Module:category tree/poscatboiler/data/non-lemma forms

local labels = {}
local raw_categories = {}



-----------------------------------------------------------------------------
--                                                                         --
--                                  LABELS                                 --
--                                                                         --
-----------------------------------------------------------------------------


labels = {
	description = "{{{langname}}} inflected forms of other ], categorized by their part of speech.",
	umbrella_parents = "Fundamental",
	parents = {{name = "{{{langcat}}}", raw = true}},
}

-- Handle specific types of non-lemma forms. Each spec is of the form {POS, PARENT_POS} where POS is the singular form
-- of a part of speech for which non-lemma forms exist (e.g. 'determiner' is listed because 'determiner forms' exist),
-- and PARENT_POS is the singular form of the parent category to add the category to. For example, the spec
-- {"circumfix", "morpheme"} means that 'circumfix forms' categories have the corresponding 'morpheme forms' categories
-- as their first parent. The second parent is the POS itself; e.g. 'morpheme forms' has 'morphemes' as its second
-- parent, with a sort key that places 'morpheme forms' at the beginning of the child categories. If PARENT_POS is
-- 'false', only the second parent appears as the parent. If a spec is a single string POS, the PARENT_POS defaults to
-- "non-lemma"; e.g. if the spec is "adjective", then 'adjective forms' categories have the corresponding
-- 'non-lemma forms' category as their first parent.
for _, spec in ipairs {
	"adjective",
	"adverb",
	{"article", "determiner"},
	{"circumfix", "morpheme"},
	{"classifier", false},
	{"comparative adjective", false},
	{"comparative adverb", false},
	"conjunction",
	"determiner",
	{"equative adjective", false},
	{"equative adverb", false},
	{"exaggerated adjective", false},
	{"excessive adjective", false},
	{"infinitive", false},
	"interjection",
	"morpheme",
	"noun",
	"numeral",
	"participle",
	"particle",
	{"past participle", "participle"},
	{"prefix", "morpheme"},
	"proper noun",
	"postposition",
	"preposition",
	"pronoun",
	{"suffix", "morpheme"},
	"verb",
	{"superlative adjective", false},
	{"superlative adverb", false},
} do
	local pos, parent_pos
	if type(spec) == "table" then
		pos, parent_pos = unpack(spec)
	else
		pos = spec
		parent_pos = "non-lemma"
	end
	local plpos = require("Module:string utilities").pluralize(pos)
	local parents = {}
	if parent_pos ~= false then
		table.insert(parents, parent_pos .. " forms")
	end
	table.insert(parents, {name = plpos, sort = "**"})
	labels = {
		description = ("{{{langname}}} %s that are inflected to display grammatical relations other than the main form."):format(plpos),
		parents = parents,
	}
end

labels = {
	description = "{{{langname}}} participles that express an agent actively performing an action.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that function as adjectives and express an agent actively performing an action.",
	parents = {"participles", "adjectives"},
}

labels = {
	description = "{{{langname}}} participles that function as adjectives.",
	parents = {"participles", "adjectives"},
}

labels = {
	description = "{{{langname}}} adjectives that are inflected to modify a noun in a genitive construction.",
	parents = {{name = "adjective forms", sort = "construct"}},
}

labels = {
	description = "{{{langname}}} adjectives that are inflected to give attributes to specific and identifiable nouns.",
	parents = {{name = "adjective forms", sort = "definite forms"}},
}

labels = {
	description = "{{{langname}}} adjectives that are inflected to be quantified as two.",
	parents = {{name = "adjective forms", sort = "dual forms"}},
}

labels = {
	description = "{{{langname}}} adjectives that are inflected to give attributes to nouns associated to female beings.",
	parents = {{name = "adjective forms", sort = "feminine forms"}},
}

labels = {
	description = "{{{langname}}} participles that function as adverbs.",
	parents = {"participles", "adverbs"},
}

labels = {
	description = "{{{langname}}} participles that express an agent by who something is done.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that function as adverbs and express an action completed before the one expressed by the primary verb.",
	parents = {"participles", "adverbs"},
}

labels = {
	description = "{{{langname}}} classifiers that are inflected for singular number.",
	parents = {"classifier forms"},
}

labels = {
	description = "{{{langname}}} nominal plural forms with specifically collective semantics.",
	parents = {"noun forms"},
}

