Module:User:Kritixilithos
Jump to navigation
Jump to search
- This module lacks a documentation subpage. You may create it.
- Useful links: subpage list • links • transclusions • testcases • user page • user talk page • userspace
This is a private module sandbox of Kritixilithos, for his own experimentation. Items in this module may be added and removed at Kritixilithos's discretion; do not rely on this module's stability.
-- tamil noun declension, work in progress
-- following https://archive.org/details/progressivegramm00ardeiala
-- for sandhi rules
-- and https://www.sai.uni-heidelberg.de/tamil/pdf/modgram.pdf
-- for case declension
--க் ங் ச் ஞ் ட் ண் த் ந் ப் ம் ய் ர் ல் வ் ழ் ள் ற் ன்
--k G c C T N t M p m y r l v z L R n
-- TODO: animate, pronouns
--local us=require("ustring/ustring")
local us=mw.ustring
local u=us.char
local s=us.sub
local gsub=us.gsub
local link = require("Module:links")
local lang = require("Module:languages").getByCode("ta")
local export={}
--vowel letters
local _A=u(0xb85);local _AA=u(0xb86);local _I=u(0xb87);local _II=u(0xb88);local _U=u(0xb89);local _UU=u(0xb8a);local _E=u(0xb8e);local _EE=u(0xb8f);local _AI=u(0xb90);local _O=u(0xb92);local _OO=u(0xb93);local _AU=u(0xb94)
--vowel combining chars
local AA=u(0xbbe);local I=u(0xbbf);local II=u(0xbc0);local U=u(0xbc1);local UU=u(0xbc2);local E=u(0xbc6);local EE=u(0xbc7);local AI=u(0xbc8);local O=u(0xbca);local OO=u(0xbcb);local AU=u(0xbcc);local DOT=u(0xbcd)
--consonants without dots
local kA=u(0xb95);local GA=u(0xb99);local cA=u(0xb9a);local CA=u(0xb9e);local TA=u(0xb9f);local NA=u(0xba3);local tA=u(0xba4);local MA=u(0xba8);local nA=u(0xba9);local pA=u(0xbaa);local mA=u(0xbae);local yA=u(0xbaf);local rA=u(0xbb0);local RA=u(0xbb1);local lA=u(0xbb2);local LA=u(0xbb3);local zA=u(0xbb4);local vA=u(0xbb5)
--convenient regices
local vls='['.._A.._AA.._I.._II.._U.._UU.._E.._EE.._AI.._O.._OO.._AU..']'
local vcs='['..AA..I..II..U..UU..E..EE..AI..O..OO..AU..']'
local svls='['.._A.._I.._U.._E.._O..']'
local svcs='['..I..U..E..O..']?'
local ks='['..kA..GA..cA..CA..TA..NA..tA..MA..pA..mA..yA..rA..lA..vA..zA..LA..RA..nA..']'
--match
function m(q,r)
return us.find(q,r)~=nil
end
--2 short open
function oo(w)
k_v=ks..'['..E..I..O..U..']?'
return(m(s(w,1,1),'['.._A.._I.._E.._U.._O..']')and m(s(w,2),'^'..k_v..'$'))or(m(w,'^'..k_v..k_v..'$'))
end
--combine word ending in konsonant with word beginning in vowel
function kv(wk,vw)
tr={[_A]='',[_AA]=AA,[_I]=I,[_II]=II,[_U]=U,[_UU]=UU,[_E]=E,[_EE]=EE,[_AI]=AI,[_O]=O,[_OO]=OO,[_AU]=AU}
return wk..tr[s(vw,1,1)]..s(vw,2)
end
--combine via sandhi
---some rules redundant for our purposes
---chap2, rules 1-8 done (mostly)
function c(q,w)
if not oo(q)and m(q,U..'$')and m(w,'^'..vls)then
return kv(s(q,1,-2),w)
elseif m(q,U..'$')and oo(q)and m(w,'^'..vls)then
--TODO: atu, itu, etu are exceptions and follow rule 1 instead
return kv(q..vA,w)
elseif not m(q,U..'$')and(m(q,vcs..'$')or m(q,vls..'$'))and m(w,'^'..vls)then
--note: if m(q,'E$'), either 'y' or 'v' can be inserted
if m(q,'['..I..II..AI.._I.._II.._AI..']$')then
return kv(q..yA,w)
else
return kv(q..vA,w)
end
elseif(m(q,'^'..ks..svcs..ks..DOT..'$')or m(q,'^'..svls..ks..DOT..'$'))and m(w,'^'..vls)then
return kv(gsub(q,'(.)'..DOT..'$','%0%1'),w)
elseif m(q,ks..DOT..'$')and m(w,'^'..vls)then
return kv(s(q,0,-2),w)
----doesn't seem relevant
--elseif(m(q,vcs..'$')or m(q,vls..'$'))and m(w,'^['..kA..cA..tA..pA..']')and(q=='aMta'or q=='inta'or q=='enta'or q=='appaTi'or q=='ippaTi'or q=='eppaTi'
----or m(w,'^[^ ]') --irrelevant as not dealing with compounds
--or m(q,'Ay$')or m(q,'Aka$')
----TODO:check q for dat, acc, inf respectively for below (if needed)
--or m(q,'ku$')or m(q,'Y$')or m(q,'a$'))then
--return q..gsub(w,'^ ?([kctp])','%1%0')
elseif m(q,'['..nA..lA..NA..LA..']'..DOT..'$')and m(w,'^['..kA..cA..tA..pA..']')then
if m(q,'['..nA..lA..']'..DOT..'$')then
return gsub(q,'..$',RA..DOT)..w
else
return gsub(q,'..$',TA..DOT)..w
end
elseif m(q,mA..DOT..'$')then
if m(w,'^'..kA)then
return gsub(q,'..$',GA..DOT)..w
elseif m(w,'^'..cA)then return gsub(q,'..$',CA..DOT)..w
elseif m(w,'^'..tA)then return gsub(q,'..$',MA..DOT)..w
else return gsub(q,'..$','')..w end
else
--XXX:my addition, not sure if right
return q..w
end
end
--oblique stem
-- TODO:override for oblique suffix for 1.8 words
function obl(s)
if m(s,ks..mA..DOT..'$')or m(s,_A..mA..DOT..'$')then
return gsub(s,'..$',tA..DOT..tA..U)..''
