Module:la-IPA/testcases

local tests = require "Module:UnitTests"

local unpack = unpack or table.unpack -- Lua 5.2 compatibility

local to_IPA = require "Module:la-pronunc".convert_words

local Latin = require "Module:languages".getByCode "la"
local function link(word)
	return require "Module:links".full_link { term = word, lang = Latin }
end

local function format_IPA(phonemic, phonetic)
	return ('<span class="IPA">/%s/</span> <span class="IPA"></span>')
		:format(tostring(phonemic), tostring(phonetic))
end

function tests:test_all()
	local examples = {
		-- spelling, Classical phonemic and phonetic, Ecclesiastical phonemic and phonetic
		{ "ecclēsia", { "ekˈkleː.si.a", "ɛkˈkɫeː.si.a" }, { "ekˈkle.si.a", "ekˈklɛː.s̬i.a" } },
		{ "magnus", { "ˈmaɡ.nus", "ˈmaŋ.nʊs" }, { "ˈmaɲ.ɲus", "ˈmaɲ.ɲus" } },
		{ "coepī", { "ˈkoe̯.piː", "ˈkoe̯.piː" }, { "ˈt͡ʃe.pi", "ˈt͡ʃɛː.pi" } },
		{ "Caesar", { "ˈkae̯.sar", "ˈkae̯.sar" }, { "ˈt͡ʃe.sar", "ˈt͡ʃɛː.s̬ar" } },
		{ "adhāmō", { "aˈdaː.moː", "aˈdaː.moː" }, { "aˈda.mo", "aˈdaːmo" } },
        --Not implemented any more 
		--"rhotic allophones",
		--{ "rēs", { "ˈreːs", "ˈreːs" }, {"ˈres", "ˈrɛs"} },
		--{ "ferre", { "ˈfer.re", "ˈfɛr.rɛ" }, {"ˈfer.re", "ˈfɛr.re"} },
	    --{ "corporis", { "ˈkor.po.ris", "ˈkɔr.pɔ.rɪs" }, {"ˈkor.po.ris", "ˈkɔr.po.ris"} },
	    --{ "inrīdeō", { "inˈriː.de.oː", "ɪnˈriː.de.oː" }, {"inˈri.de.o", "inˈriː.de.o"} },
	    --{ "abrumpō", { "abˈrum.poː", "abˈrʊm.poː" }, false, },
        --Not implemented yet: phrases
        --{ "bona rēs", { "ˈbo.na ˈreːs", "ˈbɔ.naˈreːs" }, false, },
        --{ "ācer Amor", { "ˈaː.keˈr‿a.mɔr", "ˈaː.kɛˈra.mɔɾ" }, false, },
		"lateral allophones",
        { "similis", { "ˈsi.mi.lis", "ˈsɪ.mɪ.lɪs" }, false, },
        { "simile", { "ˈsi.mi.le", "ˈsɪ.mɪ.ɫɛ" }, false, },
        { "simul", { "ˈsi.mul", "ˈsɪ.mʊɫ" }, false, },
        { "alga", { "ˈal.ɡa", "ˈaɫ.ɡa" }, false, },
        { "nōlle", { "ˈnoːl.le", "ˈnoːl.lɛ" }, false, },
        --Not implemented yet: phrases
        --{ "simul ipse", { "ˈsi.muˈl‿ip.se", "ˈsɪ.mʊˈlɪp.sɛ" }, false, },
        --{ "difficilī argūmentō", { "difˈfi.ki.l‿ar.guːˈmen.toː", "dɪfˈfɪ.kɪ.ɫar.guːˈmɛn.toː" }, false, },
		"semivowels between vowels",
		{ "majus", { "ˈmaj.jus", "ˈmaj.jʊs" }, {"ˈma.jus", "ˈmaː.jus"} },
		{ "peius", { "ˈpej.jus", "ˈpɛj.jʊs" }, {"ˈpe.jus", "ˈpɛː.jus"} },
		{ "cuius", { "ˈkuj.jus", "ˈkʊj.jʊs" }, {"ˈku.jus", "ˈkuː.jus"} },
		{ "trāiectus", { "traːˈjek.tus", "traːˈjɛk.tʊs" }, {"traˈjek.tus", "traˈjɛk.tus"} },
		{ "dēiectus", { "deːˈjek.tus", "deːˈjɛk.tʊs" }, {"deˈjek.tus", "deˈjɛk.tus"} },
        { "prōiectus", { "proːˈjek.tus", "proːˈjɛk.tʊs" }, {"proˈjek.tus", "proˈjɛk.tus"} },
		{ "cavus", { "ˈka.wus", "ˈka.wʊs" } },
		{ "cauus", { "ˈka.wus", "ˈka.wʊs" } },
