Module:User:Benwing2/cs-adjective

local export = {}

--[=[

Authorship: Ben Wing

]=]

--[=[

TERMINOLOGY:

-- "slot" = A particular combination of case/gender/number. Example slot names for adjectives are "gen_f" (genitive feminine singular) and "nom_mp_an" (animate nominative masculine plural). Each slot is filled with zero or more forms.

-- "form" = The declined Czech form representing the value of a given slot.

-- "lemma" = The dictionary form of a given Czech term. Generally the nominative masculine singular, but may occasionally be another form if the nominative masculine singular is missing. ]=]

local lang = require("Module:languages").getByCode("cs") local m_links = require("Module:links") local m_table = require("Module:table") local m_string_utilities = require("Module:string utilities") local iut = require("Module:inflection utilities") local com = require("Module:User:Benwing2/cs-common")

local current_title = mw.title.getCurrentTitle local NAMESPACE = current_title.nsText local PAGENAME = current_title.text

local u = mw.ustring.char local rsplit = mw.text.split local rfind = mw.ustring.find local rmatch = mw.ustring.match local rgmatch = mw.ustring.gmatch local rsubn = mw.ustring.gsub local ulen = mw.ustring.len local uupper = mw.ustring.upper

-- version of rsubn that discards all but the first return value local function rsub(term, foo, bar) local retval = rsubn(term, foo, bar) return retval end

-- version of rsubn that returns a 2nd argument boolean indicating whether -- a substitution was made. local function rsubb(term, foo, bar) local retval, nsubs = rsubn(term, foo, bar) return retval, nsubs > 0 end

-- All slots that are used by any of the different tables. The key is the slot and the value is a list of the tables -- that use the slot. "" = regular, "plonly" = special=plonly in, "dva" = special=dva in --. local input_adjective_slots = { nom_m = {""}, nom_f = {""}, nom_n = {""}, nom_mp_an = {"", "plonly"}, nom_fp = {"", "plonly"}, nom_np = {"", "plonly"}, nom_mp = {"dva"}, nom_fnp = {"dva"}, gen_m = {""}, gen_f = {""}, gen_p = {"", "plonly", "dva"}, dat_m = {""}, dat_f = {""}, dat_p = {"", "plonly", "dva"}, acc_m_an = {""}, acc_m_in = {""}, acc_f = {""}, acc_n = {""}, acc_mfp = {"", "plonly"}, acc_np = {"", "plonly"}, acc_mp = {"dva"}, acc_fnp = {"dva"}, ins_m = {""}, ins_f = {""}, ins_p = {"", "plonly", "dva"}, loc_m = {""}, loc_f = {""}, loc_p = {"", "plonly", "dva"}, short_m = {""}, short_f = {""}, short_n = {""}, short_mp_an = {""}, short_fp = {""}, short_np = {""}, }

local output_adjective_slots = { nom_m = "nom|m|s", nom_m_linked = "nom|m|s", -- used in Module:cs-noun? nom_f = "nom|f|s", nom_n = "nom|n|s", nom_mp_an = "an|nom|m|p", nom_fp = "in|nom|m|p|;|nom|f|p", nom_np = "nom|n|p", nom_mp = "nom|m|p", nom_fnp = "nom|f//n|p", gen_m = "gen|m//n|s", gen_f = "gen|f|s", gen_p = "gen|p", dat_m = "dat|m//n|s", dat_f = "dat|f|s", dat_p = "dat|p", acc_m_an = "an|acc|m|s", acc_m_in = "in|acc|m|s", acc_f = "acc|f|s", acc_n = "acc|n|s", acc_mfp = "acc|m//f|p", acc_np = "acc|n|p", acc_mp = "acc|m|p", acc_fnp = "acc|f//n|p", ins_m = "ins|m//n|s", ins_f = "ins|f|s", ins_p = "ins|p", loc_m = "loc|m//n|s", loc_f = "loc|f|s", loc_p = "loc|p", short_m = "short|m|s", short_f = "short|f|s", short_n = "short|n|s", short_mp_an = "short|an|m|p", short_fp = "short|in|m|p|;|short|f|p", short_np = "short|n|p", }

