Module:qhv-pron: Difference between revisions
Jump to navigation
Jump to search
This module generates IPA pronunciation for High Valyrian words. Backend to
No edit summary |
(Thanks Juelos!) |
||
(75 intermediate revisions by the same user not shown) | |||
Line 7: | Line 7: | ||
local split = mw.text.split | local split = mw.text.split | ||
local gsplit = mw.text.gsplit | local gsplit = mw.text.gsplit | ||
local m_table = require("Module:table") | |||
local export = {} | local export = {} | ||
local palatal = "ɲʎɟj" | |||
local velar = "kɡxɣ" | local velar = "kɡxɣ" | ||
local uvular = "q" | local uvular = "q" | ||
local consonants = "[ | local consonants = "[mpbvwθntdszlrɾŋɴhɥṛ" .. palatal .. velar .. uvular .. "]" | ||
local NONSYLLABIC = u(0x032F) -- non-syllabic, combining inverted breve below | local NONSYLLABIC = u(0x032F) -- non-syllabic, combining inverted breve below | ||
local DIPHTHONG = u(0x035C) -- double articulation, combining double breve below | local DIPHTHONG = u(0x035C) -- double articulation, combining double breve below | ||
local vowels = "[ | local vowels = "[aeiouyáéíóúýàèìòùỳː" .. NONSYLLABIC .. DIPHTHONG .. "]" | ||
local rules = { | local rules = { | ||
{"rh", "ṛ"}, {"th", "θ"}, {"lj", "ʎ"}, {"ñ", "ɲ"}, {"kh", "x"}, {"[gɡ]h", "ɣ"}, {"g", "ɡ"}, | {"rh", "ṛ"}, {"mh", "m"}, {"th", "θ"}, {"lj", "ʎ"}, {"ñ", "ɲ"}, {"kh", "x"}, {"[gɡ]h", "ɣ"}, {"g", "ɡ"}, | ||
{"([ | {"([aāáà][eo])", "%1" .. NONSYLLABIC}, {"([iíuú])([aāeēoōáéóàèò])", "%1" .. DIPHTHONG .. "%2"}, | ||
{"ā", "aː"}, {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"}, {"ȳ", "yː"}, | {"ā", "aː"}, {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"}, {"ȳ", "yː"}, | ||
{"à", "áː"}, {"è", "éː"}, {"ì", "íː"}, {"ò", "óː"}, {"ù", "úː"}, {"ỳ", "ýː"}, | |||
{"n([" .. velar .. "])", "ŋ%1"}, {"n([" .. uvular .. "])", "ɴ%1"}, {"(" .. vowels .. ")r(" .. vowels .. ")", "%1ɾ%2"}, | {"n([" .. palatal .. "])", "ɲ%1"}, {"n([" .. velar .. "])", "ŋ%1"}, {"n([" .. uvular .. "])", "ɴ%1"}, | ||
{"(" .. vowels .. ")r(" .. vowels .. ")", "%1ɾ%2"}, {"ɲi", "ni"}, {"ʎi", "li"}, | |||
} | |||
local deacuter = { | |||
["á"] = "a", ["é"] = "e", ["í"] = "i", ["ó"] = "o", ["ú"] = "u", ["ý"] = "y", | |||
} | } | ||
function export.write_stress(term, from_module) | |||
local pattern = "(" .. consonants .. "*".. vowels .. "*" .. consonants .. "-)" | local pattern = "(" .. consonants .. "*".. vowels .. "*" .. consonants .. "-)" | ||
local | local weight = {} | ||
term = gsub(term, pattern, "·%1") | term = gsub(term, pattern, "·%1") | ||
term = gsub(term, "^·", "") | term = gsub(term, "^·", "") | ||
term = gsub(term, "·$", "") | |||
term = gsub(term, "·(" .. consonants .. ")·", "·%1") | term = gsub(term, "·(" .. consonants .. ")·", "·%1") | ||
term = gsub(term, "·(" .. consonants .. ")$", "%1") | term = gsub(term, "·(" .. consonants .. ")$", "%1") | ||
term = gsub(term, "·(" .. consonants .. ")(" .. consonants .. ")", "%1·%2") | term = gsub(term, "·(" .. consonants .. ")(" .. consonants .. ")", "%1·%2") | ||
term = gsub(term, "([ptkbdɡ])·([rlṛsz])", "·%1%2") | |||
term = gsub(term, "·(" .. consonants .. "?" .. consonants .. ")$", "%1") | |||
local syllables = split(term, "·") | |||
if from_module then return #syllables end -- allow other modules to know the number of syllables | |||
syllables = | if #syllables == 1 then return table.concat(syllables) end -- account for monosyllables | ||
for i, syllable in ipairs(syllables) do | for i, syllable in ipairs(syllables) do | ||
