Module:Citation/CS1/COinS

local coins = {};


--[[--------------------------< F O R W A R D   D E C L A R A T I O N S >--------------------------------------
]]
local is_set, in_array, remove_wiki_link;										-- functions in Module:Citation/CS1/Utilities

local cfg;																		-- table of configuration tables that are defined in Module:Citation/CS1/Configuration


--[[--------------------------< S T R I P _ A P O S T R O P H E _ M A R K U P >--------------------------------

Strip wiki italic and bold markup from argument so that it doesn't contaminate COinS metadata.
This function strips common patterns of apostrophe markup.  We presume that editors who have taken the time to
markup a title have, as a result, provided valid markup. When they don't, some single apostrophes are left behind.

]]

local function strip_apostrophe_markup (argument)
	if not is_set (argument) then return argument; end

	if argument:find ( "''", 1, true ) == nil then								-- Is there at least one double apostrophe?  If not, exit.
		return argument;
	end

	while true do
		if argument:find ( "'''''", 1, true ) then								-- bold italic (5)
			argument=argument:gsub("%'%'%'%'%'", "");							-- remove all instances of it
		elseif argument:find ( "''''", 1, true ) then							-- italic start and end without content (4)
			argument=argument:gsub("%'%'%'%'", "");
		elseif argument:find ( "'''", 1, true ) then							-- bold (3)
			argument=argument:gsub("%'%'%'", "");
		elseif argument:find ( "''", 1, true ) then								-- italic (2)
			argument=argument:gsub("%'%'", "");
		else
			break;
		end
	end
	return argument;															-- done
end


--[[--------------------------< M A K E _ C O I N S _ T I T L E >----------------------------------------------

Makes a title for COinS from Title and / or ScriptTitle (or any other name-script pairs)

Apostrophe markup (bold, italics) is stripped from each value so that the COinS metadata isn't corrupted with strings
of %27%27...

]]

local function make_coins_title (title, script)
	if is_set (title) then
		title = strip_apostrophe_markup (title);								-- strip any apostrophe markup
	else
		title='';																-- if not set, make sure title is an empty string
	end
	if is_set (script) then
		script = script:gsub ('^%l%l%s*:%s*', '');								-- remove language prefix if present (script value may now be empty string)
		script = strip_apostrophe_markup (script);								-- strip any apostrophe markup
	else
		script='';																-- if not set, make sure script is an empty string
	end
	if is_set (title) and is_set (script) then
		script = ' ' .. script;													-- add a space before we concatenate
	end
	return title .. script;														-- return the concatenation
end


--[[--------------------------< E S C A P E _ L U A _ M A G I C _ C H A R S >----------------------------------

Returns a string where all of lua's magic characters have been escaped.  This is important because functions like
string.gsub() treat their pattern and replace strings as patterns, not literal strings.
]]

local function escape_lua_magic_chars (argument)
	argument = argument:gsub("%%", "%%%%");										-- replace % with %%
	argument = argument:gsub("([%^%$%(%)%.%[%]%*%+%-%?])", "%%%1");				-- replace all other lua magic pattern characters
	return argument;
end


--[[--------------------------< G E T _ C O I N S _ P A G E S >------------------------------------------------

Extract page numbers from external wikilinks in any of the |page=, |pages=, or |at= parameters for use in COinS.

]]

local function get_coins_pages (pages)
	local pattern;
	if not is_set (pages) then return pages; end								-- if no page numbers then we're done
	
	while true do
		pattern = pages:match("%[(%w*:?//[^ ]+%s+)[%w%d].*%]");					-- pattern is the opening bracket, the url and following space(s): "[url "
		if nil == pattern then break; end										-- no more urls
		pattern = escape_lua_magic_chars (pattern);								-- pattern is not a literal string; escape lua's magic pattern characters
		pages = pages:gsub(pattern, "");										-- remove as many instances of pattern as possible
	end
	pages = pages:gsub("[%[%]]", "");											-- remove the brackets
	pages = pages:gsub("–", "-" );							-- replace endashes with hyphens
	pages = pages:gsub("&%w+;", "-" );						-- and replace html entities (&ndash; etc.) with hyphens; do we need to replace numerical entities like &#32; and the like?
	return pages;
end


--[=[-------------------------< C O I N S _ R E P L A C E _ M A T H _ S T R I P M A R K E R >------------------

There are three options for math markup rendering that depend on the editor's math preference settings.  These
settings are at [[Special:Preferences#mw-prefsection-rendering]] and are
	PNG images
	TeX source
	MathML with SVG or PNG fallback

All three are heavy with html and css which doesn't belong in the metadata.

