Module:Cite IUCN

require('strict');
local getArgs = require ('Module:Arguments').getArgs;

local amendment_pattern = '%s*%(amended version of (%d%d%d%d) assessment%)';
local errata_pattern = '%s*%(errata version published in (%d%d%d%d)%)';
local green_status_pattern = '%s*%((Green Status assessment)%)';


--[[--------------------------< I U C N _ I D E N T I F I E R S _ G E T >--------------------------------------

cs1|2 templates cite single sources;  when the identifiers in |doi=, |id=, and |page= are different from each other
then the template is attempting to cite multiple sources.  This function evaluates the identifier portions of these
parameters. returns seven values: identifyier parts (or nil when parameter not used) and a message (nil on success,
error message else)

the identifier portions of the several parameters must be properly formed

]]

local function iucn_identifiers_get (args, error_msgs_t)
	local doi_taxon_ID, doi_assesment_ID
	local page_taxon_ID, page_assesment_ID
	local url_taxon_ID, url_assesment_ID

	if args.doi then
		local lang_tag
		doi_taxon_ID, doi_assesment_ID, lang_tag = args.doi:match ('[Tt](%d+)[Aa](%d+)%.(%l%l)$')
		if not doi_taxon_ID or not ({['en'] = true, ['es'] = true, ['fr'] = true, ['pt'] = true})[lang_tag] then
			table.insert (error_msgs_t, 'malformed |doi= identifier');
		end
	end
	if args.page then
		page_taxon_ID, page_assesment_ID = args.page:match ('^[eE]%.[Tt](%d+)[Aa](%d+)$')
		if not page_taxon_ID then
			table.insert (error_msgs_t, 'malformed |page= identifier');
		end
	end
	if args.url then
		if args.url:match ('https://www.iucnredlist.org/species/') then			-- must be a 'new-form' url
			url_taxon_ID, url_assesment_ID = args.url:match ('/species/(%d+)/(%d+)')
			if not url_taxon_ID then
				table.insert (error_msgs_t, 'malformed |url= identifier');
			end
		end
	end

	if not error_msgs_t[1] then
		if doi_taxon_ID and page_taxon_ID then
			if (doi_taxon_ID ~= page_taxon_ID or ((doi_assesment_ID ~= page_assesment_ID) and not args.errata)) then
				table.insert (error_msgs_t, '|doi= / |page= mismatch');
			end
		end
		if doi_taxon_ID and url_taxon_ID then
			if (doi_taxon_ID ~= url_taxon_ID or ((doi_assesment_ID ~= url_assesment_ID) and not args.errata)) then
				table.insert (error_msgs_t, '|doi= / |url= mismatch');
			end
		end
		
		if page_taxon_ID and url_taxon_ID then
			if (page_taxon_ID ~= url_taxon_ID or ((page_assesment_ID ~= url_assesment_ID) and not args.errata)) then
				table.insert (error_msgs_t, '|page= / |url= mismatch');
			end
		end
	end

	return doi_taxon_ID, doi_assesment_ID, page_taxon_ID, page_assesment_ID;
end


--[[--------------------------< I U C N _ V O L U M E _ C H E C K >--------------------------------------------

compares volume in |volume= (if present) against year in |date= or |year= (if present) against volume in |doi= (if present)

returns nil if all that are present are correct; message else

]]

local function iucn_volume_check (args, maint_msgs_t)
	local vol = args.volume;
	local date = args.date or args.year;
	local doi = args.doi and args.doi:match ('[Ii][Uu][Cc][Nn]%.[Uu][Kk]%.(%d%d%d%d)')

	if vol and date and (vol ~= date) then
		table.insert (maint_msgs_t, '|volume= / |date= mismatch');
	end

	if vol and doi and ((vol ~= doi) and not args.amends) then
		table.insert (maint_msgs_t, '|volume= / |doi= mismatch');
	end

	if date and doi and ((doi ~= date) and not args.amends) then
		table.insert (maint_msgs_t, '|date= / |doi= mismatch');
	end
end


