Module:it-pronunciation/sandbox/testcases

local tests = require "Module:UnitTests"

setmetatable(_G, { __index = require "Module:it-pronunciation/sandbox" })

local function tag_IPA(transcription)
	if transcription:find "^Error" then
		return transcription
	else
		return '<span class="IPA">' .. transcription .. '</span>'
	end
end

local examples = {
	{ "càne", "ˈka.ne", "ˈkäː.n̺e̞" },
	{ "bène", "ˈbɛ.ne", "ˈbɛː.n̺e" },
	{ "sóno", "ˈso.no", "ˈs̪oː.n̺o" },
	{ "sòno", "ˈsɔ.no", "ˈs̪ɔː.n̺o" },
	{ "ho", "o", "" },
	"English letter sequences",
	{ "smoking", "ˈzmo.kiŋ", "" },
	{ "gangster", "ˈgaŋ.ster", "" },
	{ "hòckey", "ˈɔ.ke.i", "" },
	{ "hockeìstico", "o.keˈis.ti.ko", "" },
	{ "crawl", "krɔl", "" },
	{ "shòck", "ʃɔk", "" },
	{"Oahu", "oˈa.u", ""},
	"y",
	{ "yo-yo", "ˈjo.jo", "" },
	{ "whisky", "ˈwis.ki", "" },
	"x",
	{ "Craxi", "ˈkrak.si", "" },
	"-izzare",
	{ "analizzare", "a.na.lidˈd͡za.re", ""},
	{ "nazionalizzare", "nat.t͡sjo.na.lidˈd͡za.re", "" },
	{ "zonizzare", "d͡zo.nidˈd͡za.re", "", true },
	{ "rizzare", "ritˈt͡sa.re", ""},
	"c, g, sc",
	{ "cèlo", "ˈt͡ʃɛ.lo", "ˈt͡ʃɛː.lo" },
	{ "ciaccòna", "t͡ʃakˈkɔ.na", "t͡ʃäkˈkɔː.n̺ä" },
	{ "fàccio", "ˈfat.t͡ʃo", "ˈfät͡ʃ.t͡ʃo" }, -- ˈfat.t͡ʃo is too phonetic; other alternative, ˈfa.t͡ʃːo
	{ "che", "ke", "k̟e" },
	{ "òcchio", "ˈɔk.kjo", "ˈɔk̟.k̟jo" }, -- Good test for syllabification.
	{ "òcchi", "ˈɔk.ki", "ˈɔk̟.k̟i" },
	{ "òggi", "ˈɔd.d͡ʒi", "ˈɔd͡ʒ.d͡ʒi" }, -- ˈɔd.d͡ʒi is too phonetic; other alternative, ˈɔ.d͡ʒːi 
	{ "Sciàlpi", "ˈʃal.pi", "ˈʃäl.pi" },
	{ "lasciàre", "laʃˈʃa.re", "läʃˈʃäː.r̺e̞" },
	{ "scleràre", "skleˈra.re", "s̪kleˈr̺äː.r̺e̞" },
	{ "ùscio", "ˈuʃ.ʃo", "ˈuʃ.ʃo" },
	"gn, gl",
	{ "ógni", "ˈoɲ.ɲi", "ˈoɲ.ɲi" },
	{ "gnòcco", "ˈɲɔk.ko", "ˈɲɔk.ko" },
	{ "gnagnà", "ɲaɲˈɲa", "ɲäɲˈɲä" }, -- probably not a real word
	{ "vòglio", "ˈvɔʎ.ʎo", "ˈvɔʎ.ʎo" },
	{ "gli", "ʎi", "ʎi" },
	{ "inglése", "inˈɡle.ze", "iŋˈɡleː.z̪e" },
	"ng, nq, nch",
	{ "anche", "ˈan.ke", "ˈäŋ.k̟e" },
	{ "dúnque", "ˈdun.kwe", "ˈduŋ.kwe"  },
	{ "vóngola", "ˈvon.ɡo.la", "ˈvoŋ.ɡo.lä" }, 
	"s",
	{ "sguardo", "ˈzɡwar.do", "ˈzɡwär.do" },
	{ "sregolato", "zre.ɡoˈla.to", "zre.ɡoˈläː.to" },
	{ "gira-sole", "d͡ʒi.raˈso.le", "" },
	{ "desìdero", "deˈzi.de.ro", "" },
	"z, zz",
	{ "fòrza", "ˈfɔr.t͡sa", "ˈfɔr̺.t̪͡s̪ä" },
	{ "ragàzza", "raˈɡat.t͡sa", "r̺äˈɡät̪͡s̪.t̪͡s̪ä" },
	{ "mèzzo", "ˈmɛt.t͡so", "ˈmɛt̪͡s̪.t̪͡s̪o" },
	{ "mèzzo", "ˈmɛd.d͡zo", "ˈmɛd̪͡z̪.d̪͡z̪o", { 1 } },
	{ "zanzara", "d͡zanˈd͡za.ra", "d̪͡z̪änˈd̪͡z̪äː.r̺ä", true },
	-- ] says the z is geminate, while the English
	-- and Italian Wiktionaries say it isn't. It sounds geminate in the soundfile.
	{ "nazione", "natˈt͡sjo.ne" , "n̺ätˈt͡sjoː.n̺e" },
	{ "spazzi", "ˈspat.t͡si", "ˈspät.t͡si" },
	{ "zìo", "ˈt͡si.o", "" },
	{ "agenzìa", "a.d͡ʒenˈt͡si.a", "" },
	{ "grazie", "ˈɡrat.t͡sje", "" },
	{ "azienda", "adˈd͡zjen.da", "", { 1 } },
	{ "romanzière", "ro.manˈd͡zjɛ.re", "",  { 1 } },
	{ "Anzio", "ˈan.t͡sjo", ""},
	