Without this function, the metadata saved in the raw wikitext contained the rendering determined by the settings
of the last editor to save the page.

This function gets the rendered form of an equation according to the editor's preference before the page is saved.  It
then searches the rendering for the text equivalent of the rendered equation and replaces the rendering with that so
that the page is saved without extraneous html/css markup and with a reasonably readable text form of the equation.

When a replacement is made, this function returns true and the value with replacement; otherwise false and the intital
value.  To replace multipe equations it is necesary to call this function from within a loop.

]=]

local function coins_replace_math_stripmarker (value)
	local stripmarker = cfg.stripmarkers['math'];
	local rendering = value:match (stripmarker);								-- is there a math stripmarker

	if not rendering then														-- when value doesn't have a math stripmarker, abandon this test
		return false, value;
	end
	
	rendering = mw.text.unstripNoWiki (rendering);								-- convert stripmarker into rendered value (or nil? ''? when math render error)
	
	if rendering:match ('alt="[^"]+"') then										-- if PNG math option
		rendering = rendering:match ('alt="([^"]+)"');							-- extract just the math text
	elseif rendering:match ('$%s+.+%s+%$') then									-- if TeX math option; $ is legit character that is escapes as \$
		rendering = rendering:match ('$%s+(.+)%s+%$')							-- extract just the math text
	elseif rendering:match ('<annotation[^>]+>.+</annotation>') then			-- if MathML math option
		rendering = rendering:match ('<annotation[^>]+>(.+)</annotation>')		-- extract just the math text
	else
		return false, value;													-- had math stripmarker but not one of the three defined forms
	end
	
	return true, value:gsub (stripmarker, rendering, 1);
end


--[[--------------------------< C O I N S _ C L E A N U P >----------------------------------------------------

Cleanup parameter values for the metadata by removing or replacing invisible characters and certain html entities.

2015-12-10: there is a bug in mw.text.unstripNoWiki ().  It replaces math stripmarkers with the appropriate content
when it shouldn't.  See https://phabricator.wikimedia.org/T121085 and Wikipedia_talk:Lua#stripmarkers_and_mw.text.unstripNoWiki.28.29

TODO: move the replacement patterns and replacement values into a table in /Configuration similar to the invisible
characters table?

]]

local function coins_cleanup (value)
	local replaced = true;														-- default state to get the do loop running

	while replaced do															-- loop until all math stripmarkers replaced
		replaced, value = coins_replace_math_stripmarker (value);				-- replace math stripmarker with text representation of the equation
	end

	value = value:gsub (cfg.stripmarkers['math'], "MATH RENDER ERROR");						-- one or more couldn't be replaced; insert vague error message
	
	value = mw.text.unstripNoWiki (value);										-- replace nowiki stripmarkers with their content
	value = value:gsub ('<span class="nowrap" style="padding%-left:0%.1em;">&#39;(s?)</span>', "'%1");	-- replace {{'}} or {{'s}} with simple apostrophe or apostrophe-s
	value = value:gsub ('&nbsp;', ' ');											-- replace &nbsp; entity with plain space
	value = value:gsub ('\226\128\138', ' ');									-- replace hair space with plain space
	if not mw.ustring.find (value, cfg.indic_script) then						-- don't remove zero width joiner characters from indic script
		value = value:gsub ('&zwj;', '');											-- remove &zwj; entities
		value = mw.ustring.gsub (value, '[\226\128\141\226\128\139\194\173]', '');	-- remove zero-width joiner, zero-width space, soft hyphen
	end
	value = value:gsub ('[\009\010\013]', ' ');									-- replace horizontal tab, line feed, carriage return with plain space
	return value;
end


--[[--------------------------< C O I N S >--------------------------------------------------------------------

COinS metadata (see <http://ocoins.info/>) allows automated tools to parse the citation information.

