local export = {}
local lang = require("Module:languages").getByCode( "mn" )
local links = require( "Module:links" )
table = require( "Module:table" ).reverse
table = require( "Module:table" ).removeDuplicates
local char = mw.ustring.char
local find = mw.ustring.find
local len = mw.ustring.len
local match = mw.ustring.match
local gmatch = mw.ustring.gmatch
local sub = mw.ustring.sub
local gsub = mw.ustring.gsub
local lower = mw.ustring.lower
local reverse = require( "Module:string" ).reverse
export.FVS1 = char( 0x180B )
export.FVS2 = char( 0x180C )
export.FVS3 = char( 0x180D )
export.FVS4 = char( 0x180F )
export.MVS = char( 0x180E )
export.NNBSP = char( 0x202F )
export.stem_barrier = char( 0xF000 )
local function format_Mongolian_text(text) return "<span class=\"Mong\" lang=\"mn\">" .. text .. "</span>" end
function export.see(frame)
local params = {
= {},
}
local args = require("Module:parameters").process(frame:getParent().args, params)
local title = args
local curr_title = mw.title.getCurrentTitle().subpageText
local content = mw.title.new(title):getContent()
local senses = {}
local sense_id = 0
local j, pos, s, section
if title == curr_title then
return error("The soft-directed item is the same as the page title.")
end
if content then
if not match(content, "==Mongolian==") then
categories = categories .. "]"
elseif not match(content, "mn%-IPA") and not match(content, "mn%-see") then
require("Module:debug").track("mn-see/unidirectional reference to variant")
elseif not match(content, curr_title) then
require("Module:debug").track("mn-see/unidirectional reference variant→orthodox")
end
end
while true do
_, j, language_name, s = content:find("%f==%s*(+)%s*==(\n.-)\n==%f", pos)
if j == nil then
i, j, language_name, s = content:find("%f==%s*(+)%s*==(\n.+)", pos)
end
if j == nil then
break
else
pos = j - 1
end
if language_name == "Mongolian" then
section = s
end
end
if not section then
return ""
end
section = section:gsub("\n===+Etymology.-(\n==)", "%1")
local text = {}
for sense in section:gmatch("\n# (+)") do
if not sense:match("rfdef") and not sense:match("defn") then
sense_id = sense_id + 1
table.insert(senses, sense)
end
end
table.insert(text, table.concat(senses, "\n# "))
table.insert(text, "</div>")
return frame:preprocess( table.concat(text) )
end
function export.ipasee(frame)
local params = {
= {},
}
local args = require("Module:parameters").process(frame:getParent().args, params)
local title = args
local curr_title = mw.title.getCurrentTitle().subpageText
local content = mw.title.new(title):getContent()
local senses = {}
local sense_id = 0
local j, pos, s, section
if title == curr_title then
return error("The soft-directed item is the same as the page title.")
end
if content then
if not match(content, "==Mongolian==") then
categories = categories .. "]"
elseif not match(content, "mn%-IPA") and not match(content, "mn%-see") and not match(content, "mn%-IPA-see") then
require("Module:debug").track("mn-IPA-see/unidirectional reference to variant")
elseif not match(content, curr_title) then
require("Module:debug").track("mn-IPA-see/unidirectional reference variant→orthodox")
end
end
while true do
_, j, language_name, s = content:find("%f==%s*(+)%s*==(\n.-)\n==%f", pos)
if j == nil then
i, j, language_name, s = content:find("%f==%s*(+)%s*==(\n.+)", pos)
end
if j == nil then
break
else
pos = j - 1
end
if language_name == "Mongolian" then
section = s
end
end
if not section then
return ""
end
section = section:gsub("\n===+Etymology.-(\n==)", "%1")
local text = {}
if section:match("{{mn-IPA\|(+)}}") then
return frame:expandTemplate{ title = "mn-IPA", args = { section:match("{{mn-IPA\|(+)}}") } }
else
return frame:expandTemplate{ title = "mn-IPA", args = { title } }
end
end
--Breaks down a string into vowel harmonic segments
function export.vowelharmony( text, params )
if not params then params = {} end
local vh = {}
local breaks = { 1 }
local switchers = { "ау", "оу", "уу", "иу", "яу", "ёу", "еу", "юу", "уй", "эү", "өү", "үү", "иү", "еү", "юү", "үй", "%s" }