local function get_output_adjective_slots(alternant_multiword_spec) return output_adjective_slots end

local function combine_stem_ending(stems, ending) if stems == "?" then return "?" else return stems .. ending end end

local function add(base, slot, stems, endings, footnote) if stems then stems = iut.combine_form_and_footnotes(stems, footnote) end iut.add_forms(base.forms, slot, stems, endings, combine_stem_ending) end

local function add_normal_decl(base, stems,	nom_m, nom_f, nom_n, nom_mp_an, nom_fp, nom_np,	gen_m, gen_f, gen_p,	dat_m, dat_f, dat_p,	acc_f,	loc_m, loc_f, loc_p,	ins_m, ins_f, ins_p,	footnote) if stems then stems = iut.combine_form_and_footnotes(stems, footnote) end add(base, "nom_m", stems, nom_m) add(base, "nom_f", stems, nom_f) add(base, "nom_n", stems, nom_n) add(base, "nom_mp_an", stems, nom_mp_an) add(base, "nom_fp", stems, nom_fp) add(base, "nom_np", stems, nom_np) add(base, "gen_m", stems, gen_m) add(base, "gen_f", stems, gen_f) add(base, "gen_p", stems, gen_p) add(base, "dat_m", stems, dat_m) add(base, "dat_f", stems, dat_f) add(base, "dat_p", stems, dat_p) add(base, "acc_f", stems, acc_f) add(base, "loc_m", stems, loc_m) add(base, "loc_f", stems, loc_f) add(base, "loc_p", stems, loc_p) add(base, "ins_m", stems, ins_m) add(base, "ins_f", stems, ins_f) add(base, "ins_p", stems, ins_p) end

local function add_short_decl(base, stems, m, f, n, mp_an, fp, np, footnote) if stems then stems = iut.combine_form_and_footnotes(stems, footnote) end add(base, "short_m", stems, m)	add(base, "short_f", stems, f)	add(base, "short_n", stems, n)	add(base, "short_mp_an", stems, mp_an) add(base, "short_fp", stems, fp) add(base, "short_np", stems, np) end

local decls = {}

decls["normal"] = function(base) local stem, suffix

-- hard in -ý stem, suffix = rmatch(base.lemma, "^(.*)(ý)$") if stem then add_normal_decl(base, stem,			"ý", "á", "é", {}, "é", "á",			"ého", "é", "ých",			"ému", "é", "ým",			"ou",			"ém", "é", "ých",			"ým", "ou", "ými"		) -- Do the nominative masculine animate plural separately since it may have a different stem (with the second		-- palatalization applied). add_normal_decl(base, com.apply_second_palatalization(stem), nil, nil, nil, "í") if base.short then -- Examples of short adjectives: -- bledý "pale" -> bled -- bosý "barefoot" -> bos -- pilný "hardworking, diligent" -> pilen (reducible) -- veselý "funny, jolly" -> vesel -- jistý "certain, sure" -> jist -- vinný "guilty" -> vinen (reducible) -- živý "alive, living" -> živ -- tichý "quiet" -> tich; mp_an = tiši -- vědomý "conscious, aware" -> vědom -- rád "glad" (short only) -- chudý "poor" -> chud -- nevinný "innocent" -> nevinen (reducible) -- silný "strong" -> silen (reducible) -- známý "known" -> znám -- mladý "young" -> mlád (note length); other forms have mlád- or mlad- -- starý "old" -> stár (note length); other forms have stár- or star-; mp_an = stáři, staři -- slabý "weak" -> sláb (note length); other forms only have sláb- (FIXME: check against a grammar) -- zdravý "healthy" -> zdráv (note length); other forms only have zdráv- (FIXME: check against a grammar) -- nemocný "ill" -> nemocen (reducible) -- plný "full" -> pln -- schopný "able, capable" -> schopen (reducible) -- vděčný "grateful" -> vděčen (reducible) -- věrný "faithful" -> věren (reducible) -- šťastný "happy" -> šťasten (reducible) -- křepký "strong" -> křepek (reducible); mp_an = křepci -- hotový "ready, finished" -> hotov -- němý "mute" -> něma -- smutný "sad" -> smuten (reducible) -- samotný "lonely" -> samoten (reducible) -- bohatý "rich" -> bohat -- chladný "cold" -> chladen (reducible) -- dlužný "necessary?" -> dlužen (reducible) -- povinný "mandatory" -> povinen (reducible) -- zodpovědný "responsible" -> zodpověden (reducible) -- udatný "brave" -> udaten (reducible) -- náchylný "susceptible" -> náchylen (reducible) for _, short_stem_obj in ipairs(base.short) do				add_short_decl(base, short_stem_obj.base, "") add_short_decl(base, short_stem_obj.stem, nil, "a", "o", nil, "y", "a") add_short_decl(base, com.apply_second_palatalization(short_stem_obj.stem.form), nil, nil, nil, "i",					short_stem_obj.stem.footnotes) end end return end