	-- initial z followed by two vowels is usually voiced
	{ "zàino", "ˈd͡zai.no", "" },
	
	-- initial z followed by syllable beginning in voiceless consonant is usually voiceless
	{ "zampa", "ˈt͡sam.pa", "" },
	{ "zòccolo", "ˈt͡sɔk.ko.lo", "" },
	{ "zùfolo", "ˈt͡su.fo.lo", "" },
	-- exceptions
	{ "zàffiro", "ˈd͡zaf.fi.ro", "", { 1 } },
	
	-- initial z followed by syllable beginning in voiced consonant is usually voiced
	{ "zèbra", "ˈd͡zɛ.bra", "" },
	{ "zuzzurellone", "d͡zud.d͡zu.relˈlo.ne", "" },
	-- exceptions
	{ "zanna", "ˈt͡san.na", "" },
	{ "zigano", "t͡siˈɡa.no", "" },
	
	-- intervocalic "autogeminazione" ]
	{ "spazî", "ˈspat.t͡si", "ˈspät.t͡si" },
	{ "ozòno", "odˈd͡zɔ.no", "", true },
	"semivowels",
	{ "quéllo", "ˈkwel.lo", "ˈkwel.lo" },
	{ "grànchio", "ˈɡran.kjo", "ˈɡräŋ.k̟jo" },
	{ "capiènza", "kaˈpjɛn.t͡sa", "käˈpjɛn̺.t͡sä" },
	{ "niènte", "ˈnjɛn.te", "ˈnjɛn̪.t̪e" }, -- or ?
	{ "sùo", "ˈsu.o", "ˈsuː.o" },
	{ "può", "ˈpwɔ", "ˈpwɔ" },
	{ "nuòvo", "ˈnwɔ.vo", "ˈn̺wɔː.vo" },
	{ "devi-are", "de.viˈa.re", "" },
	"Handling of words without accents",
	{ "sono", "ˈso.no", "ˈsoː.n̺o" },
	{ "cane", "ˈka.ne", "ˈkäː.n̺e̞" },
	{ "carta", "ˈkar.ta", "ˈkär̺.t̪ä" },
	"Syllabification",
	{ "sottosvillupo", "sot.to.zvilˈlu.po", "sot.to.zvilˈluː.po" },
	{ "manùbrio", "maˈnu.brjo", "mäˈn̺u.brjo" } -- Is u long here?
	
	--[[
	-- For new examples:
	{ "", "", "" },
	-- respelling, phonemic, phonetic
	]]
}

function tests:test_to_phonemic()
	self:iterate(examples,
		function (self, example, expected, _, voiced_z)
			local success, result = pcall(to_phonemic, example, voiced_z)
			if not success then
				result = ("Error: %s"):format(result)
			end
			
			local voiced_text
			if not voiced_z then
				voiced_text = ""
			elseif type(voiced_z) == "boolean" then
				if voiced_z then
					voiced_text = " (all z sequences are voiced)"
				else
					voiced_text = " (no z sequences are voiced)"
				end
			else
				voiced_text = (" (z sequence%s %s %s voiced)"):format(
					voiced_z and "s" or "",
					table.concat(voiced_z, ", "),
					voiced_z and "are" or "is")
			end
			
			self:equals(example .. voiced_text, result, expected, { display = tag_IPA })
		end)
end

function tests:test_to_phonetic()
	self:iterate(examples,
		function (self, example, _, expected, voiced_z)
			if expected == "" then
				return
			end
			
			local success, result = pcall(to_phonetic, example, voiced_z)
			if not success then
				result = ("Error: %s"):format(result)
			end
			
			local voiced_text
			if not voiced_z then
				voiced_text = ""
			elseif type(voiced_z) == "boolean" then
				if voiced_z then
					voiced_text = " (all z sequences are voiced)"
				else
					voiced_text = " (no z sequences are voiced)"
				end
			else
				mw.logObject(voiced_z, "voiced_z")
				voiced_text = (" (z sequence%s %s %s voiced)"):format(
					voiced_z and "s" or "",
					table.concat(voiced_z, ", "),
					voiced_z and "are" or "is")
			end
			
			self:equals(example .. voiced_text, result, expected, { display = tag_IPA })
		end)
end

return tests