local export = {}
export.STATUS = {
OK = "ok",
INLINE = "inline",
MISSING = "missing",
REDLINK = "redlink",
AMBIGUOUS = "ambiguous",
}
export.TRANSITIVE = {
ALWAYS = "always", -- always recurse into children
NEVER = "never", -- never recurse into children
CROSS_LANG = "cross_lang", -- only recurse when source lang differs from target lang (but pos chain continues)
CROSS_LANG_NO_INTERNAL_SOURCE = "cross_lang_no_internal_source", -- like CROSS_LANG, but source breaks for internal derivations in the same language context
}
-- Deep merge tables (nested tables are merged recursively, later values override earlier)
local function deep_merge(...)
local result = {}
for _, t in ipairs({ ... }) do
for k, v in pairs(t) do
if type(v) == "table" and type(result[k]) == "table" then
result[k] = deep_merge(result[k], v)
else
result[k] = v
end
end
end
return result
end
local function make_glossary_link(term, display_text)
if not term then return display_text end
return "[[Appendix:Glossary#" .. term:gsub(" ", "_") .. "|" .. display_text .. "]]"
end
-- Extract base word and connector from text like "Borrowed from" or "calque of"
local function split_glossary_text(text)
for _, pattern in ipairs({ "^(.-)(%s+[Oo][Ff])$", "^(.-)(%s+[Ff][Rr][Oo][Mm])$" }) do
local base, rest = text:match(pattern)
if base then return base, rest end
end
return text, ""
end
local TRANSITIVE = export.TRANSITIVE
local function create_keyword(opts)
local entry = {
is_group = opts.is_group or false,
abbrev = opts.abbrev,
glossary = opts.glossary,
transitive = opts.transitive or TRANSITIVE.ALWAYS, -- default "always"
inherited_chain = opts.inherited_chain or false,
affix_categories = opts.affix_categories or false,
borrowing_type = opts.borrowing_type,
specialized_borrowing = opts.specialized_borrowing,
toplevel_category = opts.toplevel_category,
no_child_categories = opts.no_child_categories or false,
source_category_type = opts.source_category_type,
invisible = opts.invisible or false,
pos_override = opts.pos_override,
new_sentence = opts.new_sentence or false,
separate_clause = opts.separate_clause or false,
aliases = opts.aliases,
}
-- Only set text/phrase for visible keywords
if not opts.invisible then
entry.phrase = opts.phrase
if opts.text then
if opts.glossary then
local base_word, rest = split_glossary_text(opts.text)
entry.text = make_glossary_link(opts.glossary, base_word) .. rest
else
entry.text = opts.text
end
end
end
return entry
end
-- Shared defaults for keyword groups
local DEFAULTS = {
-- Keywords that pass through inheritance chain
inheritance = {
transitive = TRANSITIVE.ALWAYS,
inherited_chain = true,
},
-- Standard transitive derivation
transitive = {
transitive = TRANSITIVE.ALWAYS,
},
-- Standard for internal derivations: transitive across languages, but not within them
internal_derivation = {
transitive = TRANSITIVE.CROSS_LANG,
},
-- Borrowing keywords
borrowing = {
transitive = TRANSITIVE.ALWAYS,
},
-- Affix group keywords (compound words, blends, etc.)