-- soft in -í stem, suffix = rmatch(base.lemma, "^(.*)(í)$") if stem then add_normal_decl(base, stem,			"í", "í", "í", "í", "í", "í",			"ího", "í", "ích",			"ímu", "í", "ím",			"í",			"ím", "í", "ích",			"ím", "í", "ími"		) return end

-- possessive in -ův stem, suffix = rmatch(base.lemma, "^(.*)(ův)$") if stem then add_normal_decl(base, stem,			"ův", "ova", "ovo", "ovi", "ovy", "ova",			"ova", "ovy", "ových",			"ovu", "ově", "ovým",			"ovu",			"ově", "ově", "ových",			"ovým", "ovou", "ovými"		) return end

-- possessive in -in stem, suffix = rmatch(base.lemma, "^(.*)(in)$") if stem then add_normal_decl(base, stem,			"in", "ina", "ino", "ini", "iny", "ina",			"ina", "iny", "iných",			"inu", "ině", "iným",			"inu",			"ině", "ině", "iných",			"iným", "inou", "inými"		) return end

error("Unrecognized adjective lemma, should end in '-ý', '-í', '-ův' or '-in': '" .. base.lemma .. "'") end

decls["irreg"] = function(base) local stem, suffix

-- determiner like můj stem, suffix = rmatch(base.lemma, "^(.*)(ůj)$") if stem then add_normal_decl(base, stem,			"ůj", {"á", "oje"}, {"e", "oje"}, {"í", "oji"}, {"é", "oje"}, {"á", "oje"},			"ého", {"é", "ojí"}, "ých",			"ému", {"é", "ojí"}, "ým",			{"ou", "oji"},			"ém", {"é", "ojí"}, "ých",			"ým", {"ou", "ojí"}, "ými"		) return end

-- determiner like ten, tamten, tamhleten, tuten, jeden, onen -- tento uses 'ten to' -- tenhle uses 'ten hle' -- tenhleten uses 'ten hleten ' stem, suffix = rmatch(base.lemma, "^(.*)(en)$") if stem then local nom_stem = stem .. suffix if nom_stem == "jeden" then stem = "jedn" end add_normal_decl(base, nom_stem, "") add_normal_decl(base, stem,			nil, "a", "o", "i", "y", "a",			"oho", "é", "ěch",			"omu", "é", "ěm",			"u",			"om", "é", "ěch",			"ím", "ou", "ěmi"		) return end