weight[i] = match(syllable, | if match(syllable, "[áéíóúý]") then -- if the user inputted manual stress, ignore all the rest | ||
end | table.insert(syllables, i, "ˈ") | ||
return table.concat(syllables) | |||
end | |||
if match(syllable, consonants .. "$") or match(syllable, "ː$") then weight[i] = "h" | |||
elseif match(syllable, NONSYLLABIC .. "$") then weight[i] = "h" | |||
else weight[i] = "l" | |||
end | |||
end | |||
local a, p | local a, p = weight[#weight-2], weight[#weight-1] | ||
if p == nil then table.insert(syllables, #syllables, "ˈ") | |||
elseif p == " | elseif a == nil then table.insert(syllables, #syllables-1, "ˈ") | ||
elseif p == "l" and a == "h" then table.insert(syllables, #syllables-2, "ˈ") | |||
else | 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 table.concat(syllables) | ||
Line 56: | Line 80: | ||
end | end | ||
return write_stress(term) | return export.write_stress(term, false) | ||
end | end | ||
function export.antique_crux(term) | function export.antique_crux(term) | ||
local oldrules = { | local oldrules = { | ||
{"v([ou])", "w%1"}, {"v([ay])", "ɥ%1"}, {"(" .. vowels .. ")v(" .. consonants .. ")", " | {"v([ou])", "w%1"}, {"v([ay])", "ɥ%1"}, {"(" .. vowels .. ")v(ˈ?" .. consonants .. ")", "%1u%2"}, | ||
{"j([iy])", "ɟ%1"}, | {"j([iy])", "ɟ%1"}, {"sˈ?ri(".. vowels .. ")", "ˈɟ%1"}, {"sr", "ɟ"}, {"sˈ?r", "ˈɟ"}, | ||
} | } | ||
Line 113: | Line 137: | ||
local ipa = "* " | local ipa = "* " | ||
ipa = ipa .. line_format(separate_word(term, false), {' | ipa = ipa .. line_format(separate_word(term, false), {'Post-Classical'}) | ||
if export.crux(term) ~= export.antique_crux(term) then | if export.crux(term) ~= export.antique_crux(term) then | ||
ipa = ipa .. "\n* " .. line_format(separate_word(term, true), {' | ipa = ipa .. "\n* " .. line_format(separate_word(term, true), {'Classical'}) | ||
end | end | ||
ipa = gsub(ipa, "ṛ", "r̥") | |||
ipa = gsub(ipa, "([áéíóú])", function(v) return deacuter[v] end) | |||
return ipa | return ipa |
Latest revision as of 18:44, 27 November 2021
- The following documentation is located at Module:qhv-pron/doc.[edit]
- Useful links: subpage list • links • transclusions • testcases • sandbox
This module generates IPA pronunciation for High Valyrian words. Backend to
{{qhv-IPA}}
.
local sub = mw.ustring.sub
local find = mw.ustring.find
local gmatch = mw.ustring.gmatch
local gsub = mw.ustring.gsub
local match = mw.ustring.match
local u = mw.ustring.char
local split = mw.text.split
local gsplit = mw.text.gsplit
local m_table = require("Module:table")
local export = {}
local palatal = "ɲʎɟj"
local velar = "kɡxɣ"
local uvular = "q"
local consonants = "[mpbvwθntdszlrɾŋɴhɥṛ" .. palatal .. velar .. uvular .. "]"
local NONSYLLABIC = u(0x032F) -- non-syllabic, combining inverted breve below
local DIPHTHONG = u(0x035C) -- double articulation, combining double breve below
local vowels = "[aeiouyáéíóúýàèìòùỳː" .. NONSYLLABIC .. DIPHTHONG .. "]"
local rules = {
{"rh", "ṛ"}, {"mh", "m"}, {"th", "θ"}, {"lj", "ʎ"}, {"ñ", "ɲ"}, {"kh", "x"}, {"[gɡ]h", "ɣ"}, {"g", "ɡ"},
{"([aāáà][eo])", "%1" .. NONSYLLABIC}, {"([iíuú])([aāeēoōáéóàèò])", "%1" .. DIPHTHONG .. "%2"},
{"ā", "aː"}, {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"}, {"ȳ", "yː"},
{"à", "áː"}, {"è", "éː"}, {"ì", "íː"}, {"ò", "óː"}, {"ù", "úː"}, {"ỳ", "ýː"},
{"n([" .. palatal .. "])", "ɲ%1"}, {"n([" .. velar .. "])", "ŋ%1"}, {"n([" .. uvular .. "])", "ɴ%1"},
