Wikipedia

Module:Citation/CS1/Configuration: Difference between revisions

Browse history interactively
← Previous editNext edit →
Content deleted Content added
bump ssrn;
sync from sandbox;
Line 1:
local lang_obj = mw.language.getContentLanguage(); -- make a language object for the local language; used here for languages and dates
 
 
--[[--------------------------< S E T T I N G S >--------------------------------------------------------------
 
boolean settings used to control various things. these setting located here to make them easy to find
 
]]
-- these settings local to this module only
local local_digits_from_mediawiki = false; -- for i18n; when true, module fills date_names['local_digits'] from MediaWiki; manual fill required else; always false at en.wiki
local local_date_names_from_mediawiki = false; -- for i18n; when true, module fills date_names['local']['long'] and date_names['local']['short'] from MediaWiki;
-- manual translation required else; ; always false at en.wiki
 
-- these settings exported to other modules
local use_identifier_redirects = true; -- when true use redirect name for identifier label links; always true at en.wiki
local local_lang_cat_enable = false; -- when true categorizes pages where |language=<local wiki's language>; always false at en.wiki
local date_name_auto_xlate_enable = false; -- when true translates English month-names to the local-wiki's language month names; always false at en.wiki
local date_digit_auto_xlate_enable = false; -- when true translates Western date digit to the local-wiki's language digits (date_names['local_digits']); always false at en.wiki
 
 
--[[--------------------------< U N C A T E G O R I Z E D _ N A M E S P A C E S >------------------------------
Line 34 ⟶ 52:
['archived-dead'] = 'Archived from $1 on $2',
['archived-live'] = '$1 from the original on $2',
['archived-missing'] = 'Archived from the original $1 on $2',
['archived-unfit'] = 'Archived from the original on ',
['archived'] = 'Archived',
Line 48 ⟶ 65:
['inset'] = '$1 inset',
['interview'] = 'Interviewed by $1',
['lay summary'] = 'Lay summary',
['mismatch'] = '<code class="cs1-code">&#124;$1=</code> / <code class="cs1-code">&#124;$2=</code> mismatch', -- $1 is year param name; $2 is date param name
['newsgroup'] = '[[Usenet newsgroup|Newsgroup]]:&nbsp;$1',
Line 125 ⟶ 141:
 
local citation_class_map_t = { -- TODO: if kept, these and all other config.CitationClass 'names' require some sort of i18n
['arxiv'] = 'arXiv',
['audio-visual'] = 'AV media',
['AV-media-notes'] = 'AV media notes',
['biorxiv'] = 'bioRxiv',
['citeseerx'] = 'CiteSeerX',
['encyclopaedia'] = 'encyclopedia',
['mailinglist'] = 'mailing list',
['pressreleasemedrxiv'] = 'press releasemedRxiv',
['pressrelease'] = 'press release',
['ssrn'] = 'SSRN',
['techreport'] = 'tech report',
}
 
Line 275 ⟶ 297:
['Issue'] = {'issue', 'number'},
['Language'] = {'language', 'lang'},
['LayDate'] = 'lay-date',
['LayFormat'] = 'lay-format',
['LaySource'] = 'lay-source',
['LayURL'] = 'lay-url',
['MailingList'] = {'mailing-list', 'mailinglist'}, -- cite mailing list only
['Map'] = 'map', -- cite map only
Line 326 ⟶ 344:
['Title'] = 'title', -- Used by InternetArchiveBot
['TitleLink'] = {'title-link', 'episode-link', 'episodelink'}, -- Used by InternetArchiveBot
['TitleNote'] = {'title-note', 'department'},
['TitleType'] = {'type', 'medium'},
['TransChapter'] = {'trans-article', 'trans-chapter', 'trans-contribution',
Line 332 ⟶ 350:
['Transcript'] = 'transcript',
['TranscriptFormat'] = 'transcript-format',
['TranscriptURL'] = {'transcript-url', 'transcripturl'}, -- Used by InternetArchiveBot
['TransMap'] = 'trans-map', -- cite map only
['TransPeriodical'] = {'trans-journal', 'trans-magazine', 'trans-newspaper',
Line 347 ⟶ 365:
['Year'] = 'year',
 