--[[--------------------------< C I T E >----------------------------------------------------------------------

Wraps {{cite journal}}:
     takes cite journal parameters but updates old style url using electronic page number
     page should be in format e.T13922A45199653
     the url uses                13922/45199653
     so we need to extract the number between T and A (taxon ID) and the number after A (assessment ID)
     the target url is https://www.iucnredlist.org/species/13922/45199653
     usage: {{#invoke:iucn|cite}}
     template: {{Template:Cite iucn}}

]]

local function cite (frame)
	local error_msgs_t = {};													-- holds error messages for rendering
	local maint_msgs_t = {};														-- holds hidden maint messages for rendering
	local namespace = mw.title.getCurrentTitle().namespace;						-- used for categorization
	local args = getArgs (frame);												-- local copy of template arguments

	if args.title and (args.title:match (errata_pattern) or args.title:match (amendment_pattern)) then
		table.insert (error_msgs_t, 'title has extraneous text');					-- announce that this template has has errata or amendment text
	end

	local doi_taxon_ID, doi_assesment_ID;										-- all of these contain the same identifying info in slightly
	local page_taxon_ID, page_assesment_ID;										-- different forms. when any combination of these is present,

	doi_taxon_ID, doi_assesment_ID, page_taxon_ID, page_assesment_ID = iucn_identifiers_get (args, error_msgs_t);

	args.id = nil																-- unset; not supported

	local url_taxon_ID = page_taxon_ID or doi_taxon_ID;							-- select for use in url that we will create
	local url_assesment_ID = page_assesment_ID  or doi_assesment_ID;
	
	local url = args.url;
	if url then
		if url:find ('iucnredlist.org/details/', 1, true) then					-- old-form url
			if url_taxon_ID then												-- when there is an identifier
				url = nil														-- unset; we'll create new url below
			else																-- here when old-form but no identifier that we can use to create new url
				args.url = args.url:gsub ("http:", "https:")					-- sometimes works with redirect on iucn site
			end
			table.insert (maint_msgs_t, 'old-form url')							-- announce that this template has has an old-form url
		elseif url:find ('iucnredlist.org/species/', 1, true) then				-- new-form url
--			table.insert (maint_msgs_t, 'new-form url')				--TODO: restore this line when most new-form urls have been removed from article space		-- announce that this template has has an new-form url
		else
			table.insert (error_msgs_t, 'unknown url')							-- emit error message
		end
	end

	if not url then																-- when no url or unset old-form url
		if url_taxon_ID then
			args.url = "https://www.iucnredlist.org/species/" .. url_taxon_ID .. '/' .. url_assesment_ID
		else
			table.insert (error_msgs_t, 'no identifier')						-- emit error message
		end
	end

	-- add journal if not provided (TODO decide if this should override provided value)
	if not args['journal'] and not args['work'] then
		args['journal'] = "[[IUCN Red List|IUCN Red List of Threatened Species]]"
	end
	
	iucn_volume_check (args, maint_msgs_t);										-- |volume=, |year= (|date=), |doi= must all refer to the same volume

	if not args.volume and (args.year or args.date) then
		args.volume = args.year or args.date
	end
	
	if args.errata then
		args['orig-date'] = 'errata version of ' .. (args.year or args.date or args.volume) .. ' assessment';
		args.date = args.errata;												-- update publication data to errata year
		args.year = nil;														-- unset these as no longer needed
		args.errata = nil;
	elseif args.amends then
		args['orig-date'] = 'amended version of ' .. args.amends .. ' assessment';
		args.amends = nil;														-- unset as no longer needed
	end
																				-- add free-to-read icon to mark a correctly formed doi
	args['doi-access'] = args.doi and args.doi:match ('10%.2305/[Ii][Uu][Cc][Nn].+[Tt]%d+[Aa]%d+%.%a%a') and 'free' or nil
	
	local out_t = {};
	if error_msgs_t[1] then
		table.insert (out_t, ' <span class="error" style="font-size:100%">{{[[Template:cite iucn|cite iucn]]}}: error: ');
		table.insert (out_t, table.concat (error_msgs_t, ', '));
		table.insert (out_t, ' ([[Template:Cite iucn#Error messages|help]])</span>');
		if (0 == namespace) then
			table.insert (out_t, '[[Category:cite IUCN errors]]');
		end
	end

