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

Revision as of 00:17, 21 May 2020

Revision as of 15:37, 21 May 2020

Line 1:

require('Module:No globals');

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

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

local anchor_id_list = {};

--local redirect_patterns_anchor = {

-- '{{%s*[Aa]nchor',

-- '{{%s*[Aa]nchor for redirect',

-- '{{%s*[Aa]nchors',

-- '{{%s*[Aa]nchro',

-- '{{%s*[Aa]ncor',

-- }

--local redirect_patterns_harvc = {

-- '{{%s*[Hh]arvc',

-- '{{%s*[Cc]itec',

-- }

local redirect_patterns_sfn_whitelist = {

'{{%s*[Ss]fn whitelist',

'{{%s*[Hh]arv whitelist',

}

--local redirect_patterns_vcite = {

-- '{{%s*[Vv]cite',

-- '{{%s*[Vv]ancite',

---- '{{%s*[Cc]it ', -- disabled 'cit journal & cit paper' redirect to vcite journal but 'cit book', 'cit new', 'cit web' are cs1 redirects

-- }

local redirects_citation = {

['citation'] = true,

['cite'] = true,

['cite citation'] = true,

['cite study'] = true,

['cite technical standard'] = true,

}

local redirects_date = {

['date'] = true,

Line 38 ⟶ 10:

['isotomos'] = true,

}

local redirects_patent = { -- special case cs1-like templates because uses different parameters for name and date in anchor ID

--local redirects_harvc = {

-- ['harvc'] = true,

-- ['citec'] = true,

-- }

local redirects_patent = {

['cite patent'] = true,

['citeref patent'] = true,

Line 118 ⟶ 86:

'<source.->.-</source>', -- deprecated alias of syntaxhighlight tag

}

local template_skip = { -- TODO: necessary to keep this? with known template tables templates not in those tables will be skipped

local template_skip = {

['citation-attribution'] = true,

}

local Article_content;

local anchor_id_list = {}; -- exported tables

local template_list = {};

local article_whitelist = {};

Line 406 ⟶ 377:

local function template_name_get (template)

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

local template_name = template:match ('{{%s*([^#][^#/|]+)'); -- get template name; ignore subpages ~/new, ~/sandbox; parser functions, magic words don't count as templates

Line 433 ⟶ 403:

for param, value in template:gmatch ('|%s*([^=]-)%s*=%s*([^|}]+)') do -- build a table of template parameters and their values

if value then -- there must be a value but when

if '' ~= value and not value:match ('^%s$') then -- skip when value is empty string or only whitespace

params[param] = mw.text.trim (value); -- add trimmed value else

Line 449 ⟶ 419:

local function anchor_id_make_harvc (template)

local date = date_get (template, alias_patterns_harvc_date); -- get date; done here because might be in {{date}}; return date if valid; empty string else

local anchor_id;

local params = {}; -- table of harvc parameters

Line 465 ⟶ 435:

end

anchor_id = names_get (params, aliases_harvc_author); -- get the harvc contributor names

if anchor_id then -- if names were gotten

return 'CITEREF' .. anchor_id .. date;

end

Line 474 ⟶ 444:

--[[--------------------------< CA IN TC EH O R E_ FI D _ M A K E _ C S 1 2 >------------------------------------------------------

for cs1|2 template and cs1-like templates

inspect |ref= to decide what to do:

|ref= - empty or missing: get names and date from template parameters because; all cs1|2 will soon create CITEREF anchor IDs

|ref=harv - get names and date from template parameters

|ref={{SfnRef|name|name|name|name|year}} - assemble an anchor id from {{sfnref}} positional parameters

Line 483 ⟶ 455:

|ref=none - skip; do nothing because an anchor id intentionally suppressed; TODO: keep with a type code of '0'?

|ref=<text> - save param value because may match an anchor id override value in {{harv}} template |ref= parameter or {{harvc}} |id= parameter

this no longer applies; all cs1|2 will soon create CITEREF anchor IDs

|ref= - empty or missing

for cs1: skip

if |mode=cs2: spoof |ref=harv

for cs2: get names and date from template parameters

if |mode=cs1: skip

]]

local function anchor_id_makeanchor_id_make_cs12 (template)

local ref; -- content of |ref=

local template_name; -- name of the template for cs2 detection

local anchor_id; -- the assembled anchor id from this template

local date;

local params = {}; -- table of cs1|2template parameters

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

Line 506 ⟶ 471:

if redirects_patent[template_name] then

date = date_get (template, alias_patterns_patent_date); -- get date; done here because might be in {{date}}

else

date = date_get (template, alias_patterns_date);

Line 522 ⟶ 487:

end

end

-- this disabled because all cs1|2 templates will create CITEREF anchor IDs after next cs1|2 module-suite update

-- if not ref then -- here when |ref= missing or empty

-- if redirects_citation[template_name] then -- could be cs2

-- if template:match ('|%s*mode%s*=%s*cs1') then

-- return nil; -- |ref= missing or empty; citation template but |mode=cs1

-- else

-- ref = 'harv'; -- spoof to handle cs2 as if it were cs1 with |ref=harv

-- end

-- else -- |ref= missing or empty; not a cs2 template

-- if template:match ('|%s*mode%s*=%s*cs2') then

-- ref = 'harv'; -- |ref= missing or empty; not a cs2 template; |mode=cs2; spoof as if it were cs1 with |ref=harv

-- end

-- end

-- end

end

Line 545 ⟶ 495:

return 'CITEREF' .. (params.last or '') .. (params.year or ''); -- cite LSA always creates an anchor id using only |last= and |year= (no aliases)

end

-- all cs1|2 templates will create CITEREF anchor IDs after next cs1|2 module-suite update so keep going

