Module:Footnotes/anchor id list/sandbox: Difference between revisions

Revision as of 12:34, 5 December 2024

Latest revision as of 10:15, 24 May 2025

(19 intermediate revisions by 2 users not shown)

Line 2:

local data = mw.loadData ('Module:Footnotes/anchor id list/data');

local whitelist = mw.loadData ('Module:Footnotes/whitelist/sandbox');

local Lang_obj = mw.language.getContentLanguage(); -- used by template_list_addtemplate_name_get() to uppercase first letter of template name TODO: better way to do that?

local redirects_date = {

Line 92:

['Citation-attribution'] = true,

}

local global_article_content = nil

local Article_content;

local anchor_id_listglobal_anchor_id_list = {}; nil -- exported tables

local global_template_list = nil

local global_article_whitelist = nil

local anchor_id_list = {}; -- exported tables

local template_list = {};

local article_whitelist = {};

local template_list_done = false

Line 108:

local function article_content_get ()

if global_article_content then return global_article_content end

if not Article_content then

Article_content local article_content = mw.title.getCurrentTitle():getContent() or ''; -- get the content of the article or ''; new pages edited w/ve do not have 'content' until saved; ve does not preview; phab:T221625

for _, tag in ipairs (patterns_tags) do

Article_content article_content = Article_contentarticle_content:gsub (tag, ''); -- remove certain html-like tags and their content

end

end

global_article_content = article_content

return article_content

end

Line 373 ⟶ 374:

end

--[[--------------------------< T E M P L A T E _ LN IA SM TE _ AF DR O M _ M O D U L E >--------------------------------------------

if passed a module invocation, return the name of the template represented. Otherwise return the input.

{{#invoke:cite|foo|...}} or {{#invoke:cite bar||...}} will return "cite foo" and "cite bar", respectively.

]]

local function template_name_from_module (template, template_name)

if template_name and template_name:match ('^#invoke%s*:') then -- handle directly-invoked citation modules

template_name = template_name:match ('^#invoke%s*:%s*(.+)'); -- get module name

local func_name = template:match ('^{{[^|}]+%|%s*([^/|}]*)'); -- get function name

if template_name and func_name then -- valid module invocation

return template_name:gsub ('%s+$', '') .. ' ' .. func_name; -- ensure exactly one trailing whitespace between module and function name

end;

return nil -- could not get module and function name

end;

return template_name

end

--[[--------------------------< T E M P L A T E _ N A M E _ G E T >--------------------------------------------

Line 386 ⟶ 406:

local template_name = template:match ('^{{%s*([^/|}]+)'); -- get template name; ignore subpages ~/new, ~/sandbox; parser functions

template_name = template_name_from_module (template, template_name); -- if passed a module invocation, return the name of the template represented

if not template_name or template_name:match ('^#') then -- parser functions, magic words don't count as templates

return nil; -- could not get template name from (possibly corrupt) template; extraneous opening { mid template can cause this;

Line 521 ⟶ 543:

date_default = wrap_data[vol][2]; -- get the default date

end

if 'harv' == ref or not ref then -- |ref=harv specified or |ref= missing or empty

anchor_id = names_get (params, aliases_contributor) or -- get contributor, author, or editor names

Line 550 ⟶ 573:

return anchor_id; -- anchor_id text; nil else

end

Line 689 ⟶ 709:

list_add (anchor_id, anchor_id_list, true); -- add anchor ID to the list

end

end

end

--[[--------------------------< T E M P L A T E _ L I S T _ A D D >--------------------------------------------

makes a list of templates use in the article.

]]

local function template_list_add (template)

local template = template:match ('{{%s*(.-)[|}]'); -- keep the case of the template - this is different from template_name_get()

if template and not template:match ('^#') then -- found a template or magic word; ignore magic words

template=mw.text.trim (template); -- trim whitespace

template = Lang_obj:ucfirst (template); -- first character in template name must be uppercase (same as canonical template name) TODO: better way to do this?