	if maint_msgs_t[1] then
		table.insert (out_t, '<span class="citation-comment" style="display: none; color: #33aa33; margin-left: 0.3em;">');
		if not error_msgs_t[1] then
			table.insert (out_t, '{{[[Template:cite iucn|cite iucn]]}}: ')
			table.insert (out_t, table.concat (maint_msgs_t, ', '));
			table.insert (out_t, ' ([[Template:Cite iucn#Maintenance messages|help]])</span>');
			if (0 == namespace) then
				table.insert (out_t, '[[Category:cite IUCN maint]]');
			end
		end
	end

	if (not args['doi-access']) and (0 == namespace) then
		table.insert (out_t, '[[Category:cite IUCN without doi]]');
	end
	
	return frame:expandTemplate{ title = 'cite journal', args = args } ..							-- the template
		table.concat (out_t);																		-- error and maint messages and categories
end


--[=[-------------------------< E T _ A L _ P A T T E R N S >--------------------------------------------------

This adapted from Module:Citation/CS1/Configuration

This table provides Lua patterns for the phrase "et al" and variants in a name.

]=]

local et_al_patterns = {
	"[;,]? *[\"']*%f[%a][Ee][Tt]%.? *[Aa][Ll][%.;,\"']*$",						-- variations on the 'et al' theme
	"[;,]? *[\"']*%f[%a][Ee][Tt]%.? *[Aa][Ll][Ii][AaIi][Ee]?[%.;,\"']*$",		-- variations on the 'et alia', 'et alii' and 'et aliae' themes (false positive 'et aliie' unlikely to match)
	"[;,]? *%f[%a]and [Oo]thers",												-- an alternative to et al.
	}


--[[---------------------< N A M E _ H A S _ E T A L >--------------------------

This adapted from Module:Citation/CS1

Evaluates the content of a name for variations on the theme of et al.  If found,
returns true; nil else

]]

local function name_has_etal (name)
	local etal;

	if name then																-- name can be nil in which case just return
		name = name:gsub ('%b<>', '');											-- remove any html markup (typically <i>...</i>)
		for _, pattern in ipairs (et_al_patterns) do							-- loop through all of the patterns
			if name:match (pattern) then										-- if this 'et al' pattern is found in name
				return true;													-- has etal, so return true
			end
		end
	end
end


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

creates a list of individual |authorn= parameters from the list of names provided in the raw iucn citation.  names
must have the form: Surname, I. (more than one 'I.' pair allowed but no spaces between I. pairs)

assumes that parenthetical text at the end of the author-name-list is a collaboration
	Name, I.I., & Name, I.I. (Colaboration name)

assumes that <i>et al.</i> is the last name in a list of names

]]

--local function author_names_get (raw_iucn_cite)
local function author_names_get (raw_iucn_cite, params_t)						-- EXPERIMENT
	local list = {};															-- table that holds name list parts
	local author_names = raw_iucn_cite:match ('^([^%d]-)%s+%d%d%d%d');			-- extract author name-list from raw iucn citation
	local collaboration = author_names:match ('%s*(%b())$');					-- get collaboration name if it exists

	if collaboration then														-- when there is a colaboration
		collaboration = collaboration:gsub ('[%(%)]', '');						-- remove bounding parentheses
		author_names = author_names:gsub ('%s*(%b())$', '');					-- and remove collaboration from author-name-list
	end
	
	local names = author_names:gsub ('%.?,?%s+&%s+', '.|');						-- replace 'separators' (<optional dot><optional comma><space><ampersand><space>) with <dot><pipe>
	names = names:gsub ('%.,%s*', '.|');										-- replace 'separators' (<dot><comma><optional space>) with <dot><pipe>
	names = names:gsub ('(%.%u),', '%1.|');										-- special case for when last initial is missing its trailing dot
	list = mw.text.split (names, '|');											-- split the string on the pipes into entries in list{}
	