--XXX:is katti 2 short syllables?
elseif m(s,'['..TA..RA..']'..U..'$')and not(m(s,'^'..ks..svcs..ks..svcs..'('..ks..DOT..')?$')or m(s,'^'..svls..ks..svcs..'('..ks..DOT..')?$'))then
return gsub(s,'(.).$','%1'..DOT..'%0')..''
else
return s
end
end
--plural
-- TODO:epicene suffix, and gendered nouns
function pl(s)
--XXX:is katti 2 short syllables?
if m(s,'['..AA..EE..II..UU..OO.._AA.._EE.._II.._UU.._OO..']$')or m(s,'^'..ks..svcs..ks..U..'$')or m(s,'^'..svls..ks..U..'$')then
return c(s,kA..DOT..kA..LA..DOT)
else
return c(s,kA..LA..DOT)
end
end
--conditional plural or oblique
function plobl(s,p)if p then return pl(s)else return obl(s)end end
function acc(s,p)return c(plobl(s,p),_AI)end
function dat(s,p)
stem=plobl(s,p)
--XXX: not sure about AU and _AU
if(stem~=s or m(s,'['..I..II..AI..AU..U.._I.._II.._AI.._AU.._U..']$'))and not p then
return c(stem,kA..DOT..kA..U)
else
return c(stem,_U..kA..DOT..kA..U)
end
end
function ins(s,p)return c(plobl(s,p),_AA..lA..DOT)end
--TODO: also _I..lA..DOT for animate
function loc(s,p,an)
if an then
return c(plobl(s,p),_I..TA..mA..DOT)
else
return c(plobl(s,p),_I..lA..DOT)
end
end
--TODO: also _I..lA..DOT for animate
function abl(s,p,an)
return c(loc(s,p,an),_I..rA..U..nA..DOT..tA..U)
end
--TODO: also _U..TA..nA..DOT
function soc(s,p)
return c(plobl(s,p),_OO..TA..U)
end
--TODO: also _U..TA..AI..yA and _A..tA..U
-- note: potential exceptions with former for some
-- also, check when euphonic increment is compulsory
-- because currently does not match Arden's gen(maram)
function gen(s,p)
return c(plobl(s,p),_I..nA..DOT)
end
function voc(s,p)return c(plobl(s,p),_EE)end
function ln(w)
return link.full_link({lang=lang,alt=w},"term")
end
function export.conj(frame)
mw.log("test")
local args = frame:getParent().args
local page_title = mw.title.getCurrentTitle().text
local w = args[1] or page_title
for k, v in pairs(args) do
mw.log(k,v)
end
mw.log(ln(w))
local ret=[=[
{| class="inflection-table vsSwitcher" data-toggle-category="inflection" style="background:#FEFEFE; text-align:center; border: 1px solid #CCC;"
|- style="background: #d9ebff;"
! class="vsToggleElement" style="text-align: left;" colspan="3" | Declension of ]=]..frame:expandTemplate{title = 'l-self', args = {[1] = "ta", [2] = w}}.."\n"..[=[
|- class="vsHide"
! style="background:#eff7ff" |
! style="background:#eff7ff" | Singular
! style="background:#eff7ff" | Plural
|- class="vsHide"
!style="background:#eff7ff"| ''Nominative''
|]=]..ln(w)..[=[
|]=]..ln(pl(w))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Vocative''
|]=]..ln(voc(w,1<0))..[=[
|]=]..ln(voc(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Accusative''
|]=]..ln(acc(w,1<0))..[=[
|]=]..ln(acc(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Dative''
|]=]..ln(dat(w,1<0))..[=[
|]=]..ln(dat(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Genitive''
|]=]..ln(gen(w,1<0))..[=[
|]=]..ln(gen(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Locative''
|]=]..ln(loc(w,1<0))..[=[
|]=]..ln(loc(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Sociative''
|]=]..ln(soc(w,1<0))..[=[
|]=]..ln(soc(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Instrumental''
|]=]..ln(ins(w,1<0))..[=[
|]=]..ln(ins(w,1>0))..[=[
|- class="vsHide"
!style="background:#eff7ff"| ''Ablative''
|]=]..ln(abl(w,1<0))..[=[
|]=]..ln(abl(w,1>0))..[=[
|}
]=]
return ret
end
--maram=mA..rA..mA..DOT
--print(pl(maram,1>0))
--print(acc(maram,1<0))
--print(dat(maram,1<0))
--print(ins(maram,1<0))
--print(loc(maram,1<0,1<0))
--print(abl(maram,1<0,1<0))
--print(soc(maram,1<0))
--print(gen(maram,1<0))
--print(voc(maram,1<0))
return export