মডিউল:links

উইকিঅভিধান, মুক্ত অভিধান থেকে

This module provides many useful utility functions for creating and processing wikilinks within Wiktionary.

full_link[সম্পাদনা]

full_link(term, alt, lang, sc, face, id, annotations, curtitle)

Creates a full link, with annotations, in the style of {{l}} or {{term}}. The parameters are: term (page name), alternative display, language object, script code, script face (see Module:script utilities#tag_text), senseid, table of annotations (see format_link_annotations). All parameters may be nil except for the annotations, but an error will be shown if the term, alt and transliteration (in the annotations) are all missing.

The function will:

  • Try to determine the script, based on the characters found in the term or alt parameter, if the script was not given.
  • Call language_link on the term/alt forms, to remove diacritics in the page name, process any embedded wikilinks and create links to appendix pages when necessary.
  • Call Module:script utilities#tag_text to add the appropriate language and script tags to the term.
  • Generate a transliteration, based on the alt or term parameters, if the script is not Latin and no transliteration was provided.
  • Add the annotations (transliteration, gender, gloss etc.) after the link.

format_link_annotations[সম্পাদনা]

format_link_annotations(lang, info)

Formats annotations that are displayed with a link. Annotations are the extra bits of information that are displayed following the linked term, and include things such as gender, transliteration, gloss and so on. The first parameter is the language object, the second is a table possessing some or all of the following keys:

genders
Table containing a list of gender specifications in the style of Module:gender and number.
tr
Transliteration.
gloss
Gloss that translates the term in the link, or gives some other descriptive information.
pos
Part of speech of the linked term. If the given parameter matches one of the templates in Category:Part of speech tags, then call that to show a part-of-speech tag. Otherwise, just show the given text as it is.
lit
Literal meaning of the term, if the usual meaning is figurative or idiomatic.

Any of the above values can be omitted from the info parameter. If a completely empty table is given (with no annotations at all), then an empty string is returned.

language_link[সম্পাদনা]

language_link(text, alt, lang, id, curtitle)

Creates a basic link to the given term. It links to the language section (such as #English), but it does not add language and script wrappers, so any code that uses this function should call Module:script utilities#tag_text to add such wrappers itself at some point.

It accepts the following parameters:

text
The text to turn into a link. This is generally the name of a page. The string can contain wikilinks already embedded in them, which can also have their own alternative display text. These are processed individually just like a single link would be. The alt parameter is ignored in this case.
alt
The alternative display for the link, if this is different from the linked page. If this is nil, the text parameter is used instead (much like regular wikilinks). If text contains wikilinks in it, this parameter is ignored and has no effect.
lang
The language object for the term being linked.
id
An additional sense ID to insert into the link. This is used in conjunction with {{senseid}}. This is optional.
curtitle
When provided, the function won't link to this page, but generate a bolded "self link" instead. It is used in inflection tables.

The following special options are processed for each link (both simple text and with embedded wikilinks):

  • The target page name will be processed to generate the correct entry name. This is done via Module:languages#makeEntryName and is generally used to automatically strip dictionary-only diacritics that are not part of the normal written form of a language.
  • If the text starts with *, then the term is considered a reconstructed term, and a link to the Appendix: namespace will be created. If the text contains embedded wikilinks, then * is automatically applied to each one individually, while preserving the displayed form of each link as it was given. This allows linking to phrases containing multiple reconstructed terms, while only showing the * once at the beginning.
  • If the text starts with :, then the link is treated as "raw" and the above steps are skipped. This can be used in rare cases where the page name begins with * or if diacritics should not be stripped. For example:
  • A link to the correct language section is added, along with the sense ID.

remove_links[সম্পাদনা]

remove_links(text)

Replaces all [[wikilinks]] with their linktitle as simple text. This function can be invoked from either a template or from another module.


local export = {}

--[=[
	[[Unsupported titles]], pages with high memory usage,
	extraction modules and part-of-speech names are listed
	at [[Module:links/data]].