	if 0 == #list then
		params_t['author'] = author_names;										-- EXPERIMENT
		return table.concat ({'|author=', author_names}), params_t;				-- EXPERIMENT		-- no 'names' of the proper form; return the original as a single |author= parameter
--		return table.concat ({'|author=', author_names})						-- no 'names' of the proper form; return the original as a single |author= parameter
	else
		for i, name in ipairs (list) do											-- spin through the list and 
			if name_has_etal (name) then										-- if this name has some form of 'et al'
				params_t['display-authors'] = 'etal';							-- EXPERIMENT
				list[i] = '|display-authors=etal';								-- add |dispaly-authors=etal parameter and 
				break;															-- assume that the etal was the last 'name' so stop processing names
			else
				params_t['author' .. i] = name;									-- EXPERIMENT
				list[i] = table.concat ({'|author', (i == 1) and '' or i, '=', name});	-- add |authorn= parameter names; create |author= instead of |author1=
			end
		end
		if collaboration then
			params_t['collaboration'] = collaboration;							-- EXPERIMENT
			table.insert (list, table.concat ({'|collaboration', '=', collaboration}));	-- add |collaboration= parameter
		end
		return table.concat (list, ' ');										-- make a big string and return that
	end
end


--[[--------------------------< T I T L E _ G E T >------------------------------------------------------------

extract and format citation title; attempts to get the italic right

''binomen'' (amended or errata title)
''binomen''
''binomen'' ssp. ''subspecies''
''binomen'' subsp. ''subspecies''
''binomen'' var. ''variety''
''binomen'' subvar. ''subvariety''

all of the above may have trailing amended or errata text in parentheses

TODO: are there others?

]]

local function title_get (raw_iucn_cite)
	local title = raw_iucn_cite:match ('%d%d%d%d%.%s+(.-)%s*%. The IUCN Red List of Threatened Species');

	local patterns = {															-- tables of string.match patterns [1] and string.gsub patterns [2]
		{'(.-)%sssp%.%s+(.-)%s(%b())$', "''%1'' ssp. ''%2'' %3"},				-- binomen ssp. subspecies (zoology) with errata or amended text
		{'(.-)%sssp%.%s+(.+)', "''%1'' ssp. ''%2''"},							-- binomen ssp. subspecies (zoology)
		{'(.-)%ssubsp%.%s+(.-)%s(%b())$', "''%1'' subsp. ''%2'' %3"},			-- binomen subsp. subspecies (botany) with errata or amended text
		{'(.-)%ssubsp%.%s+(.+)', "''%1'' subsp. ''%2''"},						-- binomen subsp. subspecies (botany)
		{'(.-)%svar%.%s+(.-)%s+(%b())$', "''%1'' var. ''%2'' %3"},				-- binomen var. variety (botany) with errata or amended text
		{'(.-)%svar%.%s+(.+)', "''%1'' var. ''%2''"},							-- binomen var. variety (botany)
		{'(.-)%ssubvar%.%s+(.-)%s(%b())$', "''%1'' subvar. ''%2'' %3"},			-- binomen subvar. subvariety (botany) with errata or amended text
		{'(.-)%ssubvar%.%s+(.+)', "''%1'' subvar. ''%2''"},						-- binomen subvar. subvariety (botany)
		{'(.-)%s*(%b())$', "''%1'' %2"},										-- binomen with errata or amended text
		{'(.+)', "''%1''"},														-- binomen
		}
	
	for i, v in ipairs (patterns) do											-- spin through the patterns
		if title:match (v[1]) then												-- when a match
			title = title:gsub (v[1], v[2]);									-- add italics 
			break;																-- and done
		end
	end

--	return table.concat ({' |title=', title});									-- return the |title= parameter
	return title;																-- return the formatted title
end


