Module:is-noun/builtin

From Wiktionary, the free dictionary
Jump to navigation Jump to search


local export = {}

--[=[

Authorship: Ben Wing <benwing2>

]=]

-- This contains "built-in" nouns, i.e. nouns (mostly irregular) whose declension is built into the module.
--
-- Individual entries in this table are of the form {KEY, VALUE, ENGLISH_DESC}, as follows:
-- 1. KEY is one of the following:
--    a. A string, the infinitive of the noun in question. If the string is preceded by ^, the user-specified noun
--       must exactly match this string; otherwise, it can contain an extra prefix, which is appended onto all
--       generated forms. For example, "[[fingur]]" will also match "[[baugfingur]]", "[[vísifingur]]", etc.; but
--       "^fingur" will only match [[fingur]] (FIXME: better example).
--    b. An object of the form {term = "TERM", prefixes = {"PREFIX", "PREFIX", ...}}. In this case, `term` specifies
--       the term to match, and `prefixes` specifies the allowed prefixes that can be prepended onto the term in order
--       for this entry to be considered a match. If a prefix is preceded by ^, it must match exactly; otherwise, longer
--       prefixes ending in the specified string can match. For example, the spec (FIXME: Update)
--      {term = "vedere", prefixes = {"prov", "pro"}} will match "provvedere", "provedere", and prefixed derivatives of
--      these nouns such as "riprovvedere"; but not "rivedere", "prevedere", "vedere" by itself, etc. Similarly, the
--      spec {term = "bere", prefixes = {"^", "ri", "tra"}} will match "bere" by itself (because of the "^"), as well as
--      "ribere", "trabere", "strabere" (because of "tra"), etc.; but not "iubere" or "ebere".
-- 2. VALUE is a string, which is of the same format as is normally used in {{is-ndecl}} and {{is-noun}} (inside of
--    angle brackets).
-- 3. ENGLISH_DESC is an English-language description of the nouns handled by this entry, such as
--    "compounds of <<altari|altar>> (not <<altari>> by itself)" or "!gust, squall" or
--    "patronymics with <<dóttir|daughter>> in them (defaults to {{cd|.both}}) for this reason". This is used in
--    generating the documentation describing the built-in nouns handled by the module. Double angle bracket specs such
--    as <<altari>> are converted into links, and are equivalent to e.g. {{m|is|altari}}. Text after the vertical bar
--    is a gloss, so that e.g. <<dóttir|daughter>> is converted to {{m|is|dóttir||daughter}}; but this doesn't happen if
--    there's an equal sign after the vertical bar, such as <<Brattavellir|pos=town name>>, which converts directly to
--    {{m|is|Brattavellir|pos=town name}}. Template references in the description will be expanded. The default is
--    "<<NOUN>> and compounds". If the description begins with an exclamation point, what follows is a gloss, and the 
--    default becomes "<<NOUN|GLOSS>> and compounds".
--
-- The order of the entries matters, as the entries are processed sequentially. The general ordering used is
-- alphabetical, but exceptions may be made to ensure the right precedence, e.g. "andare" precedes "dare" (FIXME:
-- Update).
export.builtin_nouns = {
	{"alin", "f.vstem:áln", "!ell"},
	{"altari", "n,,u/u.uUmut", "compounds of <<altari|altar>> (not <<altari>> by itself)"},
	{"auga", "n.genplna", "!eye"},
	{"bær", "m,jar,ir.j", "!farm; town"},
