Module:User:Benwing2/it-verb

local export = {}

--[=[

Authorship: Ben Wing

]=]

--[=[

TERMINOLOGY:

-- "slot" = A particular combination of tense/mood/person/number/etc. Example slot names for verbs are "pres1p" (present indicative first-person plural), "impsub12s" (imperfect	 subjunctive first/second-person singular form) and "pp" (past participle). Each slot is filled with zero or more forms.

-- "form" = The conjugated Italian form representing the value of a given slot.

-- "lemma" = The dictionary form of a given Italian term. For Italian, always the infinitive. ]=]

--[=[

RULES FOR CONJUGATION:

1. Present indicative: -- Individual form overrides always take precedence. -- Otherwise, if presrow: if given, this specifies all six forms (1s, 2s, 3s, 1p, 2p, 3p). -- Otherwise: 1. The 1s form is generated from the explicitly specified pres1s principal part. 2. The 3p form is generated from the explicitly specified pres1s principal part by replacing -o with -ano (for -are verbs) or with -ono (for -ere/-ire verbs). 3. The 2s and 3s forms are generated as follows: a. If an explicit pres3s principal part is given using PRES1S^PRES3S, the 2s/3s stem is generated by removing final -a/-e. b. Otherwise, if an explicit stem was given using stem:, this serves as the 2s/3s stem. c. Otherwise, if the infinitive is root-stressed, the 2s/3s stem comes from the infinitive. This is used e.g.		 with togliere (tòlgo, tògli, tòglie, togliàmo, togliéte, tòlgono), where the stem of the 1s and 3p forms is different from the stem of the 2s and 3s forms. d. Otherwise, the pres1s stem is used. From this stem, the 2s form is generated by adding -i (possibly with some adjustments; two unstressed i's	 together are compressed to one, and -h- is inserted after the stem if the verb is an -are verb and the stem ends	  with -c or -g), and the 3s form is generated by adding -a (for -are verbs) or -e (for -ere/-ire verbs). 4. The 1p and 2p forms are generated from the infinitive stem (or explicit stem given by stem:) by adding -iàmo (1p form) and -àte/-éte/-ìte (2p form, -are/-ere/-ire verbs). We don't use the pres1s form because it might have a different stem (e.g. +isc verbs).

2. Present subjunctive: -- Individual form overrides always take precedence. -- Otherwise, if subrow: if given, this specifies all four forms (123s, 1p, 2p, 3p). -- Otherwise: 1. If sub: is given, this specifies the 123s form; otherwise the 123s form is generated from the pres1s form by  changing -o to -a (for -ere/-ire verbs), or to -i (for -are verbs). E.g. for venire, vèngo -> vènga; for potere, pòsso -> pòssa. sub: needs to be given for e.g. essere (pres1s sóno, sub1s sìa), sapere (pres1s sò* but sub1s sàppia), fare (pres1s fàccio:fò*[archaic or dialectal] but sub1s just fàccia), andare (similar to fare), dovere. 2. The 3p form is generated from the 123s form by adding -no. 3. The 1p form is copied from the pres1p form. 4. The 2p form is generated from the 1p form by replacing -mo with -te.

3. Imperative: -- Individual form overrides always take precedence. -- Otherwise, if improw: if given, this specifies 2s, 2p. -- Otherwise: 1. If imp: given, this specifies the 2s form; otherwise the 2s form is generated by copying the pres2s form (for -ere/-ire verbs) or by copying the pres3s form (for -are verbs). We use the present indicative 2s/3s forms to preserve vowel and other alternations in the root-stressed forms (for all -are verbs, for	 venire and tenere, for +isc verbs, etc.). 2. The 2p form is generated by copying the pres2p form. 3. The 3s, 1p, 3p forms are copied from the corresponding present subjunctive forms. ]=]

--[=[

EXAMPLES OF CONJUGATION:


 * Including archaic and literary forms:


 * Including archaic and poetic forms:


 * Third-person only verbs:


 * Defective verbs:


 * Multiword expressions:

]=]

--[=[

FIXME:

1. Fix inf_linked and lemma_linked to work like in Module:es-verb. (DONE) 2. Finish support for reflexive and pronominal verbs. (DONE) 3. Finish support for reflexive and pronominal imperatives. (DONE) 4. Finish support for negative imperatives. (DONE) 5. Fix handling of third-only verbs; require that irregular forms be specified in the first person. Remove existing half-implemented support for specifying principal parts in the third person. (DONE) 6. Support defective verbs specified using e.g. redire. Include categorization; but if row overrides or single overrides of all forms given, don't categorize as defective for that row. (DONE) 7. Fix handling of aux; snarf code from Module:de-verb to handle aux with multiword expressions. (DONE) 8. Add automatic support for common irregular verbs: essere, avere, andare, fare, dare, dire, venire, vedere, tenere, bere, etc. Should make combinations of these verbs with clitics, and multiword expressions with these verbs, easier to handle. (DONE) 9. Add support for calling from in Module:it-headword. (DONE) 10. Throw an error if forms missing accents are specified (perhaps except with some special signal, to make   it easier to port the old Module:it-conj). 11. Consider adding combined clitic tables like in Module:es-verb. 12. Consider adding automatic support for prefixed -fare verbs. (DONE) 13. Consider displaying irregular forms in a different color, as with the old Module:es-conj, or adding a triangle next to them, as with Module:ru-verb. 14. Consider supporting replace_reflexive_indicators. 15. Add late (post-clitic-addition) overrides. (DONE) 16. PRES/PAST/PP spec should be required to come first to avoid ambiguities. (DONE) 17. Add variant codes to avoid mismatching variants in the conditional -èbbe/-ébbe, -éttero vs. érono, etc. 18. If explicit fut: given, it should control the conditional as well. (DONE) 19. If present -, sub:- or imp:-, it should suppress the whole row in the absence of row or individual overrides. 20. 'ci ci vuole' should maybe -> 'a noi ci vuole' instead of 'ci vuole'. 21. Instead of * before, use ! after so that * before can be used for reconstructed terms. (DONE) 22. When handling built-in verbs, automatically add + after vowel in cases like comporre. (DONE; HANDLED AUTOMATICALLY) 23. When handling built-in verbs, make sure we correctly handle root-stressed infinitives. (DONE) 24. When handling built-in verbs, make sure explicit ! afterwards is handled as if automatically there when a   prefix is added, and make sure final accent is handled correct when a prefix is added. In both cases, verify both sfare and rifare, sdarsi and ridare. (DONE) 25. When handling built-in verbs, make sure we handle prefixes correctly w.r.t. negative imperatives. (DONE) 26. Support ref: in footnotes. (DONE) 27. Finish built-in -ire verbs. 28. Implement error("If past participle given as '-', auxiliary must be explicitly specified as '-'"). (DONE) 29. Make present participles default to enabled. (DONE) 30. Instead of a qualifier for syntactic gemination, use a superscripted symbol with a tooltip, as for. Do this automatically for multisyllabic terms ending in a stressed vowel, but don't do it if the verb ends up non-final, as in andare a letto. 31. darsi: imp2s is wrong; it generates dàtti and also dàtti (which is wrong) (DONE) 32. liquefare: forms like liquefà missing the indication of following syntactic gemination; it should be automatic in multisyllabic words. 33. ridare has ridò with a syntactic gemination footnote, but liquefà doesn't have it (see #32). It should be a   tooltip. 34. Add 'addnote[SLOT_SPEC]' to make tangere with disused 1s/2s/1p/2p easier to handle. (DONE) 35. Throw an error if comma seen in single form specs like 'imp:'. (DONE) 36. Overrides like phis:+[rare] of affarsi using built-in @ (from fare) don't properly pick up the irregular built-in forms. (DONE) 37. Overrides like phis:afféci[rare] of affarsi using built-in @ (from fare) get the prefix duplicated. (DONE) 38. Support negated addnote like 'addnote[-pres3s][rare]'. (DONE) 39. Support comma-separated addnote spec for the equivalent of alternation. (DONE) 40. Issue an error if addnote spec has no effect. (DONE) 41. Support verbs in -gliela like fargliela. (DONE) 42. Support /\ notation for optional root-stressed infinitive with ending first, for suadere and derivatives. (DONE) 43. Support verbs in -gli like mancargli qualche rotella. (DONE) 44. Correctly incorporate links in multiword expressions from Module:it-headword so the autolinking algorithm there works (e.g. for portare il cervello all'ammasso) and head=~... specs are correctly propagated. (DONE) 45. Support head= with the verb as part of a larger bracketed expression, e.g. for stare a vedere, . (DONE) 46. Expand addnote[] notation to support references (maybe needs no work)? 47. Support verbs in -glire (boglire, inorgoglire, saglire, etc.) correctly. (DONE) 48. Support. (DONE) --]=]

local lang = require("Module:languages").getByCode("it") local m_string_utilities = require("Module:string utilities") local m_links = require("Module:links") local m_table = require("Module:table") local iut = require("Module:inflection utilities") local com_module = "Module:it-common" local m_builtin

local force_cat = false -- set to true for debugging local check_for_red_links = false -- set to false for debugging

local u = mw.ustring.char local rfind = mw.ustring.find local rsubn = mw.ustring.gsub local rmatch = mw.ustring.match local rsplit = mw.text.split local ulower = mw.ustring.lower local uupper = mw.ustring.upper local usub = mw.ustring.sub local ulen = mw.ustring.len local unfd = mw.ustring.toNFD local unfc = mw.ustring.toNFC

-- 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

local GR = u(0x0300) local V = "[aeiou]" local NV = "[^aeiou]" local AV = "[àèéìòóù]" local MAV = "[aeiouàèéìòóù]" -- maybe-accented vowel local NMAV = "[^aeiouàèéìòóù]" -- not maybe-accented vowel

local PRESERVE_ACCENT = u(0xFFF0)

local full_person_number_list = {"1s", "2s", "3s", "1p", "2p", "3p"} local imp_person_number_list = {"2s", "3s", "1p", "2p", "3p"}

-- Used only in error messages concerning the principal part for a given row, to generate the English description -- (by concatenating the following person/number description to the overall description for the row, taken from -- the `.desc` element of the rowspec). local principal_part_person_number_desc = { ["1s"] = "first-singular ", ["12s"] = "first/second-singular ", ["123s"] = "first/second/third-singular ", ["2s"] = "second-singular ", [""] = "", }

export.all_verb_slots = { {"aux", "-"}, }

-- Used to create the accelerator entries in all_verb_slots. local person_number_tag_prefix = { ["1s"] = "1|s|", ["2s"] = "2|s|", ["3s"] = "3|s|", ["1p"] = "1|p|", ["2p"] = "2|p|", ["3p"] = "3|p|", [""] = "", -- used for non-finite forms such as the past participle }

substitutable_reflexive_pronoun = { ["si"] = { ["1s"] = "mi", ["2s"] = "ti", ["3s"] = "si", ["1p"] = "ci", ["2p"] = "vi", ["3p"] = "si", ["nf"] = "si", },	["se"] = { ["1s"] = "me", ["2s"] = "te", ["3s"] = "se", ["1p"] = "ce", ["2p"] = "ve", ["3p"] = "se", ["nf"] = "se", },	["si_no_ci"] = { ["1s"] = "mi", ["2s"] = "ti", ["3s"] = "si", ["1p"] = "", ["2p"] = "vi", ["3p"] = "si", ["nf"] = "si", },	["si_no_vi"] = { ["1s"] = "mi", ["2s"] = "ti", ["3s"] = "si", ["1p"] = "ci", ["2p"] = "", ["3p"] = "si", ["nf"] = "si", },	["space_no_ci"] = { ["1s"] = " ", ["2s"] = " ", ["3s"] = " ", ["1p"] = "", ["2p"] = " ", ["3p"] = " ", ["nf"] = " ", },	["space_no_vi"] = { ["1s"] = " ", ["2s"] = " ", ["3s"] = " ", ["1p"] = " ", ["2p"] = "", ["3p"] = " ", ["nf"] = " ", }, }

-- FIXME, the following broken and not yet used local reflexive_forms = { ["si"] = {"mi", "ti", "si", "ci", "vi", "si"}, ["suo"] = {"mìo", "tùo", "sùo", "nòstro", "vòstro", "sùo"}, ["sua"] = {"mìa", "tùa", "sùa", "nòstra", "vòstra", "sùa"}, ["suoi"] = {"mièi", "tuòi", "suòi", "nòstri", "vòstri", "suòi"}, ["sue"] = {"mìe", "tùe", "sùe", "nòstre", "vòstre", "sùe"}, }

local function remove_accents(form) return rsub(form, AV, function(v) return usub(unfd(v), 1, 1) end) end

-- Map a function over `forms` (a list of form objects of the form {form=FORM, footnotes=FOOTNOTES}) and modify the -- form objects in `forms` in-place. If `include_footnotes` is given, the function is called with two arguments -- (the existing form and footnotes) and should return two arguments, the new form and footnotes; otherwise, the -- function is called with one argument (the existing form) and should return the new form. -- -- WARNING: This is dangerous and should only be done near the end. local function map_side_effecting_forms(forms, fun, include_footnotes) for _, form in ipairs(forms) do if form.form ~= "?" then if include_footnotes then local new_form, new_footnotes = fun(form.form, form.footnotes) form.form = new_form form.footnotes = new_footnotes else local new_form = fun(form.form) form.form = new_form end end end end

-- Add links around words. If multiword_only, do it only in multiword forms. local function add_links(form, multiword_only) if form == "" or form == " " then return form end if not form:find("%[%[") then if rfind(form, "[%s%p]") then --optimization to avoid loading Module:headword on single-word forms local m_headword = require("Module:headword") if m_headword.head_is_multiword(form) then form = m_headword.add_multiword_links(form) end end if not multiword_only and not form:find("%[%[") then form = "" .. form .. "" end end return form end

-- Add suffix to form, removing the PRESERVE_ACCENT notation if present (as for datti of dare). local function add_suffix_to_form(form, suffix) if suffix ~= "" then return rsub(form, PRESERVE_ACCENT, "") .. suffix else return form end end

local function convert_accented_links_in_text(text) local need_preserve_accent_note = false local retval = rsub(text, "%[%[([^%[%]|]+)%]%]",		function(linktext)			if rfind(linktext, PRESERVE_ACCENT) then				linktext = rsub(linktext, PRESERVE_ACCENT, "")				need_preserve_accent_note = true			elseif rfind(linktext, "^.*" .. MAV .. ".*" .. AV .. "$") then				-- Final accented vowel with preceding vowel; keep accent.			else				local unaccented = remove_accents(linktext)				if unaccented == linktext then					-- keep linktext				else					linktext = unaccented .. "|" .. linktext				end			end			return "" .. linktext .. ""		end) return retval, need_preserve_accent_note end

-- Convert links around accented words to two-part links without extra accents. Remove PRESERVE_ACCENT characters, -- and if necessary, insert a footnote indicating that a written accent on a monosyllabic verb is preserved. local function convert_accented_links(alternant_multiword_spec) for slot, forms in pairs(alternant_multiword_spec.forms) do -- Side-effect the forms to save memory. map_side_effecting_forms(forms, function(form, footnotes)			local new_form, need_preserve_accent_note = convert_accented_links_in_text(form)			if need_preserve_accent_note then				footnotes = iut.combine_footnotes(footnotes, {"[with written accent on monosyllabic verb]"})			end			return new_form, footnotes		end, "include footnotes") end end

local function strip_spaces(text) return rsub(text, "^%s*(.-)%s*$", "%1") end

-- Like iut.split_alternating_runs but strips spaces from both ends of the odd-numbered elements (only in -- odd-numbered runs if preserve_splitchar is given). Effectively we leave alone the footnotes and splitchars -- themselves, but otherwise strip extraneous spaces. Spaces in the middle of an element are also left alone. local function split_alternating_runs_and_strip_spaces(segment_runs, splitchar, preserve_splitchar) local split_runs = iut.split_alternating_runs(segment_runs, splitchar, preserve_splitchar) local function strip_run(run) end for i, run in ipairs(split_runs) do		if not preserve_splitchar or i % 2 == 1 then for j, element in ipairs(run) do				if j % 2 == 1 then run[j] = strip_spaces(element) end end end end return split_runs end

local function check_not_null(base, form, spec, principal_part_desc) if not form then error(("Spec '%s' cannot be used because default %s cannot be derived from '%s'"):format( spec, principal_part_desc, base.lemma)) end end

