--
----------------------------< F O R W A R D D E C L A R A T I O N S >--------------------------------------
local is_set, in_array, set_error, select_one, add_maint_cat, substitute, make_wikilink; -- functions in Module:Citation/CS1/Utilities
local z; -- table of tables defined in Module:Citation/CS1/Utilities
local cfg; -- table of configuration tables that are defined in Module:Citation/CS1/Configuration
--
--
local function wikidata_article_name_get (q) if not is_set (q) or (q and not mw.wikibase) then -- when no q number or when a q number but mw.wikibase not installed on this wiki return nil; -- abandon end
local wd_article; local this_wiki_code = cfg.this_wiki_code; -- wikipedia subdomain; 'en' for en.wikipedia.org
wd_article = mw.wikibase.getSitelink (q, this_wiki_code .. 'wiki'); -- fetch article title from wd; nil when no title available at this wiki
if wd_article then wd_article = table.concat ; -- interwiki-style link without brackets if taken from wd; leading colon required end
return wd_article; -- article title from wd; nil elseend
----------------------------< E X T E R N A L _ L I N K _ I D >----------------------------------------------
Formats a wiki style external link
local function external_link_id(options) local url_string = options.id; local ext_link; local this_wiki_code = cfg.this_wiki_code; -- wikipedia subdomain; 'en' for en.wikipedia.org local wd_article; -- article title from wikidata if options.encode
nil then url_string = mw.uri.encode(url_string); end
ext_link = mw.ustring.format ('[%s%s%s %s]', options.prefix, url_string, options.suffix or "", mw.text.nowiki(options.id)); if is_set(options.access) then ext_link = substitute (cfg.presentation['ext-link-access-signal'],); -- add the free-to-read / paywall lock end
if not (cfg.use_identifier_redirects and is_set (options.redirect)) then -- redirect has priority so if enabled and available don't fetch from wikidata because expensive wd_article = wikidata_article_name_get (options.q); -- if wikidata has an article title for this wiki, get it; end local label_link = (cfg.use_identifier_redirects and is_set (options.redirect) and options.redirect) or wd_article or options.link;
return table.concat ;end
----------------------------< I N T E R N A L _ L I N K _ I D >----------------------------------------------
Formats a wiki style internal link
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
if not (cfg.use_identifier_redirects and is_set (options.redirect)) then -- redirect has priority so if enabled and available don't fetch from wikidata because expensive wd_article = wikidata_article_name_get (options.q); -- if wikidata has an article title for this wiki, get it; end
local label_link = (cfg.use_identifier_redirects and is_set (options.redirect) and options.redirect) or wd_article or options.link;
return table.concat ;end
--embargo= against today's date. If embargo date isin the future, returns the content of |embargo=; otherwise, returns and empty string because the embargo has expired or because|embargo= was not set in this cite.
local function is_embargoed (embargo) if is_set (embargo) then local lang = mw.getContentLanguage; local good1, embargo_date, good2, todays_date; good1, embargo_date = pcall(lang.formatDate, lang, 'U', embargo); good2, todays_date = pcall(lang.formatDate, lang, 'U'); if good1 and good2 then -- if embargo date and today's date are good dates if tonumber(embargo_date) >= tonumber(todays_date) then -- is embargo date is in the future? return embargo; -- still embargoed else add_maint_cat ('embargo') return ; -- unset because embargo has expired end end end return ; -- |embargo= not set return empty stringend