affix_group = {
is_group = true,
transitive = TRANSITIVE.CROSS_LANG,
affix_categories = true,
},
-- Calque-like keywords (calque, partial calque, semantic loan)
calque_like = {
transitive = TRANSITIVE.NEVER,
no_child_categories = true,
new_sentence = true,
},
-- Non-transitive influence
influence_like = {
transitive = TRANSITIVE.NEVER,
no_child_categories = true,
},
}
export.keywords = {
--
-- Inheritance keywords
--
["from"] = create_keyword(deep_merge(DEFAULTS.internal_derivation, {
text = "From", phrase = "from",
})),
["inherited"] = create_keyword(deep_merge(DEFAULTS.inheritance, {
text = "Inherited from",
phrase = "from",
glossary = "inherited",
aliases = { "inh" },
})),
--
-- Basic derivation keywords
--
["uder"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "From",
phrase = "from",
toplevel_category = "undefined derivations",
})),
["derived"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Derived from",
phrase = "from",
abbrev = "der.",
glossary = "derived terms",
aliases = { "der" },
})),
--
-- Affix/compound group keywords
--
["affix"] = create_keyword(deep_merge(DEFAULTS.affix_group, {
text = "From",
phrase = "from",
aliases = { "af" },
})),
["blend"] = create_keyword(deep_merge(DEFAULTS.affix_group, {
text = "Blend of",
phrase = "a blend of",
abbrev = "blend",
glossary = "blend",
toplevel_category = "blends",
})),
["univerbation"] = create_keyword(deep_merge(DEFAULTS.affix_group, {
text = "Univerbation of",
phrase = "univerbation of",
abbrev = "univ.",
glossary = "univerbation",
toplevel_category = "univerbations",
aliases = { "univ" },
})),
--
-- Borrowing keywords
--
["bor"] = create_keyword(deep_merge(DEFAULTS.borrowing, {
text = "Borrowed from",
phrase = "borrowed from",
abbrev = "bor.",
glossary = "loanword",
borrowing_type = "borrowed",
aliases = { "borrowed" },
})),
["lbor"] = create_keyword(deep_merge(DEFAULTS.borrowing, {
text = "Learned borrowing from",
phrase = "a learned borrowing from",
abbrev = "lbor.",
glossary = "learned borrowing",
specialized_borrowing = "learned",
})),
["obor"] = create_keyword(deep_merge(DEFAULTS.borrowing, {
text = "Orthographic borrowing from",
phrase = "an orthographic borrowing from",
abbrev = "obor.",
glossary = "orthographic borrowing",
specialized_borrowing = "orthographic",
})),
["slbor"] = create_keyword(deep_merge(DEFAULTS.borrowing, {
text = "Semi-learned borrowing from",
phrase = "a semi-learned borrowing from",
abbrev = "slbor.",
glossary = "semi-learned borrowing",
specialized_borrowing = "semi-learned",
})),
["ubor"] = create_keyword(deep_merge(DEFAULTS.borrowing, {
text = "Unadapted borrowing from",
phrase = "an unadapted borrowing from",
abbrev = "ubor.",
glossary = "unadapted borrowing",
specialized_borrowing = "unadapted",
})),
--
-- Calque-like keywords (non-transitive, start new sentence)
--
["calque"] = create_keyword(deep_merge(DEFAULTS.calque_like, {
text = "Calque of",
phrase = "a calque of",
abbrev = "calq.",
glossary = "calque",
specialized_borrowing = "calque",
aliases = { "cal", "clq" },
})),
["partial calque"] = create_keyword(deep_merge(DEFAULTS.calque_like, {
text = "Partial calque of",
phrase = "a partial calque of",
abbrev = "pcalq.",
glossary = "partial calque",
specialized_borrowing = "partial-calque",
aliases = { "pcal" },
})),
["semantic loan"] = create_keyword(deep_merge(DEFAULTS.calque_like, {
text = "Semantic loan of",
phrase = "a semantic loan of",
abbrev = "sl.",
glossary = "semantic loan",
specialized_borrowing = "semantic-loan",
aliases = { "sl" },
})),
--
-- Influence keywords (non-transitive, separate clause)
--
["influence"] = create_keyword(deep_merge(DEFAULTS.influence_like, {
text = "Influenced by",
phrase = "influenced by",
abbrev = "influ.",
glossary = "contamination",
separate_clause = true,
})),
--
-- Morphological derivation keywords
--
["clipping"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Clipping of",
phrase = "clipping of",
abbrev = "clip.",
glossary = "clipping",
toplevel_category = "clippings",
aliases = { "clip" },
})),
["ellipsis"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Ellipsis of",
phrase = "ellipsis of",
abbrev = "ellip.",
glossary = "ellipsis",
toplevel_category = "ellipses",
aliases = { "ellip" },
})),
["back-formation"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Back-formation from",
phrase = "a back-formation from",
abbrev = "bf.",
glossary = "back-formation",
toplevel_category = "back-formations",
aliases = { "bf" },
})),
["transliteration"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Transliteration of",
phrase = "borrowed from",
abbrev = "translit.",
glossary = "transliteration",
aliases = { "translit" },
})),
["vrd"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Vṛddhi derivative of",