]]

local function COinS(data, class)
	if 'table' ~= type(data) or nil == next(data) then
		return '';
	end

	for k, v in pairs (data) do													-- spin through all of the metadata parameter values
		if 'ID_list' ~= k and 'Authors' ~= k then								-- except the ID_list and Author tables (author nowiki stripmarker done when Author table processed)
			data[k] = coins_cleanup (v);
		end
	end

	local ctx_ver = "Z39.88-2004";
	
	-- treat table strictly as an array with only set values.
	local OCinSoutput = setmetatable( {}, {
		__newindex = function(self, key, value)
			if is_set(value) then
				rawset( self, #self+1, table.concat{ key, '=', mw.uri.encode( remove_wiki_link( value ) ) } );
			end
		end
	});
	
	if in_array (class, {'arxiv', 'biorxiv', 'citeseerx', 'journal', 'news'}) or (in_array (class, {'conference', 'interview', 'map', 'press release', 'web'}) and is_set(data.Periodical)) or 
		('citation' == class and is_set(data.Periodical) and not is_set (data.Encyclopedia)) then
			OCinSoutput.rft_val_fmt = "info:ofi/fmt:kev:mtx:journal";			-- journal metadata identifier
			if in_array (class, {'arxiv', 'biorxiv', 'citeseerx'}) then			-- set genre according to the type of citation template we are rendering
				OCinSoutput["rft.genre"] = "preprint";							-- cite arxiv, cite biorxiv, cite citeseerx
			elseif 'conference' == class then
				OCinSoutput["rft.genre"] = "conference";						-- cite conference (when Periodical set)
			elseif 'web' == class then
				OCinSoutput["rft.genre"] = "unknown";							-- cite web (when Periodical set)
			else
				OCinSoutput["rft.genre"] = "article";							-- journal and other 'periodical' articles
			end
			OCinSoutput["rft.jtitle"] = data.Periodical;						-- journal only
			OCinSoutput["rft.atitle"] = data.Title;								-- 'periodical' article titles

																				-- these used only for periodicals
			OCinSoutput["rft.ssn"] = data.Season;								-- keywords: winter, spring, summer, fall
			OCinSoutput["rft.chron"] = data.Chron;								-- free-form date components
			OCinSoutput["rft.volume"] = data.Volume;							-- does not apply to books
			OCinSoutput["rft.issue"] = data.Issue;
			OCinSoutput["rft.pages"] = data.Pages;								-- also used in book metadata

	elseif 'thesis' ~= class then												-- all others except cite thesis are treated as 'book' metadata; genre distinguishes
		OCinSoutput.rft_val_fmt = "info:ofi/fmt:kev:mtx:book";					-- book metadata identifier
		if 'report' == class or 'techreport' == class then						-- cite report and cite techreport
			OCinSoutput["rft.genre"] = "report";
		elseif 'conference' == class then										-- cite conference when Periodical not set
			OCinSoutput["rft.genre"] = "conference";
			OCinSoutput["rft.atitle"] = data.Chapter;							-- conference paper as chapter in proceedings (book)
		elseif in_array (class, {'book', 'citation', 'encyclopaedia', 'interview', 'map'}) then
			if is_set (data.Chapter) then
				OCinSoutput["rft.genre"] = "bookitem";
				OCinSoutput["rft.atitle"] = data.Chapter;						-- book chapter, encyclopedia article, interview in a book, or map title
			else
				if 'map' == class or 'interview' == class then
					OCinSoutput["rft.genre"] = 'unknown';						-- standalone map or interview
				else
					OCinSoutput["rft.genre"] = 'book';							-- book and encyclopedia
				end
			end
		else	--{'audio-visual', 'AV-media-notes', 'DVD-notes', 'episode', 'interview', 'mailinglist', 'map', 'newsgroup', 'podcast', 'press release', 'serial', 'sign', 'speech', 'web'}
			OCinSoutput["rft.genre"] = "unknown";
		end
		OCinSoutput["rft.btitle"] = data.Title;									-- book only
		OCinSoutput["rft.place"] = data.PublicationPlace;						-- book only
		OCinSoutput["rft.series"] = data.Series;								-- book only
		OCinSoutput["rft.pages"] = data.Pages;									-- book, journal
		OCinSoutput["rft.edition"] = data.Edition;								-- book only
		OCinSoutput["rft.pub"] = data.PublisherName;							-- book and dissertation
		
