Module:Citation/CS1/Identifiers: Difference between revisions

m
(sync from sandbox;)
m (47 revisions imported from templatewiki:Module:Citation/CS1/Identifiers)
Line 1: Line 1:
local identifiers = {};


--[[--------------------------< F O R W A R D  D E C L A R A T I O N S >--------------------------------------
--[[--------------------------< F O R W A R D  D E C L A R A T I O N S >--------------------------------------
Line 8: Line 11:


local cfg; -- table of configuration tables that are defined in Module:Citation/CS1/Configuration
local cfg; -- table of configuration tables that are defined in Module:Citation/CS1/Configuration
local wd_int_lang = (mw.site.server:match ('wikidata') and mw.getCurrentFrame():preprocess('{{int:lang}}')) or '';




Line 39: Line 44:
end
end
if is_set (options.q) and mw.wikibase then -- wikibase test here avoids script errors in third party wikis that aren't using mw.wikibase extension
if is_set (options.q) and mw.wikibase then
wd_article = mw.wikibase.getEntity (options.q):getSitelink (this_wiki_code .. 'wiki'); -- fetch article title from wd
wd_article = mw.wikibase.getEntity (options.q):getSitelink (this_wiki_code .. 'wiki'); -- fetch article title from wd
if wd_article then
if wd_article then
Line 61: Line 66:


local function internal_link_id(options)
local function internal_link_id(options)
local id = mw.ustring.gsub (options.id, '%d', cfg.date_names.local_digits); -- translate 'local' digits to Western 0-9


return table.concat (
return table.concat (
Line 71: Line 75:
{
{
options.prefix,
options.prefix,
id, -- translated to western digits
options.id,
options.suffix or ''
options.suffix or ''
}),
}),
substitute (cfg.presentation['bdi'], {'', mw.text.nowiki (options.id)}) -- bdi tags to prevent Latn script identifiers from being reversed at rtl language wikis
mw.text.nowiki (options.id)
); -- nowiki because MediaWiki still has magic links for ISBN and the like; TODO: is it really required?
);
});
});
end
end
Line 281: Line 285:
1–4 must be digits and must represent a year in the range of 1000 – next year
1–4 must be digits and must represent a year in the range of 1000 – next year
5 must be a letter
5 must be a letter
6–8 must be letter, digit, ampersand, or dot (ampersand cannot directly precede a dot; &. )
6 must be letter, ampersand, or dot (ampersand cannot directly precede a dot; &. )
7–8 must be letter, digit, ampersand, or dot (ampersand cannot directly precede a dot; &. )
9–18 must be letter, digit, or dot
9–18 must be letter, digit, or dot
19 must be a letter or dot
19 must be a letter or dot
Line 299: Line 304:
err_type = 'length';
err_type = 'length';
else
else
year = id:match ("^(%d%d%d%d)[%a][%w&%.][%w&%.][%w&%.][%w.]+[%a%.]$") --  
year = id:match ("^(%d%d%d%d)[%a][%a&%.][%a&%.%d][%a&%.%d][%a%d%.]+[%a%.]$") --  
if not year then -- if nil then no pattern match
if not year then -- if nil then no pattern match
err_type = 'value'; -- so value error
err_type = 'value'; -- so value error
Line 390: Line 395:
if is_set(inactive) then
if is_set(inactive) then
local inactive_year = inactive:match("%d%d%d%d") or ''; -- try to get the year portion from the inactive date
local inactive_year = inactive:match("%d%d%d%d") or ''; -- try to get the year portion from the inactive date
local inactive_month, good;
if is_set(inactive_year) then
 
table.insert( z.error_categories, "Pages with DOIs inactive since " .. inactive_year );
if is_set (inactive_year) then
if 4 < inactive:len() then -- inactive date has more than just a year (could be anything)
local lang_obj = mw.getContentLanguage(); -- get a language object for this wiki
good, inactive_month = pcall (lang_obj.formatDate, lang_obj, 'F', inactive); -- try to get the month name from the inactive date
if not good then
inactive_month = nil; -- something went wrong so make sure this is unset
end
end
else
else
inactive_year = nil; -- |doi-broken= has something but it isn't a date
table.insert( z.error_categories, "Pages with inactive DOIs" ); -- when inactive doesn't contain a recognizable year
end
end
inactive = " (" .. cfg.messages['inactive'] .. " " .. inactive .. ")"
if is_set(inactive_year) and is_set (inactive_month) then
table.insert( z.error_categories, 'Pages with DOIs inactive as of ' .. inactive_year .. ' ' .. inactive_month); -- use inactive month in category name
elseif is_set(inactive_year) then
table.insert( z.error_categories, 'Pages with DOIs inactive as of ' .. inactive_year);
else
table.insert( z.error_categories, 'Pages with inactive DOIs'); -- when inactive doesn't contain a recognizable date
end
inactive = " (" .. cfg.messages['inactive'] .. ' ' .. inactive .. ')';
end
end
text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=access}) .. (inactive or '')
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=access}) .. (inactive or '')


if nil == id:match("^10%.[^%s–]-/[^%s–]-[^%.,]$") or -- doi must begin with '10.', must contain a fwd slash, must not contain spaces or endashes, and must not end with period or comma
if nil == id:match("^10%.[^%s–]-/[^%s–]-[^%.,]$") then -- doi must begin with '10.', must contain a fwd slash, must not contain spaces or endashes, and must not end with period or comma
id: match ('^10.5555') then -- test doi will never resolve
cat = ' ' .. set_error( 'bad_doi' );
cat = ' ' .. set_error ('bad_doi');
end
end