	Other modules used:
		[[Module:script utilities]]
		[[Module:scripts]]
		[[Module:languages]] and its submodules
		[[Module:gender and number]]
		[[Module:debug]]
]=]

-- These are prefixed with u to avoid confusion with the default string methods
-- of the same name.
local usub = mw.ustring.sub

local table_insert = table.insert
local table_concat = table.concat

local ignore_cap
local phonetic_extraction
local pos_tags
local unsupported_titles
function export.getLinkPage(target, lang, sc)
	unsupported_titles = unsupported_titles or mw.loadData("Module:links/data").unsupported_titles
	if unsupported_titles[target] then
		return "অসমর্থিত শিরোনাম/" .. unsupported_titles[target]
	end

	-- If the link contains unexpanded template parameters, then don't create a link.
	if target:find("{{{") then
		return nil
	end

	if target:sub(1, 1) == ":" or target:sub(1, 2) == "w:" or target:sub(1, 10) == "wikipedia:" then
		return target
	end
	
	-- Remove diacritics from the page name
	target = lang:makeEntryName(target, sc)

	if target:sub(1, 1) == "/" then
		return ":" .. target

	-- Link to appendix for reconstructed terms and terms in appendix-only languages
	elseif target:sub(1, 1) == "*" and #target > 1 then
		if lang:getCode() == "und" then
			return nil
		end

		target = "Reconstruction:" .. lang:getCanonicalName() .. "/" .. usub(target, 2)
	elseif lang:getType() == "reconstructed" then
		error("The specified language " .. lang:getCanonicalName()
			.. " is unattested, while the given word is not marked with '*' to indicate that it is reconstructed")
	elseif lang:getType() == "appendix-constructed" then
		target = "পরিশিষ্ট:" .. lang:getCanonicalName() .. "/" .. target
	end

	return target
end

-- Make a language-specific link from given link's parts
local function makeLangLink(link, lang, sc, id, allow_self_link)
	
	-- Find fragments (when link didn't come from parseLink).
	-- Prevents {{l|en|word#Etymology 2|word}} from linking to [[word#Etymology 2#English]].
	if link.fragment == nil then
		-- Replace numeric character references with the corresponding character ( → '),
		-- as they contain #, which causes the numeric character reference to be
		-- misparsed (wa'a → waa → pagename wa&, fragment 29;a).
		link.target = link.target:gsub("&#(%d+);",
			function(number) return mw.ustring.char(tonumber(number)) end)
		local first, second = link.target:match("^([^#]+)#(.+)$")
		if first then
			link.target, link.fragment = first, second
		end
	end

	-- If there is no display form, then create a default one.
	if not link.display then link.display = link.target end
	
	-- Process the display form.
	link.display = lang:makeDisplayText(link.display, sc)
	
	-- Process the target
	link.target = export.getLinkPage(link.target, lang, sc)

	if not link.target then
		return link.display
	end

	-- If the target is the same as the current page and there is no sense id
	-- and linking to the same page hasn't been turned on, then return a "self-link"
	-- like the software does.
	if not (allow_self_link or id) and link.target:gsub("^:", "") == mw.title.getCurrentTitle().prefixedText then
		return "<strong class=\"selflink\">" .. link.display .. "</strong>"
	end

	--[[
		Add fragment
		Do not add a section link to "Undetermined", as such sections do not exist and are invalid.
		TabbedLanguages handles links without a section by linking to the "last visited" section,
		but adding "Undetermined" would break that feature.
		For localized prefixes that make syntax error, please use the format: ["xyz"] = true,
	]]
	local prefix = link.target:match("^:?([^:]+):")
	local prefixes = {
		w = true,
		wikipedia = true,
		Category = true,
	}

	if not prefixes[prefix] then
		if link.fragment or link.target:find("#$") then
			require("Module:debug/track") {
				"links/fragment",
				"links/fragment/" .. lang:getCode()
			}
		end

		if not link.fragment and lang:getCode() ~= "und" then
			if id then
				link.fragment = require("Module:senseid").anchor(lang, id)
			elseif not mw.ustring.find(link.target, "^Appendix:")
					and not mw.ustring.find(link.target, "^Reconstruction:") then
				link.fragment = lang:getCanonicalName()
			end
		end

