--[[--------------------------< 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',
}
--[[--------------------------< C S 1 2 _ S T R I P P E D _ L I S T >------------------------------------------
This table is created from cs1_template_patterns and cs2_template_patterns. To make this table, entries in
the source tables are evaluated to replace lua patterns with the appropriate characters and create names for
this list. For example:
[Cc]ite ar[Xx]iv
becomes
Cite arXiv
Cite arxiv
cite arXiv
cite arxiv
]]
local cs12_stripped_list = {};
local function add_stripped (name)
if not cs12_stripped_list[name] then
cs12_stripped_list[name] = true;
end
end
local function pattern_convert (name)
if name:match ('(.-)%[(%a)(%a)%](.*)') then -- mixed case optional letters
local l;
lead, c, l, tail = name:match ('(.-)%[(%a)(%a)%](.*)');
add_stripped (lead .. c .. tail); -- uppercase
add_stripped (lead .. l .. tail); -- lowercase
elseif name:match ('^([^%[]+)(%[ %%%-%]%?)(.+)$') then -- [ %-]?
lead, c, tail = name:match ('^([^%[]+)(%[ %%%-%]%?)(.+)$');
add_stripped (lead .. tail); -- neither char
add_stripped (lead .. ' ' .. tail); -- space
add_stripped (lead .. '-' .. tail); -- hyphen
elseif name:match ('^([^%[]+)(%[%%%-%]%?)(.+)$') then -- [%-]?
lead, c, tail = name:match ('^([^%[]+)(%[%%%-%]%?)(.+)$');
add_stripped (lead .. tail); -- no hyphen
add_stripped (lead .. '-' .. tail); -- hyphen
elseif name:match ('^([^%[]+)(%[ %%%-%])(.+)$') then -- [ %-]
lead, c, tail = name:match ('^([^%[]+)(%[ %%%-%])(.+)$');
add_stripped (lead .. ' ' .. tail); -- space
add_stripped (lead .. '-' .. tail); -- hyphen
elseif name:match ('^([^%?]+)(%%%-%?)(.+)$') then -- %-?
lead, c, tail = name:match ('^([^%?]+)(%%%-%?)(.+)$');
add_stripped (lead .. tail); -- no hyphen
add_stripped (lead .. '-' .. tail); -- hyphen
elseif name:match ('^(.-)(%%%-)(.+)$') then -- %-
lead, c, tail = name:match ('^(.-)(%%%-)(.+)$');
add_stripped (lead .. '-' .. tail); -- hyphen
elseif name:match ('^(.-)(.)%?(.*)$') then -- .?
lead, c, tail = name:match ('^(.-)(.)%?(.*)$');
add_stripped (lead .. tail); -- no character
add_stripped (lead .. c .. tail); -- character
else
add_stripped (name); -- no patterns so save as is
end
end
for _, list in ipairs ({cs1_template_patterns, cs2_template_patterns}) do
for _, pattern in ipairs (list) do
local name = pattern:gsub ('^%[(%a)%a%]', '%1'); -- leading character (usually uppercase)
pattern_convert (name);
name = pattern:gsub ('^%[%a(%a)%]', '%1'); -- leading character (usually lowercase)
pattern_convert (name);
end
end
--[[--------------------------< H A R V _ T E M P L A T E _ P A T T E R N S >----------------------------------
These are patterns for the harv family of templates and their redirects.
]]
local harv_template_patterns = { -- lua patterns of the cannonical names and redirects
'[Hh]arvard citation no brackets', -- canonical
'[Hh]arnvb', '[Hh]arvardnb', '[Hh]arvnb',
'[Hh]arvard citation', -- canonical
'[Hh]arv', '[Hh]arvsp',
'[Hh]arvard citation text', -- canonical
'[Hh]arvtxt',
'[Hh]arvcoltxt', -- canonical
'[Hh]arvcol', -- canonical
'[Hh]rvcoln',
'[Hh]arvcolnb', -- canonical
'[Hh]arvard citations', -- canonical
'[Hh]arvs',
'[Hh]arvp', -- canonical
};
--[[--------------------------< S F N _ T E M P L A T E _ P A T T E R N S >------------------------------------
These are patterns for the sfn family of templates and their redirects.
