Module:Footnotes/anchor id list/data

-- copied from Module:Ref info/data - maybe overkill here?
-- for this application make lists from the redirect lists and force all template names first character uppercase

require('strict');

--[[--------------------------< C S 1 _ T E M P L A T E _ P A T T E R N S >------------------------------------

These are patterns for cs1 templates and their redirects.  These patterns exclude redirects that are vcite-like
which redirects should be deleted because vcite is not cs1.

]]

local cs1_template_patterns = {             -- lua patterns of the cannonical names and redirects
 '[Cc]ite ar[Xx]iv',               -- arXiv is the canonical name
  '[Cc]ite ArXiv',

 '[Cc]ite AV media',               -- canonical
  '[Cc]ite audio', '[Cc]ite AV', '[Cc]ite AV ?Media', '[Cc]ite av media', 
  '[Cc]ite cd', '[Cc]ite DVD', '[Cc]ite dvd', '[Cc]ite film',
  '[Cc]ite image', '[Cc]ite media', '[Cc]ite movie',
  '[Cc]ite music video', '[Cc]ite radio', '[Cc]ite song',
  '[Cc]ite ?video', '[Cc]ite visual', '[Cc]ite You[Tt]ube',
  '[Cc]ita vídeo',              -- non-English redirect; TODO: tally separately?

 '[Cc]ite AV media notes',             -- canonical
  '[Cc]ite album[ %-]notes', '[Cc]ite av media notes',
  '[Cc]ite DVD[ %-]notes', '[Cc]ite dvd%-notes', '[Cc]ite liner notes',
  '[Cc]ite music release notes', '[Ll]iner notes',

 '[Cc]ite bio[Rr]xiv',              -- bioRxiv is the canonical form
 '[Cc]ite [Bb]ook',               -- book is the canonical form
  '[Bb]ook cite', '[Bb]ook reference', '[Bb]ook reference url',
  '[Cc] book', '[Cc]it book', '[Cc]ite books', '[Cc]ite chapter',
  '[Cc]ite ebook', '[Cc]ite manual', '[Cc]ite page',
  '[Cc]ite publication', '[Cc]ite score',
  '[Cc]ite work', '[Cc]ite%-?book', 
  '[Bb]okref', '[Cc]itace monografie', '[Cc]itar livro',     -- non-English redirects; TODO: tally separately?
  '[Cc]iteer boek', '[Oo]uvrage', '[Rr]ef%-llibre', '서적 인용',

 '[Cc]ite citeseerx',              -- canonical
 '[Cc]ite conference',              -- canonical
  '[Cc]ita conferenza', '[Cc]ite proceedings', '[Cc]onference reference', -- cita conferenza is non-English; TODO: tally separately?
  
 '[Cc]ite ?encyclopedia',             -- cite encyclopedia is the canonical name
  '[Cc]ite contribution', '[Cc]ite dic', '[Cc]ite dictionary',
  '[Cc]ite encyclopaedia', '[Cc]ite encyclopædia', '[Ee]ncyclopedia',

 '[Cc]ite [Ee]pisode',              -- episode is the canonical form
  '[Cc]ite show',

 '[Cc]ite interview',              -- canonical
 '[Cc]ite ?journal',               -- cite journal is the canonical form
  '[Cc] journal', '[Cc]itation journal', '[Cc]ite abstract',
  '[Cc]ite document', '[Cc]ite Journal', '[Cc]ite journal zh',
  '[Cc]ite ?paper', '[Vv]cite2 journal',
  '[Cc]ita pubblicazione', '[Cc]itace periodika', '[Cc]itar jornal',  -- non-English redirects; TODO: tally separately?
  '[Cc]itar publicació', '[Cc]ytuj pismo', '[Tt]idskriftsref',
  'Навод из стручног часописа', '저널 인용',

 '[Cc]ite [Mm]agazine',              -- magazine is the canonical form
  '[Cc]ite mag', '[Cc]ite magazine article', '[Cc]ite newsletter',
  '[Cc]ite periodical',

 '[Cc]ite mailing ?list',             -- mailing list is the canonical form
  '[Cc]ite list',

 '[Cc]ite maps?',               -- map is the canonical form
 '[Cc]ite[ %-]?news',              -- cite news is the canonical form
  '[Cc] news', '[Cc]it news', '[Cc]itation news', '[Cc]ite article',
  '[Cc]ite n', '[Cc]ite new', '[Cc]ite newspaper', '[Cc]ite News',
  '[Cc]ite news%-q', '[Cc]ite news2', '[Cc]itenewsauthor', '[Cc]ute news',
  '[Cc]itar notícia', '[Hh]aber kaynağı', '[Tt]idningsref', 'استشهاد بخبر', -- non-English redirects; TODO: tally separately?
  '뉴스 인용',