['AuthorList-First'] = {"first#", "author-first#", "author#-first", "author-given#", "author#-given",
"authorsubject-first#", "subject#-first", "subject-given#", "authorsubject#-given"},
"given#"},
['AuthorList-Last'] = {"last#", "author-last#", "author#-last", "surname#",
"author ['AuthorList-surnameLast'] = {"last#", "author#-surnamelast#", "author#-last", "subjectauthor-surname#", 'host"author#'}-surname",
"subject-last#", "subject#-last", "subject-surname#", "subject#-surname",
"author#", 'host#', "subject#", "surname#"},
['AuthorList-Link'] = {"author-link#", "author#-link", "subject-link#",
"subject#-link", "authorlink#", "author#link"},
Line 394 ⟶ 414:
'AuthorList-Mask', 'ContributorList-Mask', 'EditorList-Mask', 'InterviewerList-Mask', 'TranslatorList-Mask', -- name-list mask may have name separators
'PostScript', 'Quote', 'ScriptQuote', 'TransQuote', 'Ref', -- miscellaneous
'ArchiveURL', 'ChapterURL', 'ConferenceURL', 'LayURL', 'MapURL', 'TranscriptURL', 'URL', -- URL-holding parameters
}
 
local url_meta_params = { -- table of aliases[] keys (meta parameters); each key has a table of parameter names for a value
'ArchiveURL', 'ChapterURL', 'ConferenceURL', 'ID', 'LayURL', 'MapURL', 'TranscriptURL', 'URL', -- parameters allowed to hold urls
'Page', 'Pages', 'At', 'QuotePage', 'QuotePages', -- insource locators allowed to hold urls
}
Line 617 ⟶ 637:
date_names[invert_t[2]][i] = name; -- invert to get [i] = 'name' for conversions from ymd
end
end
 
if local_digits_from_mediawiki then -- if fetching local digits from MediaWiki is enabled
local digits_t = {};
for i=0, 9 do -- loop 10x and
digits_t [lang_obj:formatNum (i)] = tostring (i); -- format the loop indexer as local lang table index and assign loop indexer (a string) as the value
end
date_names['local_digits'] = digits_t;
end
 
Line 638 ⟶ 666:
'{{ *([Mm]DY) *[|}]', -- 0
}
 
local title_object = mw.title.getCurrentTitle();
local content = title_object: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
 
local function get_date_format ()
local title_object = mw.title.getCurrentTitle();
if title_object.namespace == 10 then -- not in template space so that unused templates appear in unused-template-reports;
return nil; -- auto-formatting does not work in Template space so don't set global_df
end
local content = title_object: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 _, pattern in ipairs (df_template_patterns) do -- loop through the patterns looking for {{Use dmy dates}} or {{Use mdy dates}} or any of their redirects
local start, _, match = content:find(pattern); -- match is the three letters indicating desired date format
if match then
contentlocal use_dates_template = content:match ('%b{}', start); -- get the whole template
if contentuse_dates_template:match ('| *cs1%-dates *= *[lsy][sy]?') then -- look for |cs1-dates=publication date length access-/archive-date length
return match:lower() .. '-' .. contentuse_dates_template:match ('| *cs1%-dates *= *([lsy][sy]?)');
else
return match:lower() .. '-all'; -- no |cs1-dates= k/v pair; return value appropriate for use in |df=
Line 658 ⟶ 687:
end
 
local global_df; -- TODO: add this to <global_cs1_config_t>?
 
 
Line 848 ⟶ 877:
['id-access'] = make_keywords_list ({keywords.free}),
}
 
 
--[[--------------------------< C S 1 _ C O N F I G _ G E T >--------------------------------------------------
 
fetch and validate values from {{cs1 config}} template to fill <global_cs1_config_t>
 
no error messages; when errors are detected, the parameter value from {{cs1 config}} is blanked.
 
Supports all parameters and aliases associated with the metaparameters: DisplayAuthors, DisplayContributors,
DisplayEditors, DisplayInterviewers, DisplayTranslators, NameListStyle, and Mode. The DisplayWhatever metaparameters
accept numeric values only (|display-authors=etal and the like is not supported).
 