]]
local sfn_template_patterns = { -- lua patterns of the cannonical names and redirects
'[Ss]fn', -- canonical
'[Hh]f', '[Ss]f', '[Ss]hortened footnote',
'[Ss]hortened footnote template', '[Ss]nf',
'[Ss]fnp', -- canonical
'[Ss]fb', '[Ss]fnb',
'[Ss]fnm', -- canonical
'[Ss]fnmp', -- canonical
};
--[[--------------------------< R E F L I S T _ T E M P L A T E _ P A T T E R N S >----------------------------
These are patterns for the reflist template and its redirects.
]]
local reflist_template_patterns = {
'[Rr]ef[Ll]ist', -- reflist is the canonical form
'[Ff]ootnotes?', '[Ff]ootnotesSmall', '[Rr]ealist', '[Rr]ef [Ll]ist',
'[Rr]ef%-list', '[Rr]eference', '[Rr]eference list', '[Rr]efIist',
'[Rr]EFLIST', '[Rr]efs', '[Rr]FS', '[Rr]fs',
'[Jj]egyzetek', '[Ll]istaref', '[Rr]efer[eê]ncias', '[Rr]éférences', -- non-English redirects
'[Tt]ham khảo', 'Примечания', 'مراجع', 'پانویس',
};
--[[--------------------------< C L E A N U P _ T E M P L A T E _ P A T T E R N S >----------------------------
These are patterns for some of the cleanup templates and their redirects.
]]
local cleanup_template_patterns = {
'[Cc]itation [Nn]eeded', -- Citation needed is the canonical form
'[Aa]re you sure%?', '[Cc][Bb]', '[Cc]cn', '[Cc]iation needed', '[Cc]it',
'[Cc]itaiton needed', '[Cc]itation missing', '[Cc]itation need',
'[Cc]itation [Rr]equested', '[Cc]itation ?required',
'[Cc]itation[-Nn]?eeded', '[Cc]itationeeded', '[Cc]ite missing',
'[Cc]ite[ %-]?needed', '[Cc]ite source', '[Cc]itesource', '[Cc]itn',
'[Cc]N', '[Cc]n', '[Cc]tn', '[Ff]ACT', '[Ff]act[s%?]?',
'[Ff]citation needed', '[Mm]e%-fact', '[Nn]eed [Cc]itation',
'[Nn]eed sources', '[Nn]eed%-ref', '[Nn]eedcitation', '[Nn]eedcite',
'[Nn]eeds citations?', '[Nn]eeds reference', '[Nn]eedsref',
'[Nn]o source given', '[Pp]ROV%-statement', '[Pp]rove ?it',
'[Rr]ef%-?needed', '[Rr]ef%?', '[Rr]eference needed', '[Rr]efplease',
'[Rr]equest [Cc]itation', '[Rr]éférence nécessaire', '[Ss]ource needed',
'[Ss]ource%?', '[Ss]ourceme', '[Uu]ncited', '[Uu]nreferenced inline',
'[Uu]nsourced%-inline',
'[Cc]ita requerida', '[Cc]itazione necessaria', '[Kk]älla behövs', -- non-English redirects
'[Dd]isputed[ %-]inline', -- Disputed inline is canonical form
'[Dd]ispute[ %-]inline',
'[Dd]ubious', -- canonical
'[Dd]ebatable', '[Dd]isputable', '[Dd]isputed?Assertion',
'[Dd]isputed Point', '[Dd]oubtful', '[Dd]UB', '[Dd]ub',
'[Dd]ubious %- [Dd]iscuss', '[Dd]ubious [Ii]nline',
'[Dd]ubious%-inline', '[Oo]dd',
'[Ii]frågasatt uppgift', -- non-English redirect
'[Ff]ailed ?verification', -- Failed verification is the canonical form
'[Bb]adref', '[Ff]ailed ref', '[Ff]ailed reference',
'[Ff]ails verification', '[Ff][Vv]', '[Nn]cg', '[Nn]ICG', '[Nn]icg',
'[Nn]IGC', '[Nn]igc', '[Nn]ot in citation', '[Nn]ot in citation given',
'[Nn]ot in cited source', '[Nn]ot in ref', '[Nn]ot in ref given',
'[Nn]ot in reference', '[Nn]ot in reference given', '[Nn]ot in source',
'[Nn]ot in source given', '[Nn]ot specifically in source',
'[Nn]otincitation', '[Nn]otincitationgiven', '[Nn]otinref',
'[Nn]otinsource', '[Nn]otinsourcegiven', '[Vv]erification[ %-]failed',
};
--[[--------------------------< D E A D _ L I N K _ T E M P L A T E _ P A T T E R N S >------------------------
These are patterns for the dead link template and its redirects.