 '[Cc]ite newsgroup',              -- canonical
  '[Cc]ite usenet',
  
 '[Cc]ite podcast',               -- canonical
 '[Cc]ite [Pp]ress release',             -- press release is the canonical form
  '[Cc]ite media release', '[Cc]ite news release', '[Cc]ite pr',
  '[Cc]ite press', '[Cc]ite press release\.', '[Cc]ite press[%-]?release',

 '[Cc]ite report',               -- canonical
 '[Cc]ite serial',               -- canonical
 '[Cc]ite sign',                -- canonical
  '[Cc]ite plaque',

 '[Cc]ite speech',               -- canonical
 '[Cc]ite ssrn',                -- canonical
  '[Cc]ite SSRN',
  
 '[Cc]ite tech ?report',              -- techreport is the canonical form
  '[Cc]ite standard', '[Cc]ite technical report', '[Tt]echrep reference',

 '[Cc]ite thesis',               -- canonical
  '[Cc]ite dissertation',
  '[Cc]itar tese',              -- non-English redirect; TODO: tally separately?

 '[Cc]ite [Ww]eb',               -- web is the canonical form
  '[Cc] web', '[Cc]it web', '[Cc]ite blog', '[Cc]ite URL', '[Cc]ite url',
  '[Cc]ite w', '[Cc]ite wb', '[Cc]ite we', '[Cc]ite web\.',
  '[Cc]ite webpage', '[Cc]ite website', '[Cc]ite website article',
  '[Cc]ite%-?web', '[Cc]itweb', '[Cc]w', '[Rr]ef web', '[Ww]eb citation',
  '[Ww]eb cite', '[Ww]eb link', '[Ww]eb[ %-]reference', '[Ww]eblink',
  '[Cc]hú thích web', '[Cc]ita web', '[Cc]itace elektronické monografie', -- non-English redirects; TODO: tally separately?
  '[Cc]itat web', 'مرجع ويب', 'یادکرد وب', '웹 인용',
 }


--[[--------------------------< C S 2 _ T E M P L A T E _ P A T T E R N S >------------------------------------

These are patterns for cs2 templates redirects.

]]

local cs2_template_patterns = {             -- lua patterns of the cannonical names and redirects
 '[Cc]itation',
  '[Cc]ite', '[Cc]ite citation', '[Cc]ite study',
  '[Cc]ite [Tt]echnical standard',
 }


--[[--------------------------< V C I T E _ T E M P L A T E _ P A T T E R N S >--------------------------------

These are patterns for Vcite-family templates and their redirects.

]]

local vcite_template_patterns = {
 '[Vv]cite book',               -- canonical
  '[Vv]ancite book', '[Vv]ancite report', '[Vv]cite encyclopedia',
  '[Vv]cite report',

 '[Vv]cite journal',               -- canonical
  '[Cc]it journal', '[Cc]it paper', '[Vv]ancite journal', '[Vv]cite paper',

 '[Vv]cite news',               -- canonical
  '[Vv]ancite news',

 '[Vv]cite web',                -- canonical
  '[Vv]ancite web',
 }


--[[--------------------------< H A R V C _ T E M P L A T E _ P A T T E R N S >--------------------------------

These are patterns for the harvc template and its redirects.

]]

local harvc_template_patterns = {
 '[Hh]arvc',                 -- canonical
  '[Cc]itec',
 }


--[[--------------------------< C S 1 _ L I K E _ T E M P L A T E _ P A T T E R N S >--------------------------

These are patterns for miscellaneous templates and their redirects that 'look like' cs1 templates (begin with cite ...)

Because they 'look like' cs1 templates they are handled as if they were cs1 templates.  These templates are NOT
wrapper templates; names and dates are always to be extracted from the article instantiation of the template.

]]

local cs1_like_template_patterns = {
 '[Cc]ite LSA',                -- canonical
 }


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

These are patterns for the wikicite template and its redirects.

]]

local wikicite_template_patterns = {
 '[Ww]ikicite',                -- canonical
 }


--[[--------------------------< A N C H O R _ T E M P L A T E _ P A T T E R N S >------------------------------

These are patterns for the anchor template and its redirects.

]]

local anchor_template_patterns = {
 '[Aa]nchors?',                -- anchor is canonical form
 '[Aa]nchor for redirect',
 '[Aa]nchro',
 '[Aa]ncor',
 }


--[[--------------------------< S F N _ W H I T E L I S T _ P A T T E R N S >----------------------------------

These are patterns for the anchor template and its redirects.