local switchers2 = { "аа", "оо", "өө", "ээ", "яа", "ёо", "еө", "еэ", "Аа", "Оо", "Өө", "Ээ", "Яа", "Ёо", "Еө", "Еэ" }
for _, v in pairs( switchers ) do
v = gsub( v, "(.)(.)", "%1" .. char( 0x301 ) .. "?" .. char( 0x300 ) .. "?%2" )
local c = 0
while c ~= nil do
c = find( lower( text ), v, c + 1 )
if c ~= nil and c ~= 1 then
table.insert( breaks, c )
end
end
end
if params.bor then
for _,v in pairs( switchers2 ) do
if match( text, v .. "$" ) then
table.insert( breaks, len( text ) - 1 )
end
end
end
table.sort( breaks )
for i, b in ipairs( breaks ) do
table.insert( vh, { Cyrl = {}, Mong = {} } )
if i == #breaks then
vh.substring = sub( text, b, len( text ) )
else
vh.substring = sub( text, b, breaks - 1 )
end
end
if params.bor then
for i,s in ipairs( vh ) do
vh = { Cyrl = {}, Mong = {} }
local orig_text = s.substring
s.substring = lower( s.substring )
if params.bor == "Russian" then s.substring = gsub( s.substring, "у", "ү" ) end
local substring_nostress = gsub( s.substring, "", "" )
if match( s.substring, "кило" .. char( 0x301 ) .. "?$" ) then -- irregular
vh.Cyrl.a = "э"
vh.location = find( s.substring, "" )
vh.position = "front"
vh.quality = "unrounded"
elseif match( substring_nostress, "у" ) or match( substring_nostress, "уй" ) then
vh.Cyrl.a = "а"
vh.location = 1
vh.position = "back"
vh.quality = "unrounded"
elseif match( substring_nostress, "ү" ) or match( substring_nostress, "үй" ) then
vh.Cyrl.a = "э"
vh.location = 1
vh.position = "front"
vh.quality = "unrounded"
elseif match( s.substring, "" .. char( 0x301 ) ) then
vh.Cyrl.a = "а"
vh.location = find( s.substring, char( 0x301 ) ) - 1
vh.position = "back"
vh.quality = "unrounded"
elseif match( s.substring, "" .. char( 0x301 ) ) then
vh.Cyrl.a = "о"
vh.location = find( s.substring, char( 0x301 ) ) - 1
vh.position = "back"
vh.quality = "rounded"
elseif ( match( substring_nostress, "" ) and ( find( substring_nostress, "" ) == find( substring_nostress, "" ) ) ) or match( substring_nostress, "" ) then
vh.Cyrl.a = "а"
vh.location = find( substring_nostress, "" )
vh.position = "back"
vh.quality = "unrounded"
elseif match( substring_nostress, "" ) and find( substring_nostress, "" ) == find( substring_nostress, "" ) then
vh.Cyrl.a = "о"
vh.location = find( substring_nostress, "" )
vh.position = "back"
vh.quality = "rounded"
elseif ( match( substring_nostress, "" ) and find( substring_nostress, "" ) == find( substring_nostress, "" ) ) or match( substring_nostress, "" ) then
vh.Cyrl.a = "э"
vh.location = find( substring_nostress, "" )
vh.position = "front"
vh.quality = "unrounded"
elseif match( substring_nostress, "" ) and find( substring_nostress, "" ) == find( substring_nostress, "ө" ) then
vh.Cyrl.a = "ө"
vh.location = find( substring_nostress, "ө" )
vh.position = "front"
vh.quality = "rounded"
else
vh.Cyrl.a = "э"
vh.location = find( substring_nostress, "и" ) or 1
vh.position = "front"
vh.quality = "unrounded"
end
if match( vh.Cyrl.a, "" ) then
vh.Cyrl.ii = "ы"
vh.Cyrl.u = "у"
vh.Mong.a = "ᠠ"
vh.Mong.u = "ᠤ"
else
vh.Cyrl.ii = "ий"
vh.Cyrl.u = "ү"
vh.Mong.a = "ᠡ"
vh.Mong.u = "ᠦ"
end
if match( vh.Cyrl.a, "ө" ) then -- ө takes the diphthong эй not өй
vh.Cyrl.ai = "эй"
else
vh.Cyrl.ai = vh.Cyrl.a .. "й"
end
vh.Cyrl.aa = vh.Cyrl.a .. vh.Cyrl.a
vh.Cyrl.uu = vh.Cyrl.u .. vh.Cyrl.u
vh.substring = orig_text
end
else
local location
local pattern
for i,s in ipairs( vh ) do
local orig = s.substring
s.substring = reverse( lower( s.substring ) )
local vowel = match( s.substring, "" )
location,_ = find( s.substring, "" )
if vh.Cyrl.a == nil then
vh.Cyrl.a = "э"
vh.position = "front"
vh.quality = "unrounded"
pattern = "и"
end
vh.Cyrl.a = lower( vh.Cyrl.a )
if vowel == "а" or vowel == "у" or vowel == "я" then
vh.Cyrl.a = "а"
vh.position = "back"
vh.quality = "unrounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
elseif vowel == "о" or vowel == "ё" then
vh.Cyrl.a = "о"
vh.position = "back"
vh.quality = "rounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
