initial commit

1 files changed, 171 insertions, 0 deletions

diff --git a/language.lua b/language.lua
new file mode 100644
index 0000000..cf6725f
--- /dev/null
+++ b/language.lua
@@ -0,0 +1,171 @@
+local language = {}
+setmetatable(language, {__index=_G})
+setfenv(1, language)
+
+
+-- local utility functions
+
+-- convert string to array of characters
+local function string_to_array(str)
+	local arr = {}
+	for i=1,#str do
+		table.insert(arr, string.sub(str, i, i))
+	end
+	return arr
+end
+
+
+-- iterate over lines in string
+local function lines(str)
+	local clean = function(s) return string.gsub(s, "\n", "") end
+	return coroutine.wrap(function()
+		for line in string.gmatch(str, "[^\n]*\n") do
+			coroutine.yield(clean(line))
+		end
+		local lastline = string.match(str, "\n?[^\n]+$")
+		if lastline then
+			coroutine.yield(clean(lastline))
+		end
+	end)
+end
+
+
+-- parse phonemes from string
+local function parse_phonemes(str)
+	local phonemes = {}
+	for line in lines(str) do
+		if string.match(line, "^%s*#") then
+			-- this is a comment, ignore
+		elseif string.match(line, "^%s*$") then
+			-- ignore blank lines
+		else
+			class, phones = string.match(line, "(.)=(.*)")
+			-- remove whitespace from phones
+			phones = string.gsub(phones, "%s", "")
+			phonemes[class] = string_to_array(phones)
+		end
+	end
+	return phonemes
+end
+
+
+-- parse syllables from string
+local function parse_syllables(str)
+	local syllables = {}
+	for line in lines(str) do
+		line = string.gsub(line, "%s", "")
+		table.insert(syllables, string_to_array(line))
+	end
+	return syllables
+end
+
+
+-- parse orthography patterns from string
+local function parse_orthography(str, phonemes)
+	local rules = {}
+	for line in lines(str) do
+		if string.match(str, '^%s*$') then
+			-- ignore blank lines
+		else
+			-- trim whitespace
+			line = string.gsub(line, "%s*$", "")
+			local pattern, replace = string.match(line, "(.*)=(.*)")
+			pattern = string.gsub(pattern, "%%(.)", function(class)
+				if phonemes[class] == nil then
+					return '%' .. class
+				end
+				return '[' .. table.concat(phonemes[class], '') .. ']'
+			end)
+			table.insert(rules, { pattern=pattern, replace=replace })
+		end
+	end
+	return rules
+end
+
+
+-- randomly choose an element from an array with a given distribution
+local function random_choice(tbl, distribution)
+	local x = distribution(math.random())
+	local bin = math.floor(x * #tbl) + 1
+	return tbl[bin]
+end
+
+
+-- probability distributions
+
+-- uniform (flat) distribution
+function uniform(x)
+	return x
+end
+
+
+function quadratic(x)
+	return x*x
+end
+
+
+
+
+
+
+Language = {}
+
+function Language.new(_, phonemes, syllables, orthography, len_min, len_max, p_dist, s_dist, l_dist)
+	print(len_min, len_max)
+	local self = {}
+	setmetatable(self, {__index=Language})
+	self.phonemes = parse_phonemes(phonemes)
+	self.p_dist = p_dist or uniform
+	self.syllables = parse_syllables(syllables)
+	self.s_dist = s_dist or uniform
+	self.orthography = parse_orthography(orthography, self.phonemes)
+	self.length = {}
+	for len=len_min,len_max do
+		table.insert(self.length, len)
+	end
+	self.l_dist = l_dist or uniform
+	return self
+end
+setmetatable(Language, {__call=Language.new})
+
+
+function Language.random_phone(self, class)
+	return random_choice(self.phonemes[class], self.p_dist)
+end
+
+
+function Language.syllable(self)
+	-- pick a random syllable type
+	local syl = random_choice(self.syllables, self.s_dist)
+	local result = ''
+	for _, ch in ipairs(syl) do
+		if self.phonemes[ch] then
+			result = result .. self:random_phone(ch)
+		else
+			result = result .. ch
+		end
+	end
+	return result
+end
+
+
+function Language.word(self)
+	local len = random_choice(self.length, self.l_dist)
+	local word = ''
+	for i=1,len do
+		word = word .. self:syllable()
+	end
+	return word
+end
+
+
+function Language.romanize(self, text)
+	local result = text
+	for _, rule in ipairs(self.orthography) do
+		result = string.gsub(result, rule.pattern, rule.replace)
+	end
+	return result 
+end
+
+
+return language