	{"baugur", "m.dati:-/-:i", "!ring"},
	{"bekkur", "m,jar:s,ir.j", "!bench"},
	{"belgur", "m,s:jar,ir.j", "!bellows; skin, hide"},
	{"bítur", "m,,ir:ar.dat-/-", "<<bítur|biter>> (mostly in compounds)"},
	{"^%u.*björn", "m,s:ar.unumut.imut", "names with the element <<björn!bear>>"},
	{"björn", "m,ar,ir.unumut.imut", "!bear"},
	{"bóla", "f.genplna:a", "!bubble; stud, tack"},
	{"bolur", "m,,ir.dat-/-", "!torso, trunk"},
	{"bót", "f,,^ur", "!cure; bight, inlet"},
	{"brella", "f.genplna", "!trick"},
	{"bróðir", "m.rstem", "!brother"},
	{"brú", "f,,^r", "!bridge"},
	{"bugur", "m,,ir:ar.dati/-", "!bend, curve"},
	{"bylur", "m,s:jar,jir.j", "!gust, squall"},
	{"dagur", "m.imut.dati/i", "!day"},
	{"dómur", "m.dati/-:i", "<<dómur|opinion; judgment; verdict, sentence>>, <<-dómur|-dom>> and compounds/derivatives of both"},
	{"dóttir", "f.rstem.both.plstem:dót", "patronymics with <<dóttir|daughter>> in them (defaults to {{cd|.both}}) for this reason"},
	{"drykkur", "m,jar,ir.j", "!drink"},
	{"dyr", "f.pl.decllemma:dyrir", "!door(way)"},
	{"eyra", "n.genplna", "!ear"},
	{"faðir", "m.rstem", "!father"},
	{"fé", "n.já.sg", "!sheep; cattle; money"},
	{"fingur", "m,,!#.#", "!fingur"},
	{"fiskur", "m.dati/i:-", "!fish"},
	{"fjörður", "m,ar,ir.unumut.imut", "!fjord"},
	{"flokkur", "m.dati/-", "!group; class, category"},
	{"fótur", "m,ar,^ur.imut.dati/i", "!foot"},
	{"friður", "m,ar.sg.dati/-:i[less common]", "!peace"},
	{"frömuður", "m,ar,ir.dati/i", "!promoter"},
	{"fruma", "f.genplna:a", "!cell"},
	{"fundur", "m,ar,ir", "!discovery, find; meeting"},
	{"fylli", "f,i:ar,ir:ar", "!fill, satiation"},
	{"gígur", "m,,ar:ir.dat-/-", "!crater"},
	{"grein", "f,,ar:ir", "<<grein|branch; article, piece; subject, field>> and compounds with two nominative/accusative plurals; those with plurals only in ''-ar'' should use {{cd|f,,ar}}"},
	{"gripur", "m,,ir.dat-/-", "!object, artifact; valuable object, precious item; domestic animal"},
	{"guð", "m,,ir.dati/i", "<<guð|god>> when masculine, and all compounds"},
	{"hamar", "m.con", "!hammer"},
	{"hérað", "n.uUmut,umut", "!region, district, hundred"},
	{"hjarta", "n.genplna", "!heart"},
	{"hnappur", "m.dat-:i/-:i", "!button; key (on a keyboard or typewriter)"},
	{"hnífur", "m.dat-:i/-", "!knife"},
	{"hönd", "f,,^ur.dat^i:-", "compounds of <<hönd|hand>> (not <<hönd>> by itself)"},
	{"hópur", "m.dat-:i/-", "!group, crowd"},
	{"hríð", "f,,ir:ar", "!storm, blizzard"},
	{"hringur", "m,,ir:ar.j,-j.dat-/-", "compounds of <<hringur|ring>> (not <<hringur>> by itself)"},
	{"hvalur", "m,,ir.dat-/-", "!whale"},
	{"hylur", "m,s:jar,jir:ir.j.dat-:i/-", "!stream pool"},
	{"jaðar", "m.con", "!edge"},
	{"jöfnuður", "m,ar.unuUmut,-unuUmut.dati/i.sg", "!equal distribution"},
	{"klauf", "f,,ir:ar", "!hoof; claw; slit, slash"},
	{"kona", "f.genpl!kvenna", "!woman"},
	{"könnuður", "m,ar,ir.dati/i", "!investigator"},
	{"læti", "n.pl.unimut", "''only'' for plural-only <<læti|behavior, demeanor>> and compounds; not for compounds in singular-only <<-læti>>"},
	{"laukur", "m.dat-:i/-", "!onion, leek, garlic; bulb"},
	{"leggur", "m,jar:s,ir.j", "!leg; stalk; bone"},
	{"leikur", "m,,ir.j", "compounds of <<leikur|game; play>> (not <<leikur>> by itself)"},
	{"lifur", "f,,ar.#.con.defcon", "!liver"},
	{"lús", "f,,^", "!louse"},
	{"mær", "f,,jar.decllemma:mey.j.datju:-", "!maiden"},
	{"má[fv]ur", "m.dati:-/i", "!gull, seagull"},