list_add (template, template_list); -- add to list with (unused) tally

end

end

Line 725 ⟶ 729:

local function anchor_id_list_make ()

return local anchor_id_list; = {}

article_content_get (); -- attempt to get this article's content

local template_list = {};

local article_whitelist = {};

local article_content = article_content_get (); -- attempt to get this article's content

if ''article_content == Article_content'' then -- when there is no article content

return ''; -- no point in continuing

end

Line 736 ⟶ 743:

local find_pattern = '%f[{]{{[^{]';

local tstart, tend = Article_contentarticle_content:find (find_pattern); -- find the first template; do not find template variables: {{{template var|}}}

while tstart do

template = Article_contentarticle_content:match ('%b{}', tstart); -- get the whole template

if not template then

break; -- template is nil for some reason (last template missing closing }} for example) so declare ourselves done

end

template_name = template_name_get (template); -- get first char uppercase trimmed template name; ignore subpages ~/new, ~/sandbox

list_add (templatetemplate_name, template_list); -- add tothis listtemplate's withname (unused)to tallythe list

mw.logObject(template_name,'DEBUG template_name')

template_list_add (template); -- add this template's name to the list

if data.known_templates_cs12 [template_name] then

mw.log('make_cs12')

anchor_id = anchor_id_make_cs12 (template); -- extract an anchor id from this template

list_add (anchor_id, anchor_id_list, true)

elseif data.known_templates_vcite [template_name] then

mw.log('vcite')

local ref = template:match ('|%s*ref%s*=%s*(%b{})'); -- first look for |ref={{sfnref}} or |ref={{harvid}} because we will strip templates from the vcite template

if ref then -- |ref={{template}}

Line 772 ⟶ 775:

elseif data.known_templates_harvc [template_name] then

mw.log('harvc')

anchor_id = anchor_id_make_harvc (template); -- extract an anchor id from this template

list_add (anchor_id, anchor_id_list, true);

elseif data.known_templates_wikicite [template_name] then

mw.log('wikicite')

local ref = template:match ('|%s*ref%s*=%s*(%b{})'); -- first look for |ref={{sfnref}} or |ref={{harvid}}

Line 802 ⟶ 803:

elseif data.known_templates_anchor [template_name] then

mw.log('anchor')

anchor_id_make_anchor (template, anchor_id_list); -- extract anchor ids from this template if any

elseif data.known_templates_sfn_whitelist [template_name] then

mw.log('sfn_whitelist')

template = template:gsub ('^{{[^|]+|', ''):gsub ('}}$', '', 1); -- remove outer {{ and }} and template name

template = mw.text.trim (template, '%s|'); -- trim leading trailing white space and pipes

Line 819 ⟶ 818:

elseif template_name and whitelist.wrapper_templates[template_name] then

mw.log('wrapper_template')

anchor_id = anchor_id_make_wrapper (template); -- extract an anchor id from this template if possible

list_add (anchor_id, anchor_id_list, true);

Line 825 ⟶ 823:

elseif template_name and template_name:match ('^Cit[ea]') then -- not known, not known wrapper; last gasp, try as cs1-like

mw.log('cite like')

anchor_id = anchor_id_make_cs12 (template); -- extract an anchor id from this template if possible

list_add (anchor_id, anchor_id_list, true);

end

tstart, tend = Article_contentarticle_content:find (find_pattern, tend); -- search for another template; begin at end of last search

end

template_list_done = true

mw.logObject (anchor_id_list, 'anchor_id_list');

Line 838 ⟶ 834:

mw.logObject (article_whitelist, 'article_whitelist');

return anchor_id_list;

global_anchor_id_list = anchor_id_list

global_template_list = template_list

global_article_whitelist = article_whitelist

end

Line 850 ⟶ 849:

local function citeref_patterns_make()

if not template_list_doneglobal_template_list then error("Template list not yet created")return end

local citeref_patterns = {}

local template_patterns = whitelist.wrapper_template_patterns

for _, p in ipairs(template_patterns) do

for _, t in ipairs(p[1]) do -- loop through list of template wrappers

if template_listglobal_template_list[t] then -- if wrapper is found in article, record corresponding patterns

for _, pat in ipairs(p[2]) do

table.insert(citeref_patterns, pat)

Line 871 ⟶ 870:

]]

-- First create global_anchor_id_list, global_template_list, global_article_whitelist

anchor_id_list_make()

-- Then stuff them (and derived tables) into return table

return {

anchor_id_list = anchor_id_list_make()global_anchor_id_list or {}, -- table of anchor ids available in this article

article_whitelist = article_whitelistglobal_article_whitelist or {}, -- table of anchor ids with false-positive error message to be suppressed

template_list = template_listglobal_template_list or {}, -- table of templates used in this article

citeref_patterns = citeref_patterns_make() or {}, -- table of Lua patterns to search for citeref from wrappers

}