-- náš, váš stem, suffix = rmatch(base.lemma, "^(.*)(áš)$") if stem then local nom_stem = stem .. suffix stem = stem .. "aš" add_normal_decl(base, nom_stem, "") add_normal_decl(base, stem,			nil, "e", "e", "i", "e", "e",			"eho", "í", "ich",			"emu", "í", "im",			"i",			"em", "í", "ich",			"ím", "í", "imi"		) return end

if base.lemma == "jenž" then local preposition_footnote = "the leading letter j- is changed to n- when the pronoun is preceded by a preposition, e.g., , " preposition_footnote = "[" .. mw.getCurrentFrame:preprocess(preposition_footnote) .. "]"		-- Add the non-prepositional forms. add_normal_decl(base, "",			"jenž", "jež", "jež", "již", "jež", "jež",			"jehož", "jíž", "jichž",			"jemuž", "jíž", "jimž",			"již",			nil, nil, nil,			"jímž", "jíž", "jimiž"		) -- Add the prepositional forms. (FIXME: Maybe should go in a separate column in a special table.) add_normal_decl(base, "",			nil, nil, nil, nil, nil, nil,			"něhož", "níž", "nichž",			"němuž", "níž", "nimž",			"niž",			"němž", "níž", "nichž",			"nímž", "níž", "nimiž",			preposition_footnote		) -- Unusually, the accusative masculine animate singular is not the same as the genitive masculine singular, -- and the accusative masculine inanimate singular is not the same as the nominative masculine singular. add(base, "acc_m_an", "", {"jejž", "jehož"}) add(base, "acc_m_an", "", {"nějž", "něhož"}, preposition_footnote) add(base, "acc_m_in", "", "jejž") add(base, "acc_m_in", "", "nějž", preposition_footnote) return end

if base.lemma == "tentýž" then add_normal_decl(base, "",			"tentýž", "tatáž", "totéž", "titíž", "tytéž", "tatáž",			"téhož", "téže", "týchž",			{"témuž", "tomutéž"}, "téže", "týmž",			"tutéž",			"tomtéž", "téže", "týchž",			"tímtéž", "toutéž", "týmiž"		) return end

if base.lemma == "týž" then add_normal_decl(base, "",			"týž", "táž", nil, "tíž", nil, "tatáž",			"téhož", "téže", "týchž",			"témuž", "téže", "týmž",			"touž",			"témž", "téže", "týchž",			"týmž", "touž", "týmiž"		) return end

if base.lemma == "sám" then -- This mixes long and short endings. add_normal_decl(base, "sám", "") add_normal_decl(base, "sam",			nil, "a", "o", "i", "y", "a",			"ého", "é", "ých",			"ému", "é", "ým",			"u",			"ém", "é", "ých",			"ým", "ou", "ými"		) -- Unusually, the accusative masculine animate singular is not the same as the genitive masculine singular. add(base, "acc_m_an", "sam", {"a", "ého"}) return end

if base.lemma == "jejíž" then add_normal_decl(base, "jej",			"íž", "íž", "íž", nil, "íž", "íž",			"íhož", "íž", "íchž",			"ímuvz", "íž", "ímž",			"íž",			"ímž", "íž", "íchž",			"ímž", "íž", "ímiž"		) add_normal_decl(base, "jij", nil, nil, nil, "íž") return end

if base.lemma == "všechen" then add_normal_decl(base, "",			"všechen", "všechna", {"všechno", "vše"}, "všichni", "všechny", "všechna",			"všeho", "vší", "všech",			"všemu", "vší", "všem",			{"všechnu", "vši"},			"všem", "vší", "všech",			"vším", "vší", "všemi"		) return end

if base.lemma == "všecek" then add_normal_decl(base, "",			"všecek", "všecka", {"všecko", "vše"}, "všicci", "všecky", "všecka",			"všeho", "vší", "všech",			"všemu", "vší", "všem",			{"všecku", "vši"},			"všem", "vší", "všech",			"vším", "vší", "všemi"		) return end

if base.lemma == "všecken" then add_normal_decl(base, "",			"všecken", "všeckna", {"všeckno", "vše"}, "všickni", "všeckny", "všeckna",			"všeho", "vší", "všech",			"všemu", "vší", "všem",			{"všecknu", "vši"},			"všem", "vší", "všech",			"vším", "vší", "všemi"		) return end