--[[--------------------------< M A K E _ C I T E _ I U C N >--------------------------------------------------

parses apart an iucn-format citation copied from their webpage and reformats that into a {{cite iucn}} template for substing

automatic substing by User:AnomieBOT/docs/TemplateSubster

]]

local function make_cite_iucn (frame)
	local args_t = getArgs (frame);
	local raw_iucn_cite = args_t[1];

	local template_t = {'{{cite iucn '};										-- sequence that holds the {{cite iucn}} template as it is being assembled; for nowiki'd output
	local params_t = {};														-- table of parameter/value pairs for substing
	
	local year, volume, page, doi, accessdate;

	year = raw_iucn_cite:match ('^%D+(%d%d%d%d)');
	volume, page = raw_iucn_cite:match ('(%d%d%d%d):%s+(e%.T%d+A+%d+)%.%s?');
	doi = raw_iucn_cite:match ('10%.2305/IUCN%.UK%.[%d%-]+%.RLTS%.T%d+A%d+%.%a%a');

	accessdate = raw_iucn_cite:match ('Accessed on (.-)%.?$') or raw_iucn_cite:match ('Downloaded on (.-)%.?$');	-- 'Downloaded' → 'Accessed' change occured December 2021;
	accessdate = accessdate:gsub ('^0', '');									-- strips leading 0 in day 01 January 2020 -> 1 January 2020

	table.insert (template_t, author_names_get (raw_iucn_cite, params_t));		-- add author name parameters; as a single string to <template_t>; as individual entries to <params_t>

	table.insert (template_t, table.concat ({' |year=', year}));				-- add formatted year
	params_t.year = year;
	
	local title = title_get (raw_iucn_cite);
	local type_p = title:match (green_status_pattern);
	if type_p then
		title = title:match ('^([^%(]+)%s*%(');
		table.insert (template_t, table.concat ({' |type=', type_p}));			-- add formatted errata
		params_t.type = type_p;
	end

	local errata = title:match (errata_pattern);								-- nil unless IUCN citation has errata annotation; else year that this errata published (|date=)
	if errata then
		table.insert (template_t, table.concat ({' |errata=', errata}));		-- add formatted errata
		params_t.errata = errata;
		title = title:gsub (errata_pattern, '');								-- remove errata annotation
	end
	local amends = title:match (amendment_pattern);								-- nil unless IUCN citation has amendment annotation; else year that this assessment amends (|orig-date=)
	if amends then
		table.insert (template_t, table.concat ({' |amends=', amends}));		-- add year of assessment that this assessment amends
		params_t.amends = amends;
		title = title:gsub (amendment_pattern, '');								-- remove amendment annotation
	end

	table.insert (template_t, table.concat ({' |title=', title}));				-- add formatted title
	params_t.title = title;
	table.insert (template_t, table.concat ({' |volume=', volume}));			-- add formatted volume
	params_t.volume = volume;
	table.insert (template_t, table.concat ({' |page=', page}));				-- add formatted page
	params_t.page = page;
	table.insert (template_t, table.concat ({' |doi=', doi}));					-- add formatted doi
	params_t.doi = doi;

	table.insert (template_t, table.concat ({' |access-date=', accessdate}));	-- add formatted access-date
	params_t['access-date'] = accessdate;

	table.insert (template_t, '}}');											-- close the template

	if args_t[2] then															-- if anything in args_t[2], write a nowiki'd version that editors can copy into <ref> tags
		return frame:preprocess (table.concat ({'<syntaxhighlight lang="wikitext" inline="1">', table.concat (template_t), '</syntaxhighlight>'})); -- caveat lector: if left long enough anomiebot will subst this
	end

	if args_t['ref'] then														-- enable subst of ref tags with name
		return frame:preprocess ('<ref name=' .. args_t['ref'] .. '>' .. table.concat (template_t) .. '</ref>')
	end

	return frame:preprocess (table.concat (template_t));						-- render {{cite iucn}} template; substable
end


--[[--------------------------< E X P O R T E D   F U N C T I O N S >------------------------------------------
]]

return {
	cite = cite,
	make_cite_iucn = make_cite_iucn,
	}
Module:Cite IUCN

Usage

Background information

Old-style and new-style IUCN urls

Current IUCN citations

Legacy issue with old-style urls

New taxon IDs