elseif vowel == "э" then
vh.position = "front"
if location and sub( s.substring, location-1, location-1 ) == "й" and match( s.substring, "", location+1 ) == "ө" then
vh.Cyrl.a = "ө"
vh.quality = "rounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
else
vh.Cyrl.a = "э"
vh.quality = "unrounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
end
elseif vowel == "ү" then
vh.Cyrl.a = "э"
vh.position = "front"
vh.quality = "unrounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
elseif vowel == "ө" then
vh.Cyrl.a = "ө"
vh.position = "front"
vh.quality = "rounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
elseif vowel == "е" then
vh.position = "front"
if match( s.substring, "ө", location+1 ) then
vh.Cyrl.a = "ө"
vh.quality = "rounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
else
vh.Cyrl.a = "э"
vh.quality = "unrounded"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
end
elseif vowel == "ю" then
vh.quality = "unrounded"
if match( s.substring, "", location+1 ) then
vh.Cyrl.a = "а"
vh.position = "back"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
else
vh.Cyrl.a = "э"
vh.position = "front"
pattern = ""
if match( s.substring, "" ) then vh.violation = true else vh.violation = false end
end
end
location = 0
local function prev_vowel( n ) return match( s.substring, "", n+1 ) end
local function prev_hvowel( n ) return match( s.substring, pattern, n+1 ) end
while prev_vowel( location ) and ( prev_vowel( location ) == prev_hvowel( location ) or prev_vowel( location ) == "и" ) do
if prev_vowel( location ) == prev_hvowel( location ) then
location = find( s.substring, pattern, location+1 )
else -- if и
local icheck = location+1
while prev_vowel( icheck ) == "и" do
icheck = find( s.substring, "и", icheck+1 )
end
if prev_vowel( icheck ) and prev_vowel( icheck ) == prev_hvowel( icheck ) then
location = icheck
elseif not prev_vowel( icheck ) and vowel == "э" then
location = icheck
elseif vowel == "э" and sub( s.substring, icheck-1, icheck-1 ) == "й" then
location = icheck
else
break
end
end
end
if match( vh.Cyrl.a, "" ) then
vh.Cyrl.ii = "ы"
vh.Cyrl.u = "у"
vh.Mong.a = "ᠠ"
vh.Mong.u = "ᠤ"
else
vh.Cyrl.ii = "ий"
vh.Cyrl.u = "ү"
vh.Mong.a = "ᠡ"
vh.Mong.u = "ᠦ"
end
if match( vh.Cyrl.a, "ө" ) then -- ө takes the diphthong эй not өй
vh.Cyrl.ai = "эй"
else
vh.Cyrl.ai = vh.Cyrl.a .. "й"
end
vh.Cyrl.aa = vh.Cyrl.a .. vh.Cyrl.a
vh.Cyrl.uu = vh.Cyrl.u .. vh.Cyrl.u
s.substring = orig
vh.location = len( s.substring ) + breaks - location
end
end
return vh
end
--Breaks down a string into syllables and returns a table
function export.syllables( text )
local consonant = ""
local vowel = ""
local sign = ""
local iotated = ""
local punctuation = ""
local final_clusters = require( "Module:mn/data" ).syll_final_cons
local chars = {}
for v in gmatch( text, "" ) do -- strip diacritics
table.insert( chars, v )
end
text = table.concat( chars ) -- reform text without diacritics
local breaks = {}
for i,v in pairs( chars ) do
if i == 1 or match( chars, punctuation ) then -- first letter of word
table.insert( breaks, i )
elseif match( chars, export.stem_barrier ) then -- stem barrier (used by inflection templates)
table.insert( breaks, i )
elseif i ~= #chars and match( v, consonant ) and match( chars, vowel ) then -- if followed by vowel (i.e. if lone/cluster-final), count backwards until vowel, punctuation/space or start of string is found; if a vowel is found first, then letter must be medial and lone/cluster-final, so is the break; if punctuation/start of string found first, letter is part of word-initial cluster, so is not the break (occurs in loanwords, e.g. трол|лей|бус ("trolleybus"))
local j = i
while j > 1 and ( match( chars, consonant ) or match( chars, sign ) ) do
j = j - 1
if match( chars, vowel ) then
table.insert( breaks, i )
end
end
