Module:qay-pron: Difference between revisions

← Older edit Newer edit →

@@ Line 14: / Line 14: @@
 local export = {}
-local consonants = "[mpbvwθntdszlrɾŋɴhɥṛ]"
+local consonants = "[pbmvstdnrɾljkɡŋhxçʤʧ]"
-local NONSYLLABIC = u(0x032F) -- non-syllabic, combining inverted breve below
-local DIPHTHONG = u(0x035C) -- double articulation, combining double breve below
 local front = "ie"
 local back = "ou"
-local vowels = "[a" .. front .. back .. "]"
+local vowels = "[a" .. front .. back .. "ː]"
--- ʤʧ
 local function open_to_closed(v)
@@ Line 32: / Line 29: @@
 	return table.concat(otc)
 end
+local function same(foo, bar)
+	foo, bar = mw.ustring.toNFD(foo), mw.ustring.toNFD(bar) -- decompose diacritics
+	foo, bar = match(foo, "^."), match(bar, "^.") -- sort out the letter
+	return foo == bar and true or false
+end
+local shared_rules = {
+	{"n[kg]", {["nk"] = "ŋk", ["ng"] = "ŋ"}}, {"c", "ʧ"}, {"j", "ʤ"}, {"y", "j"}, {"g", "ɡ"},
+	-- Long vowels
+	{"ā", "aː"},  {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"},
+}
 local phonemic_rules = {
-	{"ng", "ŋ"}, {"y", "j"}, {"g", "ɡ"}, {"([tkdɡ])j", "%1ʲ"}, {"ā", "aː"},  {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"},
+	{"([tkdɡ])j", "%1ʲ"}, {"(" .. consonants .. consonants .. ")", function(c1, c2) return same(c1,c2) and c1 or c1 .. c2 end},
 }
 local phonetic_rules = {
-	{"ng", "ŋ"}, {"y", "j"}, {"g", "ɡ"}, {"nk", "ŋk"}, {"([^nŋ])[tk]j", "%1ʧ"}, {"([^nŋ])[dɡ]j", "%1ʤ"}, {"r", "ɾ"}, {"h([" .. front .. "])", "ç%1"},
+	{"([^nŋ])[tk]j", "%1ʧ"}, {"([^nŋ])[dɡ]j", "%1ʤ"}, {"r", "ɾ"},
-	{"h([" .. back .. "])", "x%1"},	{"ā", "aː"},  {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"}, {"a", "ä"},
+	{"h([" .. front .. "])", "ç%1"}, {"h([" .. back .. "])", "x%1"},
+	--{"(" consonants .. vowel)", ""},
 }
-local function write_stress(term)
+local function syllabify(term, pos)
-	--[=[local pattern = "(" .. consonants .. "*".. vowels .. "*" .. consonants .. "-)"
+	term = gsub(term, "(" .. consonants .. "*)(" .. vowels .. "*)", "%1%2·")
-	local weight = {}
+	term = gsub(term, "··", "·"); term = gsub(term, "·$", "")
-	term = gsub(term, pattern, "·%1")
+	term = gsub(term, "·(" .. consonants .. ")(" .. consonants .. ")(" .. vowels .. "*)", "%1·%2%3")
-	term = gsub(term, "^·", "")
-	term = gsub(term, "·$", "")
-	term = gsub(term, "·(" .. consonants .. ")·", "·%1")
 	term = gsub(term, "·(" .. consonants .. ")$", "%1")
-	term = gsub(term, "·(" .. consonants .. ")(" .. consonants .. ")", "%1·%2")
+	term = gsub(term, "·(" .. consonants .. ")·", "%1·")
-	term = gsub(term, "([ptkbdɡ])·([rlṛsz])", "·%1%2")
-	term = gsub(term, "·(" .. consonants .. "?" .. consonants .. ")$", "%1")
-	local syllables = split(term, "·")
+	local syll = split(term, "·"); local noa = {}
-	if from_module then return #syllables end -- allow other modules to know the number of syllables
+	local monosyll = {["n"] = "ˈ", ["pron"] = "", ["particle"] = "(ˈ)", ["prep"] = "(ˈ)", ["conj"] = "(ˈ)"}
-	if #syllables == 1 then return table.concat(syllables) end -- account for monosyllables
+	if #syll ~= 1 then
-	for i, syllable in ipairs(syllables) do
+		if match(term, "[áéíóúĄ]") then
-		if match(syllable, "[áéíóúý]") then -- if the user inputted manual stress, ignore all the rest
+			for _, s in ipairs(syll) do
-			table.insert(syllables, i, "ˈ")
+				s = remove_acute(s, match(s, "[áéíóúĄ]") and true or false)
-			return table.concat(syllables)
+				table.insert(noa, s)
-		end
+			end
-		if match(syllable, consonants .. "$") or match(syllable, "ː$") then weight[i] = "h"
+		else
-		elseif match(syllable, NONSYLLABIC .. "$") then weight[i] = "h"
+			syll[#syll - 1] = "ˈ" .. syll[#syll - 1]
-		else weight[i] = "l"
 		end
 	end
-	local a, p = weight[#weight-2], weight[#weight-1]
+	return table.concat(#noa > 1 and noa or syll, "·")
-	if p == nil then table.insert(syllables, #syllables, "ˈ")
-	elseif a == nil then table.insert(syllables, #syllables-1, "ˈ")
-	elseif p == "l" and a == "h" then table.insert(syllables, #syllables-2, "ˈ")
-	elseif (p == "l" and a == "l") or (p == "h") then table.insert(syllables, #syllables-1, "ˈ")
-	else table.insert(syllables, #syllables-1, "ˈ") end
-	return table.concat(syllables)]=]
-	return term
 end
 function export.phonemic(term)
 	term = mw.ustring.lower(term)
+	for _, rule in ipairs(shared_rules) do
+		term = gsub(term, rule[1], rule[2])
+	end
 	for _, micrule in ipairs(phonemic_rules) do
@@ Line 93: / Line 97: @@
 function export.phonetic(term)
 	term = mw.ustring.lower(term)
+	for _, rule in ipairs(shared_rules) do
+		term = gsub(term, rule[1], rule[2])
+	end
 	for _, ticrule in ipairs(phonetic_rules) do
 		term = gsub(term, ticrule[1], ticrule[2])
 	end
+	term = gsub(term, "a", "ä")
 	return write_stress(term)