-- Indicate whether to skip `slot` when conjugating. If `checking_defective` is given, we are checking for defective -- rows and only want certain user-specified indicators respected. local function skip_slot(base, slot, checking_defective) if not checking_defective then if base.props.nofinite and slot:find("[123]") then -- Skip all finite (1/2/3-person) slots. return true end if base.props.presonly and slot:find("[123]") and not slot:find("^pres") then -- Skip all finite (1/2/3-person) slots except the present indicative. return true end if base.props.no_root_stressed and (slot:find("[123]s") or slot:find("3p")) and (			slot:find("^pres") or slot:find("^sub") or slot:find("^imp[123]") or slot:find("^negimp")		) then -- Skip all 1s/2s/3s/3p slots in the present indicative and subjunctive and the imperative. -- Beware of impsub slots. return true end end if base.props.impers or base.props.thirdonly then -- Impersonal or third-person-only verb. if slot:find("[12]") and not slot:find("3") then -- Skip slots for 1/2 person that don't also reference 3rd person (hence we don't skip sub123s). return true end if slot:find("^imp[123]") or slot:find("^negimp") then -- Skip all imperative slots, including third-person ones. Beware of impsub slots. return true end end if base.props.impers and slot:find("3p") then -- Skip third plural slots for impersonal verbs. return true end return false end

local function erase_suppressed_slots(base) for _, slot_spec in ipairs(export.all_verb_slots) do		local slot, accel = unpack(slot_spec) if skip_slot(base, slot) then base.forms[slot] = nil end end end

local function substitute_reflexive_pronoun(text, persnum) return rsub(text, "<(.-)>", function(refl)		return substitutable_reflexive_pronoun[refl][persnum]	end) end

local function escape_reflexive_indicators(arg1) -- FIXME, broken and not yet used if not arg1:find("pron>") then return arg1 end local segments = iut.parse_balanced_segment_run(arg1, "<", ">") -- Loop over every other segment. The even-numbered segments are angle-bracket specs while -- the odd-numbered segments are the text between them. for i = 2, #segments - 1, 2 do		if segments[i] == " " then segments[i] = "⦃⦃pron⦄⦄" end end return table.concat(segments) end

local function undo_escape_form(form) -- FIXME, broken and not yet used return rsub(rsub(form, "⦃⦃", "<"), "⦄⦄", ">") end

local function remove_reflexive_indicators(form) -- FIXME, broken and not yet used return rsub(form, "⦃⦃.-⦄⦄", "") end

local function replace_reflexive_indicators(slot, form) -- FIXME, broken and not yet used if not form:find("⦃") then return form end error("Internal error: replace_reflexive_indicators not implemented yet") end

-- Add the `stem` to the `ending` for the given `slot` and apply any phonetic modifications. local function combine_stem_ending(base, slot, stem, ending) -- Add h after c/g in -are forms to preserve the sound. if base.conj_vowel == "à" and stem:find("[cg]$") and rfind(ending, "^[eèéiì]") then stem = stem .. "h" end

-- Handle boglire, inorgoglire, saglire, etc.	if base.conj_vowel == "ì" and stem:find("gl$") and not rfind(ending, "^[iì]") then stem = stem .. "i" end

-- Two unstressed i's coming together compress to one. if ending:find("^i") then stem = rsub(stem, "i$", "") end

-- Remove accents from stem if ending is accented. if rfind(ending, AV) then stem = remove_accents(stem) end

return stem .. ending end

local function add_forms(base, slot, stems, endings, allow_overrides) local function do_combine_stem_ending(stem, ending) return combine_stem_ending(base, slot, stem, ending) end iut.add_forms(base.forms, slot, stems, endings, do_combine_stem_ending) end

local function insert_forms(base, slot, forms) iut.insert_forms(base.forms, slot, forms) end

local function copy_forms(base, slot, forms) -- When copying forms, clone the form objects because in various later places, we side-effect existing form objects -- and don't want any shared objects. insert_forms(base, slot, mw.clone(forms)) end

local function general_list_form_contains_form(list, form, process_form) if not list then -- general_list_form_contains_form is used to determine whether the defaults in `list` contain a given form; if not, that -- row is irregular. If there are no defaults, any form is irregular. return false end for _, formobj in ipairs(list) do		local formobj_form = formobj.form if process_form then formobj_form = process_form(formobj_form) end if formobj_form == form then return true end end return false end

-- Parse a raw form that may be decorated with !, # or similar before it and * or ! after it. Return three values: -- the prespec, bare form and postspec. local function parse_decorated_form(form) return rmatch(form, "^([!#]*)(.-)([*!]*)$") end

-- Process the user-given specs in `specs` in order to generate verb forms, and insert the resulting forms into -- `destforms`[`slot`]. If the destination slot has no forms yet (i.e. it is nil), it will be first set to {}. -- Duplicate forms will not be inserted. -- -- `specs` is a list of objects of the form {form = SPEC, footnotes = FOOTNOTES}, where FOOTNOTES is a list of -- user-specified footnotes/qualifiers (given after the spec using brackets) or nil if no footnotes. `specs` can be -- nil, which is equivalent to specifying "+" to request the default. A given SPEC may be "-" to indicate that the -- corresponding forms are missing, or be decorated with preceding decorators like !! (= "elevated style"), -- ! (= "careful style") or # (= "traditional"), or with following decorators like ! (= preserve monosyllabic accent), -- * (= verb form triggers syntactic gemination of following consonant), ** (= verb form triggers optional syntactic -- gemination). -- -- Processing is done by calling `generate_forms` on each spec after stripping decorators. If spec is "+", that will be -- passed directly to `generate_forms`, but if spec is "-", `generate_forms` will not be called. The return value of -- `generate_forms` can be anything that is convertible to a list of {form = SPEC, footnotes = FOOTNOTES} objects using -- iut.convert_to_general_list_form. In other words, it can be a single string (a form), a single object of the form -- {form = FORM, footnotes = FOOTNOTES, ...}, or a list of either of these. local function process_specs(base, destforms, slot, specs, generate_forms) specs = specs or 	for _, spec in ipairs(specs) do		local decorated_form = spec.form -- Skip "-"; effectively, no forms get inserted into destforms[slot]. if decorated_form ~= "-" then local prespec, form, postspec =	parse_decorated_form(decorated_form) local forms = form == "?" and "?" or generate_forms(form) -- If `generate_forms` return nil, no forms get inserted into destforms[slot]. This happens e.g. when -- fut:- is given and no explicit conditional principal part is given. In that case, -- generate_default_conditional_principal_part fetches the future principal parts, which don't exist, -- so it returns nil, and the surrounding generate_principal_part_forms also returns nil. The effect is -- that the conditional principal part ends up nil and no conditional parts get inserted. if forms then forms = iut.convert_to_general_list_form(forms, spec.footnotes) if base.all_footnotes then forms = iut.convert_to_general_list_form(forms, base.all_footnotes) end for _, formobj in ipairs(forms) do					local qualifiers = formobj.footnotes local form = formobj.form if prespec:find("!!") then qualifiers = iut.combine_footnotes({"[elevated style]"}, qualifiers) prespec = rsub(prespec, "!!", "") end if prespec:find("!") then qualifiers = iut.combine_footnotes({"[careful style]"}, qualifiers) prespec = rsub(prespec, "!", "") end if prespec:find("#") then qualifiers = iut.combine_footnotes({"[traditional]"}, qualifiers) prespec = rsub(prespec, "#", "") end local preserve_monosyllabic_accent if postspec:find("!") then preserve_monosyllabic_accent = true postspec = rsub(postspec, "!", "") end if preserve_monosyllabic_accent and rfind(form, "^" .. NMAV .. "*" .. AV .. "$") then -- final accented vowel without preceding vowel, and "!" after form; add PRESERVE_ACCENT form = PRESERVE_ACCENT .. form end if postspec == "*" then qualifiers = iut.combine_footnotes(qualifiers, {"[with syntactic gemination after the verb]"}) elseif postspec == "**" then qualifiers = iut.combine_footnotes(qualifiers, {"[with optional syntactic gemination after the verb]"}) elseif postspec ~= "" then error("Decorated form '" .. decorated_form .. "' has too many asterisks after it, use '*' for syntactic gemination and '**' for optional syntactic gemination") end iut.insert_form(destforms, slot, {form = form, footnotes = qualifiers}) end end end end end

local function set_up_base_verb(base) local ret = base.verb local raw_verb = ret.raw_verb

if rfind(raw_verb, "r$") then if rfind(raw_verb, "[ou]r$") or base.props.rre then ret.verb = raw_verb .. "re" else ret.verb = raw_verb .. "e" end else ret.verb = raw_verb end end

local function add_default_verb_forms(base) local ret = base.verb

-- Need to call combine_stem_ending to combine stem and ending rather than just pasting them together to handle -- cases like boglire, where 'bógl' + 'o' becomes 'bóglio'. local function comb(slot, stem, ending) return combine_stem_ending(base, slot, stem, ending) end

-- Process 'phisstem:...' spec. if base.principal_part_specs.explicit_phis_stem_spec then -- Put the explicit past historic stem in ret.phisstem (i.e. base.verb.phisstem). process_specs(base, ret, "phisstem", base.principal_part_specs.explicit_phis_stem_spec, iut.identity) end

ret.default_stem, ret.default_ending_vowel = rmatch(base.verb.verb, "^(.-)([aeir])re$") if not ret.default_stem then error("Unrecognized verb '" .. base.verb.verb .. "', doesn't end in -are, -ere, -ire or -rre") end base.props.syncopated = base.props.rre or ret.default_ending_vowel == "r"

-- Process 'stem:...' spec. local ending_vowel if base.principal_part_specs.explicit_stem_spec then local function generate_explicit_stem_forms(form) local stem, this_ending_vowel if form == "+" then stem = ret.default_stem this_ending_vowel = ret.default_ending_vowel if base.props.syncopated then error("Can't use + with 'stem:' in syncopated verbs; specify an explicit stem") end else base.explicit_non_default_stem_spec = true stem, this_ending_vowel = rmatch(form, "^(.*)([aeiàéì])$") if not stem then error("Unrecognized stem '" .. form .. "', should end in -a, -e, -i, -à, -é or -ì") end end if ending_vowel and ending_vowel ~= this_ending_vowel then error("Can't currently specify explicit stems with two different conjugation vowels (" .. ending_vowel .. " and " .. this_ending_vowel .. ")")			end ending_vowel = this_ending_vowel return stem end -- Put the explicit stem in ret.stem (i.e. base.verb.stem). process_specs(base, ret, "stem", base.principal_part_specs.explicit_stem_spec, generate_explicit_stem_forms) else if base.props.syncopated then error("With syncopated verb '" .. ret.verb .. "', must use 'stem:' to specify an explicit stem") end -- Convert to general list form so we can call iut.map_forms over it. ret.stem = iut.convert_to_general_list_form(ret.default_stem) ending_vowel = ret.default_ending_vowel end

base.conj_vowel = ending_vowel == "a" and "à" or		ending_vowel == "e" and "é" or		ending_vowel == "i" and "ì" or		ending_vowel

-- Process 'unstressed_stem:...' spec. if base.principal_part_specs.explicit_unstressed_stem_spec then local function generate_explicit_stem_forms(form) local stem if form == "+" then stem = ret.default_stem if base.props.syncopated then error("Can't use + with 'unstressed_stem:' in syncopated verbs; specify an explicit stem") end else base.explicit_non_default_unstressed_stem_spec = true stem = rmatch(form, "^(.*)" .. ret.default_ending_vowel .. "$") if not stem then error("Stem vowel of stem '" .. form .."' specified with 'unstressed_stem:' must end in conjugation vowel -" .. default_ending_vowel) end end return stem end -- Put the explicit stem in ret.unstressed_stem (i.e. base.verb.unstressed_stem). process_specs(base, ret, "unstressed_stem", base.principal_part_specs.explicit_unstressed_stem_spec,			generate_explicit_stem_forms) else ret.unstressed_stem = iut.map_forms(ret.stem, function(stem) return remove_accents(stem) end) end

ret.pres = iut.map_forms(ret.stem, function(stem) return comb("pres1s", stem, "o") end) ret.pres3s = iut.map_forms(ret.stem, function(stem) return comb("pres3s", stem, ending_vowel == "a" and "a" or "e") end) if ending_vowel == "i" then ret.isc_pres = iut.map_forms(ret.unstressed_stem, function(stem) return comb("pres1s", stem, "ìsco") end) ret.isc_pres3s = iut.map_forms(ret.unstressed_stem, function(stem) return comb("pres3s", stem, "ìsce") end) end ret.phis = iut.map_forms(ret.unstressed_stem, function(stem)		if ending_vowel == "a" then			return comb("phis1s", stem, "ài")		elseif ending_vowel == "e" then			-- Per Anna M. Thornton, "Overabundance: Multiple Forms Realizing the Same Cell" in Morphological Autonomy, p. 366			-- ,			-- -éi tends to occur only with stems ending in -t, while -étti/-ètti occurs with stems not ending in -t.			-- The opposite combinations are almost vanishingly rare, and should not be included.			if stem:find("t$") then				return comb("phis1s", stem, "éi")			else				return comb("phis1s", stem, "étti")			end		else			return comb("phis1s", stem, "ìi")		end	end) ret.pp = iut.map_forms(ret.unstressed_stem, function(stem)		if ending_vowel == "a" then			return comb("pp", stem, "àto")		elseif ending_vowel == "e" then			return comb("pp", stem, rfind(stem, "[cg]$") and "iùto" or "ùto")		else			return comb("pp", stem, "ìto")		end	end) end

local function is_single_vowel_spec(spec) return rfind(spec, "^" .. AV .. "[+-]?$") or rfind(spec, "^" .. AV .. "%-%-$") or rfind(spec, "^" .. AV .. "%+%+$") end

-- Given an unaccented stem, pull out the last two vowels as well as the in-between stuff, and return -- before, v1, between, v2, after as 5 return values. `unaccented` is the full verb and `unaccented_desc` -- a description of where the verb came from; used only in error messages. local function analyze_stem_for_last_two_vowels(unaccented_stem, unaccented, unaccented_desc) local before, v1, between, v2, after = rmatch(unaccented_stem, "^(.*)(" .. V .. ")(" .. NV .. "*)(" .. V .. ")(" .. NV .. "*)$")	if not before then before, v1 = "", "" between, v2, after = rmatch(unaccented_stem, "^(.*)(" .. V .. ")(" .. NV .. "*)$")	end if not between then error("No vowel in " .. unaccented_desc .. " '" .. unaccented .. "' to match") end return before, v1, between, v2, after end

-- Apply a single-vowel spec in `form`, e.g. é+, to `unaccented_stem`. `unaccented` is the full verb and -- `unaccented_desc` a description of where the verb came from; used only in error messages. local function apply_vowel_spec(unaccented_stem, unaccented, unaccented_desc, vowel_spec) local function vowel_spec_doesnt_match error("Vowel spec '" .. vowel_spec .. "' doesn't match vowel of " .. unaccented_desc .. " '" .. unaccented .. "'") end local raw_spec_vowel = usub(unfd(vowel_spec), 1, 1) local form local spec_vowel = rmatch(vowel_spec, "^(.)%-%-$") if spec_vowel then -- a spec like ò-- local before, v1, between1, v2, between2, v3, after = rmatch(unaccented_stem,			"^(.*)(" .. V .. ")(" .. NV .. "*)(" .. V .. ")(" .. NV .. "*)(" .. V .. ")(" .. NV .. "*)$")		if not before then error(mw.getContentLanguage:ucfirst(unaccented_desc) .. " '" .. unaccented ..				"' must have at least three vowels to use the vowel spec '" .. vowel_spec .. "'") end if raw_spec_vowel ~= v1 then vowel_spec_doesnt_match end form = before .. spec_vowel .. between1 .. v2 .. between2 .. v3 .. after else local before, v1, between, v2, after = analyze_stem_for_last_two_vowels(unaccented_stem, unaccented, unaccented_desc) if v1 == v2 then local first_second spec_vowel, first_second = rmatch(vowel_spec, "^(.)([+-]+)$") -- include ++ if not spec_vowel then error("Last two stem vowels of " .. unaccented_desc .. " '" .. unaccented ..					"' are the same; you must specify + (second vowel) or - (first vowel) after the vowel spec '" ..					vowel_spec .. "'") end if raw_spec_vowel ~= v1 then vowel_spec_doesnt_match end if first_second == "-" then form = before .. spec_vowel .. between .. v2 .. after else form = before .. v1 .. between .. spec_vowel .. after end else if rfind(vowel_spec, "%+%+$") then vowel_spec = rsub(vowel_spec, "%+%+$", "") elseif rfind(vowel_spec, "[+-]$") then error("Last two stem vowels of " .. unaccented_desc .. " '" .. unaccented ..					"' are different; specify just an accented vowel, without a following + or -: '" .. vowel_spec .. "'") end if raw_spec_vowel == v1 then form = before .. vowel_spec .. between .. v2 .. after elseif raw_spec_vowel == v2 then form = before .. v1 .. between .. vowel_spec .. after elseif before == "" then vowel_spec_doesnt_match else error("Vowel spec '" .. vowel_spec .. "' doesn't match either of the last two vowels of " .. unaccented_desc ..					" '" .. unaccented .. "'") end end end return form end

local function add_prefixed_reflexive_variant(base, slot, persnum) insert_forms(base, slot .. "_variant", iut.map_forms(base.forms[slot], function(form) return substitute_reflexive_pronoun(base.verb.finite_pref, persnum) .. "... " .. form .. "" end)) end

local function add_non_finite_prefixed_reflexive_variants(base, rowslot) add_prefixed_reflexive_variant(base, rowslot, "nf") end

local function add_finite_reflexive_clitics(base, rowslot) for _, persnum in ipairs(full_person_number_list) do base.forms[rowslot .. persnum] = iut.map_forms(base.forms[rowslot .. persnum], function(form)			return substitute_reflexive_pronoun(base.verb.finite_pref, persnum) .. "" .. form .. ""		end) end end

