr|ca}} (instead of after the default Wikimedia sort order, which is approximately
based on Unicode sort order and places ç after z) and {{m|fr|côté}} sorts after {{m|fr|coté}} (instead of between
c and d). Similarly, in Russian the normalization process converts Cyrillic ё to a string consisting of Cyrillic е
followed by U+10FFFF, so that effectively ё sorts after е instead of the default Wikimedia sort, which (I think)
puts ё after я, the last letter of the Cyrillic alphabet.
* `force_output` forces normal output in all namespaces. Normally, nothing is output if the page isn't in the main,
Appendix:, Thesaurus:, Reconstruction: or Citations: namespaces.
* `sc` is a script object; if nil, the default will be derived from the sort base (or its default value, the
subpage name) by calling {lang:findBestScript()}. The value of `sc` is used during the sort base normalization
process; for example, languages with multiple scripts will often have script-specific normalization processes.
]==]
function export.format_categories(categories, lang, sort_key, sort_base, force_output, sc)
if type(lang) == "table" and not lang.getCode then
error("The second argument to format_categories should be a language object.")
end
title_obj = title_obj or mw.title.getCurrentTitle()
category_namespaces = category_namespaces or mw.loadData("Module:utilities/data").category_namespaces
if not (
force_output or
category_namespaces[title_obj.namespace] or
title_obj.prefixedText == "Wiktionary:Sandbox"
) then
return ""
elseif not page_data then
page_data = mw.loadData("Module:headword/data").page
pagename = page_data.encoded_pagename
pagename_defaultsort = page_data.pagename_defaultsort
end
local extra_categories
local function generate_sort_key(lang, sort_key, sort_base, sc)
-- Generate a default sort key.
-- If the sort key is "-", bypass the process of generating a sort key altogether. This is desirable when categorising (e.g.) translation requests, as the pages to be categorised are always in English/Translingual.
if sort_key == "-" then
sort_key = sort_base and sort_base:uupper() or pagename_defaultsort
else
lang = lang or require("Module:languages").getByCode("und")
sort_base = lang:makeSortKey(sort_base or pagename, sc) or pagename_defaultsort
if not sort_key or sort_key == "" then
sort_key = sort_base
elseif lang:getCode() ~= "und" then
if not extra_categories then
extra_categories = {}
end
-- insert(extra_categories, lang:getFullName() .. " terms with " .. (
-- sort_key:uupper() == sort_base and "redundant" or
-- "non-redundant non-automated"
-- ) .. " sortkeys")
end
end
if not sort_key or sort_key == "" then
sort_key = pagename_defaultsort
end
return sort_key
end
local ret = {}
local default_sort_key = generate_sort_key(lang, sort_key, sort_base, sc)
local ins_point = 0
local function process_category(cat)
local this_sort_key
if type(cat) == "string" then
this_sort_key = default_sort_key
else
this_sort_key = generate_sort_key(cat.lang or lang, cat.sort_key or sort_key,
cat.sort_base or sort_base, cat.sc or sc)
cat = cat.cat
end
ins_point = ins_point + 1
ret[ins_point] = "[[Category:" .. cat .. "|" .. this_sort_key .. "]]"
end
for _, cat in ipairs(categories) do
process_category(cat)
end
if extra_categories then
for _, cat in ipairs(extra_categories) do
process_category(cat)
end
end
return concat(ret)
end
end
do
local catfix_scripts
--[==[
Add a "catfix", which is used on language-specific category pages to add language attributes and often script
classes to all entry names. The addition of language attributes and script classes makes the entry names display
better (using the language- or script-specific styles specified in [[MediaWiki:Common.css]]), which is particularly
important for non-English languages that do not have consistent font support in browsers.
Language attributes are added for all languages, but script classes are only added for languages with one script
listed in their data file, or for languages that have a default script listed in the {catfix_script} list in
[[Module:utilities/data]]. Some languages clearly have a default script, but still have other scripts listed in
their data file and therefore need their default script to be specified. Others do not have a default script.
* Serbo-Croatian is regularly written in both the Latin and Cyrillic scripts. Because it uses two scripts,
Serbo-Croatian cannot have a script class applied to entries in its category pages, as only one script class
can be specified at a time.
* Russian is usually written in the Cyrillic script ({{cd|Cyrl}}), but Braille ({{cd|Brai}}) is also listed in
its data file. So Russian needs an entry in the {catfix_script} list, so that the {{cd|Cyrl}} (Cyrillic) script
class will be applied to entries in its category pages.
To find the scripts listed for a language, go to [[Module:languages]] and use the search box to find the data file
for the language. To find out what a script code means, search the script code in [[Module:scripts/data]].
]==]
function export.catfix(lang, sc)
if not lang or not lang.getCanonicalName then
error('The first argument to the function "catfix" should be a language object from [[Module:languages]] or [[Module:etymology languages]].')
end
if sc and not sc.getCode then
error('The second argument to the function "catfix" should be a script object from [[Module:scripts]].')
end
local canonicalName = lang:getCanonicalName()
local nonEtymologicalName = lang:getFullName()
-- To add script classes to links on pages created by category boilerplate templates.
if not sc then
catfix_scripts = catfix_scripts or mw.loadData("Module:utilities/data").catfix_scripts
sc = catfix_scripts[lang:getCode()] or catfix_scripts[lang:getFullCode()]
if sc then
sc = require("Module:scripts").getByCode(sc)
end
end
local catfix_class = "CATFIX-" .. mw.uri.anchorEncode(canonicalName)
if nonEtymologicalName ~= canonicalName then
catfix_class = catfix_class .. " CATFIX-" .. mw.uri.anchorEncode(nonEtymologicalName)
end
return "<span id=\"catfix\" style=\"display:none;\" class=\"" .. catfix_class .. "\">" ..
require("Module:script utilities").tag_text(" ", lang, sc, nil) ..
"</span>"
end
end
--[==[
Implementation of the {{tl|catfix}} template.
]==]
function export.catfix_template(frame)
local params = {
[1] = { type = "language", required = true },
[2] = { alias_of = "sc" },
["sc"] = { type = "script" },
}
local args = require("Module:parameters").process(frame:getParent().args, params)
return export.catfix(args[1], args.sc)
end
--[==[
Given a type (as a string) and an arbitrary number of entities, checks whether all of those entities are language,
family, script, writing system or Wikimedia language objects. Useful for error handling in functions that require
one of these kinds of object.
If `noErr` is set, the function returns false instead of throwing an error, which allows customised error handling to
be done in the calling function.
]==]
function export.check_object(typ, noErr, ...)
local function fail(message)
if noErr then
return false
else
error(message, 3)
end
end
local objs = {...}
if #objs == 0 then
return fail("Must provide at least one object to check.")
end
for _, obj in ipairs(objs) do
if type(obj) ~= "table" or type(obj.hasType) ~= "function" then
return fail("Function expected a " .. typ .. " object, but received a " .. type(obj) .. " instead.")
elseif not (typ == "object" or obj:hasType(typ)) then
for _, wrong_type in ipairs{"family", "language", "script", "Wikimedia language", "writing system"} do
if obj:hasType(wrong_type) then
return fail("Function expected a " .. typ .. " object, but received a " .. wrong_type .. " object instead.")
end
end
return fail("Function expected a " .. typ .. " object, but received another type of object instead.")
end
end
return true
end
return export