		-- This allows linking to pages like [[sms:a]] without it being treated weirdly.
		link.target = link.target:gsub(":", "&#x3a;")
	end

	return "[[" .. link.target .. (link.fragment and "#" .. link.fragment or "") .. "|" .. link.display .. "]]"
end


-- Split a link into its parts
local function parseLink(linktext)
	local link = { target = linktext }
	local first, second = link.target:match("^([^|]+)|(.+)$")

	if first then
		link.target = first
		link.display = second
	else
		link.display = link.target
	end

	first, second = link.target:match("^(.+)#(.+)$")

	if first then
		link.target = first
		link.fragment = second
	else
		-- So that makeLangLink does not look for a fragment again
		link.fragment = false
	end

	return link
end


-- Creates a basic wikilink to the given term. If the text already contains
-- links, these are replaced with links to the correct section.
function export.language_link(data, allow_self_link)
	if type(data) ~= "table" then
		error("The first argument to the function language_link must be a table. See Module:links/documentation for more information.")
	end

	local text = data.term

	ignore_cap = ignore_cap or mw.loadData("Module:links/data").ignore_cap
	if ignore_cap[data.lang:getCode()] and text then
		text = text:gsub("%^", "")
	end

	-- If the text begins with * and another character,
	-- then act as if each link begins with *
	local allReconstructed = false

	if text:find("^*.") then
		allReconstructed = true
	end

	-- Do we have embedded wikilinks?
	if text:find("[[", nil, true) then
		--[=[
		[[Special:WhatLinksHere/Template:tracking/links/alt-ignored]]
		[[Special:WhatLinksHere/Template:tracking/links/id-ignored]]
		]=]

		if data.alt then
			require("Module:debug/track")("links/alt-ignored")
			mw.log("(from Module:links)", "text with embedded wikilinks:", text,
				"ignored alt:", data.alt, "lang:", data.lang:getCode())
		end

		if data.id then
			require("Module:debug/track")("links/id-ignored")
			mw.log("(from Module:links)", "text with embedded wikilinks:", text,
				"ignored id:", data.id, "lang:", data.lang:getCode())
		end

		-- Begins and ends with a wikilink tag
		if text:find("^%[%[(.+)%]%]$") then
			-- There are no [ ] in between.
			-- This makes the wikilink tag redundant.
			if text:find("^%[%[[^%[%]]+%]%]$") then
				require("Module:debug/track")("links/redundant wikilink")
			else
				local temp = text:gsub("^%[%[(.+)%]%]$", "%1")
				temp = temp:gsub("%]%], %[%[", "|")

				if not temp:find("[%[%]]") then
					require("Module:debug/track")("links/list")
				end
			end
		end

		text = text:gsub("%[%[([^%]]+)%]%]",
			function(linktext)
				local link = parseLink(linktext)

				if allReconstructed then
					link.target = "*" .. link.target
				end

				return makeLangLink(link, data.lang, data.sc, data.id, allow_self_link)
			end)

		-- Remove the extra * at the beginning if it's immediately followed
		-- by a link whose display begins with * too
		if allReconstructed then
			text = text:gsub("^%*%[%[([^|%]]+)|%*", "[[%1|*")
		end
	else
		-- There is no embedded wikilink, make a link using the parameters.
		text = makeLangLink({ target = text, display = data.alt }, data.lang, data.sc, data.id, allow_self_link)
	end

	return text
end

function export.mark(text, itemType, face, lang)
	local tag = { "", "" }

	if itemType == "gloss" then
		tag = { '<span class="mention-gloss-double-quote">“</span><span class="mention-gloss">',
			'</span><span class="mention-gloss-double-quote">”</span>' }
	elseif itemType == "tr" then
		if face == "term" then
			tag = { '<span lang="' .. lang:getCode() .. '" class="tr mention-tr Latn">',
				'</span>' }
		else
			tag = { '<span lang="' .. lang:getCode() .. '" class="tr Latn">', '</span>' }
		end
	elseif itemType == "ts" then
		tag = { '<span class="ts mention-ts Latn">/', '/</span>' }
	elseif itemType == "pos" then
		tag = { '<span class="ann-pos">', '</span>' }
	elseif itemType == "annotations" then
		tag = { '<span class="mention-gloss-paren annotation-paren">(</span>',
			'<span class="mention-gloss-paren annotation-paren">)</span>' }
	end

	if type(text) == "string" then
		return tag[1] .. text .. tag[2]
	else
		return ""
	end
end