-- return nil; -- not cite LSA so done

end

if 'harv' == ref or not ref then -- |ref=harv specified or |ref= missing or empty (new cs1|2 default is not default for other templates handled here)

if redirects_patent[template_name] then -- if this is a cite patent template

anchor_id = names_get (params, aliases_inventor); -- inventor names only

Line 580 ⟶ 528:

--[[--------------------------< L I S T _ A D D >--------------------------------------------------------------

adds an <item> to the <list> table; for anchor IDs, the boolean <encode> argument must be set true; no return value

]]

local function list_add (item, list, encode)

if item then -- if there was an anchor id extracteditem

if encode then -- for anchor IDs ...

item = mw.uri.anchorEncode (item); -- encode to remove wikimarkup, convert spaces to underscores etc

end

if not list[item] then -- if not already saved

list[item] = 1; -- save it

else -- here when this anchor iditem already saved

list[item] = list[item] + 1; -- to indicate that there are multiple same name/date citationsitems

end

end

Line 607 ⟶ 555:

local function anchor_id_make_anchor (template, anchor_id_list)

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

template = template:gsub ('^[^|]+|', ''); -- remove template name and first pipe

template = wikilink_strip (template); -- strip any wikilink markup (there shouldn't be any but just in case)

Line 614 ⟶ 562:

local anchor_id;

for param in template:gmatch ('%b{}') do -- loop through the template; remove and save templates (presumed to be sfnref or harvid)

table.insert (params, param); -- save it

template = template:gsub ('%b{}', '', 1); -- remove it from source template

Line 621 ⟶ 569:

for _, t in ipairs (params) do -- spin through the templates in params

anchor_id = sfnref_get (t); -- attempt to decode {{sfnref}} and {{harvid}}

if anchor_id then -- nil when not {{sfnref}} andor {{harvid}}

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

end

Line 645 ⟶ 593:

]]

local Lang_obj = mw.language.getContentLanguage();

local function template_list_add (template)

Line 652 ⟶ 598:

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

--mw.logObject (template_list, 'template_list')

-- return template_list;

end

--[[--------------------------< CA IN TC EH O R E_ I FD _ L I S T _ M A K E >--------------------------------------------

makes a list of anchor ids from cs1|2, cs1|2-like, vcite xxx, and harvc, anchor, wikicite templates

Because cs1|2 wrapper templates can, and often do, hide |ref=, the author and date parameters inside the wrapper,

these parameters are not available in the article's wikisource so {{harv}}, {{sfn}}, and {{harvc}} templates that

link correctly to those wrapper templates will incorrectly show error messages. Use |ignore-err=yes in the {{harv}},

{{sfn}}, and {{harvc}} templates to supress the error message.

creates a list of templates used in the article for use with the whitelist

creates a list of article-local whitelisted anchor IDs from {{sfn whitelist}}

]]

Line 689 ⟶ 636:

if data.known_templates_cs12 [template_name] then

anchor_id = anchor_id_makeanchor_id_make_cs12 (template); -- extract an anchor id from this template

list_add (anchor_id, anchor_id_list, true)

Line 739 ⟶ 686:

elseif data.known_templates_anchor [template_name] then

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

end

elseif data.known_templates_sfn_whitelist [template_name] then

end

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

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

template = mw.text.split (template, '%s*|%s*'); -- make a table of the template's parameters

for _, anchor_id in ipairs (template) do -- spin through this template's parameter

mw.logObject (anchor_id_list, 'anchor_id_list')

if '' ~= anchor_id and not article_whitelist[anchor_id] then

mw.logObject (template_list, 'template_list')

article_whitelist[anchor_id] = 1; -- add to the whitelist

return anchor_id_list;

end

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

makes a list of templates use in the article.

]]

--[[

local Lang_obj = mw.language.getContentLanguage();

local function template_list_make ()

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

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

return ''; -- no point in continuing

end

local template_list = {};

for template in Article_content:gmatch ('{{%s*(.-)[|}]') do

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)

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

end

end

mw.logObject (template_list, 'template_list')

return template_list;

end

]]

--[[--------------------------< A R T I C L E _ L O C A L _ W H I T E L I S T _ M A K E >----------------------

makes a list of templates use in the article.

]]

local function article_local_whitelist_make ()

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

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

return ''; -- no point in continuing

end

local article_whitelist = {};

local tstart, tend;

local template;

for _, pattern in ipairs (redirect_patterns_sfn_whitelist) do

tstart, tend = Article_content:find (pattern); -- find the first whitelist template

while tstart do -- nil when whitelist template not found

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

if template then -- necessary?

template = template:gsub (pattern, ''):gsub ('}}$', '', 1); -- remove outer {{ and }} and template name

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

template = mw.text.split (template, '%s*|%s*'); -- make a table of the template's parameters

for _, anchor_id in ipairs (template) do -- spin through this template's parameter

if '' ~= anchor_id and not article_whitelist[anchor_id] then

article_whitelist[anchor_id] = 1; -- add to the whitelist

end

end

end

tstart = tend; -- reset the search starting index

tstart, tend = Article_content:find (pattern, tstart); -- search for another whitelist template

end

end

mw.logObject (article_whitelistanchor_id_list, 'article_whitelistanchor_id_list');

mw.logObject (template_list, 'template_list');

return article_whitelist;

mw.logObject (article_whitelist, 'article_whitelist');

return anchor_id_list;

end

Line 828 ⟶ 713:

return {

anchor_id_list = anchor_id_list_make(), -- table of anchor ids available in this article

article_whitelist = article_local_whitelist_make()article_whitelist, -- table of anchor ids with false-positive error message to be suppressed

-- template_list = template_list_make(), -- table of templates used in this article

template_list = template_list, -- table of templates used in this article

}