Module:Unicode data/testcases

-- Example Unit tests for [[Module:Bananas]]. Click talk page to run tests.
local p = require 'Module:UnitTests'
local Unicode_data = require 'Module:Unicode data'

function p:iterate(examples, func)
	if type(examples) ~= 'table' then
		error('First argument of iterate should be a table, not ' .. type(examples) .. '.')
	end
	if type(func) == 'string' then
		func = self[func]
		for i, example in ipairs(examples) do
			if type(example) == 'table' then
				func(self, unpack(example))
			else
				error('iterate does not know what to do with example number ' .. i .. ', whose type is ' .. type(example) .. '.')
			end
		end
	else
		error('Second argument of iterate should be a string, not ' .. type(func) .. '.')
	end
end

local U = mw.ustring.char
local function show(codepoint)
	if Unicode_data.is_printable(codepoint) then
		local printed_codepoint = U(codepoint)
		if mw.ustring.toNFC(printed_codepoint) ~= printed_codepoint then
			printed_codepoint = ("&#x%X;"):format(codepoint)
		end
		if Unicode_data.is_combining(codepoint) then
			printed_codepoint = "◌" .. printed_codepoint
		end
		return ("U+%04X: %s"):format(codepoint, printed_codepoint)
	else
		return ("U+%04X"):format(codepoint)
	end
end

function p:check_lookup_name(codepoint, name)
	self:equals(show(codepoint), Unicode_data.lookup_name(codepoint), name)
end

function p:test_lookup_name()
	local examples = {
		{   0x0000, "<control-0000>" },
		{   0x007F, "<control-007F>" },
		{   0x00C1, "LATIN CAPITAL LETTER A WITH ACUTE" },
		{   0x0300, "COMBINING GRAVE ACCENT" },
		{   0x0378, "<reserved-0378>" },
		{   0x1B44, "BALINESE ADEG ADEG" },
		{   0x1F71, "GREEK SMALL LETTER ALPHA WITH OXIA" },
		{   0x3555, "CJK UNIFIED IDEOGRAPH-3555" },
		{   0xAC01, "HANGUL SYLLABLE GAG" },
		{   0xD5FF, "HANGUL SYLLABLE HEH" },
		{   0xDC00, "<surrogate-DC00>", },
		{   0xEEEE, "<private-use-EEEE>" },
		{   0xFDD1, "<noncharacter-FDD1>", },
		{   0xFFFD, "REPLACEMENT CHARACTER" },
		{   0xFFFF, "<noncharacter-FFFF>" },
		{  0x1F4A9, "PILE OF POO" },
		{  0xE0000, "<reserved-E0000>" },
		{  0xF0F0F, "<private-use-F0F0F>" },
		{ 0x10FFFF, "<noncharacter-10FFFF>" },
	}
	
	self:iterate(examples, "check_lookup_name")
end

function p:check_is_combining(codepoint, expected)
	self:equals(
		("%s (%s)"):format(show(codepoint), Unicode_data.lookup_name(codepoint)),
		Unicode_data.is_combining(codepoint),
		expected
	)
end

function p:test_is_combining()
	local examples = {
		{ 0x0300, true },
		{ 0x0060, false },
	}
	
	self:iterate(examples, "check_is_combining")
end

function p:check_lookup_script(codepoint, expected)
	self:equals(
		("%s (%s)"):format(show(codepoint), Unicode_data.lookup_name(codepoint)),
		Unicode_data.lookup_script(codepoint),
		expected)
end

function p:test_lookup_script()
	local examples = {
		{ 0x0061, "Latn" },
		{ 0x002F, "Zyyy" },
		{ 0x0300, "Zinh" },
		{ 0x0378, "Zzzz" },
		{ 0x0398, "Grek" },
		{ 0x03E2, "Copt" },
		{ 0x2014, "Zyyy" },
	}
	
	self:iterate(examples, "check_lookup_script")
end

function p:check_get_best_script(str, expected)
	self:equals(
		str,
		Unicode_data.get_best_script(str),
		expected)
end

function p:test_get_best_script()
	local examples = {
		-- Two examples from [[Template talk:Lang#Italicisation of Halkomelem]]
		{ "lá:yelhp", "Latn" },
		{ "xʷməθkʷəy̓əm", nil },
	}
	
	self:iterate(examples, "check_get_best_script")
end

local new_p = {}
for k, v in pairs(p) do
	if type(k) == "string" then
		new_p[k:gsub("^test_", "test ")] = v
	else
		new_p[k] = v
	end
end
p = new_p

return p