-- Format the annotations (things following the linked term)
function export.format_link_annotations(data, face)
	local output = {}

	-- Interwiki link
	if data.interwiki then
		table_insert(output, data.interwiki)
	end

	-- Genders
	if type(data.genders) ~= "table" then
		data.genders = { data.genders }
	end

	if data.genders and #data.genders > 0 then
		local m_gen = require("Module:gender and number")
		table_insert(output, "&nbsp;" .. m_gen.format_list(data.genders, data.lang))
	end

	local annotations = {}

	-- Transliteration and transcription
	if data.tr or data.ts then
		local kind
		if face == "term" then
			kind = face
		else
			kind = "default"
		end

		if data.tr and data.ts then
			table_insert(annotations,
				require("Module:script utilities").tag_translit(data.tr, data.lang, kind)
				.. " " .. export.mark(data.ts, "ts"))
		elseif data.ts then
			table_insert(annotations, export.mark(data.ts, "ts"))
		else
			table_insert(annotations,
				require("Module:script utilities").tag_translit(data.tr, data.lang, kind))
		end
	end

	-- Gloss/translation
	if data.gloss then
		table_insert(annotations, export.mark(data.gloss, "gloss"))
	end

	-- Part of speech
	if data.pos then
		-- debug category for pos= containing transcriptions
		if data.pos:find("/[^><]*/") then
			data.pos = data.pos .. "[[Category:links likely containing transcriptions in pos]]"
		end

		pos_tags = pos_tags or mw.loadData("Module:links/data").pos_tags
		table_insert(annotations, export.mark(pos_tags[data.pos] or data.pos, "pos"))
	end

	-- Literal/sum-of-parts meaning
	if data.lit then
		table_insert(annotations, "literally " .. export.mark(data.lit, "gloss"))
	end

	if #annotations > 0 then
		table_insert(output, " " .. export.mark(table_concat(annotations, ", "), "annotations"))
	end

	return table_concat(output)
end

-- A version of {{l}} or {{m}} that can be called from other modules too
function export.full_link(data, face, allow_self_link, no_check_redundant_translit)
	if type(data) ~= "table" then
		error("The first argument to the function full_link must be a table. "
			.. "See Module:links/documentation for more information.")
	end

	-- Create the link
	local output = {}
	local categories = {}
	local link = ""
	local annotations

	phonetic_extraction = phonetic_extraction or mw.loadData("Module:links/data").phonetic_extraction

	-- Is there any text to show?
	if (data.term or data.alt) then
		-- Try to detect the script if it was not provided
		if not data.sc then
			data.sc = require("Module:scripts").findBestScript(data.alt or data.term, data.lang)
		end
		
		-- If using a discouraged character sequence, add to maintenance category
		if data.sc:hasNormalizationFixes() == true then
			local toNFC = mw.ustring.toNFC
			if (data.term and data.sc:fixDiscouragedSequences(toNFC(data.term)) ~= toNFC(data.term)) or (data.alt and data.sc:fixDiscouragedSequences(toNFC(data.alt)) ~= toNFC(data.alt)) then
				table.insert(categories, "[[Category:Pages using discouraged character sequences]]")
			end
		end

		local class = ""

		-- Encode certain characters to avoid various delimiter-related issues at various stages. We need to encode < and >
		-- because they end up forming part of CSS class names inside of <span ...> and will interfere with finding the end
		-- of the HTML tag. I first tried converting them to URL encoding, i.e. %3C and %3E; they then appear in the URL as
		-- %253C and %253E, which get mapped back to %3C and %3E when passed to [[Module:accel]]. But mapping them to &lt;
		-- and &gt; somehow works magically without any further work; they appear in the URL as < and >, and get passed to
		-- [[Module:accel]] as < and >. I have no idea who along the chain of calls is doing the encoding and decoding. If
		-- someone knows, please modify this comment appropriately!
		local encode_accel_char_map = {
			["%"] = ".",
			[" "] = "_",
			["<"] = "&lt;",
			[">"] = "&gt;",
		}
		local function encode_accel_param_chars(param)
			local retval = param:gsub("[% <>]", encode_accel_char_map) -- discard second return value
			return retval
		end

