--[===[
MODULE "MAVKFORMO" (avk form)
"eo.wiktionary.org/wiki/Modulo:mavkformo" <!--2021-Dec-30-->
Purpose: analyzes a quasi-base or non-base form of an -avk- verb,
and returns selected one of 8 types of information
Utilo: analizas kvazaux-bazan aux nebazan formon de -avk- verbo,
kaj redonas unu elektitan tipon de informo el 8 disponeblaj
Manfaat: menganalisis sebuah bentuk kvasi-dasar atau fleksi verba -avk-,
dan memberikan kembali satu dari 8 ...
Syfte: analyserar en kvasi-grundform eller boejningsform hos ett -avk- verb,
och levererar en utvald typ av information av 8 tillgaengliga
Used by templates / Uzata far sxablonoj:
- "avk-konjugacioj" and "avk-verba-formo" (not to be
called from any other place)
Required submodules / Bezonataj submoduloj / Submodul yang diperlukan: none
This module is special in that it takes parameters both those sent
to itself (own frame) and those sent to the caller (caller's frame).
This module needs parameters that are additional to or different from
parameters submitted to the calling template.
The one hidden parameter can be on our own or on the caller's frame. If the
parameter is not found on our own frame, the caller's one is peeked. This
way we can avoid forwarding that does not work for non-existent parameters
(non-existent ie nil becomes empty string, not appreciated).
Parameters: * 1 anonymous obligatory parameter :
type of information requested :
* "R" for root without the final vowel
* "D" for description
* "B" for quasi-base form (1st singular present)
* "V" for the final vowel (a,e,i,u) without decoration
* "W" for the final vowel with decoration
* "F" for the inflexion affix including the final vowel
with decoration if applicable
* "T" for tense code "0" ... "2"
* "Z" for root plus final vowel without decoration
anything else is an error
* 1 hidden named parameter :
* "pagenameoverridetestonly="
Returned: * desired text, or empty on error
This module is unbreakable (when called with correct module name
and function name).
Cxi tiu modulo estas nerompebla (kiam vokita kun gxustaj nomo de modulo
kaj nomo de funkcio).
: ---------------------------------------
* #T00 ("R", page "platitit")
* expected result: "plat"
* actual result: "{{#invoke:mavkformo|ek|R|pagenameoverridetestonly=platitit}}"
::* #T01 ("D", page "platitit", holy vowel "i")
::* expected result: "unua persono de pluralo estonteco"
::* actual result: "{{#invoke:mavkformo|ek|D|pagenameoverridetestonly=platitit}}"
* #T02 ("B", page "platitit")
* expected result: "plati" (i with comma above)
* actual result: "{{#invoke:mavkformo|ek|B|pagenameoverridetestonly=platitit}}"
::* #T03 ("V", page "platitit")
::* expected result: "i" (without decoration)
::* actual result: "{{#invoke:mavkformo|ek|V|pagenameoverridetestonly=platitit}}"
* #T04 ("W", page "platitit")
* expected result: "i" (with comma above)
* actual result: "{{#invoke:mavkformo|ek|W|pagenameoverridetestonly=platitit}}"
::* #T05 ("F", page "platitit")
::* expected result: "itit" (without decoration)
::* actual result: "{{#invoke:mavkformo|ek|F|pagenameoverridetestonly=platitit}}"
* #T06 ("Z", page "platitit")
* expected result: "plati" (without decoration)
* actual result: "{{#invoke:mavkformo|ek|Z|pagenameoverridetestonly=platitit}}"
: ---------------------------------------
* #T10 ("D", page "dankayac", holy vowel "a")
* expected result: "dua persono de pluralo estinteco"
* actual result: "{{#invoke:mavkformo|ek|D|pagenameoverridetestonly=dankayac}}"
::* #T11 ("T", page "dankayac", holy vowel "a")
::* expected result: "1" (code for past)
::* actual result: "{{#invoke:mavkformo|ek|T|pagenameoverridetestonly=dankayac}}"
* #T12 ("R", page "gi", shortest root, nonstandard quasi-base form without decoration)
* expected result: "g"
* actual result: "{{#invoke:mavkformo|ek|R|pagenameoverridetestonly=gi}}"
::* #T13 ("R", page "i", invalid too short quasi-base form without decoration)
::* expected result: ""
::* actual result: "{{#invoke:mavkformo|ek|R|pagenameoverridetestonly=i}}"
: ---------------------------------------
]===]
local avkformo = {}
------------------------------------------------------------------------
---- CONSTANTS ----
------------------------------------------------------------------------
local contabaffixes = {}
contabaffixes ["l"] = 1 -- sing 2
contabaffixes ["r"] = 2 -- sing 3
contabaffixes ["t"] = 3 -- plur 1
contabaffixes ["c"] = 4 -- plur 2
contabaffixes ["d"] = 5 -- plur 3
contabaffixes ["v"] = 6 -- plur 4
local contabpersons = {}
contabpersons [0] = "unua persono de singularo"
contabpersons [1] = "dua persono de singularo"
contabpersons [2] = "tria persono de singularo"
contabpersons [3] = "unua persono de pluralo"
contabpersons [4] = "dua persono de pluralo"
contabpersons [5] = "tria persono de pluralo"
contabpersons [6] = "kvara persono de pluralo"
local contabtenses = {}
contabtenses [0] = "prezenco"
contabtenses [1] = "estinteco"
contabtenses [2] = "estonteco"
constrbaza = " (kvaza" .. string.char(197,173) .. "baza formo)"
------------------------------------------------------------------------
---- HIGH LEVEL FUNCTIONS ----
------------------------------------------------------------------------
local function lfisholy (numcode)
return ((numcode==97) or (numcode==101) or (numcode==105) or (numcode==117))
end--function lfisholy
------------------------------------------------------------------------
-- Local function LFAEIU
-- Add or remove decoration (only comma above a,e,i,u) in the
-- last letter of a string.