--[=[-------------------------< I S _ V A L I D _ B I O R X I V _ D A T E >------------------------------------ returns true if: 2019-12-11T00:00Z <= biorxiv_date < today + 2 days The dated form of biorxiv identifier has a start date of 2019-12-11. The unix timestamp for that date is {{#time:U|2019-12-11}} = 1576022400 biorxiv_date is the date provided in those |biorxiv= parameter values that are dated at time 00:00:00 UTC today is the current date at time 00:00:00 UTC plus 48 hours if today is 2015-01-01T00:00:00 then adding 24 hours gives 2015-01-02T00:00:00 – one second more than today adding 24 hours gives 2015-01-03T00:00:00 – one second more than tomorrow This function does not work if it is fed month names for languages other than English. Wikimedia #time: parser apparently doesn't understand non-Engish date month names. This function will always return false when the date contains a non-English month name because good1 is false after the call to lang.formatDate. To get around that call this function with YYYY-MM-DD format dates. ]=]
local function is_valid_biorxiv_date (biorxiv_date) local good1, good2; local biorxiv_ts, tomorrow_ts; -- to hold unix time stamps representing the dates local lang_object = mw.getContentLanguage;
good1, biorxiv_ts = pcall (lang_object.formatDate, lang_object, 'U', biorxiv_date); -- convert biorxiv_date value to unix timesatmp good2, tomorrow_ts = pcall (lang_object.formatDate, lang_object, 'U', 'today + 2 days'); -- today midnight + 2 days is one second more than all day tomorrow if good1 and good2 then -- lang.formatDate returns a timestamp in the local script which which tonumber may not understand biorxiv_ts = tonumber (biorxiv_ts) or lang_object:parseFormattedNumber (biorxiv_ts); -- convert to numbers for the comparison; tomorrow_ts = tonumber (tomorrow_ts) or lang_object:parseFormattedNumber (tomorrow_ts); else return false; -- one or both failed to convert to unix time stamp end
return ((1576022400 <= biorxiv_ts) and (biorxiv_ts < tomorrow_ts)) -- 2012-12-11T00:00Z <= biorxiv_date < tomorrow's dateend
----------------------------< IS _ V A L I D _ I S X N >-----------------------------------------------------
ISBN-10 and ISSN validator code calculates checksum across all isbn/issn digits including the check digit.ISBN-13 is checked in isbn.
If the number is valid the result will be 0. Before calling this function, issbn/issn must be checked for lengthand stripped of dashes, spaces and other non-isxn characters.
local function is_valid_isxn (isxn_str, len) local temp = 0; isxn_str = ; -- make a table of byte values '0' → 0x30 .. '9' → 0x39, 'X' → 0x58 len = len+1; -- adjust to be a loop counter for i, v in ipairs(isxn_str) do -- loop through all of the bytes and calculate the checksum if v
0; -- returns true if calculation result is zeroend
----------------------------< IS _ V A L I D _ I S X N _ 1 3 >-----------------------------------------------
ISBN-13 and ISMN validator code calculates checksum across all 13 isbn/ismn digits including the check digit.If the number is valid, the result will be 0. Before calling this function, isbn-13/ismn must be checked for lengthand stripped of dashes, spaces and other non-isxn-13 characters.
local function is_valid_isxn_13 (isxn_str) local temp=0; isxn_str = ; -- make a table of byte values '0' → 0x30 .. '9' → 0x39 for i, v in ipairs(isxn_str) do temp = temp + (3 - 2*(i % 2)) * tonumber(string.char(v)); -- multiply odd index digits by 1, even index digits by 3 and sum; includes check digit end return temp % 10
--
local function normalize_lccn (lccn) lccn = lccn:gsub ("%s", ""); -- 1. strip whitespace
if nil ~= string.find (lccn,'/') then lccn = lccn:match ("(.-)/"); -- 2. remove forward slash and all character to the right of it end
local prefix local suffix prefix, suffix = lccn:match ("(.+)%-(.+)"); -- 3.a remove hyphen by splitting the string into prefix and suffix
if nil ~= suffix then -- if there was a hyphen suffix=string.rep("0", 6-string.len (suffix)) .. suffix; -- 3.b.2 left fill the suffix with 0s if suffix length less than 6 lccn=prefix..suffix; -- reassemble the lccn end return lccn; end
--