	else																		-- cite thesis
		OCinSoutput.rft_val_fmt = "info:ofi/fmt:kev:mtx:dissertation";			-- dissertation metadata identifier
		OCinSoutput["rft.title"] = data.Title;									-- dissertation (also patent but that is not yet supported)
		OCinSoutput["rft.degree"] = data.Degree;								-- dissertation only
		OCinSoutput['rft.inst'] = data.PublisherName;							-- book and dissertation
	end
																				-- and now common parameters (as much as possible)
	OCinSoutput["rft.date"] = data.Date;										-- book, journal, dissertation
	
	for k, v in pairs( data.ID_list ) do										-- what to do about these? For now assume that they are common to all?
--		if k == 'ISBN' then v = clean_isbn( v ) end
		if k == 'ISBN' then v = v:gsub( "[^-0-9X]", "" ); end
		local id = cfg.id_handlers[k].COinS;
		if string.sub( id or "", 1, 4 ) == 'info' then							-- for ids that are in the info:registry
			OCinSoutput["rft_id"] = table.concat{ id, "/", v };
		elseif string.sub (id or "", 1, 3 ) == 'rft' then						-- for isbn, issn, eissn, etc that have defined COinS keywords
			OCinSoutput[ id ] = v;
		elseif id then															-- when cfg.id_handlers[k].COinS is not nil
			OCinSoutput["rft_id"] = table.concat{ cfg.id_handlers[k].prefix, v };	-- others; provide a url
		end
	end

--[[	
	for k, v in pairs( data.ID_list ) do										-- what to do about these? For now assume that they are common to all?
		local id, value = cfg.id_handlers[k].COinS;
		if k == 'ISBN' then value = clean_isbn( v ); else value = v; end
		if string.sub( id or "", 1, 4 ) == 'info' then
			OCinSoutput["rft_id"] = table.concat{ id, "/", v };
		else
			OCinSoutput[ id ] = value;
		end
	end
]]
	local last, first;
	for k, v in ipairs( data.Authors ) do
		last, first = coins_cleanup (v.last), coins_cleanup (v.first or '');	-- replace any nowiki strip markers, non-printing or invisible characers
		if k == 1 then															-- for the first author name only
			if is_set(last)  and is_set(first) then								-- set these COinS values if |first= and |last= specify the first author name
				OCinSoutput["rft.aulast"] = last;								-- book, journal, dissertation
				OCinSoutput["rft.aufirst"] = first;								-- book, journal, dissertation
			elseif is_set(last) then 
				OCinSoutput["rft.au"] = last;									-- book, journal, dissertation -- otherwise use this form for the first name
			end
		else																	-- for all other authors
			if is_set(last) and is_set(first) then
				OCinSoutput["rft.au"] = table.concat{ last, ", ", first };		-- book, journal, dissertation
			elseif is_set(last) then
				OCinSoutput["rft.au"] = last;									-- book, journal, dissertation
			end
		end
	end

	OCinSoutput.rft_id = data.URL;
	OCinSoutput.rfr_id = table.concat{ "info:sid/", mw.site.server:match( "[^/]*$" ), ":", data.RawPage };
	OCinSoutput = setmetatable( OCinSoutput, nil );
	
	-- sort with version string always first, and combine.
	--table.sort( OCinSoutput );
	table.insert( OCinSoutput, 1, "ctx_ver=" .. ctx_ver );  -- such as "Z39.88-2004"
	return table.concat(OCinSoutput, "&");
end


--[[--------------------------< S E T _ S E L E C T E D _ M O D U L E S >--------------------------------------

Sets local cfg table and imported functions table to same (live or sandbox) as that used by the other modules.

]]

local function set_selected_modules (cfg_table_ptr, utilities_page_ptr)
	cfg = cfg_table_ptr;

	is_set = utilities_page_ptr.is_set;											-- import functions from select Module:Citation/CS1/Utilities module
	in_array = utilities_page_ptr.in_array;
	remove_wiki_link = utilities_page_ptr.remove_wiki_link;
end



return {
	make_coins_title = make_coins_title,
	get_coins_pages = get_coins_pages,
	COinS = COinS,
	set_selected_modules = set_selected_modules,
	}