local function do_ending_stressed_inf(base) if rfind(base.verb.verb, "rre$") then error("Use a backslash (\\) with -rre verbs") end -- Add acute accent to -ere, grave accent to -are/-ire. local accented = rsub(base.verb.verb, "ere$", "ére") accented = unfc(rsub(accented, "([ai])re$", "%1" .. GR .. "re")) iut.insert_form(base.forms, "inf", {form = accented}) end

local function do_root_stressed_inf(base, specs) local function generate_root_stressed_inf_forms(base, spec, form_to_do, from_defaulted_pres) if spec == "-" then error("Spec '-' not allowed as root-stressed infinitive spec") end if spec == "+" then if from_defaulted_pres then error("Can't use + for present tense with root-stressed infinitive, would trigger infinite loop") end local rre_vowel = rmatch(base.verb.verb, "([aiu])rre$") if rre_vowel then -- do_root_stressed_inf is used for verbs in -ere and -rre. If the root-stressed vowel isn't explicitly -- given and the verb ends in -arre, -irre or -urre, derive it from the infinitive since there's only -- one possibility. If the verb ends in -erre or -orre, this won't work because we have both scérre -- (= scegliere) and disvèrre (= disvellere), as well as pórre and tòrre (= togliere). local before, v1, between, v2, after = analyze_stem_for_last_two_vowels(					rsub(base.verb.verb, "re$", ""), base.verb.verb, "root-stressed infinitive") local vowel_spec = unfc(rre_vowel .. GR) if v1 == v2 then vowel_spec = vowel_spec .. "+"				end spec = vowel_spec else -- Use the single-vowel spec(s) in the present tense principal part. local temp = {} process_specs(base, temp, "temp", base.principal_part_specs.pres, function(form)					return generate_root_stressed_inf_forms(base, form, form_to_do, "from defaulted pres") end) if not temp.temp then error("Unable to generate infinitive from present tense") end return temp.temp end end local verb_stem, verb_suffix = rmatch(base.verb.verb, "^(.-)([er]re)$") if not verb_stem then error("Verb '" .. base.verb.verb .. "' must end in -ere or -rre to use backslash (\\) notation") end if not is_single_vowel_spec(spec) then if from_defaulted_pres then error("When defaulting root-stressed infinitive vowel to present, present spec must be a single-vowel spec, but saw '"					.. spec .. "'") else error("Explicit root-stressed infinitive spec '" .. spec .. "' should be a single-vowel spec") end end

local expanded = apply_vowel_spec(verb_stem, base.verb.verb, "root-stressed infinitive", spec) if form_to_do == "stem" then return expanded else return expanded .. verb_suffix end end

process_specs(base, base.principal_part_forms, "root_stressed_stem", specs, function(form)		return generate_root_stressed_inf_forms(base, form, "stem") end) process_specs(base, base.forms, "inf", specs, function(form)		return generate_root_stressed_inf_forms(base, form, "inf") end) end

local function add_infinitive(base, rowslot) -- When do_root_stressed_inf is called, this also sets base.principal_part_forms.root_stressed_stem, which is needed -- by add_present_indic, so we have to do this before conjugating the present indicative. local function root do_root_stressed_inf(base, base.principal_part_specs.root_stressed_inf) end local function ending do_ending_stressed_inf(base) end if not base.principal_part_specs.root_stressed_inf then ending elseif base.props.opt_root_stressed_inf == "root-first" then root ending elseif base.props.opt_root_stressed_inf == "ending-first" then ending root else root end end

local function add_infinitive_reflexive_clitics(base, rowslot) base.forms[rowslot] = iut.map_forms(base.forms[rowslot], function(form)		local unaccented_form = remove_accents(form)		form = rsub(form, "r?re$", "r")		return "" .. form .. "" .. substitute_reflexive_pronoun(base.verb.linked_suf, "nf")	end) end

local function generate_pres_forms(base, form) local principal_part_desc = "first-singular present indicative" if form == "+" then check_not_null(base, base.verb.pres, form, principal_part_desc) return base.verb.pres elseif form == "+isc" then check_not_null(base, base.verb.isc_pres, form, principal_part_desc) return base.verb.isc_pres elseif is_single_vowel_spec(form) then check_not_null(base, base.verb.pres, form, principal_part_desc) return iut.map_forms(base.verb.pres, function(defform)			defform = remove_accents(defform) -- in case we specified 'stem:...' with an accent			local pres = rmatch(defform, "^(.*)o$")			if not pres then				error("Internal error: Default present '" .. defform .. "' doesn't end in -o")			end			return apply_vowel_spec(pres, defform, "default present", form) .. "o"		end) elseif not rfind(form, "[oò]$") then error("Present first-person singular form '" .. form .. "' should end in -o") else local unaccented_form = remove_accents(form) if not general_list_form_contains_form(base.verb.pres, unaccented_form, remove_accents) and (base.verb.isc_pres and not general_list_form_contains_form(base.verb.isc_pres, unaccented_form, remove_accents)) then base.rowprops.irreg.pres = true -- FIXME! Here we are encoding knowledge of the algorithm in add_present_indic to determine how to -- propagate irregular present 1s to other forms. This duplicates the logic of that algorithm, and if that -- code ever changes, this code needs to change too. In practice, it doesn't currently matter so much -- because the values in `is_irreg` are currently used only by the code in Module:it-headword to -- determine whether to show irregular principal parts, and the present tense is always shown in any case. -- But in the future, the values in `is_irreg` could be used for other purposes. base.is_irreg.pres1s = true base.is_irreg.pres3p = true if not base.principal_part_forms.pres3s and not base.explicit_non_default_stem_spec and not base.principal_part_forms.root_stressed_stem then base.is_irreg.pres2s = true base.is_irreg.pres3s = true end end return form end end

local function generate_pres3s_forms(base, form) local principal_part_desc = "third-singular present indicative" if form == "+" then check_not_null(base, base.verb.pres3s, form, principal_part_desc) return base.verb.pres3s elseif form == "+isc" then check_not_null(base, base.verb.isc_pres3s, form, principal_part_desc) return base.verb.isc_pres3s elseif is_single_vowel_spec(form) then check_not_null(base, base.verb.pres3s, form, principal_part_desc) return iut.map_forms(base.verb.pres3s, function(defform)			defform = remove_accents(defform) -- in case we specified 'stem:...' with an accent			local pres3s, final_vowel = rmatch(defform, "^(.*)([ae])$")			if not pres3s then				error("Internal error: Default third-person singular present '" .. defform .. "' doesn't end in -a or -e")			end			return apply_vowel_spec(pres3s, defform, "default third-person singular present", form) .. final_vowel		end) elseif not rfind(form, "[aàeè]") then error("Present third-person singular form '" .. form .. "' should end in -a or -e") else local unaccented_form = remove_accents(form) if not general_list_form_contains_form(base.verb.pres3s, unaccented_form, remove_accents) and (base.verb.isc_pres3s and not general_list_form_contains_form(base.verb.isc_pres3s, unaccented_form, remove_accents)) then base.rowprops.irreg.pres = true base.is_irreg.pres3s = true -- pres3s is copied to pres2s. base.is_irreg.pres2s = true end return form end end

-- Generate the present indicative. See "RULES FOR CONJUGATION" near the top of the file for the detailed rules. local function add_present_indic(base, rowslot) process_specs(base, base.principal_part_forms, "pres", base.principal_part_specs.pres, function(form)		return generate_pres_forms(base, form) end) if base.principal_part_specs.pres3s then process_specs(base, base.principal_part_forms, "pres3s", base.principal_part_specs.pres3s, function(form)			return generate_pres3s_forms(base, form) end) end

-- If no present indicative principal parts (user specified 'pres:-'), don't generate any present indicative forms. -- Otherwise we will end up generating pres23s and pres12p forms based on the overall verb stem(s). if not base.principal_part_forms.pres and not base.principal_part_forms.pres3s then return end

local function add(pers, stems, endings) add_forms(base, rowslot .. pers, stems, endings) end

add("1s", base.principal_part_forms.pres, "") local pres1s_stem = iut.map_forms(base.principal_part_forms.pres, function(form)		if not form:find("o$") then			error("presrow: must be given in order to generate the present indicative because explicit first-person " .. "singular present indicative '" .. form .. "' does not end in -o")		end		return rsub(form, "o$", "")	end) add("3p", pres1s_stem, base.conj_vowel == "à" and "ano" or "ono") local pres23s_stem if base.principal_part_forms.pres3s then pres23s_stem = iut.map_forms(base.principal_part_forms.pres3s, function(form)			if not form:find("[ae]$") then				error("presrow: must be given in order to generate the present indicative because explicit third-person " .. "singular present indicative '" .. form .. "' does not end in -a or -e")			end			return rsub(form, "[ae]$", "")		end) else pres23s_stem = base.explicit_non_default_stem_spec and base.verb.stem or base.principal_part_forms.root_stressed_stem or pres1s_stem end add("2s", pres23s_stem, "i") add("3s", pres23s_stem, base.conj_vowel == "à" and "a" or "e") add("1p", base.verb.unstressed_stem, "iàmo") add("2p", base.verb.unstressed_stem, base.conj_vowel .. "te") end

local function generate_default_present_subj_principal_part(base, do_err) return iut.flatmap_forms(base.forms.pres1s, function(form)		if not form:find("o$") then			if do_err then				error("sub: or subrow: must be given in order to generate the singular present subjunctive " .. "because first-person singular present indicative '" .. form .. "' does not end in -o")			else				return {}			end		else			-- Need to call combine_stem_ending here to handle verbs in -care/-gare and -ciare/-giare.			return {combine_stem_ending(base, "sub123s", rsub(form, "o$", ""), base.conj_vowel == "à" and "i" or "a")}		end	end) end

-- Generate the present subjunctive. See "RULES FOR CONJUGATION" near the top of the file for the detailed rules. local function add_present_subj(base, rowslot) -- If no present subjunctive principal parts (user specified 'sub:-'), don't generate any present subjunctive forms. -- Otherwise we will end up generating sub12p forms based on the present indicative. if not base.principal_part_forms.sub then return end

local function add(pers, stems, endings) add_forms(base, rowslot .. pers, stems, endings) end local function ins(pers, forms) insert_forms(base, rowslot .. pers, forms) end

-- Generate the 123s and 3p forms. add("123s", base.principal_part_forms.sub, "") add("3p", base.principal_part_forms.sub, "no") -- Copy present indicative 1p to present subjunctive. copy_forms(base, rowslot .. "1p", base.forms.pres1p) -- Generate present subjunctive 2p from present indicative 1p by replacing -mo with -te. ins("2p", iut.map_forms(base.forms.pres1p, function(form) if not form:find("mo$") then error("subrow: must be given in order to generate the second-person plural present subjunctive "				.. "because first-person plural present indicative '" .. form .. "' does not end in -mo") end return rsub(form, "mo$", "te") end)) end

local function generate_default_imperative_principal_part(base, do_err) if base.conj_vowel == "à" then -- Copy present indicative 3s to imperative 2s. return base.forms.pres3s else -- Copy present indicative 2s to imperative 2s. return base.forms.pres2s end end

-- Generate the imperative. See "RULES FOR CONJUGATION" near the top of the file for the detailed rules. local function add_imperative(base, rowslot) if not base.principal_part_forms.imp then -- If imp:- given, suppress the whole imperative. return end

local function copy(pers, forms) copy_forms(base, rowslot .. pers, forms) end

-- Copy first imperative form (user specified or taken from present indicative 3s for conj vowel à, or from	-- present indicative 2s for other conj vowels) to imperative 2s. copy("2s", base.principal_part_forms.imp) -- Copy present indicative 2p to imperative 2p. copy("2p", base.forms.pres2p) -- Copy present subjunctive 3s, 1p, 3p to imperative. copy("3s", base.forms.sub3s) copy("1p", base.forms.sub1p) copy("3p", base.forms.sub3p) end

local function get_unlinked_clitic_suffix(base, persnum) return m_links.remove_links(substitute_reflexive_pronoun(base.verb.linked_suf, persnum)) end

local function add_imperative_prefixed_reflexive_variants(base, rowslot) -- Don't include the 2s form, which will always have an attached enclitic reflexive pronoun (in the negative,	-- the infinitive is used). for _, persnum in ipairs({"3s", "1p", "2p", "3p"}) do add_prefixed_reflexive_variant(base, rowslot .. persnum, persnum) end end

local function add_imperative_reflexive_clitics(base, rowslot) local s2suf = get_unlinked_clitic_suffix(base, "2s") local saw_form_with_apostrophe = false

-- Check if there is a 2s imperative ending in an apostrophe, e.g. dà', fà'. If so, there is probably also an -- imperative in -ài, but we don't want to generate a reflexive imperative from it (#dàiti). Otherwise, we want to -- generative a reflexive imperative as normal (e.g. distràiti from distrarsi). local imp2s_forms = base.forms[rowslot .. "2s"] if imp2s_forms then for _, form in ipairs(imp2s_forms) do			if rfind(form.form, "'$") then saw_form_with_apostrophe = true break end end end

base.forms[rowslot .. "2s"] = iut.flatmap_forms(imp2s_forms, function(form)		form = rsub(form, "'$", "") -- dà', fà', etc.		if rfind(form, AV .. "$") then -- final stressed vowel; implement syntactic gemination			if rfind(s2suf, "^gli") then				return {add_suffix_to_form(form, s2suf)}			else				return {add_suffix_to_form(form, usub(s2suf, 1, 1) .. s2suf)}			end		elseif rfind(form, "ài$") and saw_form_with_apostrophe then			-- Skip this imperative; see above.			return {}		else			return {add_suffix_to_form(form, s2suf)}		end	end) -- For the following, we shouldn't need to use add_suffix_to_form, which handles PRESERVE_ACCENT, because -- PRESERVE_ACCENT occurs only with monosyllabic forms, and the 1p/2p forms are never monosyllabic. for _, persnum in ipairs({"1p", "2p"}) do		local suf = get_unlinked_clitic_suffix(base, persnum) base.forms[rowslot .. persnum] = iut.map_forms(base.forms[rowslot .. persnum], function(form)			return form .. suf		end) end for _, persnum in ipairs({"3s", "3p"}) do base.forms[rowslot .. persnum] = iut.map_forms(base.forms[rowslot .. persnum], function(form)			return substitute_reflexive_pronoun(base.verb.finite_pref, persnum) .. "" .. form .. ""		end) end end

local function add_negative_imperative(base) if not base.principal_part_forms.imp then -- If imp:- given, suppress the whole imperative. return end for _, persnum in ipairs({"2s", "3s", "1p", "2p", "3p"}) do local from = persnum == "2s" and "inf" or "imp" .. persnum insert_forms(base, "negimp" .. persnum, iut.map_forms(base.forms[from], function(form) return "non " .. form .. "" end)) end end

local function add_negative_imperative_reflexive_clitics(base, rowslot) for _, persnum in ipairs({"2s", "1p", "2p"}) do		local suf = get_unlinked_clitic_suffix(base, persnum) local pref = substitute_reflexive_pronoun(base.verb.finite_pref, persnum) base.forms[rowslot .. persnum] = iut.flatmap_forms(base.forms[rowslot .. persnum], function(form)			local truncated = persnum == "2s" and rsub(form, "r?re%]%]$", "r") or rsub(form, "%]%]$", "")			local sufform = truncated .. suf .. "]]"			local prefform = rsub(form, "^(%[%[non%]%]) (.*)$", "%1 " .. pref .. " %2")			return {sufform, prefform}		end) end for _, persnum in ipairs({"3s", "3p"}) do		local pref = substitute_reflexive_pronoun(base.verb.finite_pref, persnum) base.forms[rowslot .. persnum] = iut.map_forms(base.forms[rowslot .. persnum], function(form)			return rsub(form, "^(%[%[non%]%]) (.*)$", "%1 " .. pref .. " %2")		end) end end

local function generate_default_past_historic_principal_part(base, do_err) if do_err then check_not_null(base, base.verb.phis, "+", "first-singular past historic") end return base.verb.phis end

local function add_past_historic(base, rowslot) if not base.principal_part_forms.phis then -- specified as "-" return end for _, form in ipairs(base.principal_part_forms.phis) do		local function add_phis(pref, s1, s2, s3, p1, p2, p3) local newform = form.footnotes and iut.convert_to_general_list_form(pref, form.footnotes) or pref local function add(pers, endings) add_forms(base, rowslot .. pers, newform, endings) end add("1s", s1) add("2s", s2) add("3s", s3) add("1p", p1) add("2p", p2) add("3p", p3) end while true do if form.form == "?" then add_phis("?", "?", "?", "?", "?", "?", "?") break end local pref = rmatch(form.form, "^(.*)ài$") if pref then add_phis(pref, "ài", "àsti", "ò", "àmmo", "àste", "àrono") break end pref = rmatch(form.form, "^(.*)éi$") if pref then add_phis(pref, "éi", "ésti", "é", "émmo", "éste", "érono") break end pref = rmatch(form.form, "^(.*)[èé]tti$") if pref then add_phis(pref, {{form = "étti"}, {form = "ètti", footnotes = {"[traditional]"}}},					"ésti", {{form = "étte"}, {form = "ètte", footnotes = {"[traditional]"}}},					"émmo", "éste", {{form = "éttero"}, {form = "èttero", footnotes = {"[traditional]"}}}) break end pref = rmatch(form.form, "^(.*)ìi$") if pref then add_phis(pref, "ìi", "ìsti", "ì", "ìmmo", "ìste", "ìrono") break end pref = rmatch(form.form, "^(.*)i$") if pref then add_phis(pref, "i", {}, "e", {}, {}, "ero") if base.verb.phisstem then add_phis(base.verb.phisstem, {}, "sti", {}, "mmo", "ste", {}) else add_phis(base.verb.unstressed_stem, {}, base.conj_vowel .. "sti", {}, base.conj_vowel .. "mmo",						base.conj_vowel .. "ste", {}) end break end error("phisrow: must be given in order to generate the past historic because explicit first-person "				.. "singular past historic '" .. form.form .. "' does not end in -i") end end end