--class= parameter which is not supported in this form the third form, valid from January 2015 is: arXiv:
local function arxiv (id, class) local handler = cfg.id_handlers['ARXIV']; local year, month, version; local err_cat = false; -- assume no error message local text; -- output text if id:match("^%a[%a%.%-]+/[90]%d[01]%d%d%d%d$") or id:match("^%a[%a%.%-]+/[90]%d[01]%d%d%d%dv%d+$") then -- test for the 9108-0703 format w/ & w/o version year, month = id:match("^%a[%a%.%-]+/([90]%d)([01]%d)%d%d%d[v%d]*$"); year = tonumber(year); month = tonumber(month); if ((not (90 < year or 8 > year)) or (1 > month or 12 < month)) or -- if invalid year or invalid month ((91
year and 3 < month)) then -- if years ok, are starting and ending months ok? err_cat = true; -- flag for error message end
elseif id:match("^%d%d[01]%d%.%d%d%d%d$") or id:match("^%d%d[01]%d%.%d%d%d%dv%d+$") then -- test for the 0704-1412 w/ & w/o version year, month = id:match("^(%d%d)([01]%d)%.%d%d%d%d[v%d]*$"); year = tonumber(year); month = tonumber(month); if ((7 > year) or (14 < year) or (1 > month or 12 < month)) or -- is year invalid or is month invalid? (doesn't test for future years) ((7
elseif id:match("^%d%d[01]%d%.%d%d%d%d%d$") or id:match("^%d%d[01]%d%.%d%d%d%d%dv%d+$") then -- test for the 1501- format w/ & w/o version year, month = id:match("^(%d%d)([01]%d)%.%d%d%d%d%d[v%d]*$"); year = tonumber(year); month = tonumber(month); if ((15 > year) or (1 > month or 12 < month)) then -- is year invalid or is month invalid? (doesn't test for future years) err_cat = true; -- flag for error message end
else err_cat = true; -- not a recognized format; flag for error message end
err_cat = err_cat and table.concat or ; -- set error message if flag is true text = external_link_id .. err_cat;
if is_set (class) then if id:match ('^%d+') then text = table.concat ; -- external link within square brackets, not wikilink else text = table.concat ; end end
return text; end
--
local function bibcode (id, access) local handler = cfg.id_handlers['BIBCODE']; local err_type; local year;
local text = external_link_id ; if 19 ~= id:len then err_type = cfg.err_msg_supl.length; else year = id:match ("^(%d%d%d%d)[%a][%w&%.][%w&%.][%w&%.][%w.]+[%a%.]$") -- if not year then -- if nil then no pattern match err_type = cfg.err_msg_supl.value; -- so value error else local next_year = tonumber(os.date ('%Y'))+1; -- get the current year as a number and add one for next year year = tonumber (year); -- convert year portion of bibcode to a number if (1000 > year) or (year > next_year) then err_type = cfg.err_msg_supl.year; -- year out of bounds end if id:find('&%.') then err_type = cfg.err_msg_supl.journal; -- journal abbreviation must not have '&.' (if it does its missing a letter) end end end
if is_set (err_type) then -- if there was an error detected text = text .. ' ' .. set_error('bad_bibcode',); end return text;end
--
local function biorxiv(id) local handler = cfg.id_handlers['BIORXIV']; local err_cat = true; -- flag; assume that there will be an error local patterns = for _, pattern in ipairs (patterns) do -- spin through the patterns looking for a match if id:match (pattern) then local y, m, d = id:match (pattern); -- found a match, attempt to get year, month and date from the identifier
if m then -- m is nil when id is the six-digit form if not is_valid_biorxiv_date (y .. '-' .. m .. '-' .. d) then -- validate the encoded date; TODO: don't ignore leapyear and actual month lengths (is a poor date validator) break; -- date fail; break out early so we don't unset the error message end end err_cat = nil; -- we found a match so unset the error message break; -- and done end end -- err_cat remains set here when no match
return external_link_id .. (err_cat and (' ' .. set_error('bad_biorxiv')) or );end
--