labels = {
	description = "{{{langname}}} adjectives that express attributes in a relatively higher degree, or serve to set apart one thing from another.",
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} adjectives that express attributes in a relatively higher degree, or in the highest possible degree.",
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} adverbs that express attributes in a relatively higher degree, or serve to set apart one thing from another.",
	parents = {"adverb forms"},
}

labels = {
	description = "{{{langname}}} participles that function as adverbs and express an action occuring simultaneously with the one expressed by the primary verb.",
	parents = {"participles", "adverbs"},
}

labels = {
	description = "{{{langname}}} verb forms that build clauses which function as adverbials.",
	parents = {"verb forms"},
}

labels = {
	description = "{{{langname}}} determiners that are inflected to display relative degrees of given qualities between nouns.",
	parents = {{name = "determiner forms", sort = "comparative forms"}},
}

labels = {
	description = "{{{langname}}} diminutives of gerunds.",
	parents = {"verb forms"},
}

labels = {
	description = "{{{langname}}} adjectives that express similar degrees of given qualities between nouns, i.e. '''A''' is as ''(insert adjective)'' as '''B'''.",
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} adverbs that express similar degrees of given qualities between verbs, i.e. '''A''' as ''(insert adverb)'' as '''B'''.",
	parents = {"adverb forms"},
}

labels = {
	description = "{{{langname}}} adjectives that express attributes in an even more magnified degree than the superlative forms.",
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} adjectives that express attributes in an excessive degree.",
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} adverbs that express attributes in an excessive degree.",
	parents = {"adverb forms"},
}

labels = {
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} participles that express an agent actively performing an action in future time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that indicate future events.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express a patient having an action performed on it in future time.",
	parents = {"participles"},
}

local function gerund_description(data)
	local default_text = "{{{langname}}} verbs that are conjugated to indicate ongoing events at unspecified moments."
	if not data.lang then
		-- This happens e.g. in ].
		return default_text
	else
		local langcode = data.lang:getCode()
		if langcode == "sq" or langcode == "kmr" or langcode == "liv" or data.lang:inFamily("gem", "itc") then
			return "{{{langname}}} forms that generally act as an action noun for the verb that they are formed from."
		elseif langcode == "sa" or langcode == "pi" then
			return "{{{langname}}} verb forms used in a clause to indicate a prior action by the subject of the sentence."
		else
			return default_text
		end
	end
end
	
labels = {
	description = gerund_description,
	additional = "{{small|] to change this description.]}}",
	parents = {"verb forms"},
}

labels = {
	description = "{{{langname}}} participles that express an ongoing or continuous action.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} non-finite verb forms of various kinds.",
	parents = {"verb forms"},
}

labels = {
	description = "{{{langname}}} adjectives that are derived from a base word to express a stronger concept than the base word.",
	parents = {"adjective forms"},
}

labels = {
	description = "{{{langname}}} words as they appear in original manuscripts.",
	parents = {"non-lemma forms"},
}

labels = {
	description = "{{{langname}}} terms whose initial consonant is mutated due to the effect of the preceding word.",
	parents = {"non-lemma forms"},
}

local inflected_pos_for_mutation = {
	"adjective",
	"adverb",
	"determiner",
	"interjection",
	"noun",
	"numeral",
	"participle",
	"particle",
	"prefix",
	"preposition",
	"pronoun",
	"proper noun",
	"verb",
}

for _, pos in ipairs(inflected_pos_for_mutation) do
	local plpos = require("Module:string utilities").pluralize(pos)
	labels = {
		description = "{{{langname}}} " .. plpos .. " whose initial consonant is mutated due to the effect of the preceding word.",
		parents = {pos .. " forms", {name = "mutated forms", sort = pos}},
	}
end

local mutation_types = {
	-- FIXME, we should add these terms to ] and link to them.
	 = "]",
	 = "]",
	 = "]",
	 = "]",
	 = "]",
	 = "]",
	 = "]",
	 = "]",
	 = "]",
}

for muttype, desc in pairs(mutation_types) do
	labels = {
		description = "{{{langname}}} terms whose initial consonant is mutated through " .. desc .. ", due to the effect of the preceding word.",
		parents = {"mutated forms"},
	}
end

labels = {
	description = "{{{langname}}} participles that express lack of an action being performed.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that function as nouns.",
	parents = {"participles", "nouns"},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to modify a noun in a genitive construction.",
	parents = {{name = "noun forms", sort = "construct"}},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to designate an identified object.",
	parents = {{name = "noun forms", sort = "definite"}},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to be associated to female beings.",
	parents = {{name = "noun forms", sort = "feminine"}},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to designate an unidentified object.",
	parents = {{name = "noun forms", sort = "indefinite"}},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to be associated to male beings.",
	parents = {{name = "noun forms", sort = "masculine"}},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to be quantified as between two (or three in some languages with dual number) and ten.",
	parents = {{name = "noun forms", sort = "paucal"}},
}