phrase = "a vṛddhi derivative of",
abbrev = "vṛd.",
glossary = "vṛddhi derivative",
})),
["apheretic"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Apheretic form of",
phrase = "an apheretic form of",
abbrev = "aph.",
glossary = "apheresis",
aliases = { "apheresis", "aphetic" },
})),
["denominal"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Denominal verb from",
phrase = "denominal verb from",
abbrev = "denom.",
glossary = "denominal",
toplevel_category = "denominal verbs",
aliases = { "denom" },
})),
["deverbal"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Deverbal from",
phrase = "deverbal from",
abbrev = "deverb.",
glossary = "deverbal",
toplevel_category = "deverbals",
})),
["reduplication"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Reduplication of",
phrase = "reduplication of",
abbrev = "redup.",
glossary = "reduplication",
toplevel_category = "reduplications",
aliases = { "redup" },
})),
["abbreviation"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Abbreviation of",
phrase = "abbreviation of",
abbrev = "abbr.",
glossary = "abbreviation",
aliases = { "abbr" },
})),
["acronym"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Acronym of",
phrase = "acronym of",
abbrev = "acronym",
glossary = "acronym",
aliases = { "acro" },
})),
["initialism"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Initialism of",
phrase = "initialism of",
abbrev = "init.",
glossary = "initialism",
aliases = { "init" },
})),
["metathesis"] = create_keyword(deep_merge(DEFAULTS.transitive, {
text = "Metathesis of",
phrase = "metathesis of",
abbrev = "meta.",
glossary = "metathesis",
toplevel_category = "words derived through metathesis",
aliases = { "meta" },
})),
--
-- Invisible keywords (no text output)
--
["root"] = create_keyword {
transitive = TRANSITIVE.ALWAYS,
invisible = true,
pos_override = "root",
},
["afeq"] = create_keyword(deep_merge(DEFAULTS.affix_group, {
text = "From",
phrase = "from",
transitive = TRANSITIVE.NEVER,
invisible = true,
})),
}
local aliases_to_register = {}
local canonical_aliases = {}
for name, keyword_data in pairs(export.keywords) do
if keyword_data.aliases then
canonical_aliases[name] = keyword_data.aliases
for _, alias in ipairs(keyword_data.aliases) do
if export.keywords[alias] then
error("Alias '" ..
alias .. "' defined in keyword '" .. name .. "' collides with existing keyword '" .. alias .. "'.")
end
if aliases_to_register[alias] then
error("Alias '" ..
alias .. "' defined in keyword '" .. name .. "' is already claimed by another keyword.")
end
aliases_to_register[alias] = keyword_data
end
keyword_data.aliases = nil
end
end
for alias, data in pairs(aliases_to_register) do
export.keywords[alias] = data
end
--
-- Language exception presets
--
local EXCEPTION_PRESETS = {
-- Fully disallowed: no tree, no text, no categories
disallowed = {
disallow = { tree = true, text = true },
suppress_categories = true,
},
-- Suppress transliteration only
no_translit = {
suppress_tr = true,
},
-- Suppress all categories only
no_categories = {
suppress_categories = true,
},
}
--[=[
Available exception options:
disallow = { Related options for disallowing output:
tree Disallow etymology trees for this language
text Disallow etymology text generation for this language
ref Reference link shown when tree/text is disallowed
}
suppress_tr Suppress transliteration in links
suppress_categories Suppress all category generation
normalize_to Normalize language code to a different code
normalize_from_families Apply normalization to languages in these families
normalize_exclude_families Exclude these families from normalization
keyword_overrides Per-keyword categorisation overrides (e.g. { ["af"] = { transitive = TRANSITIVE.NEVER } })
]=]
local function create_exception(preset, overrides)
local base = preset and EXCEPTION_PRESETS[preset] or {}
return deep_merge(base, overrides or {})
end
export.config = {
lang_exceptions = {
["zh"] = create_exception("disallowed", {
disallow = { ref = "[[Wiktionary:Beer parlour/2025/May#Template:etymon for Chinese]]" },
suppress_tr = true,
normalize_to = "zh",
normalize_from_families = { "zhx" },
normalize_exclude_families = { "qfa-cnt" },
}),
},
}
--
-- Propagate keyword overrides to aliases
--
if export.config.lang_exceptions then
for lang_code, exception in pairs(export.config.lang_exceptions) do
if exception.keyword_overrides then
for canonical, aliases in pairs(canonical_aliases) do
if exception.keyword_overrides[canonical] then
local override_data = exception.keyword_overrides[canonical]
for _, alias in ipairs(aliases) do
if not exception.keyword_overrides[alias] then
exception.keyword_overrides[alias] = override_data
end
end
end
end
end
end
end
return export