Line 745: Line 732:
--[[--------------------------< O C L C >----------------------------------------------------------------------
--[[--------------------------< O C L C >----------------------------------------------------------------------


Validate and format an oclc id.  https://www.oclc.org/batchload/controlnumber.en.html {{dead link}}
Validate and format an oclc id.  https://www.oclc.org/batchload/controlnumber.en.html
archived at: https://web.archive.org/web/20161228233804/https://www.oclc.org/batchload/controlnumber.en.html


]]
]]
Line 849: Line 835:


local function pmc(id, embargo)
local function pmc(id, embargo)
local test_limit = 7000000; -- update this value as PMCs approach
local test_limit = 6500000; -- update this value as PMCs approach
local handler = cfg.id_handlers['PMC'];
local handler = cfg.id_handlers['PMC'];
local err_cat = ''; -- presume that PMC is valid
local err_cat = ''; -- presume that PMC is valid
Line 976: Line 962:


A numerical identifier in the form nnnn.nnnnn - leading zeros in the first quartet optional
A numerical identifier in the form nnnn.nnnnn - leading zeros in the first quartet optional
format described here: http://emis.mi.sanu.ac.rs/ZMATH/zmath/en/help/search/
temporary format is apparently eight digits.  Anything else is an error


]]
]]
Line 985: Line 967:
local function zbl (id)
local function zbl (id)
local handler = cfg.id_handlers['ZBL'];
local handler = cfg.id_handlers['ZBL'];
local id_num;
local err_cat = '';
local err_cat = '';
if id:match('^%d%d%d%d%d%d%d%d$') then -- is this identifier using temporary format?
id_num = id:match ('^[Zz][Bb][Ll](.*)$'); -- identifier with zbl prefix; extract identifier
add_maint_cat ('zbl'); -- yes, add maint cat
 
elseif not id:match('^%d?%d?%d?%d%.%d%d%d%d%d$') then -- not temporary, is it normal format?
if is_set (id_num) then
err_cat = ' ' .. set_error( 'bad_zbl' ); -- no, set an error message
add_maint_cat ('zbl_format');
else -- plain number without zbl prefix
id_num = id; -- if here id does not have prefix
end
 
if id_num:match('^%d?%d?%d?%d%.%d%d%d%d%d$') then
id = id_num; -- id matches pattern
else
err_cat = ' ' .. set_error( 'bad_zbl' ); -- set an error message
end
end
Line 1,121: Line 1,112:
Parameters which have a predefined access level (e.g. arxiv) do not use this
Parameters which have a predefined access level (e.g. arxiv) do not use this
function as they are directly rendered as free without using an additional parameter.
function as they are directly rendered as free without using an additional parameter.
access-level values must match the case used in cfg.keywords['id-access'] (lowercase unless there is some special reason for something else)


]]
]]
Line 1,129: Line 1,118:
local id_accesses_list = {};
local id_accesses_list = {};
for k, v in pairs( cfg.id_handlers ) do
for k, v in pairs( cfg.id_handlers ) do
local access_param = v.custom_access; -- name of identifier's access-level parameter
local access_param = v.custom_access;
local k_lower = string.lower(k);
if is_set(access_param) then
if is_set(access_param) then
local access_level = args[access_param]; -- get the assigned value if there is one
local access_level = args[access_param];
if is_set (access_level) then
if is_set(access_level) then
if not in_array (access_level, cfg.keywords['id-access']) then -- exact match required
if not in_array (access_level:lower(), cfg.keywords['id-access']) then
table.insert( z.message_tail, { set_error( 'invalid_param_val', {access_param, access_level}, true ) } );
table.insert( z.message_tail, { set_error( 'invalid_param_val', {access_param, access_level}, true ) } );
access_level = nil; -- invalid so unset
access_level = nil;
end
end
if not is_set(id_list[k]) then -- identifer access-level must have a matching identifier
if not is_set(id_list[k]) then
table.insert( z.message_tail, { set_error( 'param_access_requires_param', {k:lower()}, true ) } ); -- param name is uppercase in cfg.id_handlers (k); lowercase for error message
table.insert( z.message_tail, { set_error( 'param_access_requires_param', {k_lower}, true ) } );
end
if is_set(access_level) then
access_level = access_level:lower();
end
end
id_accesses_list[k] = access_level;
id_accesses_list[k] = access_level;
Line 1,168: Line 1,161:
end
end


--[[--------------------------< E X P O R T E D  F U N C T I O N S >------------------------------------------
]]


return {
return {