Editing Module:Language/data/ISO 639-2/make

require('strict');

--[=[------------------------< I S O _ S Y N O N Y M _ E X T R A C T >-----------------------------------------

{{#invoke:Language/data/ISO 639-2/make|ISO_synonym_extract|file-date=2013-01-11}}

reads a local copy of data from the table at http://www.loc.gov/standards/iso639-2/php/English_list.php, extracts
the ISO 639-2 (or 639-2T) codes that have equivalent ISO 639-1 codes and creates a table to translate 639-2 to 639-1.
ISO-639-3 uses 639-2T codes

useful lines in the source table have the form:
	<English name>\t<all English names>\t<all French names>\t<639-2 code>\t<639-1 code>\n
where:
	<English name> is primary English name (not used here); one of <all English names> so duplicates code listing
	<all English names> is all of the English names (not used here)
	<all French names> is all of the French names (not used here)
	<639-2 code> is the three-character ISO 639-2 or 639-2B/639-2T language code; when 639-2T present, use that code
	<639-1 code> is the two-character ISO 639-1 language code synonym of the -2 code (if one is defined)
		
	like this (with synonym):
		Abkhazian	Abkhazian	abkhaze	abk	ab
	or (without synonym):
		Achinese	Achinese	aceh	ace	 

for the file date use the date listed at the bottom of the source page in yyyymmdd numeric format without hyphens or spaces

]=]

local function ISO_synonym_extract (frame)
	local page = mw.title.getCurrentTitle();									-- get a page object for this page
	local content = page:getContent();											-- get unparsed content
	local content_table = {};													-- table of text lines from source
	local split_table = {};														-- table of lines split at the tabs	
	local skip_table = {};														-- table of 636-2/639-2T codes that have been handled; used to prevent duplication
	local out_table = {};														-- output table
	
	local file_date = 'File-Date: ' .. frame.args["file-date"];					-- set the file date line from |file-date= (from the bottom of the source page)

	content_table = mw.text.split (content, '[\r\n]');							-- make a table of text lines
	for _, line in ipairs (content_table) do									-- for each line
		split_table = mw.text.split (line, '\t');								-- split at the table
		if split_table[5] and (' ' ~= split_table[5]) then						-- if there is a 639-1 code
			local code = split_table[4]:match ('%a+/(%a+)') or split_table[4];	-- when 639-2B/639-2T use 639-2T else use 639-2
			if not skip_table[code] then										-- skip if code already in the skip table because more than one language name
				skip_table[code] = true;										-- remember that we've handled this 636-2/639-2T code
				table.insert (out_table, "[\"" .. code .. "\"] = \"" .. split_table[5] .. "\"");		-- make new table entry
			end
		end
	end
	
	table.sort (out_table);
	
	return "<br /><pre>-- " .. file_date .. "<br />return {<br />&#9;" .. table.concat (out_table, ',<br />&#9;') .. "<br />&#9;}<br />" .. "</pre>";
end


--[[--------------------------< I S O _ 2 _ N A M E _ E X T R A C T >------------------------------------------

{{#invoke:Language/data/ISO 639-2/make|ISO_2_name_extract|file-date=2013-01-11}}

reads a local copy of data from the table at http://www.loc.gov/standards/iso639-2/php/English_list.php, extracts
the ISO 639-2 and 639-2T codes and their associated language names

useful lines in the source table have the form:
	<English name>\t<all English names>\t<all French names>\t<639-2 code>\t<639-1 code>\n
where:
	<English name> is primary English name (not used here); one of <all English names> so duplicates code listing
	<all English names> is all of the English names (used here)
	<all French names> is all of the French names (not used here)
	<639-2 code> is the three-character ISO 639-2 or 639-2B/639-2T language code; both are used
	<639-1 code> is the two-character ISO 639-1 language code synonym of the -2 code (not used here)

for the file date use the date listed at the bottom of the source page in yyyymmdd numeric format without hyphens or spaces

]]

local function ISO_2_name_extract (frame)
	local page = mw.title.getCurrentTitle();									-- get a page object for this page
	local content = page:getContent();											-- get unparsed content
	local content_table = {};													-- table of text lines from source
	local split_table = {};														-- table of lines split at the tabs	
	local skip_table = {['qaa-qtz']=true};										-- table of 636-2/639-2T codes that have been handled; used to prevent duplication; qaa-qtz reserved for local use so not supported here
	local name_table = {};														-- holds language names for processing
	local code_table = {};														-- because some languages have both -2B and -2T codes
	local out_table_T = {};														-- output table for 639-2T codes
	local out_table_B = {};														-- output table for 639-2B codes
	local out_table;															-- used as a pointer to the selected out_table_B or out_table_T
	
	local file_date = 'File-Date: ' .. frame.args["file-date"];					-- set the file date line from |file-date= (from the bottom of the source page)

	content_table = mw.text.split (content, '[\r\n]');							-- make a table of text lines
	for _, line in ipairs (content_table) do									-- for each line
		split_table = mw.text.split (line, '\t');								-- split at the tab

		if split_table[4] then													-- if a code then continue processing; skip this line else
			name_table = mw.text.split (split_table[2], ' *; *');				-- split 'all English names' at the '; ' into a table of individual names
			for i, v in ipairs (name_table) do
				name_table [i] = mw.ustring.gsub (v, '(.+)', '"%1"');			-- add double quotes around each name
			end
			
			code_table = mw.text.split (split_table[4], ' */ *');				-- split 'ISO 639-2' code at the '/' into a table of -2B and -2T individual codes
			
			for i, code in ipairs (code_table) do								-- now built a table entry for the code(s) and its(their) associated language(s)
				if not skip_table[code] then									-- source data has duplicates so check to see if we have already done this code
					out_table = (2 == #code_table and 1 == i) and out_table_B or out_table_T;	-- does this language name have both -2B and -2T codes?
					table.insert (out_table,
						table.concat ({
							'["',												-- open code index
							code,												-- the code
							'"] = {',											-- close code index; open name table
							table.concat(name_table, ', '),						-- add the names
							'}'													-- close the names table
						})
					)
					skip_table[code] = true;									-- remember that we've done this code
				end
			end
		end
	end
	
	table.sort (out_table_T);
	table.sort (out_table_B);
	
--	return "<br /><pre>-- " .. file_date .. "<br />return {<br />&#9;" .. table.concat (out_table_T, ',<br />&#9;') .. "<br />&#9;}<br />" .. "</pre>";
	return table.concat ({
		"<br /><pre>-- ",
		file_date,
		"<br />return {",
		string.rep ('&#9;', 18),
		'-- 639-2T<br />&#9;',
		table.concat (out_table_T, ',<br />&#9;'),
		"<br />&#9;}<br />" .. "</pre><br /><br />",
		"<br /><pre>-- ",
		file_date,
		"<br />return {",
		string.rep ('&#9;', 18),
		'-- 639-2B<br />&#9;',
		table.concat (out_table_B, ',<br />&#9;'),
		"<br />&#9;}<br />" .. "</pre><br /><br />",
		});
end


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

return {
	ISO_synonym_extract = ISO_synonym_extract,
	ISO_2_name_extract = ISO_2_name_extract
	};