47,726
edits
 | We're back! Sorry, bad combo of sickness, funeral and a month-long trip abroad. The site is back now.  Χρυσοφύλαξ 00:44, 12 March 2025 (UTC) |
Module:headword: Difference between revisions
Jump to navigation
Jump to search
no edit summary
Tag: Undo |
No edit summary |
||
| Line 2: | Line 2: | ||
-- Named constants for all modules used, to make it easier to swap out sandbox versions. | -- Named constants for all modules used, to make it easier to swap out sandbox versions. | ||
local debug_track_module = "Module:debug/track" | |||
local gender_and_number_module = "Module:getn" | local gender_and_number_module = "Module:getn" | ||
local headword_data_module = "Module:headword/data" | local headword_data_module = "Module:headword/data" | ||
| Line 46: | Line 47: | ||
return (rsubn(term, foo, bar)) | return (rsubn(term, foo, bar)) | ||
end | end | ||
-- Add a tracking category to track entries with certain (unusually undesirable) properties. `track_id` is an identifier | |||
-- for the particular property being tracked and goes into the tracking page. Specifically, this adds a link in the | |||
-- page text to [[Wiktionary:Tracking/headword/TRACK_ID]], meaning you can find all entries with the `track_id` property | |||
-- by visiting [[Special:WhatLinksHere/Wiktionary:Tracking/headword/TRACK_ID]]. | |||
-- | |||
-- If `lang` (a language object) is given, an additional tracking page [[Wiktionary:Tracking/headword/TRACK_ID/CODE]] is | |||
-- linked to where CODE is the language code of `lang`, and you can find all entries in the combination of `track_id` | |||
-- and `lang` by visiting [[Special:WhatLinksHere/Wiktionary:Tracking/headword/TRACK_ID/CODE]]. This makes it possible to | |||
-- isolate only the entries with a specific tracking property that are in a given language. Note that if `lang` | |||
-- references at etymology-only language, both that language's code and its full parent's code are tracked. | |||
local function track(track_id, lang) | |||
local tracking_page = "headword/" .. track_id | |||
local m_debug_track = require(debug_track_module) | |||
if lang and lang:hasType("etymology-only") then | |||
m_debug_track{tracking_page, tracking_page .. "/" .. lang:getCode(), | |||
tracking_page .. "/" .. lang:getFullCode()} | |||
elseif lang then | |||
m_debug_track{tracking_page, tracking_page .. "/" .. lang:getCode()} | |||
else | |||
m_debug_track(tracking_page) | |||
end | |||
return true | |||
end | |||
local function text_in_script(text, script_code) | local function text_in_script(text, script_code) | ||
| Line 243: | Line 269: | ||
else | else | ||
head_parts = concat(head_parts) | head_parts = concat(head_parts) | ||
end | |||
if has_manual_translits then | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/has-manual-translit]] | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/has-manual-translit/LANGCODE]] | |||
track("has-manual-translit", data.lang) | |||
end | end | ||
| Line 347: | Line 379: | ||
-- right into the 'data' table to disable inflection links of the entire headword | -- right into the 'data' table to disable inflection links of the entire headword | ||
-- when inflected forms aren't entry-worthy, e.g.: in Vulgar Latin | -- when inflected forms aren't entry-worthy, e.g.: in Vulgar Latin | ||
local nolinkinfl = data.nolinkinfl | local nolinkinfl = part.face == "hypothetical" or (part.nolink and track("nolink") or part.nolinkinfl) or ( | ||
data.nolink and track("nolink") or data.nolinkinfl) | |||
local formatted | local formatted | ||
| Line 593: | Line 626: | ||
local langcode = data.lang:getCode() | local langcode = data.lang:getCode() | ||
local full_langcode = | local full_langcode = data.lang:getFullCode() | ||
local langname = data.lang:getCanonicalName() | local langname = data.lang:getCanonicalName() | ||
local full_langname = | local full_langname = data.lang:getFullName() | ||
local raw_pagename, page = data.pagename | local raw_pagename, page = data.pagename | ||
| Line 623: | Line 656: | ||
-- convert old-style `heads`, `translits` and `transcriptions` to new-style | -- convert old-style `heads`, `translits` and `transcriptions` to new-style | ||
local maxind = math.max( | local maxind = math.max( | ||
init_and_find_maximum_index(data, "heads" | init_and_find_maximum_index(data, "heads"), | ||
init_and_find_maximum_index(data, "translits", true), | init_and_find_maximum_index(data, "translits", true), | ||
init_and_find_maximum_index(data, "transcriptions", true) | init_and_find_maximum_index(data, "transcriptions", true) | ||
| Line 643: | Line 676: | ||
------------ 4. Initialize and validate `data.categories` and `data.whole_page_categories`, and determine `pos_category` if not given, and add basic categories. ------------ | ------------ 4. Initialize and validate `data.categories` and `data.whole_page_categories`, and determine `pos_category` if not given, and add basic categories. ------------ | ||
