Module:category tree/poscatboiler/data/terms by lexical property

local labels = {}
local raw_categories = {}
local handlers = {}

local insert = table.insert

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


labels = {
	description = "{{{langname}}} terms categorized by properties relating to spelling, pronunciation or meaning.",
	umbrella_parents = "Fundamental",
	parents = {{name = "{{{langcat}}}", raw = true}}
}

labels = {
	description = "{{{langname}}} terms that have at least two meanings, one of which is a hyponym of the other.",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms categorized by their quantities of characters.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms that have two opposite meanings.",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms which include a ], either etymologically or in a definition.",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms that have different meanings depending on their etymology and/or on how they are pronounced.",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms that are spelled without a ].",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms that are spelled without a ].",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms whose characters are read equally both from left to right and vice versa, normally ignoring spaces, ]s and punctuation.",
	parents = {"terms by their sequences of characters"},
}

labels = {
	description = "{{{langname}}} compound terms where the head is a hyponym of its other part and whose other part is its synonym.",
	parents = {"terms by lexical property", "compound terms"},
}

labels = {
	description = "{{{langname}}} compound adjectives where the head is a hyponym of its other part and whose other part is its synonym.",
	parents = {"pleonastic compounds", "compound adjectives"},
}

labels = {
	description = "{{{langname}}} compound nouns where the head is a hyponym of its other part and where the head is the synonym for the whole.",
	parents = {"pleonastic compounds", "compound nouns"},
}

labels = {
	description = "{{{langname}}} terms spelled to represent a pronunciation, often a nonstandard one.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} phrases that repeat the same idea or concept using the same words.",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms categorized by properties relating to ] or ].",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms that can be spelled on a ] display, as found on pocket calculators, by turning numbers upside-down.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} words categorized by number of syllables.",
	parents = {"terms by phonemic property"},
}

labels = {
	description = "{{{langname}}} terms categorized by whether they include certain individual characters.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms categorized by whether they include certain sequences of characters.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} words categorized by the number of consecutive instances of the same letter they contain.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms containing ].",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms containing ]s.",
	parents = {"terms by orthographic property", "terms spelled with numbers"},
}

labels = {
	description = "{{{langname}}} terms where the spelling represents a variant pronunciation that differs from (one of) the current standard pronunciation(s).",
	parents = {"terms by orthographic property", "terms by phonemic property"},
}

labels = {
	description = "{{{langname}}} terms that have one or more ]: other terms that are pronounced in the same way but spelled differently.",
	parents = {"terms by lexical property"},
}

labels = {
	description = "{{{langname}}} terms where the adjective follows the noun. These adjectives within these terms are sometimes referred to as postpositive or postnominal adjectives.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms that are written in a different, non-native script.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} terms that are written using more than one script.",
	parents = {"terms by orthographic property"},
}

labels = {
	description = "{{{langname}}} individual words consisting of exactly one letter. They have meaning(s) other than their letter or the shape of their letter which are not abbreviations, names, numbers or symbols.",
	parents = {name = "character counts", sort = "1"},
}

labels = {
	description = "{{{langname}}} individual words composed of exactly two letters. They have meaning(s) beyond their component letters that are neither names nor abbreviations.",
	parents = {name = "character counts", sort = "2"},
}

labels = {
	description = "{{{langname}}} individual words composed of exactly three letters. They have meaning(s) beyond their component letters that are neither names nor abbreviations.",
	parents = {name = "character counts", sort = "3"},
}

labels = {
	description = "{{{langname}}} abbreviations composed of exactly two letters.",
	parents = {name = "character counts", sort = "2"},
}

labels = {
	description = "{{{langname}}} abbreviations composed of exactly three letters.",
	parents = {name = "character counts", sort = "3"},
}

labels = {
	description = "{{{langname}}} abbreviations composed of exactly four letters.",
	parents = {name = "character counts", sort = "4"},
}

labels = {
	description = "{{{langname}}} abbreviations composed of exactly five letters.",
	parents = {name = "character counts", sort = "5"},
}

labels = {
	description = "{{{langname}}} terms categorized by properties relating to ] and ].",
	parents = {"terms by lexical property"},
}


-- Add 'umbrella_parents' key if not already present.
for _, data in pairs(labels) do
	if not data.umbrella_parents then
		data.umbrella_parents = "Terms by lexical property subcategories by language"
	end
end


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


raw_categories = {
	description = "Umbrella categories covering topics related to terms categorized by their lexical properties, such as palindromes and number of letters or syllables in a word.",
	additional = "{{{umbrella_meta_msg}}}",
	parents = {
		"Umbrella metacategories",
		{name = "terms by lexical property", is_label = true, sort = " "},
	},
}

raw_categories = {
	description = "Umbrella categories covering topics related to words categorized by their number of syllables.",
	additional = "{{{umbrella_meta_msg}}}",
	parents = {
		"Umbrella metacategories",
		{name = "words by number of syllables", is_label = true, sort = " "},
	},
}

raw_categories = {
	description = "Umbrella categories covering topics related to terms categorized by the number of consecutive instances of the same letter they contain.",
	additional = "{{{umbrella_meta_msg}}}",
	parents = {
		"Umbrella metacategories",
		{name = "terms with consecutive instances of the same letter", is_label = true, sort = " "},
	},
}


-----------------------------------------------------------------------------
--                                                                         --
--                                 HANDLERS                                --
--                                                                         --
-----------------------------------------------------------------------------


insert(handlers, function(data)
	local number = data.label:match("^(%d*)%-syllable words$")
	if not number then
		return
	end
	return {
		description = "{{{langname}}} words that are pronounced in " .. number .. " syllable" .. (number == "1" and "" or "s") .. ".",
		breadcrumb = number,
		umbrella_parents = "Words by number of syllables subcategories by language",
		parents = {{
			name = "words by number of syllables",
			sort = ("#%03d"):format(number),
		}}
	}
end)

insert(handlers, function(data)
	local number = data.label:match("^terms with (%d*) consecutive instances of the same letter$")
	if not number then
		return
	end
	return {
		description = "{{{langname}}} terms containing " .. number .. " consecutive instances of the same letter.",
		breadcrumb = number,
		umbrella_parents = "Terms with consecutive instances of the same letter subcategories by language",
		parents = {{
			name = "terms with consecutive instances of the same letter",
			sort = ("#%03d"):format(number),
		}}
	}
end)


return {LABELS = labels, RAW_CATEGORIES = raw_categories, HANDLERS = handlers}