local function citeseerx (id) local handler = cfg.id_handlers['CITESEERX']; local matched; local text = external_link_id ; matched = id:match ("^10%.1%.1%.[1-9]%d?%d?%d?%.[1-9]%d?%d?%d?$"); if not matched then text = text .. ' ' .. set_error('bad_citeseerx'); end return text;end
--
local function doi(id, inactive, access) local cat = "" local handler = cfg.id_handlers['DOI']; local text; if is_set(inactive) then local inactive_year = inactive:match("%d%d%d%d") or ; -- try to get the year portion from the inactive date local inactive_month, good;
if is_set (inactive_year) then if 4 < inactive:len then -- inactive date has more than just a year (could be anything) local lang_obj = mw.getContentLanguage; -- get a language object for this wiki good, inactive_month = pcall (lang_obj.formatDate, lang_obj, 'F', inactive); -- try to get the month name from the inactive date if not good then inactive_month = nil; -- something went wrong so make sure this is unset end end else inactive_year = nil; -- |doi-broken= has something but it isn't a date end if is_set(inactive_year) and is_set (inactive_month) then table.insert(z.error_categories, 'Pages with DOIs inactive as of ' .. inactive_year .. ' ' .. inactive_month); -- use inactive month in category name elseif is_set(inactive_year) then table.insert(z.error_categories, 'Pages with DOIs inactive as of ' .. inactive_year); else table.insert(z.error_categories, 'Pages with inactive DOIs'); -- when inactive doesn't contain a recognizable date end inactive = " (" .. cfg.messages['inactive'] .. ' ' .. inactive .. ')'; end
text = external_link_id .. (inactive or )
local registrant = id:match ('^10%.([^/]+)/[^%s–]-[^%.,]$'); -- registrant set when doi has the proper basic form registrant_err_patterns = if registrant then -- when doi has proper form for i, pattern in ipairs (registrant_err_patterns) do -- spin through error patterns if registrant:match (pattern) then -- to validate registrant codes cat = ' ' .. set_error ('bad_doi'); -- when found, mark this doi as bad break; -- and done end end else cat = ' ' .. set_error ('bad_doi'); -- invalid directory or malformed end
return text .. cat end
--
local function hdl(id, access) local handler = cfg.id_handlers['HDL']; local query_params = local hdl, suffix, param = id:match ('(.-)(%?(%a+).+)$'); -- look for query string local found;
if hdl then -- when there are query strings, this is the handle identifier portion for _, q in ipairs (query_params) do -- spin through the list of query parameters if param:match ('^' .. q) then -- if the query string begins with one of the parameters found = true; -- announce a find break; -- and stop looking end end end
if found then id = hdl; -- found so replace id with the handle portion; this will be url encoded, suffix will not else suffix = ; -- make sure suffix is empty string for concatenation else end
local text = external_link_id
if nil
----------------------------< I S B N >----------------------------------------------------------------------
Determines whether an ISBN string is valid
local function isbn(isbn_str) if nil ~= isbn_str:match("[^%s-0-9X]") then return false, cfg.err_msg_supl.char; -- fail if isbn_str contains anything but digits, hyphens, or the uppercase X end isbn_str = isbn_str:gsub("-", ""):gsub(" ", ""); -- remove hyphens and spaces local len = isbn_str:len; if len ~= 10 and len ~= 13 then return false, cfg.err_msg_supl.length; -- fail if incorrect length end
if len
nil then -- fail if isbn_str has 'X' anywhere but last position return false, cfg.err_msg_supl.form; end return is_valid_isxn(isbn_str, 10), cfg.err_msg_supl.check; else if isbn_str:match("^%d+$")
nil then return false, cfg.err_msg_supl.prefix; -- fail when isbn13 does not begin with 978 or 979 end if isbn_str:match ('^9790') then return false, cfg.err_msg_supl.group; -- group identifier '0' is reserved to ismn end return is_valid_isxn_13 (isbn_str), cfg.err_msg_supl.check; endend
--asin= with |isbn=.Error message if not 10 characters, if not isbn10, if mixed and first character is a digit.