]]

local sfn_whitelist_patterns = {
 '[Ss]fn whitelist',               -- canonical
  '[Hh]arv whitelist',
 }


--[[--------------------------< K N O W N _ T E M P L A T E S >------------------------------------------------

These tables are created from the *_template_patterns tables.  To make these tables, entries in the source tables
are evaluated to replace lua patterns with the appropriate characters to create names for the output tables.

First charqacter is always uppercase

For example:
 [Cc]ite ar[Xx]iv
becomes
 Cite arXiv
 Cite arxiv

]]

local known_templates_cs12 = {};            -- the exported tables
local known_templates_vcite = {};
local known_templates_harvc = {};
local known_templates_wikicite = {};
local known_templates_anchor = {};
local known_templates_sfn_whitelist = {};

 local function add_stripped (list, name)
  if not list[name] then
   list[name] = true;
  end
 end


 local function pattern_convert (pattern, list)
  local lead, tail, c, l, name;
  local first_char_patterns = {
   '^%[(%a)%a%]',              -- leading character (usually uppercase)
   '^%[%a(%a)%]',              -- leading character (usually lowercase)
   }
   
  c = tostring (pattern:match ('%u'));         -- tostring() required because I don't know why; lua chokes complaining that c is not a string
  name = pattern:gsub ('^%[%a%a%]', c);         -- replace bracketed first character [Xx] with selected character from the match
  
  if name:match ('(.-)%[(%a)(%a)%](.*)') then        -- mixed case optional letters
   lead, c, l, tail = name:match ('(.-)%[(%a)(%a)%](.*)');
   add_stripped (list, lead .. c .. tail);        -- uppercase
   add_stripped (list, lead .. l .. tail);        -- lowercase
 
  elseif name:match ('^([^%[]+)(%[ %%%-%]%?)(.+)$') then     -- [ %-]?
   lead, c, tail = name:match ('^([^%[]+)(%[ %%%-%]%?)(.+)$');
   add_stripped (list, lead .. tail);         -- neither char
   add_stripped (list, lead .. ' ' .. tail);       -- space
   add_stripped (list, lead .. '-' .. tail);       -- hyphen
 
  elseif name:match ('^([^%[]+)(%[%%%-%]%?)(.+)$') then     -- [%-]?
   lead, c, tail = name:match ('^([^%[]+)(%[%%%-%]%?)(.+)$');
   add_stripped (list, lead .. tail);         -- no hyphen
   add_stripped (list, lead .. '-' .. tail);       -- hyphen
 
  elseif name:match ('^([^%[]+)(%[ %%%-%])(.+)$') then     -- [ %-]
   lead, c, tail = name:match ('^([^%[]+)(%[ %%%-%])(.+)$');
   add_stripped (list, lead .. ' ' .. tail);       -- space
   add_stripped (list, lead .. '-' .. tail);       -- hyphen
 
  elseif name:match ('^([^%?]+)(%%%-%?)(.+)$') then      -- %-?
   lead, c, tail = name:match ('^([^%?]+)(%%%-%?)(.+)$');
   add_stripped (list, lead .. tail);         -- no hyphen
   add_stripped (list, lead .. '-' .. tail);       -- hyphen
 
  elseif name:match ('^(.-)(%%%-)(.+)$') then        -- %-
   lead, c, tail = name:match ('^(.-)(%%%-)(.+)$');
   add_stripped (list, lead .. '-' .. tail);       -- hyphen

  elseif name:match ('^(.-)(.)%?(.*)$') then        -- .?
   lead, c, tail = name:match ('^(.-)(.)%?(.*)$');
   add_stripped (list, lead .. tail);         -- no character
   add_stripped (list, lead .. c .. tail);        -- character
 
  else
   add_stripped (list, name);           -- no patterns so save as is
  end
 end

 for _, t in ipairs ({
  {cs1_template_patterns, known_templates_cs12},
  {cs2_template_patterns, known_templates_cs12},
  {cs1_like_template_patterns, known_templates_cs12},
  {vcite_template_patterns, known_templates_vcite},
  {harvc_template_patterns, known_templates_harvc},
  {wikicite_template_patterns, known_templates_wikicite},
  {anchor_template_patterns, known_templates_anchor},
  {sfn_whitelist_patterns, known_templates_sfn_whitelist},
  }) do
   for _, pattern in ipairs (t[1]) do         -- for each patern in *_template_patterns
    pattern_convert (pattern, t[2])         -- convert and store in known_templates_*
   end
 end


--[[--------------------------< E X P O R T S >----------------------------------------------------------------
]]

return
 {
 known_templates_anchor = known_templates_anchor,
 known_templates_cs12 = known_templates_cs12,
 known_templates_harvc = known_templates_harvc,
 known_templates_vcite = known_templates_vcite,
 known_templates_wikicite = known_templates_wikicite,
 known_templates_sfn_whitelist = known_templates_sfn_whitelist,
 }