local function generate_default_future_principal_part(base, do_err) -- For -are verbs, we may need to make some adjustments to form the future principal part. local function are_stem_to_future_principal_part(stem) local function comb(ending) return combine_stem_ending(base, "fut1s", stem, ending) end if stem:find("[cg]$") then return {comb("herò")} elseif stem:find("[cg]i$") then if not base.forms.pres1s or base.forms.pres1s[1].form == "?" then -- missing or unknown pres1s; future still can be generated return {rsub(stem, "i$", "erò")} else -- Verbs in -ciare/-giare with the accent on the final -ì in the present singular take future in -- -cier-/-gier- not -cer-/-ger-. Compare sciare "to ski", pres1s scìo, fut1s scierò vs. -- lasciare "to let", pres1s làscio, fut1s lascerò. The only way to make this distinction -- is to check the present singular, e.g. pres1s. return iut.map_forms(base.forms.pres1s, function(form)					if rfind(form, "ìo$") then						return comb("erò")					else						return rsub(stem, "i$", "erò")					end				end) end else return {comb("erò")} end end

if base.explicit_non_default_unstressed_stem_spec then -- If user gave 'unstressed_stem:', use it here. if base.conj_vowel == "à" then return iut.flatmap_forms(base.verb.unstressed_stem, are_stem_to_future_principal_part) else return iut.map_forms(base.verb.unstressed_stem, function(form)				return form .. base.verb.default_ending_vowel .. "rò"			end) end elseif base.conj_vowel == "à" then return are_stem_to_future_principal_part(rsub(base.verb.verb, "are$", "")) else return rsub(base.verb.verb, "e$", "ò") end end

local function generate_default_conditional_principal_part(base, do_err) -- If fut:- is given, base.principal_part_forms.fut will be nil, and the following will correctly return nil, -- so that there's no conditional. return iut.map_forms(base.principal_part_forms.fut, function(form)		local pref = rmatch(form, "^(.*)ò$")		if not pref then			-- This should have been caught earlier (when processing the future) and generated an error.			error(("Internal error: When generating conditional, saw principal part for future '%s' that does not end in -ò") :format(form))		end		return combine_stem_ending(base, "cond1s", pref, "èi")	end) end

local function add_participle_reflexive_clitics(base, rowslot) -- do nothing end

local function generate_default_past_participle_principal_part(base, do_err) if do_err then check_not_null(base, base.verb.pp, "+", "first-singular past historic") end return base.verb.pp end

local function generate_default_gerund_principal_part(base, do_err) return iut.map_forms(base.verb.unstressed_stem, function(stem)		-- Need to call combine_stem_ending to handle cases like boglire, where 'bógl' + 'èndo' becomes 'boglièndo'.		return combine_stem_ending(base, "ger", stem, (base.conj_vowel == "à" and "àndo" or "èndo"))	end) end

local function add_gerund_reflexive_clitics(base, rowslot) base.forms[rowslot] = iut.map_forms(base.forms[rowslot], function(form)		return form .. get_unlinked_clitic_suffix(base, "nf")	end) end

local function generate_default_present_participle_principal_part(base, do_err) return iut.map_forms(base.verb.unstressed_stem, function(stem)		-- Need to call combine_stem_ending to handle cases like boglire, where 'bógl' + 'ènte' becomes 'bogliènte'.		return combine_stem_ending(base, "presp", stem, (base.conj_vowel == "à" and "ànte" or "ènte"))	end) end

--[=[ Data on how to conjugate individual rows (i.e. tense/aspect combinations, such as present indicative or conditional).

The order listed here matters. It determines the order of generating row forms. The order must have 'inf' < 'pres' < 'sub' < 'imp' < 'negimp' because the present indicative uses the root_stressed_stem generated by add_infinitive; the present subjunctive uses generated forms from the present indicative; the imperative uses forms from the present subjunctive and present indicative; and the negative imperative uses forms from the infinitive and the imperative. Similarly we must have 'fut' < 'cond' because the conditional uses the future principal part.

The following specs are allowed:

-- `desc` must be present and is an all-lowercase English description of the row. It is used in error messages and in  generating categories of the form 'Italian verbs with irregular ROW' and 'Italian verbs with missing ROW'. -- `tag_suffix` must be present is a string containing the tags that are appended onto the person/number tags to form the accelerator spec. For example, the spec "pres|sub" means that the accelerator spec for the third singular present subjunctive will be "3|s|pres|sub". This accelerator spec is passed to  Module:inflection utilities, which in turn passes it to Module:links when generating the link(s) for the corresponding verb form(s). The spec ultimately gets processed by Module:accel to generate the definition line for nonexistent verb forms. (FIXME: Accelerator support is currently disabled for forms with non-final accents.  We need to change the code in Module:inflection utilities so it sets the correct target not containing the   non-final accent.) -- `persnums` must be present and specifies the possible person/number suffixes to add onto the row-level slot (e.g. "phis" for the past historic) to form the individual person/number-specific slot (e.g. "phis2s" for the  second-person singular past historic). -- `row_override_persnums`, if present, specifies the person/number suffixes that are specified by a row override. If omitted, `persnums` is used. -- `row_override_persnums_to_full_persnums`, if present, specifies a mapping from the person/number suffixes specified by a row override to the person/number/suffixes used for conjugating the row. This is used, for example, with the subjunctive and imperfect subjunctive, where the first element of the row override specifies (respectively) the 123s and 12s forms, which need to be copied (respectively) to the 1s/2s/3s and 1s/3s forms. If omitted, no such copying happens. It's still possible for the row override persnums to disagree with the overall persnums. This happens, for example, with the imperative, where the 'improw:' row override spec specifies only the 2s and 2p forms; the remaining forms (3s, 1p, 3p) are generated during conjugation by copying from other forms, and can't be overridden using a row override. (They can still be overridden using a single override such  as 'imp3s:...' or a late single override such as 'imp3s!:...'. -- `generate_default_principal_part`, if present, should be a function of two arguments, `base` and `do_err`, and   should return the principal part(s) for the row. The return value can be anything that is convertible to the   "general list form" of a slot's forms, i.e. it can return a string, an object   {form = FORM, footnotes = {FOOTNOTE, FOOTNOTE, ...}}, or a list of either. It must be present if `conjugate` is   a table, but may be missing if `conjugate` is a function, in which case the function needs to generate the   principal part itself or otherwise handle things differently. For example, the present indicative does not   specify a value for `generate_default_principal_part` because there are actually two principal parts for the   present tense (first and third singular), which are processed at the beginning of the present indicative `conjugate` function. Similarly, the infinitive does not specify a value for `generate_default_principal_part` because there is no principal part to speak of; the infinitive is generated directly from the lemma in combination with the slash or backslash that follows the auxiliary and (in the case of a root-stressed infinitive) the single-vowel spec following the backslash. If `do_err` is given to this function, the function may throw an error if it can't generate the principal part; otherwise it should return nil. -- `conjugate` is either a function to conjugate the row (of two arguments, `base` and `rowslot`), or a table containing the endings to add onto the principal part to conjugate the row. In the latter case, there should be  the same number of elements in the table as there are elements in `row_override_persnums` (if given) or   `persnums` (otherwise). -- `no_explicit_principal_part` (DOCUMENT ME) -- `no_row_overrides` (DOCUMENT ME) -- `no_single_overrides` (DOCUMENT ME) -- `add_reflexive_clitics` (DOCUMENT ME) -- `dont_check_defective_status` (DOCUMENT ME) ]=] local row_conjugations = { {"inf", { desc = "infinitive", tag_suffix = "inf", persnums = {""}, -- No generate_default_principal_part; handled specially in add_infinitive. conjugate = add_infinitive, no_explicit_principal_part = true, -- because handled specially using / or \ notation no_row_overrides = true, -- useless because there's only one form; use / or \ notation no_single_overrides = true, --useless because there's only one form; use / or \ notation add_reflexive_clitics = add_infinitive_reflexive_clitics, add_prefixed_reflexive_variants = add_non_finite_prefixed_reflexive_variants, }},	{"pres", { desc = "present indicative", tag_suffix = "pres|ind", persnums = full_person_number_list, -- No generate_default_principal_part; handled specially in add_present_indic because we actually have -- two principal parts for the present indicative ("pres" and "pres3s"). conjugate = add_present_indic, -- No setting for no_explicit_principal_part here because it would never be checked; we special-case 'pres:' -- overrides before checking no_explicit_principal_part. The reason for special-casing is because there are two -- principal parts involved, "pres" and "pres3s", and we allow both to be specified using the syntax -- 'pres:PRES^PRES3S'. add_reflexive_clitics = add_finite_reflexive_clitics, }},	{"sub", { desc = "present subjunctive", tag_suffix = "pres|sub", persnums = full_person_number_list, row_override_persnums = {"123s", "1p", "2p", "3p"}, row_override_persnums_to_full_persnums = {["123s"] = {"1s", "2s", "3s"}}, generate_default_principal_part = generate_default_present_subj_principal_part, conjugate = add_present_subj, add_reflexive_clitics = add_finite_reflexive_clitics, }},	{"imp", { desc = "imperative", tag_suffix = "imp", persnums = imp_person_number_list, row_override_persnums = {"2s", "2p"}, generate_default_principal_part = generate_default_imperative_principal_part, conjugate = add_imperative, add_reflexive_clitics = add_imperative_reflexive_clitics, add_prefixed_reflexive_variants = add_imperative_prefixed_reflexive_variants, }},	{"negimp", { desc = "negative imperative", tag_suffix = "-", persnums = imp_person_number_list, -- No generate_default_principal_part because all parts are copied from other parts. conjugate = add_negative_imperative, add_reflexive_clitics = add_negative_imperative_reflexive_clitics, no_explicit_principal_part = true, -- because all parts are copied from other parts no_row_overrides = true, -- not useful; use single overrides if really needed -- We don't want a category Category:Italian verbs with missing negative imperative; doesn't make -- sense as all parts are copied from elsewhere. dont_check_defective_status = true, }},	{"phis", { desc = "past historic", tag_suffix = "phis", persnums = full_person_number_list, generate_default_principal_part = generate_default_past_historic_principal_part, conjugate = add_past_historic, add_reflexive_clitics = add_finite_reflexive_clitics, -- Set to "builtin" because normally handled specially in PRES^PRES3S,PHIS,PP spec, but when a built-in verb -- is involved, we want a way of overriding the past historic (using 'phis:'). no_explicit_principal_part = "builtin", }},	{"imperf", { desc = "imperfect indicative", tag_suffix = "impf|ind", persnums = full_person_number_list, generate_default_principal_part = function(base) return iut.map_forms(base.verb.unstressed_stem,			function(stem) return combine_stem_ending(base, "imperf1s", stem, base.conj_vowel .. "vo") end) end, conjugate = {"o", "i", "a", "àmo", "àte", "ano"}, add_reflexive_clitics = add_finite_reflexive_clitics, }},	{"impsub", { desc = "imperfect subjunctive", tag_suffix = "impf|sub", persnums = full_person_number_list, row_override_persnums = {"12s", "3s", "1p", "2p", "3p"}, row_override_persnums_to_full_persnums = {["12s"] = {"1s", "2s"}}, generate_default_principal_part = function(base) return iut.map_forms(base.verb.unstressed_stem,			function(stem) return combine_stem_ending(base, "impsub12s", stem, base.conj_vowel .. "ssi") end) end, conjugate = {"ssi", "sse", "ssimo", "ste", "ssero"}, add_reflexive_clitics = add_finite_reflexive_clitics, }},	{"fut", { desc = "future", tag_suffix = "fut", persnums = full_person_number_list, generate_default_principal_part = generate_default_future_principal_part, conjugate = {"ò", "ài", "à", "émo", "éte", "ànno"}, add_reflexive_clitics = add_finite_reflexive_clitics, }},	{"cond", { desc = "conditional", tag_suffix = "cond", persnums = full_person_number_list, generate_default_principal_part = generate_default_conditional_principal_part, conjugate = {"èi", "ésti", {"èbbe", "ébbe"}, "émmo", "éste", {"èbbero", "ébbero"}}, add_reflexive_clitics = add_finite_reflexive_clitics, }},	{"pp", { desc = "past participle", tag_suffix = "past|part", persnums = {""}, generate_default_principal_part = generate_default_past_participle_principal_part, conjugate = {""}, add_reflexive_clitics = add_participle_reflexive_clitics, -- Set to "builtin" because normally handled specially in PRES^PRES3S,PHIS,PP spec, but when a built-in verb -- is involved, we want a way of overriding the past participle (using 'pp:'). no_explicit_principal_part = "builtin", no_row_overrides = true, -- useless because there's only one form; use the PRES^PRES3S,PHIS,PP or pp: spec no_single_overrides = true, --useless because there's only one form; use the PRES^PRES3S,PHIS,PP or pp: spec }},	{"ger", { desc = "gerund", tag_suffix = "ger", persnums = {""}, generate_default_principal_part = generate_default_gerund_principal_part, conjugate = {""}, add_reflexive_clitics = add_gerund_reflexive_clitics, add_prefixed_reflexive_variants = add_non_finite_prefixed_reflexive_variants, no_row_overrides = true, -- useless because there's only one form; use explicit principal part no_single_overrides = true, -- useless because there's only one form; use explicit principal part }},	{"presp", { desc = "present participle", tag_suffix = "pres|part", persnums = {""}, generate_default_principal_part = generate_default_present_participle_principal_part, conjugate = {""}, add_reflexive_clitics = add_participle_reflexive_clitics, no_row_overrides = true, -- useless because there's only one form; use explicit principal part no_single_overrides = true, -- useless because there's only one form; use explicit principal part -- Disable this; seems most verbs do have present participles -- not_defaulted = true, -- not defaulted, user has to request it explicitly dont_check_defective_status = true, -- this is frequently missing and doesn't indicate a defective verb }}, }

local row_conjugation_map = {}

for _, rowconj in ipairs(row_conjugations) do	local rowslot, rowspec = unpack(rowconj) row_conjugation_map[rowslot] = rowspec end

local overridable_slot_set = {} local late_overridable_slot_set = {}

-- Populate all_verb_slots and overridable_slot_set. for _, rowconj in ipairs(row_conjugations) do	local rowslot, rowspec = unpack(rowconj) for _, persnum in ipairs(rowspec.persnums) do		local persnum_tag = person_number_tag_prefix[persnum] local slot = rowslot .. persnum if rowspec.tag_suffix == "-" then table.insert(export.all_verb_slots, {slot, "-"}) else table.insert(export.all_verb_slots, {slot, persnum_tag .. rowspec.tag_suffix}) end if not rowspec.no_single_overrides then overridable_slot_set[slot] = true end -- For now, we allow all slots to be late-overridable. Maybe we will rethink this later. late_overridable_slot_set[slot] = true end end

-- Add extra slots for use with reflexive verbs and. local function add_reflexive_verb_form_of_slots for _, rowconj in ipairs(row_conjugations) do		local rowslot, rowspec = unpack(rowconj) for _, persnum in ipairs(rowspec.persnums) do local slot = rowslot .. persnum table.insert(export.all_verb_slots, {slot .. "_non_reflexive", "-"}) -- If the row has prefixed reflexive variant forms (which means the clitics normally get suffixed), add -- slots for those forms. Note that this means we may add slots for variant forms never populated (e.g.			-- imp2s); but this doesn't really matter, and it's not worth it to add special-casing for this. if rowspec.add_prefixed_reflexive_variants then table.insert(export.all_verb_slots, {slot .. "_variant", "-"}) end end end end

local function handle_row_overrides_for_row(base, rowslot) if base.row_override_specs[rowslot] then for persnum, specs in pairs(base.row_override_specs[rowslot]) do local slot = rowslot .. persnum local existing_generated_form = base.forms[slot] local function generate_row_override_forms(form) if form == "+" then if not existing_generated_form then error(("Default form '+' requested in row override '%srow:' for slot %s but no default-generated form available; " .. "typically this means the principal part was given as '-'")							:format(rowslot, slot)) end return existing_generated_form end -- Check whether the row override form is the same as the default; if not, it's an irregularity. if not general_list_form_contains_form(existing_generated_form, form) then -- Note that the row has an irregularity in it. base.rowprops.irreg[rowslot] = true -- Now note that the individual form is irregular. If the row override is for a combined form like -- 123s, we have to map that to the individual forms (1s, 2s, 3s). local rowspec = row_conjugation_map[rowslot] if not rowspec then error("Internal error: No row conjugation spec for " .. rowslot) end local row_override_persnums_map = rowspec.row_override_persnums_to_full_persnums if row_override_persnums_map and row_override_persnums_map[persnum] then -- Propagate individual irregularities to actual person/number forms, as for the present and -- imperfect subjunctive. for _, full_persnum in ipairs(row_override_persnums_map[persnum]) do base.is_irreg[rowslot .. full_persnum] = true end else base.is_irreg[rowslot .. persnum] = true end end return form end base.forms[slot] = nil -- erase existing form before generating override process_specs(base, base.forms, slot, specs, generate_row_override_forms) end end end

local function handle_single_overrides_for_row(base, override_spec, rowslot) local rowspec = row_conjugation_map[rowslot] if not rowspec then error("Internal error: No row conjugation spec for " .. rowslot) end