init_and_find_maximum_index(data, "categories" | init_and_find_maximum_index(data, "categories") | ||
init_and_find_maximum_index(data, "whole_page_categories" | init_and_find_maximum_index(data, "whole_page_categories") | ||
local pos_category_already_present = false | local pos_category_already_present = false | ||
if #data.categories > 0 then | if #data.categories > 0 then | ||
local escaped_langname = pattern_escape(full_langname) | local escaped_langname = pattern_escape(full_langname) | ||
local matches_lang_pattern = "^" .. escaped_langname .. " " | local matches_lang_pattern = "^" .. escaped_langname .. " " | ||
for _, cat in ipairs(data.categories) do | |||
-- Does the category begin with the language name? If not, tag it with a tracking category. | |||
if not cat:find(matches_lang_pattern) then | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/no lang category]] | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/no lang category/LANGCODE]] | |||
track("no lang category", data.lang) | |||
end | |||
end | |||
-- If `pos_category` not given, try to infer it from the first specified category. If this doesn't work, we | -- If `pos_category` not given, try to infer it from the first specified category. If this doesn't work, we | ||
| Line 679: | Line 720: | ||
-- add an appropriate category. | -- add an appropriate category. | ||
local postype = export.pos_lemma_or_nonlemma(data.pos_category) | local postype = export.pos_lemma_or_nonlemma(data.pos_category) | ||
if not data.noposcat then | if not postype then | ||
-- We don't know what this category is, so tag it with a tracking category. | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/unrecognized pos]] | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/unrecognized pos/LANGCODE]] | |||
track("unrecognized pos", data.lang) | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/unrecognized pos/POS]] | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/unrecognized pos/POS/LANGCODE]] | |||
track("unrecognized pos/pos/" .. data.pos_category, data.lang) | |||
elseif not data.noposcat then | |||
insert(data.categories, 1, full_langname .. " " .. postype .. "s") | insert(data.categories, 1, full_langname .. " " .. postype .. "s") | ||
end | end | ||
| Line 725: | Line 774: | ||
if head_term:sub(1, 1) ~= "*" then | if head_term:sub(1, 1) ~= "*" then | ||
error("The headword '" .. head_term .. "' must begin with '*' to indicate that it is reconstructed.") | error("The headword '" .. head_term .. "' must begin with '*' to indicate that it is reconstructed.") | ||
end | |||
end | |||
------ 6b. Try to detect the script(s) if not provided. If a per-head script is provided, that takes precedence, | |||
------ otherwise fall back to the overall script if given. If neither given, autodetect the script. | |||
local auto_sc = data.lang:findBestScript(head.term) | |||
if ( | |||
auto_sc:getCode() == "None" and | |||
require(scripts_module).findBestScriptWithoutLang(head.term):getCode() ~= "None" | |||
) then | |||
insert(data.categories, full_langname .. " terms in nonstandard scripts") | |||
end | |||
if not (head.sc or data.sc) then -- No script code given, so use autodetected script. | |||
head.sc = auto_sc | |||
else | |||
if not head.sc then -- Overall script code given. | |||
head.sc = data.sc | |||
end | |||
-- Track uses of sc parameter. | |||
if head.sc:getCode() == auto_sc:getCode() then | |||
insert(data.categories, full_langname .. " terms with redundant script codes") | |||
else | |||
insert(data.categories, full_langname .. " terms with non-redundant manual script codes") | |||
end | end | ||
end | end | ||
| Line 851: | Line 924: | ||
------------ 8. Insert additional categories. ------------ | ------------ 8. Insert additional categories. ------------ | ||
if data.force_cat_output then | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/force cat output]] | |||
track("force cat output") | |||
end | |||
if has_redundant_head_param then | if has_redundant_head_param then | ||
| Line 1,075: | Line 1,153: | ||
if page.namespace == "" and not data.lang:hasType("reconstructed") then | if page.namespace == "" and not data.lang:hasType("reconstructed") then | ||
local m_links = require(links_module) | local m_links = require(links_module) | ||
for _, head in ipairs(data.heads) do | |||
if page.full_raw_pagename ~= m_links.get_link_page(remove_links(head.term), data.lang, head.sc) then | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/pagename spelling mismatch]] | |||
-- [[Special:WhatLinksHere/Wiktionary:Tracking/headword/pagename spelling mismatch/LANGCODE]] | |||
track("pagename spelling mismatch", data.lang) | |||
break | |||
end | |||
end | |||
end | end | ||