This function is positioned here because it calls isbn
local function asin(id, domain) local err_cat = ""
if not id:match("^[%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u]$") then err_cat = ' ' .. set_error ('bad_asin'); -- asin is not a mix of 10 uppercase alpha and numeric characters else if id:match("^%d%d%d%d%d%d%d%d%d[%dX]$") then -- if 10-digit numeric (or 9 digits with terminal X) if isbn(id) then -- see if asin value is isbn10 add_maint_cat ('ASIN'); elseif not is_set (err_cat) then err_cat = ' ' .. set_error ('bad_asin'); -- asin is not isbn10 end elseif not id:match("^%u[%d%u]+$") then err_cat = ' ' .. set_error ('bad_asin'); -- asin doesn't begin with uppercase alpha end end if not is_set(domain) then domain = "com"; elseif in_array (domain,) then -- Japan, United Kingdom domain = "co." .. domain; elseif in_array (domain,) then -- Australia, Brazil, Mexico domain = "com." .. domain; end local handler = cfg.id_handlers['ASIN']; return external_link_id .. err_cat;end
--
local function ismn (id) local handler = cfg.id_handlers['ISMN']; local text; local valid_ismn = true; local id_copy;
id_copy = id; -- save a copy because this testing is destructive id=id:gsub("[%s-–]", ""); -- strip spaces, hyphens, and endashes from the ismn
if 13 ~= id:len or id:match("^9790%d*$")
-- text = internal_link_id local label_link = (cfg.use_identifier_redirects and is_set (handler.redirect) and handler.redirect) or wd_article or handler.link; -- because no place to link to yet
text = table.concat (-- because no place to link to yet );
if false
--issn=0819 4327 gives: 4327 0819 4327 -- can't have spaces in an external link This code now prevents that by inserting a hyphen at the issn midpoint. It also validates the issn for lengthand makes sure that the checkdigit agrees with the calculated value. Incorrect length (8 digits), charactersother than 0-9 and X, or checkdigit / calculated value mismatch will all cause a check issn error message. Theissn is always displayed with a hyphen, even if the issn was given as a single group of 8 digits.
local function issn(id, e) local issn_copy = id; -- save a copy of unadulterated issn; use this version for display if issn does not validate local handler; local text; local valid_issn = true; if e then handler = cfg.id_handlers['EISSN']; else handler = cfg.id_handlers['ISSN']; end
id=id:gsub("[%s-–]", ""); -- strip spaces, hyphens, and endashes from the issn
if 8 ~= id:len or nil
if true
if false
----------------------------< J F M >-----------------------------------------------------------------------
A numerical identifier in the form nn.nnnn.nn
local function jfm (id) local handler = cfg.id_handlers['JFM']; local id_num; local err_cat = ; id_num = id:match ('^[Jj][Ff][Mm](.*)$'); -- identifier with jfm prefix; extract identifier
if is_set (id_num) then add_maint_cat ('jfm_format'); else -- plain number without mr prefix id_num = id; -- if here id does not have prefix end
if id_num and id_num:match('^%d%d%.%d%d%d%d%.%d%d$') then id = id_num; -- jfm matches pattern else err_cat = ' ' .. set_error('bad_jfm'); -- set an error message end return external_link_id .. err_cat;end
--
local function lccn(lccn) local handler = cfg.id_handlers['LCCN']; local err_cat = ; -- presume that LCCN is valid local id = lccn; -- local copy of the lccn
id = normalize_lccn (id); -- get canonical form (no whitespace, hyphens, forward slashes) local len = id:len; -- get the length of the lccn
if 8
len then -- LCCN should be adddddddd if nil
len then -- LCCN should be aadddddddd or dddddddddd if id:match("[^%d]") then -- if LCCN has anything but digits (nil if only digits) ... if nil
len then -- LCCN should be aaadddddddd or adddddddddd if not (id:match("^%l%l%l%d%d%d%d%d%d%d%d") or id:match("^%l%d%d%d%d%d%d%d%d%d%d")) then -- see if it matches one of our patterns err_cat = ' ' .. set_error('bad_lccn'); -- no match, set an error message end elseif 12
if not is_set (err_cat) and nil ~= lccn:find ('%s') then err_cat = ' ' .. set_error('bad_lccn'); -- lccn contains a space, set an error message end
return external_link_id .. err_cat;end
----------------------------< M R >--------------------------------------------------------------------------
A seven digit number; if not seven digits, zero-fill leading digits to make seven digits.