-- FIXME: We may need to rethink the handling of irregularity markers. If the user e.g. sets an irregular -- override using 'pres1p:' and then sets it back to regular using 'pres1p!:', it ends up irregular. for _, persnum in ipairs(rowspec.persnums) do local slot = rowslot .. persnum if base[override_spec][slot] then local existing_generated_form = base.forms[slot] local function generate_override_forms(form) if form == "+" then if not existing_generated_form then error(("Default form '+' requested in override for slot %s but no default-generated form available; " .. "typically this means the principal part was given as '-'")							:format(rowslot, slot)) end return existing_generated_form end -- Check whether the single override form is the same as the default; if not, it's an irregularity. if not general_list_form_contains_form(existing_generated_form, form) then base.rowprops.irreg[rowslot] = true base.is_irreg[slot] = true end return form end base.forms[slot] = nil -- erase existing form before generating override process_specs(base, base.forms, slot, base[override_spec][slot], generate_override_forms) end end end

local function conjugate_row(base, rowslot) local rowspec = row_conjugation_map[rowslot] if not rowspec then error("Internal error: Unrecognized row slot '" .. rowslot .. "'") end

-- Generate the principal part for this row now if it has an entry for `generate_default_principal_part`. if rowspec.generate_default_principal_part then local function generate_principal_part_forms(form) -- If form == "+", either the user did not specify a principal part override (like 'sub:') or gave the value as '+'. -- In this circumstance, and provided the user did not specify a row override (like 'subrow:'), we need the default -- principal part in order to conjugate the row, so throw an error if we can't generate it. (If the user gave a row			-- override, we may still need the default principal part if the row override contains '+', so we could check for			-- this and set 'do_err', but it seems simpler to rely on the check in `handle_row_overrides_for_row` that makes			-- sure that a default form is available when the user specifies '+' in a row override.) local do_err = form == "+" and not base.row_override_specs[rowslot] local default_principal_part = rowspec.generate_default_principal_part(base, do_err) if default_principal_part then -- There may be no default; e.g. if fut:- is given, the default conditional principal part is nil. -- process_specs calls convert_to_general_list_form on the output in any case and we need it in this form -- in order to call general_list_form_contains_form, so we may as well convert it now. default_principal_part = iut.convert_to_general_list_form(default_principal_part) end if form == "+" then return default_principal_part end -- Check whether the principal part is the same as the default; if not, the entire row is irregular. if not general_list_form_contains_form(default_principal_part, form) then base.rowprops.irreg[rowslot] = true for _, persnum in ipairs(rowspec.persnums) do base.is_irreg[rowslot .. persnum] = true end end return form end

local principal_part_specs = base.principal_part_specs[rowslot] or rowspec.not_defaulted and or process_specs(base, base.principal_part_forms, rowslot, principal_part_specs, generate_principal_part_forms) end

if type(rowspec.conjugate) == "table" then local persnums = rowspec.row_override_persnums or rowspec.persnums if #rowspec.conjugate ~= #persnums then error("Internal error: Expected " .. #persnums .. " elements for row slot '" .. rowslot				.. ", but saw " .. #rowspec.conjugate) end local stem = iut.map_forms(base.principal_part_forms[rowslot], function(form)			local principal_part_ending = rowspec.conjugate[1]			if type(principal_part_ending) ~= "string" then				error(("Internal error: First element of the `.conjugate` table of the rowspec for row '%s' is not "					.. "a single string; if this is needed, either use a conjugate function instead of a table, "					.. "generalize the code following this error message, or introduce an additional rowspec element "					.. "`principal_part_ending` containing a single string"):format(rowslot))			end			if not rfind(form, principal_part_ending .. "$") then				-- Generate the principal part description from the first person/number of the row (which should -- always be the principal part) + the overall row description.				local principal_part_desc = principal_part_person_number_desc[persnums[1]] .. rowspec.desc				error(rowslot .. "row: must be given in order to generate the " .. rowspec.desc .. " because" .. "explicit " .. principal_part_desc .. " '" .. form .. "' does not end in -" .. principal_part_ending)			end			return rsub(form, principal_part_ending .. "$", "")		end) for i, persnum in ipairs(persnums) do add_forms(base, rowslot .. persnum, stem, rowspec.conjugate[i]) end else rowspec.conjugate(base, rowslot) end

-- Now add any footnotes derived from principal part overrides of the form '+[footnote]' used in conjunction with -- built-in verbs. if base.principal_part_footnotes[rowslot] then for _, persnum in ipairs(rowspec.persnums) do local full_slot = rowslot .. persnum if base.forms[full_slot] then -- To save on memory, side-effect the existing forms. for _, formobj in ipairs(base.forms[full_slot]) do					formobj.footnotes = iut.combine_footnotes(formobj.footnotes, base.principal_part_footnotes[rowslot]) end end end end

handle_row_overrides_for_row(base, rowslot)

-- If there's a mapping from row override persnums to full persnums, copy the slots accordingly. if rowspec.row_override_persnums_to_full_persnums then for row_override_persnum, full_persnums in pairs(rowspec.row_override_persnums_to_full_persnums) do			for _, full_persnum in ipairs(full_persnums) do copy_forms(base, rowslot .. full_persnum, base.forms[rowslot .. row_override_persnum]) end end end

handle_single_overrides_for_row(base, "single_override_specs", rowslot) end

-- Process specs given by the user using 'addnote[SLOTSPEC][FOOTNOTE][FOOTNOTE][...]'. SLOTSPEC can be a Lua pattern, -- a comma-separated list of Lua patterns (any of which need to match), or a hyphen followed by one or more -- comma-separated patterns (which negates the sense of the matching). local function process_addnote_specs(base) for _, spec in ipairs(base.addnote_specs) do		for _, slot_spec in ipairs(spec.slot_specs) do			local negated = false local any_changed = false local orig_slot_spec = slot_spec if slot_spec:find("^%-") then negated = true slot_spec = usub(slot_spec, 2) end local single_specs = rsplit(slot_spec, ",") for slot, forms in pairs(base.forms) do				local matches for _, single_spec in ipairs(single_specs) do if rfind(slot, "^" .. slot_spec .. "$") then matches = true break end end if not negated and matches or negated and not matches and slot ~= "inf" and slot ~= "aux" then -- To save on memory, side-effect the existing forms. for _, form in ipairs(forms) do						form.footnotes = iut.combine_footnotes(form.footnotes, spec.footnotes) any_changed = true end end end if not any_changed then error(("addnote spec '%s' had no effect; correct it or remove it"):format(orig_slot_spec)) end end end end

local function check_for_defective_and_unknown_rows(base) for _, rowconj in ipairs(row_conjugations) do		local rowslot, rowspec = unpack(rowconj) if not rowspec.dont_check_defective_status then local row_not_entirely_unknown = false local row_not_entirely_missing = false for i, persnum in ipairs(rowspec.persnums) do local slot = rowslot .. persnum if base.forms[slot] then row_not_entirely_missing = true for _, form in ipairs(base.forms[slot]) do if form.form == "?" then base.rowprops.unknown[rowslot] = true else row_not_entirely_unknown = true end end elseif not skip_slot(base, slot, "checking defective") then base.rowprops.defective[rowslot] = true end end base.rowprops.all_unknown[rowslot] = not row_not_entirely_unknown base.rowprops.all_defective[rowslot] = not row_not_entirely_missing end end if not base.principal_part_specs.aux and not base.verb.is_reflexive then base.rowprops.defective.aux = true base.rowprops.all_defective.aux = true end if base.principal_part_specs.aux then local row_not_entirely_unknown = false for _, form in ipairs(base.principal_part_specs.aux) do if form.form == "?" then base.rowprops.unknown.aux = true else row_not_entirely_unknown = true end end base.rowprops.all_unknown.aux = not row_not_entirely_unknown end end

-- Any forms without links should get them now. Redundant ones will be stripped later. local function add_missing_links_to_forms(base) for slot, forms in pairs(base.forms) do		map_side_effecting_forms(forms, add_links) end end

local function remove_links_from_forms(base) -- Remove links from forms in case of noautolinkverb. for slot, forms in pairs(base.forms) do		map_side_effecting_forms(forms, m_links.remove_links) end end

local function conjugate_verb(base) add_default_verb_forms(base) for _, rowconj in ipairs(row_conjugations) do		local rowslot, rowspec = unpack(rowconj) conjugate_row(base, rowslot) end -- If any verb in alternant_multiword_spec is reflexive and we're being called from, we need to	-- copy the forms before adding the reflexive clitic(s) so we can properly handle non-reflexive forms of -- reflexive-only verbs. local need_extra_verb_form_of_slots = base.props.any_reflexive and base.source_module == "it-inflections" if need_extra_verb_form_of_slots then for _, rowconj in ipairs(row_conjugations) do			local rowslot, rowspec = unpack(rowconj) for _, persnum in ipairs(rowspec.persnums) do local slot = rowslot .. persnum copy_forms(base, slot .. "_non_reflexive", base.forms[slot]) end if rowspec.add_prefixed_reflexive_variants then rowspec.add_prefixed_reflexive_variants(base, rowslot) end end end if base.verb.linked_suf ~= "" then for _, rowconj in ipairs(row_conjugations) do			local rowslot, rowspec = unpack(rowconj) rowspec.add_reflexive_clitics(base, rowslot) end end erase_suppressed_slots(base) for _, rowconj in ipairs(row_conjugations) do		local rowslot, rowspec = unpack(rowconj) handle_single_overrides_for_row(base, "late_single_override_specs", rowslot) end process_addnote_specs(base) check_for_defective_and_unknown_rows(base) if base.args.noautolinkverb then remove_links_from_forms(base) else add_missing_links_to_forms(base) end end

local function analyze_verb(lemma) local is_pronominal = false local is_reflexive = false -- The particles that can go after a verb are: -- * la, le	-- * ne	-- * ci, vi (sometimes in the form ce, ve) -- * si (sometimes in the form se) -- * gli -- Observed combinations: --  * ce + la: avercela "to be angry (at someone)", farcela "to make it, to succeed", --             mettercela tutta "to put everything (into something)" --  * se + la: sbrigarsela "to deal with", bersela "to naively believe in", --             sentirsela "to have the courage to face (a difficult situation)", --             spassarsela "to live it up", svignarsela "to scurry away", --             squagliarsela "to vamoose, to clear off", cercarsela "to be looking for (trouble etc.)", --             contarsela "to have a distortedly positive self-image; to chat at length", --             dormirsela "to be fast asleep", filarsela "to slip away, to scram", --             giostrarsela "to get away with; to turn a situation to one's advantage", --             cavarsela "to get away with; to get out of (trouble); to make the best of; to manage (to do); to be good at", --             meritarsela "to get one's comeuppance", passarsela "to fare (well, badly)", --             rifarsela "to take revenge", sbirbarsela "to slide by (in life)", --             farsela/intendersela "to have a secret affair or relationship with", --             farsela addosso "to shit oneself", prendersela "to take offense at; to blame", --             prendersela comoda "to take one's time", sbrigarsela "to finish up; to get out of (a difficult situation)", --             tirarsela "to lord it over", godersela "to enjoy", vedersela "to see (something) through", --             vedersela brutta "to have a hard time with; to be in a bad situation", --             aversela "to pick on (someone)", battersela "to run away, to sneak away", --             darsela a gambe "to run away", fumarsela "to sneak away", --             giocarsela "to behave (a certain way); to strategize; to play" --  * se + ne: andarsene "to take leave", approfittarsene "to take advantage of", --             fottersene/strafottersene "to not give a fuck", --             fregarsene/strafregarsene "to not give a damn", --             guardarsene "to beware; to think twice", impiparsene "to not give a damn", --             morirsene "to fade away; to die a lingering death", ridersene "to laugh at; to not give a damn", --             ritornarsene "to return to", sbattersene/strabattersene "to not give a damn", --             infischiarsene "to not give a damn", stropicciarsene "to not give a damn", --             sbarazzarsene "to get rid of, to bump off", andarsene in acqua "to be diluted; to decay", --             nutrirsene "to feed oneself", curarsene "to take care of", --             intendersene "to be an expert (in)", tornarsene "to return, to go back", --             starsene "to stay", farsene "to matter; to (not) consider; to use", --             farsene una ragione "to resign; to give up; to come to terms with; to settle (a dispute)", --             riuscirsene "to repeat (something annoying)", venirsene "to arrive slowly; to leave" --  * ci + si: trovarcisi "to find oneself in a happy situation", --             vedercisi "to imagine oneself (in a situation)", sentircisi "to feel at ease" --  * vi + si: recarvisi "to go there" --  * glie + la: fargliela "to succeed" --	local ret = {} local linked_suf, finite_pref, finite_pref_elided_e, finite_pref_elided_ho local clitic_to_substitutable = {ce = "ce", ve = "ve", se = ""} local clitic_to_elided = { ci = "c'", vi = "vi ", si = "si ", lo = "l'", la = "l'", li = "li ", le = "le ", gli = "gli ", }	local verb, clitic, clitic2 = rmatch(lemma, "^(.-)([cvs]e)(l[oaie])$") if verb then is_pronominal = true is_reflexive = clitic == "se" clitic = clitic_to_substitutable[clitic] linked_suf = clitic .. "" .. clitic2 .. "" finite_pref = clitic .. " " .. clitic2 .. " " finite_pref_elided_e = clitic .. " " .. clitic_to_elided[clitic2] finite_pref_elided_ho = clitic .. " " .. clitic_to_elided[clitic2] end if not verb then verb, clitic = rmatch(lemma, "^(.-)glie(l[oaie])$") if verb then is_pronominal = true linked_suf = "glie" .. clitic .. "" finite_pref = linked_suf .. " "			finite_pref_elided_e = clitic_to_elided[clitic]:gsub("^%[%[", "glie")			finite_pref_elided_ho = finite_pref_elided_e		end	end	if not verb then		verb, clitic = rmatch(lemma, "^(.-)([cvs]e)ne$")		if verb then			is_pronominal = true			is_reflexive = clitic == "se"			clitic = clitic_to_substitutable[clitic]			linked_suf = clitic .. "[[ne"			finite_pref = clitic .. " ne "			finite_pref_elided_e = clitic .. " n'"			finite_pref_elided_ho = clitic .. " ne "		end	end	if not verb then		verb, clitic = rmatch(lemma, "^(.-)([cv]i)si$")		if verb then			is_pronominal = true			is_reflexive = true			local si_no_clitic, space_no_clitic			if clitic == "ci" then				si_no_clitic = ""				space_no_clitic = ""			else				si_no_clitic = ""				space_no_clitic = ""			end			linked_suf = "" .. clitic .. "" .. si_no_clitic			finite_pref = si_no_clitic .. space_no_clitic .. "" .. clitic .. " "			finite_pref_elided_e = si_no_clitic .. space_no_clitic .. clitic_to_elided[clitic]			finite_pref_elided_ho = finite_pref		end	end	if not verb then		verb, clitic = rmatch(lemma, "^(.-)([cv]i)$")		if verb then			is_pronominal = true			linked_suf = "" .. clitic .. ""			finite_pref = "" .. clitic .. " "			finite_pref_elided_e = clitic_to_elided[clitic]			finite_pref_elided_ho = finite_pref		end	end	if not verb then		verb = rmatch(lemma, "^(.-)si$")		if verb then			-- not pronominal			is_reflexive = true			linked_suf = ""			finite_pref = " "			finite_pref_elided_e = finite_pref			finite_pref_elided_ho = finite_pref		end	end	if not verb then		verb = rmatch(lemma, "^(.-)ne$")		if verb then			is_pronominal = true			linked_suf = "ne"			finite_pref = "ne "			finite_pref_elided_e = "n'"			finite_pref_elided_ho = "ne "		end	end	if not verb then		verb, clitic = rmatch(lemma, "^(.-)(gli)$")		if not verb then			verb, clitic = rmatch(lemma, "^(.-)(l[oaie])$")		end		if verb then			is_pronominal = true			linked_suf = "" .. clitic .. ""			finite_pref = "" .. clitic .. " "			finite_pref_elided_e = clitic_to_elided[clitic]			finite_pref_elided_ho = clitic_to_elided[clitic]		end	end	if not verb then		-- not pronominal		verb = lemma		linked_suf = ""		finite_pref = ""		finite_pref_elided_e = ""		finite_pref_elided_ho = ""	end

ret.raw_verb = verb ret.linked_suf = linked_suf ret.finite_pref = finite_pref ret.finite_pref_elided_e = finite_pref_elided_e ret.finite_pref_elided_ho = finite_pref_elided_ho ret.is_pronominal = is_pronominal ret.is_reflexive = is_reflexive return ret end