	{"mánuður", "m,ar,ir.unuUmut.dati/i", "!month"},
	{"markaður", "m,ar:s,ir.uUmut.dati/i:-", "!market"},
	{"maður", "m,s,^/^ir.decllemma:mannur", "!man"},
	{"mey", "f,,jar.j.datju:-", "!maiden"},
	{"móðir", "f.rstem", "!mother"},
	{"munur", "m,ar,ir.dat-/i", "compounds of <<munur|difference>> (not <<munur>> by itself)"},
	{"mús", "f,,^", "!mouse"},
	{"nátt", "f,,!nætur", "<<nátt|night>> (variant of <<nótt>>) and compounds"},
	{"nautur", "m,ar:s.dat-:i[much less common]/-:i[much less common]", "compounds of <<-nautur|companion>>"},
	{"nótt", "f,!nætur,!nætur", "compounds of <<nótt|night>> (not <<nótt>> by itself)"},
	{"öxi", "f,!axar:ar.plstem:ax", "!axe"},
	{"reið", "f,,ar:ir", "compounds of <<reið|ride>> (not <<reið>> by itself)"},
	-- NOTE: the following must follow [[brella]]
	{"rella", "f.genpla:na", "!pinwheel; small plane"},
	{"réttur", "m,ar,ir", "!law; court; course (of a meal)"},
	{"rót", "f,ar,^ur", "!root"},
	{"selur", "m,,ir.dat-/-", "!seal (animal)"},
	{"siður", "m,ar,ir.dat-/-:i", "compounds of <<siður|custom, habit; religion>> (but not <<siður>> by itself)"},
	{"skál", "f,,ar:ir", "compounds of <<skál|bowl, landscape depression>> (but not <<skál>> by itself)"},
	{"skápur", "m.dat-/-", "!closet; cabinet"},
	{"skilnaður", "m,ar,ir.both.dati/i", "!separation; divorce; departure"},
	{"skjár", "m,,ir.datplum:m[more common but sometimes proscribed]/um", "!video screen, monitor"},
	{"skógur", "m,ar.dati/i", "!forest"},
	{"skór", "m,,r.##.datplm.defgenplnna", "!shoe"},
	{"skrá", "f,r:ar", "!catalog, list; lock"},
	{"skurður", "m,ar,ir", "!cut, wound; incision"},
	{"sláttur", "m,ar,ir.imut", "!beat; pulse"},
	{"söfnuður", "m,ar,ir.-unuUmut,unuUmut.dati/i:-", "!congregration"},
	{"sóley", "f,,jar.j", "!buttercup"},
	{"son", "m,ar,ir.both.imut.imutval:y.dati/i", "patronymics with <<son|son>> in them (defaults to {{cd|.both}}) for this reason"},
	{"söngur", "m.v.dat-/-", "!song"},
	{"sonur", "m,ar,ir.imut.imutval:y.dati/i", "compounds of <<sonur|son>> (not <<sonur>> by itself)"},
	{"spónn", "m,,ir.imut.dati", "compounds of <<spónn|horn or bone spoon; veneer; (wood) shaving>> (not <<spónn>> by itself)"},
	{"stafur", "m,,ir.dat-/-", "!letter (of the alphabet); stick"},
	{"staður", "m,ar,ir.dat-/-", "!place"},
	{"stöð", "f,,var.v", "!station"},
	{"strengur", "m,s:jar,ir.j", "!string"},
	{"systir", "f.rstem", "!sister"},
	{"tá", "f,ar,^r", "!toe"},
	{"teigur", "m,,ar:ir.dat-:i/-", "!grassland, meadow"},
	{"töng", "f,,^ur:ir", "<<stöng|pole, rod>>, <<töng|pliers, tongs>> and compounds of both"},
	{"toppur", "m.dati/-:i", "!top; peak; lock (of hair)"},
	{"tré", "n.já", "!tree; wood"},
	{"þefur", "m,s:jar.sg.j", "!smell, stench"},
	{"þjófur", "m.dat-:i/-", "!thief"},
	{"þráður", "m,ar,ir.imut.dati/i", "!thread"},
	{"þröstur", "m,ar,ir.unumut.imut", "!thrush"},
	{"þurrka", "f.genplna", "!wipe; (windshield) wiper; (hair) dryer"},
	{"vangur", "m.dat-:i/-:i", "!field"},
	{"vaður", "m,ar:s,ir.dat-:i/-", "!rope; fishing line"},
	{"veiði", "f,i:ar,ar", "!catch; (plural) hunting, fishing"},
	{"verkur", "m,jar,ir.j", "!pain, ache"},
	{"vetur", "m,ar,!#.#", "!winter"},
	{"viður", "m,ar,ir.dat-:i/-:i", "<<viður|wood; forest; trees collectively>>, <<kviður|belly>> and compounds"},
	{"vinur", "m,ar,ir.dati:-/i", "!friend"},
	{"völlur", "m,ar,ir.unumut.imut", "!field, lawn"},
	{"vöndur", "m,ar,ir.unumut.imut", "!rod, birch, wand"},
}

return export