        --Note: Italian words such as Baia and Febbraio have non-doubled /j/, and presumably Ecclesiastical pronunciation of majus etc. is similar.
		"diphthongs before vowels and word-finally",
        --Note: Heavy diphthongs before vowels can alternatively be analyzed as sequences of short vowel followed by -ww- or -jj-
		{ "iūdaeus", { "juːˈdae̯.us", "juːˈdae̯.ʊs" }, { "juˈde.us", "juˈdɛː.us" } },
     	{ "Meliboea", { "me.liˈboe̯.a", "mɛ.lɪˈboe̯.a" }, { "me.liˈbe.a", "me.liˈbɛː.a" } },
	    { "Agau.ē", { "aˈɡau̯.eː", "aˈɡau̯.eː" }, false },
		{ "catasceu.a", { "ka.tasˈkeu̯.a", "ka.tasˈkɛu̯.a" }, false },
        { "nauta", { "ˈnau̯.ta", "ˈnau̯.ta" }, { "ˈnau̯.ta", "ˈnau̯.ta" } },
        { "auctus", { "ˈau̯k.tus", "ˈau̯k.tʊs" }, { "ˈau̯k.tus", "ˈau̯k.tus" } },
		{ "neu", { "ˈneu̯", "ˈnɛu̯" }, { "ˈneu̯", "ˈnɛu̯" }},
		{ "hau", { "ˈhau̯", "ˈhau̯" }, { "ˈau̯", "ˈau̯" }},
		{ "he_i", { "ˈhei̯", "ˈhɛi̯" }, { "ˈei̯", "ˈɛi̯" }},
		{ "cu_i", { "ˈkui̯", "ˈkʊi̯" }, { "ˈkui̯", "ˈkui̯" }},
		"diphthongs before consonants",
	    { "cu_idam", { "ˈkui̯.dam", "ˈkʊi̯.dãː" }, { "kui̯.dam", "kui̯.dam" }},
		{ "de_inde", { "ˈdei̯n.de", "ˈdɛi̯n.dɛ" }, { "ˈdei̯n.de", "ˈdɛi̯n.de" }},
		"palatalization of c, g, sc",
		{ "gerō", nil, { "ˈd͡ʒe.ro", "ˈd͡ʒɛː.ro" } },
		{ "certus", nil, { "ˈt͡ʃer.tus", "ˈt͡ʃɛr.tus" } },
		{ "faciō", nil, { "ˈfa.t͡ʃi.o", "ˈfaː.t͡ʃi.o" } },
		{ "caelus", nil, { "ˈt͡ʃe.lus", "ˈt͡ʃɛː.lus" } },
		{ "scelus", nil, { "ˈʃe.lus", "ˈʃɛː.lus" } },
        { "suggillo", nil, { "sudˈd͡ʒil.lo", "sudˈd͡ʒil.lo" } },
		-- palatalization of sc (as ʃ)
		{ "conscientia", nil, { "kon.ʃiˈen.t͡si.a", "kon̠ʲʃiˈɛn.t͡si.a"  } },
		{ "nescio", nil, { "ˈneʃ.ʃi.o", "ˈnɛʃ.ʃi.o" } },
		{ "masclus", nil, { "ˈmas.klus", "ˈmas.klus" } },
        { "perscisco", nil, { "perˈʃis.ko", "perˈʃis.ko" } },
        { "Volsci", nil, { "ˈvol.ʃi", "ˈvɔl̠ʲ.ʃi" } },
        { "abscindo", nil, { "abˈʃin.do", "abˈʃin.do" } },  
        --edge cases spelled with K that shouldn't display as palatalized
        { "berkelium", nil, { "berˈke.li.um", "berˈkɛː.li.um" } },
        { "alaskensis", nil, { "a.lasˈken.sis", "a.lasˈkɛn.sis" } },
		"palatalization of t",
		-- Unlike Italian, Ecclesiastical Latin is not described as regularly doubling the t͡s sound resulting from affrication of t before i.
		{ "nātiō", nil, { "ˈnat.t͡si.o", "ˈnat.t͡si.o" } },
		{ "gluttiō", nil, { "ˈɡlut.ti.o", "ˈɡlut.ti.o" } },
		{ "Christiānus", nil, { "kris.tiˈa.nus", "kris.tiˈaː.nus" } },
		{ "satietas", nil, { "satˈt͡si.e.tas", "satˈt͡siːɛ.tas" } }, 
        -- Some sources suggest that affrication occurs only in unstressed syllables (where /i/ could be reduced to /j/).
        --But I think that while this may describe the original environment of the sound change, it's not a condition for the conventional pronunciation.
        "narrow transcription of coronal plosives",