{"(" .. vowels .. ")r(" .. vowels .. ")", "%1ɾ%2"}, {"ɲi", "ni"}, {"ʎi", "li"},
}
local deacuter = {
["á"] = "a", ["é"] = "e", ["í"] = "i", ["ó"] = "o", ["ú"] = "u", ["ý"] = "y",
}
function export.write_stress(term, from_module)
local pattern = "(" .. consonants .. "*".. vowels .. "*" .. consonants .. "-)"
local weight = {}
term = gsub(term, pattern, "·%1")
term = gsub(term, "^·", "")
term = gsub(term, "·$", "")
term = gsub(term, "·(" .. consonants .. ")·", "·%1")
term = gsub(term, "·(" .. consonants .. ")$", "%1")
term = gsub(term, "·(" .. consonants .. ")(" .. consonants .. ")", "%1·%2")
term = gsub(term, "([ptkbdɡ])·([rlṛsz])", "·%1%2")
term = gsub(term, "·(" .. consonants .. "?" .. consonants .. ")$", "%1")
local syllables = split(term, "·")
if from_module then return #syllables end -- allow other modules to know the number of syllables
if #syllables == 1 then return table.concat(syllables) end -- account for monosyllables
for i, syllable in ipairs(syllables) do
if match(syllable, "[áéíóúý]") then -- if the user inputted manual stress, ignore all the rest
table.insert(syllables, i, "ˈ")
return table.concat(syllables)
end
if match(syllable, consonants .. "$") or match(syllable, "ː$") then weight[i] = "h"
elseif match(syllable, NONSYLLABIC .. "$") then weight[i] = "h"
else weight[i] = "l"
end
end
local a, p = weight[#weight-2], weight[#weight-1]
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)
end
function export.crux(term)
term = mw.ustring.lower(term)
for _, rule in ipairs(rules) do
term = gsub(term, rule[1], rule[2])
end
return export.write_stress(term, false)
end
function export.antique_crux(term)
local oldrules = {
{"v([ou])", "w%1"}, {"v([ay])", "ɥ%1"}, {"(" .. vowels .. ")v(ˈ?" .. consonants .. ")", "%1u%2"},
{"j([iy])", "ɟ%1"}, {"sˈ?ri(".. vowels .. ")", "ˈɟ%1"}, {"sr", "ɟ"}, {"sˈ?r", "ˈɟ"},
}
term = export.crux(term)
for _, oldrule in ipairs(oldrules) do
term = gsub(term, oldrule[1], oldrule[2])
end
return term
end
local function IPA_span(items)
local bits = {}
for _, item in ipairs(items) do
local bit = "<span style=\"font-size:110%;font-family:Gentium,'DejaVu Sans','Segoe UI',sans-serif>" .. item.pron .. "</span>"
table.insert(bits, bit)
end
return table.concat(bits)
end
local function format_IPA(items)
return "[[w:IPA chart|IPA]]<sup>([[IPA for High Valyrian|key]])</sup>: " .. IPA_span(items)
end
function line_format(pronunciation, register)
local full_pronunciations = {}
local IPA_args = {{pron = '[' .. pronunciation .. ']'}}
table.insert(full_pronunciations, format_IPA(IPA_args))
return "(''" .. table.concat(register, ", ") .. "'')" .. ' ' .. table.concat(full_pronunciations, ' or ')
end
function separate_word(term, b)
local result = {}
for word in gsplit(term, " ") do
if b then table.insert(result, export.antique_crux(word))
else table.insert(result, export.crux(word)) end
end
return table.concat(result, " ")
end
function export.show(frame)
local params = {
[1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "drakarys" or mw.title.getCurrentTitle().text },
}
local args = require("Module:parameters").process(frame:getParent().args, params)
local term = args[1]
local ipa = "* "
ipa = ipa .. line_format(separate_word(term, false), {'Post-Classical'})
if export.crux(term) ~= export.antique_crux(term) then
ipa = ipa .. "\n* " .. line_format(separate_word(term, true), {'Classical'})
end
ipa = gsub(ipa, "ṛ", "r̥")
ipa = gsub(ipa, "([áéíóú])", function(v) return deacuter[v] end)
return ipa
end
return export