elseif ( match( v, consonant ) or match( v, sign ) ) and ( i == #chars or match( chars, punctuation ) ) then -- if word-final consonant, count backwards until vowel, checking if each cluster is allowed as a word-final cluster; if it is, increase "stable" (the number of stable consonants (and signs) at the end) by one; if a vowel is found before an unstable cluster, the loop ends with no change; if an unstable cluster is found, "stable" will not iterate which will trigger an additional unvoweled syllable break at that consonant (occurs in loanwords, e.g. буд|ди|зм ("Buddhism"), ал|ге|бр ("algebra"))
local j = i
local check = { chars }
local stable = 1
while j > 1 and j > i - #final_clusters and stable > i - j and ( match( chars, consonant ) or match( chars, sign ) ) do
j = j - 1
table.insert( check, chars )
for k,cluster in ipairs( final_clusters ) do
if match( table.concat( table.reverse( check ) ), cluster ) then
stable = stable + 1
break
end
end
if stable == i - j then
table.insert( breaks, j )
end
end
elseif match( v, vowel ) and match( chars, sign ) then -- vowel after Ь/Ъ
table.insert( breaks, i )
elseif match( v, iotated ) and match( chars, vowel ) then -- iotated ("ya"-type) vowel after a vowel
table.insert( breaks, i )
end
end
breaks = table.removeDuplicates( breaks )
local syll = {}
for i,v in ipairs( breaks ) do
if i == #breaks then
table.insert( syll, sub( text, v ) )
else
table.insert( syll, sub( text, v, breaks - 1 ) )
end
end
return syll
end
function export.remove_final_short_vowel( text, params )
if not params then params = {} end
local vh = export.vowelharmony( text, params )
local syllables = #export.syllables( text )
local reduced = text
local no_fv = false
if ( syllables > 1 and match( text, "$" ) ) or match( text, "$" ) then
local matches = {
not params.bor,
match( text, "$" ),
match( text, "ба$" ),
match( text, "бо$" ),
match( text, "өбө$" ),
match( text, "бэ$" ),
match( text, "" .. vh.Cyrl.a .. char( 0x301 ) .. "?" .. char( 0x300 ) .. "?н" .. vh.Cyrl.a .. "$" ),
match( text, "нга$" ),
match( text, "нго$" ),
match( text, "өнгө$" ),
match( text, "нгэ$" )
}
for _,v in pairs( matches ) do
if v then
reduced = sub( text, 1, len( text ) - 1 )
no_fv = true
end
end
end
return reduced, no_fv
end
function export.remove_penultimate_short_vowel( text, params )
if not params then params = {} end
local vh = export.vowelharmony( text, params )
local syllables = table.reverse( export.syllables( text ) )
if not params.proper and ( vh.location ~= len( text ) - 1 or vh.violation == false ) and not params.bor then -- exclude proper nouns, loanwords and terms where the deleted vowel determines the vowel harmony
local check
local syllable_orig
for i,syllable in ipairs( syllables ) do
if i ~=1 and i ~= #syllables then
syllable_orig = syllable
if match( syllable, export.stem_barrier ) then
break
end
local prev_syllable = gsub( syllables, export.stem_barrier, "" )
check = ( match( prev_syllable, "*$" ) or "" ) .. syllable .. ( match( syllables, "^*" ) or "" )
local matches = {
-- CVC
match( check, "^ь?$" ), -- not VC
match( check, "^гь?$" ) and vh.position == "front", -- гV if front vowel only
-- CCVC
match( check, "^?ь?ь?$" ),
match( check, "^?ь?ь?ь?$" ),
}
local exclusions = {
match( check, "ь?в$" ), -- not Vв
match( check, "ь?и$" ), -- only иC
}
for _,v in pairs( matches ) do
if v then
syllables = sub( syllable, 1, len( syllable ) - 1 )
for _,e in pairs( exclusions ) do
if e then syllables = syllable_orig end
end
end
end
if syllables ~= syllable_orig then
break
end
end
end
end
return table.concat( table.reverse( syllables ) )
end
function export.concat_forms_in_slot( forms )
if forms then
local new_vals = {}
for _, v in ipairs( forms ) do
local val = gsub( v.form, "|", "<!>" )
table.insert( new_vals, val )
end
return table.concat( new_vals, "," )
else
return nil
end
end
function export.combine_stem_ending( stem, ending )
if stem == "?" then
return "?"
else
return stem .. ending
end
end
function export.generate_form( form, footnotes )
if type( footnotes ) == "string" then
footnotes = { footnotes }
end
if footnotes then
return { form = form, footnotes = footnotes }
else
return form
end
end
return export