-- Subfunction of find_builtin_verb. Match a single spec (which may begin with ^ to anchor against the beginning, -- otherwise anchored only at the end) against `verb`. Return the prefix and main verb. local function match_spec_against_verb(spec, verb) if spec:find("^%^") then -- must match exactly if rfind(verb, spec .. "$") then return "", verb end else local prefix, main_verb = rmatch(verb, "^(.*)(" .. spec .. ")$")		if prefix then return prefix, main_verb end end end

-- Subfunction of find_builtin_verb. Match a single prefix + spec (where the prefix may begin with ^ to anchor -- against the beginning, otherwise anchored only at the end) against `verb`. Return the prefix and main verb. local function match_prefixed_spec_against_verb(specprefix, spec, verb) if specprefix:find("^%^") then -- must match exactly specprefix = specprefix:gsub("^%^", "") if specprefix == "" then -- We can't use the second branch of the if-else statement because an empty returns the current position -- in rmatch. local main_verb = rmatch(verb, "^(" .. spec .. ")$")			if main_verb then return "", main_verb end else local prefix, main_verb = rmatch(verb, "^(" .. specprefix .. ")(" .. spec .. ")$")			if prefix then return prefix, main_verb end end else local prefix, main_verb = rmatch(verb, "^(.*" .. specprefix .. ")(" .. spec .. ")$")		if prefix then return prefix, main_verb end end end

-- Find and return the prefix, main verb and conj spec for the built-in verb matching user-specified verb `verb`. local function find_builtin_verb(verb) if not m_builtin then m_builtin = require("Module:it-verb/builtin") end for _, builtin_verb in ipairs(m_builtin.builtin_verbs) do		local spec, conj, desc = unpack(builtin_verb)

if type(spec) == "string" then local prefix, main_verb = match_spec_against_verb(spec, verb) if prefix then return prefix, main_verb, conj end else -- Of the form {term = "ergere", prefixes = {"^", "ad", "ri"}}. Note that the prefixes not preceded by ^ -- can have further prefixes before them. for _, spec_prefix in ipairs(spec.prefixes) do				local prefix, main_verb = match_prefixed_spec_against_verb(spec_prefix, spec.term, verb) if prefix then return prefix, main_verb, conj end end end end end

-- Parse the "inside" of an angle bracket spec (e.g. "a/é"), storing the results into `base`. This is the actual -- function that parses indicator specs. It is separated from, and called from, parse_indicator_spec in order to -- deal with built-in verbs, which have their own indicator specs that must be combined with the indicator spec given -- by the user. `is_builtin_verb` is true if we're processing a built-in verb spec, as opposed to a user-specified one. local function parse_inside(base, inside, is_builtin_verb) local function parse_err(msg) error((is_builtin_verb and "Internal error processing built-in verb spec: " or "") .. msg			.. ": <" .. inside .. ">") end

local function parse_qualifiers(separated_group) local qualifiers for j = 2, #separated_group - 1, 2 do			if separated_group[j + 1] ~= "" then parse_err("Extraneous text after bracketed qualifiers: '" .. table.concat(separated_group) .. "'") end if not qualifiers then qualifiers = {} end local r_spec = separated_group[j]:match("^%[r:(.*)%]$") if r_spec then local expanded_ref = require(com_module).parse_abbreviated_references_spec(r_spec) table.insert(qualifiers, "[ref:" .. expanded_ref .. "]") else table.insert(qualifiers, separated_group[j]) end end return qualifiers end

local function fetch_specs(comma_separated_group, allow_blank) local colon_separated_groups = split_alternating_runs_and_strip_spaces(comma_separated_group, ":") if allow_blank and #colon_separated_groups == 1 and #colon_separated_groups[1] == 1 and colon_separated_groups[1][1] == "" then return nil end local specs = {} for _, colon_separated_group in ipairs(colon_separated_groups) do			local form = colon_separated_group[1] if form == "" then parse_err("Blank form not allowed here, but saw '" ..					table.concat(comma_separated_group) .. "'") end if form:find(",") then parse_err("Comma in form '" .. form .. "', did you mean to use a colon?") end local new_spec = {form = form, footnotes = parse_qualifiers(colon_separated_group)} for _, existing_spec in ipairs(specs) do				if m_table.deepEquals(existing_spec, new_spec) then parse_err("Duplicate spec '" .. table.concat(colon_separated_group) .. "'") end end table.insert(specs, new_spec) end return specs end

-- Parse present-tense spec of the form PRES^PRES3S or just PRES, and set the appropriate properties in `base`. -- Used in the PRES^PRES3S,PHIS,PP spec as well as with pres:PRES^PRES3S in conjunction with built-in verbs. local function parse_present_spec(run) local cflex_separated_groups = split_alternating_runs_and_strip_spaces(run, "%^") if #cflex_separated_groups > 2 then parse_err("At most one circumflex sign (^) can appear in present tense specs") end base.principal_part_specs.pres = fetch_specs(cflex_separated_groups[1]) if #cflex_separated_groups == 2 then base.principal_part_specs.pres3s = fetch_specs(cflex_separated_groups[2]) end end

local function is_root_stressed(separator) return separator == "\\" or separator == "\\/" or separator == "/\\" end local function get_opt_root_stressed_value(separator) return separator == "\\/" and "root-first" or separator == "/\\" and "ending-first" or nil end

local segments = iut.parse_balanced_segment_run(inside, "[", "]") local dot_separated_groups = split_alternating_runs_and_strip_spaces(segments, "%.") for i, dot_separated_group in ipairs(dot_separated_groups) do		local first_element = dot_separated_group[1]

if i == 1 then -- first dot-separated group is PRES,PHIS,PP or PRES^PRES3S,PHIS,PP or similar. local comma_separated_groups = split_alternating_runs_and_strip_spaces(dot_separated_group, "[,\\/]+",				"preserve splitchar") local presind = 1 local first_separator = #comma_separated_groups > 1 and comma_separated_groups[2][1] if base.verb.is_reflexive or is_builtin_verb then if #comma_separated_groups > 1 and first_separator ~= "," then presind = 3 -- Fetch root-stressed infinitive, if given. local specs = fetch_specs(comma_separated_groups[1], "allow blank") if is_root_stressed(first_separator) then -- For verbs like scegliersi and proporsi, allow either 'é\scélgo' or '\é\scélgo' -- and similarly either 'ó+\propóngo' or '\ó+\propóngo'. if specs == nil then if #comma_separated_groups > 3 and is_root_stressed(comma_separated_groups[4][1]) then base.principal_part_specs.root_stressed_inf = fetch_specs(comma_separated_groups[3]) presind = 5 else base.principal_part_specs.root_stressed_inf = end else base.principal_part_specs.root_stressed_inf = specs end base.props.opt_root_stressed_inf = get_opt_root_stressed_value(first_separator) elseif specs ~= nil then local errpref = is_builtin_verb and "With built-in verb" or "With reflexive verb" parse_err(errpref .. ", can't specify anything before initial slash, but saw '"							.. table.concat(comma_separated_groups[1])) end elseif not is_builtin_verb and rfind(base.verb.raw_verb, "er$") then parse_err("With reflexive -ere verb, must precede present spec with / or \\ to indicate whether infinitive is root-stressed") end if not is_builtin_verb then base.principal_part_specs.aux = end else -- non-reflexive if #comma_separated_groups == 1 or first_separator == "," then parse_err("With non-reflexive verb, use a spec like AUX/PRES, AUX\\PRES, AUX/PRES,PAST,PP or similar") end presind = 3 -- Fetch auxiliary or auxiliaries. local colon_separated_groups = split_alternating_runs_and_strip_spaces(comma_separated_groups[1], ":") for _, colon_separated_group in ipairs(colon_separated_groups) do					local aux = colon_separated_group[1] if aux == "a" then aux = "avére" elseif aux == "e" then aux = "èssere" elseif aux == "-" then if #colon_separated_group > 1 then parse_err("No footnotes allowed with '-' spec for auxiliary") end aux = nil elseif aux == "?" then -- remains as-is else parse_err("Unrecognized auxiliary '" .. aux ..							"', should be 'a' (for avere), 'e' (for essere), or '-' if no past participle") end if aux then if base.principal_part_specs.aux then for _, existing_aux in ipairs(base.principal_part_specs.aux) do								if existing_aux.form == aux then parse_err("Auxiliary '" .. aux .. "' specified twice") end end else base.principal_part_specs.aux = {} end table.insert(base.principal_part_specs.aux, {form = aux, footnotes = parse_qualifiers(colon_separated_group)}) end end

-- Fetch root-stressed infinitive, if given. if is_root_stressed(first_separator) then if #comma_separated_groups > 3 and is_root_stressed(comma_separated_groups[4][1]) then base.principal_part_specs.root_stressed_inf = fetch_specs(comma_separated_groups[3]) presind = 5 else base.principal_part_specs.root_stressed_inf = end base.props.opt_root_stressed_inf = get_opt_root_stressed_value(first_separator) end end

if #comma_separated_groups == presind and comma_separated_groups[presind][1] == "@" then -- We will find the conjugation for the built-in verb later, after we've seen whether there is an -- '.rre' property. base.props.builtin = true else -- Parse present parse_present_spec(comma_separated_groups[presind])

-- Parse past historic if #comma_separated_groups > presind then if comma_separated_groups[presind + 1][1] ~= "," then parse_err("Use a comma not slash to separate present from past historic") end base.principal_part_specs.phis = fetch_specs(comma_separated_groups[presind + 2]) end

-- Parse past participle if #comma_separated_groups > presind + 2 then if comma_separated_groups[presind + 3][1] ~= "," then parse_err("Use a comma not slash to separate past historic from past participle") end base.principal_part_specs.pp = fetch_specs(comma_separated_groups[presind + 4]) end

if #comma_separated_groups > presind + 4 then parse_err("Extraneous text after past participle") end end elseif first_element == "impers" or first_element == "thirdonly" or first_element == "rre" or			first_element == "nofinite" or first_element == "no_root_stressed" or first_element == "presonly" then if #dot_separated_group > 1 then parse_err("No footnotes allowed with '" .. first_element .. "' spec") end base.props[first_element] = true elseif first_element == "addnote" then local spec_and_footnotes = parse_qualifiers(dot_separated_group) if #spec_and_footnotes < 2 then parse_err("Spec with 'addnote' should be of the form 'addnote[SLOTSPEC][FOOTNOTE][FOOTNOTE][...]'") end local slot_spec = table.remove(spec_and_footnotes, 1) local slot_spec_inside = rmatch(slot_spec, "^%[(.*)%]$") if not slot_spec_inside then parse_err("Internal error: slot_spec " .. slot_spec .. " should be surrounded with brackets") end local slot_specs = rsplit(slot_spec_inside, ",") for j, spec in ipairs(slot_specs) do				slot_specs[j] = strip_spaces(spec) end table.insert(base.addnote_specs, {slot_specs = slot_specs, footnotes = spec_and_footnotes}) else local first_element_prefix, first_element_minus_prefix = rmatch(first_element,				"^%s*([a-z0-9_!]+)%s*:%s*(.-)%s*$") if not first_element_prefix then parse_err("Dot-separated element should be either 'impers', 'thirdonly', 'nofinite', 'no_root_stressed', 'presonly', 'rre' or be of the form "					.. "'PREFIX:SPEC', but saw '" .. table.concat(dot_separated_group) .. "'") end dot_separated_group[1] = first_element_minus_prefix if first_element_prefix == "stem" then base.principal_part_specs.explicit_stem_spec = fetch_specs(dot_separated_group) elseif first_element_prefix == "phisstem" then base.principal_part_specs.explicit_phis_stem_spec = fetch_specs(dot_separated_group) elseif first_element_prefix == "unstressed_stem" then base.principal_part_specs.explicit_unstressed_stem_spec = fetch_specs(dot_separated_group) elseif first_element_prefix == "pres" then if first_element_minus_prefix == "-" and #dot_separated_group == 1 then -- Allow 'pres:-' to be given to suppress the present. We implement this using the -- equivalent of a row override for each of the present forms; this also cancels out the -- present subjunctive and the imperative. local hyphen_form = base.row_override_specs.pres = { ["1s"] = hyphen_form, ["2s"] = hyphen_form, ["3s"] = hyphen_form, ["1p"] = hyphen_form, ["2p"] = hyphen_form, ["3p"] = hyphen_form, }				else if not base.props.builtin then parse_err("Can't specify 'pres:' override except when '@' is given to request a built-in verb") end parse_present_spec(dot_separated_group) end elseif row_conjugation_map[first_element_prefix] then local no_explicit_pp = row_conjugation_map[first_element_prefix].no_explicit_principal_part if no_explicit_pp == true or not base.props.builtin and no_explicit_pp == "builtin" then parse_err("Can't specify principal part for " .. row_conjugation_map[first_element_prefix].desc						.. " using '" .. first_element_prefix .. ":'; use the specification PRES^PRES3S.PHIS.PP") else base.principal_part_specs[first_element_prefix] = fetch_specs(dot_separated_group) end elseif overridable_slot_set[first_element_prefix] then base.single_override_specs[first_element_prefix] = fetch_specs(dot_separated_group) elseif first_element_prefix:find("!$") then local late_override_slot = rmatch(first_element_prefix, "^(.*)!$") if late_overridable_slot_set[late_override_slot] then base.late_single_override_specs[late_override_slot] = fetch_specs(dot_separated_group) else parse_err("Late override " .. first_element_prefix .. " refers to an unrecognized slot in '" ..						table.concat(dot_separated_group) .. "'") end elseif first_element_prefix:find("row$") then local row_override_slot = rmatch(first_element_prefix, "^(.*)row$") if row_conjugation_map[row_override_slot] then local rowspec = row_conjugation_map[row_override_slot] if rowspec.no_row_overrides then -- This happens with e.g. pp and negimp. Doesn't make sense with pp because it's a single form -- that can be specified completely using the explicit principal part. Rarely if ever useful -- for negimp; use single overrides if absolutely necessary. parse_err("Can't specify row override for " .. rowspec.desc .. " using " .. row_override_slot							.. "row:; use an explicit principal part or single overrides (if allowed)") end local comma_separated_groups = split_alternating_runs_and_strip_spaces(dot_separated_group, ",") local persnums = rowspec.row_override_persnums or rowspec.persnums if #comma_separated_groups ~= #persnums then parse_err("For " .. row_override_slot .. "row:, expected " .. #persnums							.. " comma-separated forms but saw " .. #comma_separated_groups .. " in '"							.. table.concat(dot_separated_group) .. "'") end base.row_override_specs[row_override_slot] = {} for i, persnum in ipairs(persnums) do						base.row_override_specs[row_override_slot][persnum] = fetch_specs(comma_separated_groups[i]) end else local row_override_slots = {} for row_override_slot, _ in pairs(row_conjugation_map) do table.insert(row_override_slots, row_override_slot .. "row:") end table.sort(row_override_slots) parse_err("Row override spec should begin with one of " .. m_table.serialCommaJoin(row_override_slots)						.. ", but saw '" .. table.concat(dot_separated_group) .. "'") end else parse_err("Unrecognized prefix '" .. first_element_prefix .. "' in '"					.. table.concat(dot_separated_group) .. "'") end end end end