error("Unrecognized irregular lemma '" .. base.lemma .. "'") end

local function fetch_footnotes(separated_group) local footnotes for j = 2, #separated_group - 1, 2 do		if separated_group[j + 1] ~= "" then error("Extraneous text after bracketed footnotes: '" .. table.concat(separated_group) .. "'") end if not footnotes then footnotes = {} end table.insert(footnotes, separated_group[j]) end return footnotes end

local function parse_indicator_spec(angle_bracket_spec) local inside = rmatch(angle_bracket_spec, "^<(.*)>$") assert(inside) local base = {forms = {}} if inside ~= "" then local parts = rsplit(inside, ".", true) for _, part in ipairs(parts) do			if part == "irreg" then base.irreg = true elseif part == "short" then base.short = elseif rfind(part, "^short:") then part = rsub(part, "^short:%s*", "") base.short = {} local segments = iut.parse_balanced_segment_run(part, "[", "]") local comma_separated_groups = iut.split_alternating_runs(segments, "%s*,%s*") for _, comma_separated_group in ipairs(comma_separated_groups) do					if comma_separated_group[1] == "*" then -- reducible table.insert(base.short, {							base = {								form = "*",								footnotes = fetch_footnotes(comma_separated_group),							},							stem = {								form = "*",								footnotes = fetch_footnotes(comma_separated_group),							}						}) else local slash_separated_groups = iut.split_alternating_runs(comma_separated_group, "%s*/%s*") if #slash_separated_groups > 2 then error("Too many slash-separated stems: '" .. inside .. "'") end local short_base = slash_separated_groups[1] local short_stem = slash_separated_groups[2] local short_base_obj = { form = short_base[1], footnotes = fetch_footnotes(short_base), }						if short_stem then local short_stem_obj = { form = short_stem[1], footnotes = fetch_footnotes(short_stem), }						end table.insert(base.short, {							base = short_base_obj,							stem = short_stem_obj,						}) end iut.insert_form(forms, slot, formobj) end else error("Unrecognized indicator '" .. part .. "': '" .. inside .. "'") end end end return base end

local function normalize_all_lemmas(alternant_multiword_spec) iut.map_word_specs(alternant_multiword_spec, function(base)		base.orig_lemma = base.lemma		base.orig_lemma_no_links = m_links.remove_links(base.lemma)		base.lemma = base.orig_lemma_no_links	end) end

local function detect_indicator_spec(base) if base.short then if not base.lemma:find("ý$") then error("Short forms can only be specified for lemmas ending in -ý, but saw '" .. base.lemma .. "'") end local stem = rmatch(base.lemma, "^(.*)ý$") for _, short_spec in ipairs(base.short) do			if short_spec.base.form == "+" then short_spec.base.form = stem elseif short_spec.base.form == "*" then short_spec.base.form = com.dereduce(stem) if not short_spec.base.form then error("Unable to construct non-reduced variant of stem '" .. stem .. "'") end end if short_spec.stem.form == "+" or short_spec.stem.form == "*" then short_spec.stem.form = stem end end end

if base.irreg then base.decl = "irreg" else base.decl = "normal" end end

local function detect_all_indicator_specs(alternant_multiword_spec) iut.map_word_specs(alternant_multiword_spec, function(base)		detect_indicator_spec(base)	end) end

local function decline_adjective(base) if not decls[base.decl] then error("Internal error: Unrecognized declension type '" .. base.decl .. "'") end decls[base.decl](base) -- handle_derived_slots_and_overrides(base) end