-- "a" 97 C3 A1 161 -- "e" 101 C3 A9 169
-- "i" 105 C3 AD 173 -- "u" 117 C3 BA 186
-- $C3 = #195
local function lfaeiu (striinut,booremove)
local numlenc = 0
local numlast = 0
local numprev = 0
local numneww = 0
numlenc = string.len(striinut)
if (numlenc>=1) then
numlast = string.byte (striinut,numlenc,numlenc)
end--if
if (booremove) then -- "true" to remove -- "false" to add decoration
if (numlenc>=2) then
numprev = string.byte (striinut,(numlenc-1),(numlenc-1))
end--if
if (numprev==195) then
if (numlast==161) then
numneww = 97 -- a
end--if
if (numlast==169) then
numneww = 101 -- e
end--if
if (numlast==173) then
numneww = 105 -- i
end--if
if (numlast==186) then
numneww = 117 -- u
end--if
end--if
else
if (numlast==97) then
numneww = 161 -- a
end--if
if (numlast==101) then
numneww = 169 -- e
end--if
if (numlast==105) then
numneww = 173 -- i
end--if
if (numlast==117) then
numneww = 186 -- u
end--if
end--if (booremove) else
if ((numneww>0) and (numneww<128)) then
striinut = string.sub(striinut,1,(numlenc-2)) .. string.char(numneww) -- removing decoration
end--if
if (numneww>128) then
striinut = string.sub(striinut,1,(numlenc-1)) .. string.char(195,numneww) -- adding decoration
end--if
return striinut
end--function lfaeiu
------------------------------------------------------------------------
---- MAIN EXPORTED FUNCTION ----
------------------------------------------------------------------------
function avkformo.ek (arxframent)
-- general unknown type
local vartmp = 0 -- variable without type
-- special type "args" AKA "arx"
local arxourown = 0 -- metaized "args" from our own "frame"
local arxcaller = 0 -- metaized "args" from caller's "frame"
-- general "str"
local strpagename = ""
local strsist = "" -- only one char
local strholynd = "" -- only one ASCII char
local strholydk = "" -- only one UTF8 char
local strinflex = "" -- will get filled per "ueglstr"
local strret = "" -- result string
-- general "num"
local numlopagena = 0
local numpersono = 0 -- 0...6 -- 0 1 2 sing 3 4 5 6 plur
local numtenso = 0 -- 0...2 -- 0:present 1:past 2:future 3:invalid
local numm0 = 0
local numm1 = 0
local numm2 = 0
-- general "boo"
local booerr = false
local boohaveroot = false -- "R" or "D" or "B" or "V" or "W" "F" "T" "Z"
local boohavedeed = false
local boohaveboob = false
local boohavevaav = false
local boohavewuuw = false
local boohavefina = false
local boohavetens = false
local boohavezzzz = false
---- GET THE ARX:ES ----
arxourown = arxframent.args or {} -- "args" fr our own "frame"
arxcaller = arxframent:getParent().args or {} -- "args" fr caller's "frame"
---- PROCESS ONE HIDDEN AND ONE OBLIGATORY ANONYMOUS PARAMETER ----
-- min length is 2 to allow 2-letter verbs in base form without decoration
strpagename = ''
while (true) do -- fake loop
booerr = true -- preliminary, preassume guilt
vartmp = arxourown['pagenameoverridetestonly']
if (vartmp==nil) then
vartmp = arxcaller['pagenameoverridetestonly']
end--if
if (type(vartmp)=="string") then
numlopagena=string.len(vartmp)
if ((numlopagena>=2 and numlopagena<=100)) then
strpagename = vartmp -- no error possible here
end--if
end--if
if (arxourown[2]~=nil) then -- undesirable
break -- bad
end--if
vartmp = arxourown[1] -- mau "R" or "D" or "B" or "V" or "W" "F" "T" "Z"
boohaveroot = (vartmp=="R")
boohavedeed = (vartmp=="D")
boohaveboob = (vartmp=="B")
boohavevaav = (vartmp=="V")
boohavewuuw = (vartmp=="W")
boohavefina = (vartmp=="F")
boohavetens = (vartmp=="T")
boohavezzzz = (vartmp=="Z")
booerr = ((not boohaveroot) and (not boohavedeed) and (not boohaveboob) and (not boohavevaav) and (not boohavewuuw) and (not boohavefina) and (not boohavetens) and (not boohavezzzz))
break -- good or bad
end--while -- fake loop
---- SEIZE THE PAGENAME FROM MW AND CHECK IT ----
-- must be 2...100 octet:s keep consistent with "pagenameoverridetestonly="
-- ignore if type not "string", but otherwise do either buy it or brew error
if ((not booerr) and (strpagename=='')) then
vartmp = mw.title.getCurrentTitle().text -- without namespace prefix
if (type(vartmp)=="string") then
numlopagena = string.len(vartmp)
if ((numlopagena>=2) and (numlopagena<=100)) then
strpagename = vartmp
else
booerr = true
end--if
end--if
end--if
booerr = booerr or (strpagename=='')
---- DETECT PERSON AND TENSE ----
-- "numpersono" -- 0...6 -- 0 1 2 sing N/A l r -- 3 4 5 6 plur t c d v
-- "numtenso" -- 0...2 -- 0:present 1:past 2:future 3:invalid
-- example: present past future
-- plat-i-v plat-i-y-i-v plat-i-t-i-v
-- sux-a-v sux-a-y-a-v sux-a-t-a-v
-- the holy final vowel (a,e,i,u) is used twice in past and future tenses
if (not booerr) then
strpagename = lfaeiu (strpagename,true) -- remove deco
numlopagena = string.len(strpagename) -- 1...100 -- after removal !!!