local function create_base -- `lemma` is the verb lemma, as specified by the user. -- `verb` contains various properties of the verb derived from the lemma by analyze_verb. -- `forms` contains the final per-slot forms. This is processed further in Module:inflection-utilities. --   This is a table indexed by slot (e.g. "pres1s"). Each value in the table is a list of items of the form --   {form = FORM, footnotes = FOOTNOTES} where FORM is the actual generated form and FOOTNOTES is either nil --   or a list of footnotes (which must be surrounded by brackets, e.g. "[archaic]"). -- `principal_part_specs` contains forms specified by the user in various fashions. The value is in the same form --   as for `forms``, but the FORM contained in it is the actual user-specified form, which may be e.g. "#è" --   rather than a verb form, and needs to be processed to generate the actual form. A spec may be "+" to insert --   the default-generated form or forms, or "-" to indicate that this form doesn't exist. The source of these --   forms is either --   (a) prefixes 'imperf:', 'fut:', 'sub:', 'impsub:', 'imp:', etc. (the key is "imperf", "fut", etc.); --   (b) specs in the format e.g. "vèngo:vègno[archaic or poetic]^viène,vénni,venùto" or "é:#è" (the key is one of	--        "pres", "pres3s", "phis" or "pp" as appropriate); --   (c) an explicit stem specified using 'stem:' (the key is "explicit_stem_spec"); --   (d) a root-stressed infinitive spec such as "ó+" in "a\ó+\compóngo,compósi,compósto" (the key is	--        "root_stressed_inf"); this will have the value "+" (meaning to take the vowel spec from the present tense) --       if a spec like "a\è" is given with only one backslash, and will be missing entirely if a spec like "a/è" --       is given with a slash instead of a backslash; --   (e) an auxiliary specified using e.g. "a[transitive]:e[intransitive]/è" (the key is "aux" and the value will	--        contain the actual auxiliary in the form in place of "a" or "e"). -- `principal_part_footnotes` contains per-row footnotes derived from overriding principal part specs of the form --   '+[footnote]' used in conjunction with a built-in verb. -- `principal_part_forms` contains the processed versions of the specs contained in `principal_part_specs`. The --   keys are as in `principal_part_specs` and the values are the same as for `forms`. -- `row_override_specs` contains user-specified forms for a full tense/aspect row using 'presrow:', 'subrow:', etc. --   The key is "pres", "sub", etc. (i.e. minus the "row" suffix). The value is another table indexed by the --   person/number suffix (e.g. "1s", "2s", etc. for "pres"; "123s", "1p", "2p", etc. for "sub"), whose values --   are in the same format as `principal_part_specs`. -- `single_override_specs` contains user-specified forms using 'pres1s:', 'sub3p:', etc. The key is the slot --   ("pres1s", "sub3p", etc.) and the value is of the same format as `principal_part_specs`. -- `late_single_override_specs` contains user-specified forms using 'pres1s!:', 'sub3p!:', etc., specifying late --   overrides that take place after copying forms from one place to another and after adding reflexive clitics. --   The key is the slot ("pres1s", "sub3p", etc.) and the value is of the same format as `principal_part_specs`. -- `addnote_specs` contains specs specifying footnotes to add to individual slots or collections of slots using --   'addnote[SLOTSPEC,SLOTSPEC][FOOTNOTE][FOOTNOTE][...]'. Each element is an object of the form --   {slotspec = {SLOTSPEC, ...}, footnotes = {FOOTNOTE, ...}}. -- `is_irreg` is a table indexed by an individual form slot ("pres1s", "sub2s", "pp", etc.) whose value is true or --   false indicating whether a given form is irregular. Currently, the values in `is_irreg` are used only by the --   code in Module:it-headword to determine whether to show irregular principal parts. -- `props` is a table of miscellaneous Boolean properties. Current properties: --   - `impers` (impersonal verb, with only third-singular forms) --   - `thirdonly` (third-person only verb) --   - `nofinite` (verb is missing all finite forms) --   - `no_root_stressed` (verb is missing all root-stressed forms) --   - `presonly` (verb is missing all finite forms except the present indicative) --   - `rre` (user specified the 'rre' indicator in conjunction with a syncopated reflexive verb like	--             contrarsi reflexive of contrarre, which otherwise would get interpreted as the reflexive of	--             contrare) --   - `syncopated` (verb is syncopated, i.e. the infinitive ends in '-rre'; includes verbs with '-rre' infinitive	--                    as well as reflexive verbs ending in '-orsi' or '-ursi' and verbs where the 'rre' indicator	--                    was given by the user) --   - `builtin` (verb uses a built-in conjugation in Module:it-verb/builtin) --   - `opt_root_stressed_inf` (verb used the \/ or /\ notation to indicate an optionally root-stressed infinitive;	--                               to determine if a verb used the \ notation, look for a value in	--                               base.principal_part_specs.root_stressed_inf) -- `rowprops` is a table of tables of row-specific Boolean properties. Each subtable specifies a property of a	--   given row, such as whether the row is irregular. Specifically: --   - `rowprops.irreg`: The row is irregular, i.e. at least one slot has an irregular form. Currently used to	--      determine whether to add categories like Category:Italian verbs with irregular imperfect subjunctive, --      and whether to display the row's principal part in the headword line. --   - `rowprops.defective`: The row is defective (missing one or more forms). Forms expected to be missing due to --      'impers' or 'thirdonly' don't count. --   - `rowprops.all_defective`: The row is missing all forms. A row should be considered completely defective if	--      `rowprops.defective[row]` and `rowprops.all_defective[row]` (we need both checks in case of expected	--       missing rows, such as imperative with 'impers' or 'thirdonly'). --   - `rowprops.unknown`: The row has at least one unknown form. --   - `rowprops.all_unknown`: All forms of the row are unknown. --	-- There should be no other properties set directly at the `base` level. return {forms = {}, principal_part_specs = {}, principal_part_footnotes = {}, principal_part_forms = {}, row_override_specs = {}, single_override_specs = {}, late_single_override_specs = {}, addnote_specs = {}, is_irreg = {}, props = {}, rowprops = {irreg = {}, defective = {}, all_defective = {}, unknown = {}, all_unknown = {}}, } end

-- Parse the indicator spec of an Italian verb, e.g. ''. `angle_bracket_spec` is the indicator spec itself, -- surrounded by angle brackets. `lemma` is the verb lemma specified before the indicator spec, possibly with brackets -- if so specified by the user, and an empty string if no lemma was given. `pagename` is the pagename, either the -- actual name of the page or the value of pagename= if given. -- -- For example: -- * If the user said or  then angle_bracket_spec == "" and lemma == "". -- * If the user said then angle_bracket_spec == "" and lemma == "partecipare". -- * If the user said then angle_bracket_spec == "<ù>" and --  lemma == "annunciarsi". -- -- This function returns a `base` object (see create_base) describing the parsed spec. local function parse_indicator_spec(angle_bracket_spec, lemma, pagename) local base = create_base if lemma == "" then lemma = pagename end base.lemma = m_links.remove_links(lemma) base.verb = analyze_verb(lemma)

local inside = angle_bracket_spec:match("^<(.*)>$") assert(inside) parse_inside(base, inside, false)

local function parse_err(msg) error(msg .. ": " .. angle_bracket_spec) end

-- Set up base.verb.verb. This must be done after parse_inside because it depends on the '.rre' indicator. set_up_base_verb(base)

if base.props.builtin then local prefix, main_verb, conj = find_builtin_verb(base.verb.verb) if not prefix then parse_err("Unable to find built-in verb corresponding to '" .. base.verb.verb .. "'") end -- Create a new `base`, fill it with properties from the built-in verb, and copy over the user-specified -- properties on top of it. local nbase = create_base nbase.lemma = base.lemma nbase.verb = base.verb nbase.verb.prefix = prefix nbase.verb.verb = prefix .. main_verb nbase.verb.raw_verb = prefix .. nbase.verb.raw_verb parse_inside(nbase, conj, "is builtin") if nbase.principal_part_specs.root_stressed_inf then local base_rsi = base.principal_part_specs.root_stressed_inf -- We are dealing with a built-in verb whose spec has a backslash in it, such as scegliere with spec -- "é\scélgo,scélsi,scélto", or porre with spec "ó\póngo,pósi,pósto". The user should have specified -- "a\@" or similar. We need to merge the specs appropriately. if not base_rsi then parse_err(("Built-in verb %s has a root-stressed infinitive, and so the user specification " .. "should have a backslash (\\) preceding the @ sign"):format(main_verb)) end if #base_rsi ~= 1 or base_rsi[1].form ~= "+" or base_rsi[1].footnotes then local auxspec = base.verb.is_reflexive and "" or "a" parse_err(("Built-in verb %s has a root-stressed infinitive, and so the user specification " .. "should use a single-backslash spec like '%s\\@', not a double-backslash one like '%s\\ó\\@'"):					format(main_verb, aux_spec, aux_spec)) end -- Cancel out the user's spec so it doesn't override the built-in spec. base.principal_part_specs.root_stressed_inf = nil end

-- If there's a prefix, add it now to all the specs derived from the built-in verb. if prefix ~= "" then local function form_should_be_preserved(form) local prespec, bare_form, postspec = parse_decorated_form(form) return bare_form == "+" or bare_form == "+isc" or bare_form == "-" or bare_form == "?" or is_single_vowel_spec(bare_form) end

local function add_prefix_to_forms(tbl, slot) local saw_preserve_accent = false for _, formobj in ipairs(tbl[slot]) do					local prespec, bare_form, postspec = parse_decorated_form(formobj.form) if rfind(postspec, "!") then saw_preserve_accent = true break end end if saw_preserve_accent then -- We have to do it the "hard" way, re-inserting the forms, in case of redundancy. local existing_forms = tbl[slot] tbl[slot] = {} iut.insert_forms(tbl, slot, iut.map_forms(existing_forms, function(form) if form_should_be_preserved(form) then return form else -- If there is a ! after the form (indicating that monosyllabic accents should be							-- preserved), remove it. local prespec, bare_form, postspec = parse_decorated_form(form) return prespec .. prefix .. bare_form .. rsub(postspec, "!", "") end end)) else -- To save on memory, side-effect the existing forms. map_side_effecting_forms(tbl[slot], function(form)						if form_should_be_preserved(form) then							return form						else							local prespec, bare_form, postspec = parse_decorated_form(form)							return prespec .. prefix .. bare_form .. postspec						end					end) end end

for _, prop_table in ipairs { "principal_part_specs", "single_override_specs", "late_single_override_specs" } do				for slot, _ in pairs(nbase[prop_table]) do					add_prefix_to_forms(nbase[prop_table], slot) end end

-- nbase.row_override_specs is in a different format. for slot, prop in pairs(nbase.row_override_specs) do				for persnum, _ in pairs(prop) do					add_prefix_to_forms(prop, persnum) end end

-- We also need to hack any single-vowel specs that don't already specify +, - or -- to use ++, which is -- like + (take the right vowel of two) but won't throw an error if there aren't two matching vowels. -- Thisis needed for e.g. porre with prefix com-, where comporre ends up with two of the same vowel. local function hack_single_vowel_spec(specs) if specs then for _, spec in ipairs(specs) do						local prespec, bare_form, postspec = parse_decorated_form(spec.form) if rfind(bare_form, "^" .. AV .. "$") then spec.form = prespec .. bare_form .. "++" .. postspec end end end end hack_single_vowel_spec(nbase.principal_part_specs.root_stressed_inf) hack_single_vowel_spec(nbase.principal_part_specs.pres) hack_single_vowel_spec(nbase.principal_part_specs.pres3s) end

-- Copy user-specified principal part specs unless '+' is used, which we handle specially. The reason for this -- is that + is normally used in conjunction with footnotes such as 'phis:+[rare]' where the intention is to add -- a footnote to all the forms of that row; but if we don't do the following, the + instead requests the default -- principal part generated from the infinitive. for slot, specs in pairs(base.principal_part_specs) do			local saw_plus, saw_non_plus for _, spec in ipairs(specs) do				local prespec, bare_form, postspec = parse_decorated_form(spec.form) if bare_form == "+" then saw_plus = true else saw_non_plus = true end end if saw_plus and saw_non_plus then parse_err(("For principal part '%s:', can't specify both + and something else when overriding a built-in verb"):					format(slot)) elseif saw_plus and slot ~= "root_stressed_inf" then -- Need a special case for root_stressed_inf. We do want to copy a + from the user-specified specs for -- root_stressed_inf to `nbase`, e.g. for condurre. The built-in spec doesn't indicate that this is				-- root-stressed; this comes from the user-specified 'a\@' or similar, and if we don't copy it, the code in -- add_infinitive triggers do_ending_stressed_infinitive, which throws an error on -rre verbs. for _, spec in ipairs(specs) do					local prespec, bare_form, postspec = parse_decorated_form(spec.form) if bare_form ~= "+" then parse_err(("Internal error: Saw '%s' for principal part form for slot '%s' but expected '+'"):							format(slot, spec.form)) end if spec.footnotes then nbase.principal_part_footnotes[slot] = iut.combine_footnotes(							nbase.principal_part_footnotes[full_slot], spec.footnotes) end end else nbase.principal_part_specs[slot] = specs end end

-- Now copy remaining user-specified specs into the built-in verb `base`. for _, prop_table in ipairs { "row_override_specs", "single_override_specs", "late_single_override_specs", "props" } do			for slot, prop in pairs(base[prop_table]) do				nbase[prop_table][slot] = prop end end

for _, prop_list in ipairs { "addnote_specs" } do			for _, prop in ipairs(base[prop_list]) do				m_table.insertIfNot(nbase[prop_list], prop) end end return nbase end

return base end

-- Normalize all lemmas, substituting the pagename for blank lemmas and adding links to multiword lemmas. local function normalize_all_lemmas(alternant_multiword_spec)

-- (1) Add links to all before and after text. if not alternant_multiword_spec.args.noautolinktext then alternant_multiword_spec.post_text = add_links(alternant_multiword_spec.post_text) for _, alternant_or_word_spec in ipairs(alternant_multiword_spec.alternant_or_word_specs) do			alternant_or_word_spec.before_text = add_links(alternant_or_word_spec.before_text) if alternant_or_word_spec.alternants then for _, multiword_spec in ipairs(alternant_or_word_spec.alternants) do					multiword_spec.post_text = add_links(multiword_spec.post_text) for _, word_spec in ipairs(multiword_spec.word_specs) do						word_spec.before_text = add_links(word_spec.before_text) end end end end end

-- (2) Remove any links from the lemma. iut.map_word_specs(alternant_multiword_spec, function(base)		base.lemma = m_links.remove_links(base.lemma)	end) end

-- Detect inconsistencies in indicator specs. This checks that the properties 'impers' and 'thirdonly' are consistent -- across all verbs; checks that if the past participle is given as -, the auxiliary is also given as -; and propagates -- certain properties (the `source_module` property and the template args) down to each `base`. local function detect_all_indicator_specs(alternant_multiword_spec, source_module) alternant_multiword_spec.props = {}

local props_that_must_be_consistent = {"impers", "thirdonly"}

-- Propagate certain properties (e.g. reflexiveness) upward from individual `base` forms to the overall -- `alternant_multiword_spec`. This is done "earlier" than conjugating the verb, and is needed prior to this because -- it determines whether to generate extra slots for use with. iut.map_word_specs(alternant_multiword_spec, function(base)		-- If there is an explicit stem spec, we display the imperfect principal part explicitly even if not marked		-- as irregular.		if base.principal_part_specs.explicit_stem_spec then			alternant_multiword_spec.props.has_explicit_stem_spec = true		end		-- If any verb is pronominal, we display the overall lemma as 'pronominal'.		if base.verb.is_pronominal then			alternant_multiword_spec.props.is_pronominal = true		end		-- If any verb is non-reflexive, we show the auxiliary.		if not base.verb.is_reflexive then			alternant_multiword_spec.props.is_non_reflexive = true		end		-- If any verb is reflexive, we need to generate _non_reflexive and _variant forms for use in		-- .		if base.verb.is_reflexive then			alternant_multiword_spec.props.any_reflexive = true		end		-- `impers` and `thirdonly` need to be consistent across all verbs as they determine e.g. which forms are -- shown in the headword. for _, prop in ipairs(props_that_must_be_consistent) do			if base.props[prop] then alternant_multiword_spec.props[prop] = true end end end)

-- Propagate some settings down from `alternant_multiword_spec` to individual `base` forms. iut.map_word_specs(alternant_multiword_spec, function(base)		base.source_module = source_module		base.args = alternant_multiword_spec.args		base.props.any_reflexive = alternant_multiword_spec.props.any_reflexive	end)

for _, prop in ipairs(props_that_must_be_consistent) do		if alternant_multiword_spec.props[prop] then iut.map_word_specs(alternant_multiword_spec, function(base)				if not base.props[prop] then					error("If some alternants specify '" .. prop .. "', all must")				end			end) end end

iut.map_word_specs(alternant_multiword_spec, function(base)		if base.props.impers and base.props.thirdonly then			error("'impers' and 'thirdonly' cannot both be specified")		end

-- Check for missing past participle -> missing auxiliary. if not base.verb.is_reflexive then local pp_is_missing = base.principal_part_specs.pp and #base.principal_part_specs.pp == 0 and base.principal_part_specs.pp[1].form == "-" local aux_is_missing = not base.principal_part_specs.aux if pp_is_missing and not aux_is_missing then error("If past participle given as '-', auxiliary must be explicitly specified as '-'") end end end) end

-- Propagate indications of irregularity, defectiveness and other properties upward from individual `base` forms to the -- overall `alternant_multiword_spec`. This is done "later" than conjugating the verb. The overall indications of -- irregularity/defectiveness are used in Module:it-headword to show irregular/defective principal parts, and other -- properties are used similarly in Module:it-headword. This needs to be done later than conjugating the angle -- bracket spec because it depends on the result of conjugation. local function propagate_properties_upward_later(alternant_multiword_spec) iut.map_word_specs(alternant_multiword_spec, function(base)		local function copy_property_table(dest_table, source_table, slotprop)			if not dest_table[slotprop] then				dest_table[slotprop] = {}			end			for slot, propval in pairs(source_table[slotprop]) do				dest_table[slotprop][slot] = dest_table[slotprop][slot] or propval			end		end		copy_property_table(alternant_multiword_spec, base, "is_irreg")		if not alternant_multiword_spec.rowprops then			alternant_multiword_spec.rowprops = {}		end		for subtable_key, subtable in pairs(base.rowprops) do			copy_property_table(alternant_multiword_spec.rowprops, base.rowprops, subtable_key)		end	end) end

-- Set the overall auxiliary or auxiliaries. We can't do this using the normal inflection -- code as it will produce e.g. 'avére e avére' for conjoined verbs. local function compute_auxiliary(alternant_multiword_spec) iut.map_word_specs(alternant_multiword_spec, function(base)		iut.insert_forms(alternant_multiword_spec.forms, "aux", iut.map_forms(base.principal_part_specs.aux, function(form)				return add_links(form)			end) )	end) end