-- Process override for the arguments in `args`, storing the results into `forms`. If `do_acc`, only do accusative -- slots; otherwise, don't do accusative slots. local function process_overrides(forms, args, do_acc) for slot, _ in pairs(input_adjective_slots) do		if args[slot] and not not do_acc == not not slot:find("^acc") then forms[slot] = nil if args[slot] ~= "-" and args[slot] ~= "—" then local segments = iut.parse_balanced_segment_run(args[slot], "[", "]") local comma_separated_groups = iut.split_alternating_runs(segments, "%s*,%s*") for _, comma_separated_group in ipairs(comma_separated_groups) do					local formobj = { form = comma_separated_group[1], footnotes = fetch_footnotes(comma_separated_group), }					iut.insert_form(forms, slot, formobj) end end end end end

local function check_allowed_overrides(alternant_multiword_spec, args) local special = alternant_multiword_spec.special or alternant_multiword_spec.surname and "surname" or "" for slot, types in pairs(input_adjective_slots) do		if args[slot] then local allowed = false for _, typ in ipairs(types) do				if typ == special then allowed = true break end end if not allowed then error(("Override %s= not allowed for %s"):format(slot, special == "" and "regular declension" or "special=" .. special)) end end end end

local function set_accusative(alternant_multiword_spec) local forms = alternant_multiword_spec.forms local function copy_if(from_slot, to_slot) if not forms[to_slot] then iut.insert_forms(forms, to_slot, forms[from_slot]) end end

copy_if("nom_n", "acc_n") copy_if("gen_m", "acc_m_an") copy_if("nom_m", "acc_m_in") copy_if("nom_fp", "acc_mfp") copy_if("nom_np", "acc_np") end

local function add_categories(alternant_multiword_spec) local cats = {} local function insert(cattype) table.insert(cats, "Czech " .. cattype .. " adjectives") end if not alternant_multiword_spec.manual then iut.map_word_specs(alternant_multiword_spec, function(base)			if base.decl == "poss" then				insert("possessive")			elseif rfind(base.lemma, "ци́?й$") then				insert("ц-stem")			elseif rfind(base.lemma, "ий$") then				insert("hard-stem stem-stressed")			elseif rfind(base.lemma, "и́й$") then				insert("hard-stem ending-stressed")			elseif rfind(base.lemma, "і́?й$") then				insert("soft-stem")			elseif rfind(base.lemma, "ї́?й$") then				insert("vowel-stem")			end		end) end alternant_multiword_spec.categories = cats end

local function show_forms(alternant_multiword_spec) local lemmas = {} local lemmaform = alternant_multiword_spec.forms.nom_m or alternant_multiword_spec.forms.nom_mp or		alternant_multiword_spec.forms.nom_mp_an if lemmaform then for _, form in ipairs(lemmaform) do			table.insert(lemmas, form.form) end end local props = { lemmas = lemmas, slot_table = get_output_adjective_slots(alternant_multiword_spec), lang = lang, }	iut.show_forms(alternant_multiword_spec.forms, props) end

local function make_table(alternant_multiword_spec) local forms = alternant_multiword_spec.forms

local function template_prelude(min_width) return rsub([===[ {title}{annotation} {\op}| border="1px solid #000000" style="border-collapse:collapse;background:#F9F9F9;text-align:center; min-width:MINWIDTHem" class="inflection-table" ]===], "MINWIDTH", min_width) end

local function template_postlude return [=[ end
 * {\cl}{notes_clause}  ]=]

local table_spec_sg = [=[ ! style="background:#d9ebff" colspan=5 | singular ! style="background:#d9ebff" | ! style="background:#d9ebff" | masculine animate ! style="background:#d9ebff" | masculine inanimate ! style="background:#d9ebff" | neuter ! style="background:#d9ebff" | feminine ! style="background:#eff7ff" | nominative ! style="background:#eff7ff" | genitive ! style="background:#eff7ff" | dative ! style="background:#eff7ff" | accusative ! style="background:#eff7ff" | locative ! style="background:#eff7ff" | instrumental ]=]
 * colspan=2 | {nom_m}
 * {nom_n}
 * {nom_f}
 * colspan=3 | {gen_m}
 * {gen_f}
 * colspan=3 | {dat_m}
 * {dat_f}
 * {acc_m_an}
 * {acc_m_in}
 * {acc_n}
 * {acc_f}
 * colspan=3 | {loc_m}
 * {loc_f}
 * colspan=3 | {ins_m}
 * {ins_f}{short_sg_clause}