booerr = (numlopagena<2)-- but ONE is invalid
end--if
if (not booerr) then
numpersono = 0 -- preassume 1st any tense
strinflex = "" -- will get filled per "ueglstr"
numtenso = 3 -- preassume guilt
strsist = string.sub (strpagename,numlopagena,numlopagena) -- detect perso
vartmp = contabaffixes [strsist] -- 6 string indexes
if (type(vartmp)=="number") then
numpersono = vartmp -- default ZERO ie "sing 1"
strinflex = strsist
strpagename = string.sub (strpagename,1,(numlopagena-1))
numlopagena = numlopagena - 1 -- remove ONE processed char
end--if
numm0 = string.byte (strpagename,numlopagena,numlopagena)
numm1 = string.byte (strpagename,(numlopagena-1),(numlopagena-1))
if (numlopagena>=4) then -- test for past and future
numm2 = string.byte (strpagename,(numlopagena-2),(numlopagena-2))
if ((numm0==numm2) and lfisholy(numm0)) then
if (numm1==121) then
numtenso = 1 -- "y" -> YES it was the paaast !!!
end--if
if (numm1==116) then
numtenso = 2 -- "t" -> YES it will be the fuuuture !!!
end--if
end--if
if (numtenso~=3) then
strholynd = string.char (numm0) -- without decoration
strinflex = string.sub (strpagename,(numlopagena-2),numlopagena) .. strinflex
strpagename = string.sub (strpagename,1,(numlopagena-3))
numlopagena = numlopagena - 3 -- remove 3 processed chars
end--if
end--if
if ((numtenso==3) and (numlopagena>=2)) then -- test for present
if (lfisholy(numm0)) then
numtenso = 0 -- YES it is the preeesent !!!
strholynd = string.char (numm0) -- without decoration
if (numpersono==0) then
strinflex = lfaeiu (strholynd,false) .. strinflex -- with deco
else
strinflex = strholynd .. strinflex -- without deco
end--if
strpagename = string.sub (strpagename,1,(numlopagena-1))
numlopagena = numlopagena - 1 -- remove ONE processed char
end--if
end--if
booerr = (numtenso==3)
end--if
---- FILL ONE MORE VARIABLE ----
if (not booerr) then
strholydk = lfaeiu (strholynd,false) -- add deco
end--if
---- SELECT DESIRED OUTPUT ----
if (not booerr) then -- "R" or "D" or "B" or "V" or "W" "F" "T" "Z"
if (boohaveroot) then -- "R" -- root
strret = strpagename -- now stripped ;-)
end--if
if (boohavedeed) then -- "D" -- description
strret = contabpersons[numpersono] .. " " .. contabtenses[numtenso]
if ((numpersono==0) and (numtenso==0)) then
strret = strret .. constrbaza
end--if
end--if
if (boohaveboob) then -- "B" -- full base form
strret = strpagename .. strholydk -- with decoration
end--if
if (boohavevaav) then -- "V"
strret = strholynd
end--if
if (boohavewuuw) then -- "W"
strret = strholydk
end--if
if (boohavefina) then -- "F" -- inflexion affix
strret = strinflex
end--if
if (boohavetens) then -- "T" -- tense code "0" ... "2"
strret = tostring (numtenso)
end--if
if (boohavezzzz) then -- "Z" -- undecorated base form, root + holy vowel
strret = strpagename .. strholynd -- without decoration
end--if
end--if
---- RETURN THE JUNK STRING ----
return strret -- can be empty
end--function
---- RETURN THE JUNK LUA TABLE ----
return avkformo