]]
 
local global_cs1_config_t = {}; -- TODO: add value returned from get_date_format() to this table?
 
local function get_cs1_config ()
if title_object.namespace == 10 then -- not in template space so that unused templates appear in unused-template-reports;
return nil; -- auto-formatting does not work in Template space so don't set global_df
end
 
local start = content:find('{{ *[Cc][Ss]1 config *[|}]'); -- <start> is offset into <content> when {{cs1 config}} found; nil else
if start then
local cs1_config_template = content:match ('%b{}', start); -- get the whole template
 
if not cs1_config_template then
return nil;
end
 
local params_t = mw.text.split (cs1_config_template:gsub ('[{}]', ''), '%s*|%s*'); -- remove '{{' and '}}'; make a sequence of parameter/value pairs (split on the pipe)
table.remove (params_t, 1); -- remove the template name because it isn't a parameter/value pair
 
local config_meta_params_t = {'DisplayAuthors', 'DisplayContributors', 'DisplayEditors', 'DisplayInterviewers', 'DisplayTranslators', 'NameListStyle', 'Mode'};
local meta_param_map_t = {}; -- list of accepted parameter names usable in {{cs1 config}} goes here
for _, meta_param in ipairs (config_meta_params_t) do -- for i18n using <config_meta_params_t>, map template parameter names to their metaparameter equivalents
if 'table' == type (aliases[meta_param]) then -- if <meta_param> is a sequence,
for _, param in ipairs (aliases[meta_param]) do -- extract its contents
meta_param_map_t[param] = meta_param; -- and add to <meta_param_map_t>
end
else
meta_param_map_t[aliases[meta_param]] = meta_param; -- not a sequence so just add the parameter to <meta_param_map_t>
end
end
 
local keywords_t = {}; -- map valid keywords to their associate metaparameter; reverse form of <keyword_lists[key] for these metaparameters
for _, metaparam_t in ipairs ({{'NameListStyle', 'name-list-style'}, {'Mode', 'mode'}}) do -- only these metaparameter / keywords_lists key pairs
for _, keyword in ipairs (keywords_lists[metaparam_t[2]]) do -- spin through the list of keywords
keywords_t[keyword] = metaparam_t[1]; -- add [keyword] = metaparameter to the map
end
end
 
for _, param in ipairs (params_t) do -- spin through the {{cs1 config}} parameters and fill <global_cs1_config_t>
local k, v = param:match ('([^=]-)%s*=%s*(.+)'); -- <k> is the parameter name; <v> is parameter's assigned value
if k then
if k:find ('^display') then -- if <k> is one of the |display-<namelist>= parameters
if v:match ('%d+') then -- the assigned value must be digits; doesn't accept 'etal'
global_cs1_config_t[meta_param_map_t[k]]=v; -- add the display param and its value to globals table
end
else
if keywords_t[v] == meta_param_map_t[k] then -- keywords_t[v] returns nil or the metaparam name; these must be the same
global_cs1_config_t[meta_param_map_t[k]]=v; -- add the parameter and its value to globals table
end
end
end
end
end
end
 
get_cs1_config (); -- fill <global_cs1_config_t>
 
 
Line 994 ⟶ 1,093:
local this_wiki_code = lang_obj:getCode(); -- get this wiki's language code
if string.match (mw.site.server, 'wikidata') then
this_wiki_code = mw.getCurrentFrame():preprocesscallParserFunction('{{int:', {'lang}}'}); -- on Wikidata so use interface language setting instead
end
 
Line 1,026 ⟶ 1,125:
 
local script_lang_codes = {
'ab', 'am', 'ar', 'be', 'bg', 'bn', 'bo', 'bs', 'dv', 'dz', 'el', 'fa', 'gu', 'he',
'he', 'hi', 'hy', 'ja', 'ka', 'kk', 'km', 'kn', 'ko', 'ku', 'ky', 'lo', 'mk', 'ml', 'mn',
'ml', 'mnmni', 'mr', 'my', 'ne', 'or', 'ota', 'pa', 'ps', 'ru', 'sd', 'si', 'sr', 'syc', 'ta',
'syc', 'ta', 'te', 'tg', 'th', 'ti', 'tt', 'ug', 'uk', 'ur', 'uz', 'yi', 'yue', 'zh'
};
 
Line 1,037 ⟶ 1,136:
These tables hold language information that is different (correct) from MediaWiki's definitions
 
For each ['code<tag>'] = 'language name' in lang_code_remap{} there must be a matching ['language name'] = {'language name', 'code<tag>'} in lang_name_remap{}
 
lang_tag_remap{}:
lang_code_remap{}:
key is always lowercase ISO 639-1, -2, -3 language codetag or a valid lowercase IETF language tag
value is properly spelled and capitalized language name associated with key<tag>
only one language name per key<tag>;
key/value pair must have matching entry in lang_name_remap{}
 
lang_name_remap{}:
key is always lowercase language name
value is a table the holds correctly spelled and capitalized language name [1] and associated codetag [2] (codetag must match a codetag key in lang_code_remaplang_tag_remap{})
may have multiple keys referring to a common preferred name and codetag; For example:
['kolsch'] and ['kölsch'] both refer to 'Kölsch' and 'ksh'
 