-- Add the categories and annotation for `base` into the appropriate structures in `alternant_multiword_spec`. local function add_categories_and_annotation(alternant_multiword_spec, base, multiword_lemma, source_module) local function insert_ann(anntype, value) m_table.insertIfNot(alternant_multiword_spec.annotation[anntype], value) end

local function insert_cat(cat, also_when_multiword) -- Don't place multiword terms in categories like 'Italian verbs ending in -are' to avoid spamming the -- categories with such terms. if also_when_multiword or not multiword_lemma then m_table.insertIfNot(alternant_multiword_spec.categories, "Italian " .. cat) end end

if check_for_red_links and not source_module and not multiword_lemma then for _, slot_and_accel in ipairs(export.all_verb_slots) do			local slot = slot_and_accel[1] local forms = base.forms[slot] local must_break = false if forms then for _, form in ipairs(forms) do					if not form.form:find("%[%[") then local title = mw.title.new(form.form) if title and not title.exists then insert_cat("verbs with red links in their inflection tables") must_break = true break end end end end if must_break then break end end end

if base.props.syncopated then insert_ann("conj", "syncopated") insert_cat("syncopated verbs") elseif base.principal_part_specs.root_stressed_inf then insert_ann("conj", "root-stressed -ere") insert_cat("verbs with root-stressed infinitive") insert_cat("verbs ending in -ere") else local ending = base.conj_vowel == "à" and "are" or base.conj_vowel == "é" and "ere" or "ire" insert_ann("conj", "-" .. ending) insert_cat("verbs ending in -" .. ending) end

if base.props.impers then insert_ann("third_only", "impersonal") insert_cat("impersonal verbs") elseif base.props.thirdonly then insert_ann("third_only", "third-person only") insert_cat("third-person-only verbs") else insert_ann("third_only", "regular") end

local is_irreg = false local is_defective = false for _, rowconj in ipairs(row_conjugations) do		local rowslot, rowspec = unpack(rowconj) if base.rowprops.irreg[rowslot] then if not is_irreg then is_irreg = true insert_cat("irregular verbs") end insert_cat("verbs with irregular " .. rowspec.desc) end if base.rowprops.defective[rowslot] then if not is_defective then is_defective = true insert_cat("defective verbs") end insert_cat("verbs with missing " .. rowspec.desc) end end if not base.verb.is_reflexive and not base.principal_part_specs.aux then if not is_defective then is_defective = true insert_cat("defective verbs") end insert_cat("verbs lacking composed tenses") end

if is_irreg then insert_ann("irreg", "irregular") else insert_ann("irreg", "regular") end

if is_defective then insert_ann("defective", "defective") else insert_ann("defective", "regular") end

if not base.verb.is_reflexive and base.principal_part_specs.aux then for _, auxform in ipairs(base.principal_part_specs.aux) do if auxform.form ~= "?" then -- No auxiliaries end in a stressed vowel so this is safe. local aux_no_accents = remove_accents(auxform.form) insert_ann("aux", aux_no_accents) insert_cat("verbs taking " .. aux_no_accents .. " as auxiliary") end end end

if base.verb.is_pronominal then insert_cat("pronominal verbs") -- FIXME: Should we use this instead? This is what Spanish does. -- insert_cat("verbs with lexical clitics") end

if base.verb.is_reflexive then insert_cat("reflexive verbs") end end

-- Compute the categories to add the verb to, as well as the annotation to display in the conjugation title bar. We -- combine the code to do these functions as both categories and title bar contain similar information. local function compute_categories_and_annotation(alternant_multiword_spec, source_module) alternant_multiword_spec.categories = {} local ann = {} alternant_multiword_spec.annotation = ann ann.conj = {} ann.third_only = {} ann.irreg = {} ann.defective = {} ann.aux = {}

local multiword_lemma = false for _, form in ipairs(alternant_multiword_spec.forms.inf) do		if form.form:find(" ") then multiword_lemma = true break end end

iut.map_word_specs(alternant_multiword_spec, function(base)		add_categories_and_annotation(alternant_multiword_spec, base, multiword_lemma, source_module)	end) local ann_parts = {} local conj = table.concat(ann.conj, " or ") table.insert(ann_parts, conj) local third_only = table.concat(ann.third_only, " or ") if third_only ~= "" and third_only ~= "regular" then table.insert(ann_parts, third_only) end local irreg = table.concat(ann.irreg, " or ") if irreg ~= "" and irreg ~= "regular" then table.insert(ann_parts, irreg) end local defective = table.concat(ann.defective, " or ") if defective ~= "" and defective ~= "regular" then table.insert(ann_parts, defective) end alternant_multiword_spec.annotation = table.concat(ann_parts, "; ") end

-- Convert the forms associated with each slot into their display form (a string). local function show_forms(alternant_multiword_spec) local lemmas = iut.map_forms(alternant_multiword_spec.forms.inf, remove_reflexive_indicators) alternant_multiword_spec.lemmas = lemmas -- save for later use in make_table

local props = { lang = lang, lemmas = lemmas, slot_list = export.all_verb_slots, }	iut.show_forms(alternant_multiword_spec.forms, props) end

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

local basic_table = [=[ {description} Conjugation of {title} (See Appendix:Italian verbs) {\op}| style="background:#F0F0F0;border-collapse:separate;border-spacing:2px;width:100%" class="inflection-table" ! style="background:#e2e4c0" | infinitive ! colspan="2" style="background:#e2e4c0" | auxiliary verb ! colspan="2" style="background:#e2e4c0" | gerund ! colspan="2" style="background:#e2e4c0" | present participle ! colspan="2" style="background:#e2e4c0" | past participle ! rowspan="2" style="background:#C0C0C0" | person ! colspan="3" style="background:#C0C0C0" | singular ! colspan="3" style="background:#C0C0C0" | plural ! style="background:#C0C0C0;width:12.5%" | first ! style="background:#C0C0C0;width:12.5%" | second ! style="background:#C0C0C0;width:12.5%" | third ! style="background:#C0C0C0;width:12.5%" | first ! style="background:#C0C0C0;width:12.5%" | second ! style="background:#C0C0C0;width:12.5%" | third ! style="background:#c0cfe4" | indicative ! style="background:#c0cfe4" | io ! style="background:#c0cfe4" | tu ! style="background:#c0cfe4" | lui/lei, esso/essa ! style="background:#c0cfe4" | noi ! style="background:#c0cfe4" | voi ! style="background:#c0cfe4" | loro, essi/esse ! style="height:3em;background:#c0cfe4" | present ! style="height:3em;background:#c0cfe4" | imperfect ! style="height:3em;background:#c0cfe4" | past historic ! style="height:3em;background:#c0cfe4" | future ! style="background:#c0d8e4" | conditional ! style="background:#c0d8e4" | io ! style="background:#c0d8e4" | tu ! style="background:#c0d8e4" | lui/lei, esso/essa ! style="background:#c0d8e4" | noi ! style="background:#c0d8e4" | voi ! style="background:#c0d8e4" | loro, essi/esse ! style="height:3em;background:#c0d8e4" | present ! style="background:#c0e4c0" | subjunctive ! style="background:#c0e4c0" | che io ! style="background:#c0e4c0" | che tu ! style="background:#c0e4c0" | che lui/che lei, che esso/che essa ! style="background:#c0e4c0" | che noi ! style="background:#c0e4c0" | che voi ! style="background:#c0e4c0" | che loro, che essi/che esse ! style="height:3em;background:#c0e4c0" | present ! style="height:3em;background:#c0e4c0" | imperfect ! rowspan="2" style="height:3em;background:#e4d4c0" | imperative ! style="background:#e4d4c0" | &mdash; ! style="background:#e4d4c0" | tu ! style="background:#e4d4c0" | Lei ! style="background:#e4d4c0" | noi ! style="background:#e4d4c0" | voi ! style="background:#e4d4c0" | Loro ! style="height:3em;background:#e4d4c0" | negative imperative ]=]
 * {inf}
 * {aux}
 * colspan="2" | {ger}
 * {presp}
 * colspan="2" | {pp}
 * {pres1s}
 * {pres2s}
 * {pres3s}
 * {pres1p}
 * {pres2p}
 * {pres3p}
 * {imperf1s}
 * {imperf2s}
 * {imperf3s}
 * {imperf1p}
 * {imperf2p}
 * {imperf3p}
 * {phis1s}
 * {phis2s}
 * {phis3s}
 * {phis1p}
 * {phis2p}
 * {phis3p}
 * {fut1s}
 * {fut2s}
 * {fut3s}
 * {fut1p}
 * {fut2p}
 * {fut3p}
 * {cond1s}
 * {cond2s}
 * {cond3s}
 * {cond1p}
 * {cond2p}
 * {cond3p}
 * {sub1s}
 * {sub2s}
 * {sub3s}
 * {sub1p}
 * {sub2p}
 * {sub3p}
 * {impsub1s}
 * {impsub2s}
 * {impsub3s}
 * {impsub1p}
 * {impsub2p}
 * {impsub3p}
 * {imp2s}
 * {imp3s}
 * {imp1p}
 * {imp2p}
 * {imp3p}
 * {imp2p}
 * {imp3p}
 * {negimp2s}
 * {negimp3s}
 * {negimp1p}
 * {negimp2p}
 * {negimp3p}
 * {\cl}{notes_clause}
 * {\cl}{notes_clause}

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

local lemma_links = {} for _, lemma in ipairs(alternant_multiword_spec.lemmas) do		table.insert(lemma_links, m_links.full_link({ lang = lang, term = lemma.form }, "term")) end forms.title = table.concat(lemma_links, " or ")

if alternant_multiword_spec.annotation ~= "" then forms.title = forms.title .. " (" .. alternant_multiword_spec.annotation .. ")" end forms.description = ""

-- Format the table. forms.footnote = alternant_multiword_spec.forms.footnote forms.notes_clause = forms.footnote ~= "" and m_string_utilities.format(notes_template, forms) or "" return m_string_utilities.format(basic_table, forms) end

-- Externally callable function to conjugate a verb. Return value is ALTERNANT_MULTIWORD_SPEC, an object where the -- conjugated forms are in `ALTERNANT_MULTIWORD_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(args, source_module, headword_head) local PAGENAME = mw.title.getCurrentTitle.text local function in_template_space return mw.title.getCurrentTitle.nsText == "Template" end local pagename = source_module ~= "it-inflections" and args.pagename or PAGENAME local head = headword_head or pagename local arg1 = args[1]

if not arg1 then if (pagename == "it-conj" or pagename == "it-verb") and in_template_space then arg1 = "mettere" elseif pagename == "it-verb form of" and in_template_space then arg1 = "amare" else error("Internal error: Missing 1= and not caught earlier") end end

-- When called from, determine the non-lemma form whose inflections we're being asked to -- determine. This normally comes from the page title or the value of |pagename=. local verb_form_of_form if source_module == "it-inflections" then verb_form_of_form = args.pagename if not verb_form_of_form then if PAGENAME == "it-verb form of" and in_template_space then verb_form_of_form = "ami" else verb_form_of_form = PAGENAME end end end

local need_surrounding_angle_brackets = true local incorporated_headword_head_into_lemma = false -- Check whether we need to add <...> around the argument. If the -- argument has no < in it, we definitely do. Otherwise, we need to	-- parse the balanced [...] and <...> and add <...> only if there isn't	-- a top-level <...>. We check for [...] because there might be angle -- brackets inside of them (HTML tags in qualifiers or <> and	-- such in references). if arg1:find("<") then local segments = iut.parse_multi_delimiter_balanced_segment_run(arg1, {{"<", ">"}, {"[", "]"}}) for i = 2, #segments, 2 do			if segments[i]:find("^<.*>$") then need_surrounding_angle_brackets = false break end end end if need_surrounding_angle_brackets then if head:find(" ") then -- If multiword lemma without <...> already, try to add it after the first word. local need_explicit_angle_brackets = false if arg1:find("%(%(") then need_explicit_angle_brackets = true else -- Try to preserve the brackets in the part after the verb, but don't do it				-- if there aren't the same number of left and right brackets in the verb -- (which means the verb was linked as part of a larger expression). local refl_clitic_verb, post = rmatch(head, "^(.-)( .*)$") local left_brackets = rsub(refl_clitic_verb, "[^%[]", "") local right_brackets = rsub(refl_clitic_verb, "[^%]]", "") if #left_brackets == #right_brackets then arg1 = iut.remove_redundant_links(refl_clitic_verb) .. "<" .. arg1 .. ">" .. post incorporated_headword_head_into_lemma = true else -- Try again using the form without links. local linkless_head = m_links.remove_links(head) if linkless_head:find(" ") then refl_clitic_verb, post = rmatch(linkless_head, "^(.-)( .*)$") arg1 = refl_clitic_verb .. "<" .. arg1 .. ">" .. post else need_explicit_angle_brackets = true end end end

if need_explicit_angle_brackets then error("Multiword argument without <> and with alternants, a multiword linked verb or " ..					"unbalanced brackets; please include <> explicitly: " .. arg1) end else arg1 = "<" .. arg1 .. ">"			-- Will be incorporated through `head` below in the call to parse_indicator_spec. incorporated_headword_head_into_lemma = true end end

local function do_parse_indicator_spec(angle_bracket_spec, lemma) return parse_indicator_spec(angle_bracket_spec, lemma, head) end

local parse_props = { parse_indicator_spec = do_parse_indicator_spec, allow_default_indicator = true, allow_blank_lemma = true, }	local escaped_arg1 = escape_reflexive_indicators(arg1) local alternant_multiword_spec = iut.parse_inflected_text(escaped_arg1, parse_props) alternant_multiword_spec.pos = pos or "verbs" alternant_multiword_spec.args = args alternant_multiword_spec.verb_form_of_form = verb_form_of_form alternant_multiword_spec.incorporated_headword_head_into_lemma = incorporated_headword_head_into_lemma normalize_all_lemmas(alternant_multiword_spec) detect_all_indicator_specs(alternant_multiword_spec, source_module) if alternant_multiword_spec.props.any_reflexive and source_module == "it-inflections" then add_reflexive_verb_form_of_slots end local inflect_props = { slot_list = export.all_verb_slots, inflect_word_spec = conjugate_verb, -- We add links around the generated verbal forms rather than allow the entire multiword -- expression to be a link, so ensure that user-specified links get included as well. include_user_specified_links = true, }	iut.inflect_multiword_or_alternant_multiword_spec(alternant_multiword_spec, inflect_props) propagate_properties_upward_later(alternant_multiword_spec) compute_auxiliary(alternant_multiword_spec) convert_accented_links(alternant_multiword_spec) compute_categories_and_annotation(alternant_multiword_spec, source_module) if args.json then return require("Module:JSON").toJSON(alternant_multiword_spec) end return alternant_multiword_spec end

-- Externally callable function to parse user-specified arguments and conjugate a verb. -- Return value is ALTERNANT_MULTIWORD_SPEC, an object where the conjugated forms are in `ALTERNANT_MULTIWORD_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.parse_args_and_generate_forms(parent_args) local params = { [1] = {required = true}, ["noautolinktext"] = {type = "boolean"}, ["noautolinkverb"] = {type = "boolean"}, ["pagename"] = {}, -- for testing ["json"] = {type = "boolean"}, -- for bot use }

local args = require("Module:parameters").process(parent_args, params) return export.do_generate_forms(args) end

-- Entry point for. Template-callable function to parse and conjugate a verb given -- user-specified arguments and generate a displayable table of the conjugated forms. function export.show(frame) local parent_args = frame:getParent.args local alternant_multiword_spec = export.parse_args_and_generate_forms(parent_args) if type(alternant_multiword_spec) == "string" then return alternant_multiword_spec end show_forms(alternant_multiword_spec) return make_table(alternant_multiword_spec) .. require("Module:utilities").format_categories(alternant_multiword_spec.categories, lang, nil, nil, force_cat) end

-- Concatenate all forms of all slots into a single string of the form -- "SLOT=FORM,FORM,...|SLOT=FORM,FORM,...|...". Embedded pipe symbols (as might occur -- in embedded links) are converted to <!>. If INCLUDE_PROPS is given, also include -- additional properties (currently, none). This is for use by bots. local function concat_forms(alternant_multiword_spec, include_props) local ins_text = {} for _, slot_and_accel in ipairs(export.all_verb_slots) do		local slot = slot_and_accel[1] local formtext = iut.concat_forms_in_slot(alternant_multiword_spec.forms[slot]) if formtext then table.insert(ins_text, slot .. "=" .. formtext) end end return table.concat(ins_text, "|") end

-- Template-callable function to parse and conjugate a verb given user-specified arguments and return -- the forms as a string "SLOT=FORM,FORM,...|SLOT=FORM,FORM,...|...". Embedded pipe symbols (as might -- occur in embedded links) are converted to <!>. If |include_props=1 is given, also include -- additional properties (currently, none). This is for use by bots. function export.generate_forms(frame) local include_props = frame.args["include_props"] local parent_args = frame:getParent.args local alternant_multiword_spec = export.parse_args_and_generate_forms(parent_args) return concat_forms(alternant_multiword_spec, include_props) end

return export