		local function encode_accel_param(prefix, param)
			if not param then
				return ""
			end
			if type(param) == "table" then
				local filled_params = {}
				-- There may be gaps in the sequence, especially for translit params.
				local maxindex = 0
				for k, v in pairs(param) do
					if type(k) == "number" and k > maxindex then
						maxindex = k
					end
				end
				for i=1,maxindex do
					filled_params[i] = param[i] or ""
				end
				-- [[Module:accel]] splits these up again.
				param = table.concat(filled_params, "*~!")
			end
			-- This is decoded again by [[WT:ACCEL]].
			return prefix .. encode_accel_param_chars(param)
		end

		if data.accel then
			local form = data.accel.form and encode_accel_param_chars(data.accel.form) .. "-form-of" or ""
			local gender = encode_accel_param("gender-", data.accel.gender)
			local pos = encode_accel_param("pos-", data.accel.pos)
			local translit = encode_accel_param("transliteration-",
				data.accel.translit or (data.tr ~= "-" and data.tr or nil))
			local target = encode_accel_param("target-", data.accel.target)
			local lemma = encode_accel_param("origin-", data.accel.lemma)
			local lemma_translit = encode_accel_param("origin_transliteration-", data.accel.lemma_translit)
			local no_store = data.accel.no_store and "form-of-nostore" or ""

			local accel =
				form .. " " ..
				gender .. " " ..
				pos .. " " ..
				translit .. " " ..
				target .. " " ..
				lemma .. " " ..
				lemma_translit .. " " ..
				no_store .. " "

			class = "form-of lang-" .. data.lang:getCode() .. " " .. accel
		end
		
		-- Only make a link if the term has been given, otherwise just show the alt text without a link
		link = require("Module:script utilities").tag_text(
			data.term and export.language_link(data, allow_self_link)
			or data.alt, data.lang, data.sc, face, class)
	else
		--[[	No term to show.
				Is there at least a transliteration we can work from?	]]
		link = require("Module:script utilities").request_script(data.lang, data.sc)

		if link == "" or not data.tr or data.tr == "-" then
			-- No link to show, and no transliteration either. Show a term request.
			local category = ""

			if mw.title.getCurrentTitle().nsText ~= "Template" then
				table_insert(categories, "[[Category:" .. data.lang:getCanonicalName() .. " term requests]]")
			end

			link = "<small>[Term?]</small>"
		end
	end

	table_insert(output, link)

	if data.tr == "" or data.tr == "-" then
		data.tr = nil

	elseif phonetic_extraction[data.lang:getCode()] then
		local m_phonetic = require(phonetic_extraction[data.lang:getCode()])
		data.tr = data.tr or m_phonetic.getTranslit(export.remove_links(data.alt or data.term))

	elseif (data.term or data.alt) and not data.sc:getCode():find("Lati?n") then

		-- Try to generate a transliteration, unless transliteration has been supplied and either
		-- no_check_redundant_translit is given or we are in a high-memory entry. (Checking for redundant
		-- transliteration can use up significant amounts of memory so we don't want to do it if memory
		-- is tight. `no_check_redundant_translit` is currently set when called ultimately from
		-- {{multitrans|...|no-check-redundant-translit=1}}.)
		if not (data.tr and (
			no_check_redundant_translit or
			mw.loadData("Module:links/data").high_memory_entries[mw.title.getCurrentTitle().text]
		)) then
			local automated_tr = data.lang:transliterate(export.remove_links(data.alt or data.term), data.sc)

			if automated_tr then
				local manual_tr = data.tr

				if manual_tr then
					if manual_tr == automated_tr then
						table_insert(categories,
							"[[Category:Terms with redundant transliterations]]"
									.. "[[Category:Terms with redundant transliterations/" .. data.lang:getCode() .. "]]")
					else
						-- Prevents Arabic root categories from flooding the tracking categories.
						if mw.title.getCurrentTitle().nsText ~= "Category" then
							table_insert(categories,
								"[[বিষয়শ্রেণী:অস্বয়ংক্রিয় প্রতিবর্ণযুক্ত ভুক্তি]]]"
										.. "[[বিষয়শ্রেণী:অস্বয়ংক্রিয় প্রতিবর্ণযুক্ত ভুক্তি/" .. data.lang:getCode() .. "]]")
						end
					end
				end

				if (not manual_tr) or data.lang:overrideManualTranslit() then
					data.tr = automated_tr
				end
			end
		end
	end