]]
 
local lang_code_remaplang_tag_remap = { -- used for |language= and |script-title= / |script-chapter=
['als'] = 'Tosk Albanian', -- MediaWiki returns Alemannisch
['bh'] = 'Bihari', -- MediaWiki uses 'bh' as a subdomain name for Bhojpuri Wikipedia: bh.wikipedia.org
Line 1,059 ⟶ 1,158:
['bn'] = 'Bengali', -- MediaWiki returns Bangla
['ca-valencia'] = 'Valencian', -- IETF variant of Catalan
['fkv'] = 'Kven', -- MediaWiki returns Kvensk
['ilo'] = 'Ilocano', -- MediaWiki/IANA/ISO 639: Iloko; use en.wiki preferred name
['ksh'] = 'Kölsch', -- MediaWiki: Colognian; use IANA/ISO 639 preferred name
Line 1,077 ⟶ 1,177:
['kolsch'] = {'Kölsch', 'ksh'}, -- use IANA/ISO 639 preferred name (use non-diacritical o instead of umlaut ö)
['kölsch'] = {'Kölsch', 'ksh'}, -- use IANA/ISO 639 preferred name
['kven'] = {'Kven', 'fkv'}, -- Unicode CLDR have decided not to support English language name for these two...
['kvensk'] = {'Kven', 'fkv'}, -- ...they say to refer to IANA registry for English names
['ripuarian'] = {'Ripuarian', 'mis-x-ripuar'}, -- group of dialects; no code in MediaWiki or in IANA/ISO 639
['taiwanese hokkien'] = {'Taiwanese Hokkien', 'nan-tw'}, -- make room for MediaWiki/IANA/ISO 639 nan: Min Nan Chinese
Line 1,111 ⟶ 1,213:
local title_types = {
['AV-media-notes'] = 'Media notes',
['document'] = 'Document',
['interview'] = 'Interview',
['mailinglist'] = 'Mailing list',
Line 1,191 ⟶ 1,294:
hidden = false
},
err_archive_date_missing_url = {
message = '<code class="cs1-code">&#124;archive-date=</code> requires <code class="cs1-code">&#124;archive-url=</code>',
anchor = 'archive_date_missing_url',
category = 'CS1 errors: archive-url',
hidden = false
},
err_archive_date_url_ts_mismatch = {
message = '<code class="cs1-code">&#124;archive-date=</code> / <code class="cs1-code">&#124;archive-url=</code> timestamp mismatch',
anchor = 'archive_date_url_ts_mismatch',
category = 'CS1 errors: archive-url',
hidden = false
},
err_archive_missing_date = {
message = '<code class="cs1-code">&#124;archive-url=</code> requires <code class="cs1-code">&#124;archive-date=</code>',
Line 1,309 ⟶ 1,424:
anchor = 'bad_lccn',
category = 'CS1 errors: LCCN',
hidden = false
},
err_bad_medrxiv = {
message = 'Check <code class="cs1-code">&#124;medrxiv=</code> value',
anchor = 'bad_medrxiv',
category = 'CS1 errors: medRxiv',
hidden = false
},
Line 1,543 ⟶ 1,664:
anchor = 'invisible_char',
category = 'CS1 errors: invisible characters',
hidden = false
},
err_medrxiv_missing = {
message = '<code class="cs1-code">&#124;medrxiv=</code> required',
anchor = 'medrxiv_missing',
category = 'CS1 errors: medRxiv', -- same as bad medRxiv
hidden = false
},
Line 1,561 ⟶ 1,688:
anchor = 'missing_pipe',
category = 'CS1 errors: missing pipe',
hidden = false
},
err_missing_publisher = {
message = 'Cite $1 requires <code class="cs1-code">&#124;$2=</code>', -- $1 is cs1 template name; $2 is canonical publisher parameter name for cite $1
anchor = 'missing_publisher',
category = 'CS1 errors: missing publisher',
hidden = false
},
Line 1,585 ⟶ 1,718:
anchor = 'parameter_ignored_suggest',
category = 'CS1 errors: unsupported parameter',
hidden = false
},
err_periodical_ignored = {
message = '<code class="cs1-code">&#124;$1=</code> ignored', -- $1 is parameter name
anchor = 'periodical_ignored',
category = 'CS1 errors: periodical ignored',
hidden = false
},
Line 1,602 ⟶ 1,741:
message = '<code class="cs1-code">&#124;ssrn=</code> required',
anchor = 'ssrn_missing',
category = 'CS1 errors: SSRN', -- same as bad arxiv
hidden = false
},
Line 1,618 ⟶ 1,757:
},
err_param_unknown_empty = {
message = 'Cite has empty unknown parameter$1: $2', -- $1 is 's' or empty space; $2 is emptyemty unknown param list
anchor = 'param_unknown_empty',
category = 'CS1 errors: empty unknown parameters',
Line 1,654 ⟶ 1,793:
anchor = 'authors',
category = 'CS1 maint: uses authors parameter',
hidden = true,
},
maint_bibcode = {
message = nil,
anchor = 'bibcode',
category = 'CS1 maint: bibcode',
hidden = true,
},
maint_location_no_publisher = { -- cite book, conference, encyclopedia; citation as book cite or encyclopedia cite
message = nil,
anchor = 'location_no_publisher',
category = 'CS1 maint: ___location missing publisher',
hidden = true,
},
Line 1,756 ⟶ 1,907:
anchor = 'others_avm',
category = 'CS1 maint: others in cite AV media (notes)',
hidden = true,
},
maint_overridden_setting = {
message = nil,
anchor = 'overridden',
category = 'CS1 maint: overridden setting',
hidden = true,
},
Line 1,854 ⟶ 2,011:
q = 'Q118398',
label = 'arXiv',
prefix = 'https://arxiv.org/abs/', -- protocol-relative tested 2013-09-04
encode = false,
COinS = 'info:arxiv',
Line 1,990 ⟶ 2,147:
q = 'Q1420342',
label = 'JSTOR',
prefix = 'https://www.jstor.org/stable/', -- protocol-relative tested 2013-09-04
COinS = 'pre', -- use prefix value
encode = false,
Line 2,002 ⟶ 2,159:
q = 'Q620946',
label = 'LCCN',
prefix = 'https://lccn.loc.gov/', -- protocol-relative tested 2015-12-28
COinS = 'info:lccn',
encode = false,
separator = '&nbsp;',
},
['MEDRXIV'] = {
parameters = {'medrxiv'},
link = 'medRxiv',
redirect = 'medRxiv (identifier)',
q = 'Q58465838',
label = 'medRxiv',
prefix = 'https://www.medrxiv.org/content/',
COinS = 'pre', -- use prefix value
access = 'free', -- free to read
encode = false,
separator = '&nbsp;',
Line 2,048 ⟶ 2,217:
q = 'Q2015776',
label = 'OSTI',
prefix = 'https://www.osti.gov/biblio/', -- protocol-relative tested 2018-09-12
COinS = 'pre', -- use prefix value
encode = true,
Line 2,158 ⟶ 2,327:
 