local table_spec_pl = [=[ ! style="background:#d9ebff" colspan=5 | plural ! style="background:#d9ebff" | ! style="background:#d9ebff" | masculine animate ! style="background:#d9ebff" | masculine inanimate ! style="background:#d9ebff" | neuter ! style="background:#d9ebff" | feminine ! style="background:#eff7ff" | nominative ! style="background:#eff7ff" | genitive ! style="background:#eff7ff" | dative ! style="background:#eff7ff" | accusative ! style="background:#eff7ff" | locative ! style="background:#eff7ff" | instrumental ]=]
 * {nom_mp_an}
 * colspan=2 | {nom_fp}
 * {nom_np}
 * colspan=4 | {gen_p}
 * colspan=4 | {dat_p}
 * colspan=3 | {acc_mfp}
 * {acc_np}
 * colspan=4 | {loc_p}
 * colspan=4 | {ins_p}{short_pl_clause}

local table_spec = template_prelude("55") .. table_spec_sg .. "|-\n" .. table_spec_pl .. template_postlude

local table_spec_plonly = template_prelude("55") .. table_spec_pl .. template_postlude

local table_spec_dva = template_prelude("40") .. [=[ ! style="width:40%;background:#d9ebff" colspan="2" | ! style="background:#d9ebff" colspan="2" | plural ! style="width:40%;background:#d9ebff" colspan="2" | ! style="background:#d9ebff" | masculine ! style="background:#d9ebff" | feminine/neuter ! style="background:#eff7ff" colspan="2" | nominative ! style="background:#eff7ff" colspan="2" | genitive ! style="background:#eff7ff" colspan="2" | dative ! style="background:#eff7ff" colspan="2" | accusative ! style="background:#eff7ff" colspan="2" | locative ! style="background:#eff7ff" colspan="2" | instrumental ]=] .. template_postlude
 * {nom_mp}
 * {nom_fnp}
 * colspan="2" | {gen_p}
 * colspan="2" | {dat_p}
 * {acc_mp}
 * {acc_fnp}
 * colspan="2" | {loc_p}
 * colspan="2" | {ins_p}

local short_sg_template = [=[

! style="background:#eff7ff" | short
 * colspan=2 | {short_m}
 * {short_n}
 * {short_f}]=]

local short_pl_template = [=[

! style="background:#eff7ff" | short
 * {short_mp_an}
 * colspan=2 | {short_fp}
 * {short_np}]=]

local notes_template = [===[ {footnote} ]===]

if alternant_multiword_spec.title then forms.title = alternant_multiword_spec.title else forms.title = 'Declension of ' .. forms.lemma .. '' end

if alternant_multiword_spec.manual then forms.annotation = "" else local ann_parts = {} local decls = {} iut.map_word_specs(alternant_multiword_spec, function(base)			if base.decl == "irreg" then				m_table.insertIfNot(decls, "irregular")			elseif rfind(base.lemma, "ý$") then				m_table.insertIfNot(decls, "hard")			else				m_table.insertIfNot(decls, "soft")			end		end) table.insert(ann_parts, table.concat(decls, " // ")) forms.annotation = " (" .. table.concat(ann_parts, ", ") .. ")" end

forms.notes_clause = forms.footnote ~= "" and m_string_utilities.format(notes_template, forms) or "" forms.short_sg_clause = forms.short_m and forms.short_m ~= "—" and m_string_utilities.format(short_sg_template, forms) or "" forms.short_pl_clause = forms.short_mp_an and forms.short_mp_an ~= "—" and m_string_utilities.format(short_pl_template, forms) or "" return m_string_utilities.format(		alternant_multiword_spec.special == "plonly" and table_spec_plonly or		alternant_multiword_spec.special == "dva" and table_spec_dva or		table_spec, forms	) end