labels = {
	description = "{{{langname}}} nouns that are inflected to indicate possession.",
	parents = {{name = "noun forms", sort = "possessive"}},
}

labels = {
	description = "{{{langname}}} verbal forms that behave syntactically like adjectives (or sometimes adverbs), and in some languages are often used in compound conjugations and/or reduced relative clauses.",
	parents = {"verb forms"},
}

labels = {
	description = "{{{langname}}} participles that express a patient having an action performed on it.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that function as adjectives and express a patient having an action performed on it.",
	parents = {"participles", "adjectives"},
}

labels = {
	description = "{{{langname}}} participles that express completed events in past time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express ongoing events in past time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express an agent actively performing an action in past time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that function as adverbs and express actions in past time.",
	parents = {"adverbial participles", "adverbs"},
}

labels = {
	description = "{{{langname}}} participles that express actions in past time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express a patient having an action performed on it in past time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express having completed or undergone an action.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} pronouns that are combined with a preposition.",
	umbrella_parents = "Lemmas subcategories by language",
	parents = {"pronoun forms", {name = "prepositions", sort = "pronouns"}},
}

labels = {
	description = "{{{langname}}} participles that express an agent actively performing an action in present time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that function as adverbs and express actions in present time.",
	parents = {"adverbial participles", "adverbs"},
}

labels = {
	description = "{{{langname}}} participles that express an agent performing an action in present time in middle voice.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express actions occurring in the present time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} participles that express a patient having an action performed on it in present time.",
	parents = {"participles"},
}

labels = {
	description = "{{{langname}}} pronouns that are inflected to indicate possession.",
	parents = {{name = "pronoun forms", sort = "possessive"}},
}

labels = {
	description = "{{{langname}}} words that have alternative pronunciation-based spellings.",
	parents = {"non-lemma forms"},
}

labels = {
	description = "{{{langname}}} plural nouns that are inflected for singular number.",
	parents = {"noun forms"},
}

for _, pos in ipairs { "adjective", "adverb", "determiner" } do
	local plpos = pos .. "s"
	labels = {
		description = "{{{langname}}} " .. plpos .. " that express attributes in their highest possible degrees.",
		parents = {pos .. " forms"},
	}
end

labels = {
	description = "{{{langname}}} verbs that are conjugated to express actions caused by external conditions.",
	parents = {{name = "verb forms", sort = "causative"}},
}

labels = {
	description = "{{{langname}}} verb forms that don't follow standard patterns of conjugation.",
	parents = {{name = "verb forms", sort = "irregular"}},
}

labels = {
	description = "{{{langname}}} verbs that are conjugated to express negation.",
	parents = {{name = "verb forms", sort = "negative"}},
}

labels = {
	description = "{{{langname}}} verbs that are conjugated to direct the action to the grammatical subject, by means of passive voice.",
	parents = {{name = "verb forms", sort = "passive"}},
}

labels = {
	description = "{{{langname}}} verbs that are conjugated to indicate events before the current moment.",
	parents = {{name = "verb forms", sort = "simple past"}},
}

local labels2 = {}

-- Add "reconstructed" subcategories; add 'umbrella_parents' key if not
-- already present.
for key, data in pairs(labels) do
	labels2 = data
	if not data.umbrella_parents then
		data.umbrella_parents = "Non-lemma forms subcategories by language"
	end
	labels2 = {
		description = "{{{langname}}} " .. key .. "  that have been linguistically ].",
		umbrella_parents = "Non-lemma forms subcategories by language",
		parents = {key, {name = "reconstructed terms", sort = key}}
	}
end



-----------------------------------------------------------------------------
--                                                                         --
--                              RAW CATEGORIES                             --
--                                                                         --
-----------------------------------------------------------------------------


raw_categories = {
	description = "Umbrella categories covering topics related to non-lemma forms.",
	additional = "{{{umbrella_meta_msg}}}",
	parents = {
		"Umbrella metacategories",
		{name = "non-lemma forms", is_label = true, sort = " "},
	},
}


return {LABELS = labels2, RAW_CATEGORIES = raw_categories}