return {
use_identifier_redirects = trueuse_identifier_redirects, -- whenbooleans truedefined usein redirectthe namesettings forat identifierthe labeltop links;of always true atthis en.wikimodule
local_lang_cat_enable = local_lang_cat_enable,
local_lang_cat_enable = false; -- when true categorizes pages where |language=<local wiki's language>; always false at en.wiki
date_name_auto_xlate_enable = date_name_auto_xlate_enable,
date_name_auto_xlate_enable = false; -- when true translates English month-names to the local-wiki's language month names; always false at en.wiki
date_digit_auto_xlate_enable = date_digit_auto_xlate_enable,
date_digit_auto_xlate_enable = false; -- when true translates Western date digit to the local-wiki's language digits (date_names['local_digits']); always false at en.wiki
-- tables and variables created when this module is loaded
global_df = get_date_format (), -- this line can be replaced with "global_df = 'dmy-all'," to have all dates auto translated to dmy format.
global_cs1_config_t = global_cs1_config_t, -- global settings from {{cs1 config}}
punct_skip = build_skip_table (punct_skip, punct_meta_params),
url_skip = build_skip_table (url_skip, url_meta_params),
Line 2,188 ⟶ 2,358:
prop_cats = prop_cats,
script_lang_codes = script_lang_codes,
lang_tag_remap = lang_tag_remap,
lang_code_remap = lang_code_remap,
lang_name_remap = lang_name_remap,
this_wiki_code = this_wiki_code,
Retrieved from "https://en.wikipedia.org/wiki/Module:Citation/CS1/Configuration"