	-- Link to the transliteration entry for languages that require this
	if data.tr and data.lang:link_tr() then
		data.tr = export.language_link { lang = data.lang, term = data.tr }
	end

	table_insert(output, export.format_link_annotations(data, face))

	return table_concat(output) .. table_concat(categories)
end


--[[	Strips links: deletes category links,
		the targets of piped links,
		and all double square brackets.			]]
function export.remove_links(text)
	if type(text) == "table" then
		text = text.args[1]
	end

	if not text or text == "" then
		return ""
	end

	text = mw.ustring.gsub(text, "%[%[Category:[^|%]]-|?[^|%]]-%]%]", "")
	text = text:gsub("%[%[[^|%]]-|", "")
	text = text:gsub("%[%[", "")
	text = text:gsub("%]%]", "")

	return text
end

function export.english_links(text)
	local lang = require("Module:languages").getByCode("en")
	local sc = require("Module:scripts").findBestScript(text, lang)

	-- Parentheses around function call to remove second return value, the
	-- number of replacements.
	return (text:gsub("%[%[([^%]]+)%]%]",
		function(linktext)
			local link = parseLink(linktext)
			return makeLangLink(link, lang, sc, nil, true, false)
		end))
end

--[=[
This decodes old section encodings.
For example, Norwegian_Bokm.C3.A5l → Norwegian_Bokmål.
It isn't picky about whether the section encodings represent the UTF-8 encoding
of a real Unicode character, so it will mangle section names that contain
a period followed by two uppercase hex characters. At least such section names
are probably pretty rare.

Wiktionary adds an additional id="" attribute for sections
using a legacy encoding, if it is different from the modern minimally modified attribute.
It is like percent encoding (URI or URL encoding) except with "." instead of "%".
See [[mw:Manual:$wgFragmentMode]] and the code that does the encoding at
https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/core/+/7bf779524ab1fd8e1d74f79ea4840564d48eea4d/includes/parser/Sanitizer.php#893
]=]

-- The character class %x should not be used, as it includes the characters a-f,
-- which do not occur in these anchor encodings.
local capitalHex = "[0-9A-F]"

local function decodeAnchor(anchor)
	return (anchor:gsub("%.(" .. capitalHex .. capitalHex .. ")",
		function(hexByte)
			return string.char(tonumber(hexByte, 16))
		end))
end

function export.section_link(link)
	if type(link) ~= "string" then
		error("The first argument to section_link was a " .. type(link) .. ", but it should be a string.")
	end

	link = link:gsub("_", " ")

	local numberSigns = select(2, mw.ustring.gsub(link, "#", ""))

	if numberSigns > 1 then
		error("The section link should only contain one number sign (#).")
	end

	link = mw.uri.decode(link, "WIKI")
	local page, section = link:match("^([^#]*)#(.+)$")
	if page == "" then
		page = nil
	end

	if section then
		section = decodeAnchor(section)

		-- URI-encode (percent-encode) section to allow square brackets and
		-- other dodgy characters in section name.
		-- If not percent-encoded, they prevent the parser from creating a link.
		-- Decode percent-encoding in the displayed text
		if page then
			return "[[" .. page .. "#" .. mw.uri.encode(section, "WIKI")
				.. "|" .. page .. " §&nbsp;" .. section .. "]]"
		else
			return "[[#" .. mw.uri.encode(section, "WIKI")
				.. "|§&nbsp;" .. section .. "]]"
		end
	else
		error("The function “section_link” could not find a number sign marking a section name.")
	end
end

return export