-- Externally callable function to parse and decline an adjective given -- user-specified arguments. Return value is WORD_SPEC, an object where the -- declined forms are in `WORD_SPEC.forms` for each slot. If there are no values -- for a slot, the slot key will be missing. The value for a given slot is a -- list of objects {form=FORM, footnotes=FOOTNOTES}. function export.do_generate_forms(parent_args, pos, from_headword, def) local params = { [1] = {required = true, default = "си́ній"}, json = {type = "boolean"}, -- for use with bots title = {}, }	for slot, _ in pairs(input_adjective_slots) do		params[slot] = {} end

local args = require("Module:parameters").process(parent_args, params) local parse_props = { parse_indicator_spec = parse_indicator_spec, allow_default_indicator = true, }	local alternant_multiword_spec = iut.parse_inflected_text(args[1], parse_props) alternant_multiword_spec.title = args.title alternant_multiword_spec.forms = {} normalize_all_lemmas(alternant_multiword_spec) detect_all_indicator_specs(alternant_multiword_spec) check_allowed_overrides(alternant_multiword_spec, args) local inflect_props = { slot_table = get_output_adjective_slots(alternant_multiword_spec), inflect_word_spec = decline_adjective, }	iut.inflect_multiword_or_alternant_multiword_spec(alternant_multiword_spec, inflect_props) -- Do non-accusative overrides so they get copied to the accusative forms appropriately. process_overrides(alternant_multiword_spec.forms, args) set_accusative(alternant_multiword_spec) -- Do accusative overrides after copying the accusative forms. process_overrides(alternant_multiword_spec.forms, args, "do acc") add_categories(alternant_multiword_spec) if args.json and not from_headword then return require("Module:JSON").toJSON(alternant_multiword_spec) end return alternant_multiword_spec end

-- Externally callable function to parse and decline an adjective where all -- forms are given manually. Return value is WORD_SPEC, an object where the -- declined forms are in `WORD_SPEC.forms` for each slot. If there are no values -- for a slot, the slot key will be missing. The value for a given slot is a -- list of objects {form=FORM, footnotes=FOOTNOTES}. function export.do_generate_forms_manual(parent_args, pos, from_headword, def) local params = { special = {}, json = {type = "boolean"}, -- for use with bots title = {}, }	for slot, _ in pairs(input_adjective_slots) do		params[slot] = {} end

local args = require("Module:parameters").process(parent_args, params) local alternant_multiword_spec = { special = args.special, title = args.title, forms = {}, manual = true, }	check_allowed_overrides(alternant_multiword_spec, args) -- Do non-accusative overrides so they get copied to the accusative forms appropriately. process_overrides(alternant_multiword_spec.forms, args) set_accusative(alternant_multiword_spec) -- Do accusative overrides after copying the accusative forms. process_overrides(alternant_multiword_spec.forms, args, "do acc") add_categories(alternant_multiword_spec) if args.json and not from_headword then return require("Module:JSON").toJSON(alternant_multiword_spec) end return alternant_multiword_spec end

-- Entry point for. Template-callable function to parse and decline -- an adjective given user-specified arguments and generate a displayable table -- of the declined forms. function export.show(frame) local parent_args = frame:getParent.args local alternant_multiword_spec = export.do_generate_forms(parent_args) show_forms(alternant_multiword_spec) return make_table(alternant_multiword_spec) .. require("Module:utilities").format_categories(alternant_multiword_spec.categories, lang) end

-- Entry point for. Template-callable function to parse and -- decline an adjective given manually-specified inflections and generate a -- displayable table of the declined forms. function export.show_manual(frame) local parent_args = frame:getParent.args local alternant_multiword_spec = export.do_generate_forms_manual(parent_args) show_forms(alternant_multiword_spec) return make_table(alternant_multiword_spec) .. require("Module:utilities").format_categories(alternant_multiword_spec.categories, lang) end

return export