        { "apud", nil, { "ˈa.pud", "ˈaː.pud" } },
        { "dormiunt", nil, { "ˈdor.mi.unt", "ˈdɔr.mi.unt" } },
		"syllabification",
		{ "subigō", { "ˈsu.bi.ɡoː", "ˈsʊ.bɪ.ɡoː" } },
		{ "obsolētus", { "ob.soˈleː.tus", "ɔp.sɔˈɫeː.tʊs" } },
		{ "ȳpsīlon", { "yːpˈsiː.lon", "yːpˈsiː.ɫɔn" } },
		{ "psychē", { "ˈpsy.kʰeː", "ˈpsy.kʰeː" } },
		{ "ū̆ncia", { "ˈuːn.ki.a", "ˈuːŋ.ki.a" } },
		"hiatus",
		--{ "deeō", { "ˈde.e.oː", "ˈd̪eeoː" } }, --leaving this out as it's unclear what the intended behavior is.
	    { "vīnea", { "ˈwiː.ne.a", "ˈwiː.ne.a" }, { "ˈvi.ne.a", "ˈviː.ne.a" } },
	    { "via", { "ˈwi.a", "ˈwi.a" }, { "ˈvi.a", "ˈviː.a" } },
        { "Dryas", { "ˈdry.as", "ˈdry.as" } },
        { "duās", { "ˈdu.aːs", "ˈdu.aːs" } },   -- Allen doesn't mention this. Some sources say u has a w-like glide in hiatus.
	    { "svāvium", { "ˈsʷaː.wi.um", "ˈsʷaː.wi.ũː" }, { "ˈsʷa.vi.um", "ˈsʷaː.vi.um" } },
	    { "quantus.vīs", { "kʷanˈtus.wiːs", "kʷanˈtʊs.wiːs" }, { "kʷanˈtus.vis", "kʷanˈtuz.vis" } },
        "voicing, place assimilation, and elision in Classical Latin heterosyllabic consonant clusters",
        { "adsum", { "ˈad.sum", "ˈas.sũː" } },
        { "etsī", { "ˈet.siː", "ˈɛt.siː" } }, --no sources describe elision in the rare /t.s/ words
        { "adpetō", { "ˈad.pe.toː", "ˈap.pɛ.toː" } },
        { "quad.rāns", { "ˈkʷad.rans", "ˈkʷad.rãːs" } },  --assimilation doesn't occur in morpheme-internal heterosyllabic /d.r/
        { "adripiō", { "adˈri.pi.oː", "adˈrɪ.pi.oː" } },  --therefore, if we indicate assimilation in ad- prefixed words, it needs to be special-coded
        { "postquam", { "ˈpost.kʷam", "ˈpɔs(t).kʷãː" } }, --optional elision of coda /t/ in /st/ before a heterosyllabic following consonant
        { "stringō", { "ˈstrin.ɡoː", "ˈstrɪŋ.ɡoː" } }, --but /t/ in an onset cluster should never elide
        --Not implemented yet: "inter-word resyllabifications and assimilations for Classical Latin phrases",
        --{ "ad fontēs", { "at ˈfon.teːs", "atˈfɔn.teːs" } },
        --{ "ad astra", { "aˈd‿as.tra", "aˈdas.tra" } },
        --{ "eādem mūtāta resurgō", { "eˈaː.dem muːˈtaː.ta reˈsur.ɡoː", "eˈaːdɛm.muːˈtaː.ta.rɛˈsʊr.ɡoː" } },
        --{ "ab ōvō ūsque ad māla", { "aˈb‿oː.ˈw‿uːs.kʷ‿ad ˈmaː.la", "aˈboːˈwuːs.kʷadˈmaː.ɫa" } },
	}
	
	-- local options = { display = format_IPA }
	self:iterate(examples, function (self, word, Classical, Ecclesiastical)
		if Classical then
			self:equals(
				link(word) .. " (Classical)",
				format_IPA(to_IPA(word, false, false, false), to_IPA(word, true, false, false)),
				format_IPA(unpack(Classical)))
		end
		
		if Ecclesiastical then
			self:equals(
				link(word) .. " (Ecclesiastical)",
				format_IPA(to_IPA(word, false, true, false), to_IPA(word, true, true, false)),
				format_IPA(unpack(Ecclesiastical)))
		end
	end)
end

return tests