]]
local dead_link_template_patterns = {
'[Dd]ead [Ll]ink', -- Dead link is the canonical form
'404', '[Bb]ad ?link', '[Bb]roken', '[Bb]roken ?link', '[Dd]ead',
'[Dd]ead cite', '[Dd]ead link%-now', '[Dd]ead links', '[Dd]ead page',
'[Dd]ead URL', '[Dd]ead[ %-]?url', '[Dd]ead%-inline', '[Dd]ead%-link',
'[Dd]eadcite', '[Dd]eadlinks?', '[Dd][Ll]', '[Dd]l%-now',
'[Ll]ink ?broken',
'[Cc]ollegamento interrotto', '[Tt]oter Link', 'Недоступная ссылка', -- non-English redirects
};
--[[--------------------------< W E B A R C H I V E _ T E M P L A T E _ P A T T E R N S >----------------------
These are patterns for the webarchive template and its redirects.
]]
local webarchive_template_patterns = {
'[Ww]eb ?archive', -- Webarchive is the canonical form
'[Aa]rchive url', '[Ii]AWM', '[Ii]awm', '[Uu]rl archive', '[Ww]ayBack',
'[Ww]aybackdate', '[Ww]ebarchiv', '[Ww]ebcitation',
};
--[[--------------------------< R E F B E G I N _ T E M P L A T E _ P A T T E R N S >--------------------------
These are patterns for the refbegin template and its redirects.
]]
local refbegin_template_patterns = {
'[Rr]ef ?begin', -- Refbegin is the canonical form
'[Bb]eginref', '[Ss]ourcesstart', '[Ss]ourcestart',
}
--[[--------------------------< R P _ T E M P L A T E _ P A T T E R N S >--------------------------------------
These are patterns for the rp template and its redirects.
]]
local rp_template_patterns = {
'[Rr][Pp]', -- Rp is the canonical form
'[Pp]age reference',
'[Rr]efpages?',
}
--[[--------------------------< C S 1 | 2 A U T H O R - N A M E _ P A T T E R N S >--------------------------
These are patterns for the cs1|2 author-name parameters.
]]
local authors_param_patterns = {'authors', 'people', 'credits'};
local author_param_patterns = {'author1?', 'host1?', 'subject1?'};
local last_param_patterns = {'last1?', 'author1?%-last1?', 'surname1?'};
--[[--------------------------< E X P O R T S >----------------------------------------------------------------
]]
return
{
author_param_patterns = author_param_patterns,
authors_param_patterns = authors_param_patterns,
cleanup_template_patterns = cleanup_template_patterns,
cs1_template_patterns = cs1_template_patterns,
cs12_stripped_list = cs12_stripped_list,
cs2_template_patterns = cs2_template_patterns,
dead_link_template_patterns = dead_link_template_patterns,
harv_template_patterns = harv_template_patterns,
last_param_patterns = last_param_patterns,
refbegin_template_patterns = refbegin_template_patterns,
reflist_template_patterns = reflist_template_patterns,
rp_template_patterns = rp_template_patterns,
sfn_template_patterns = sfn_template_patterns,
webarchive_template_patterns = webarchive_template_patterns,
}