local function mr (id) local handler = cfg.id_handlers['MR']; local id_num; local id_len; local err_cat = ; id_num = id:match ('^[Mm][Rr](%d+)$'); -- identifier with mr prefix
if is_set (id_num) then add_maint_cat ('mr_format'); else -- plain number without mr prefix id_num = id:match ('^%d+$'); -- if here id is all digits end
id_len = id_num and id_num:len or 0; if (7 >= id_len) and (0 ~= id_len) then id = string.rep ('0', 7-id_len) .. id_num; -- zero-fill leading digits else err_cat = ' ' .. set_error('bad_mr'); -- set an error message end return external_link_id .. err_cat;end
--
local function oclc (id) local handler = cfg.id_handlers['OCLC']; local number; local err_msg = ; -- empty string for concatenation if id:match('^ocm%d%d%d%d%d%d%d%d$') then -- ocm prefix and 8 digits; 001 field (12 characters) number = id:match('ocm(%d+)'); -- get the number elseif id:match('^ocn%d%d%d%d%d%d%d%d%d$') then -- ocn prefix and 9 digits; 001 field (12 characters) number = id:match('ocn(%d+)'); -- get the number elseif id:match('^on%d%d%d%d%d%d%d%d%d%d+$') then -- on prefix and 10 or more digits; 001 field (12 characters) number = id:match('^on(%d%d%d%d%d%d%d%d%d%d+)$'); -- get the number elseif id:match('^%(OCoLC%)[1-9]%d*$') then -- (OCoLC) prefix and variable number digits; no leading zeros; 035 field number = id:match('%(OCoLC%)([1-9]%d*)'); -- get the number if 9 < number:len then number = nil; -- contrain to 1 to 9 digits; change this when oclc issues 10-digit numbers end elseif id:match('^%d+$') then -- no prefix number = id; -- get the number if 10 < number:len then number = nil; -- contrain to 1 to 10 digits; change this when oclc issues 11-digit numbers end end
if number then -- proper format id = number; -- exclude prefix, if any, from external link else err_msg = ' ' .. set_error('bad_oclc') -- add an error message if the id is malformed end local text = external_link_id .. err_msg;
return text;end
----------------------------< O P E N L I B R A R Y >--------------------------------------------------------
Formats an OpenLibrary link, and checks for associated errors.
local function openlibrary(id, access) local code; local handler = cfg.id_handlers['OL']; local ident; ident, code = id:gsub('^OL', ):match("^(%d+([AMW]))$"); -- optional OL prefix followed immediately by digits followed by 'A', 'M', or 'W'; remove OL prefix
if not is_set (ident) then -- if malformed return an error return external_link_id .. ' ' .. set_error('bad_ol'); end id = ident; -- use ident without the optional OL prefix (it has been removed) if (code
"M") then return external_link_id end
if (code
--pmc= PMCs are sequential numbers beginning at 1 and counting up. This code checks the PMC to see that it contains only digits and is lessthan test_limit; the value in local variable test_limit will need to be updated periodically as more PMCs are issued.
local function pmc(id, embargo)-- local test_limit = cfg.id_limits.pmc; -- update this value as PMCs approach local handler = cfg.id_handlers['PMC']; local err_cat = ; -- presume that PMC is valid local id_num; local text; id_num = id:match ('^[Pp][Mm][Cc](%d+)$'); -- identifier with pmc prefix
if is_set (id_num) then add_maint_cat ('pmc_format'); else -- plain number without pmc prefix id_num = id:match ('^%d+$'); -- if here id is all digits end
if is_set (id_num) then -- id_num has a value so test it id_num = tonumber(id_num); -- convert id_num to a number for range testing-- if 1 > id_num or test_limit < id_num then -- if PMC is outside test limit boundaries if 1 > id_num or handler.id_limit < id_num then -- if PMC is outside test limit boundaries err_cat = ' ' .. set_error('bad_pmc'); -- set an error message else id = tostring (id_num); -- make sure id is a string end else -- when id format incorrect err_cat = ' ' .. set_error('bad_pmc'); -- set an error message end if is_set (embargo) then -- is PMC is still embargoed? text = table.concat (-- still embargoed so no external link ); else text = external_link_id .. err_cat; end return text;end
----------------------------< P M I D >----------------------------------------------------------------------
Format PMID and do simple error checking. PMIDs are sequential numbers beginning at 1 and counting up. Thiscode checks the PMID to see that it contains only digits and is less than test_limit; the value in local variabletest_limit will need to be updated periodically as more PMIDs are issued.
local function pmid(id)-- local test_limit = cfg.id_limits.pmid; -- update this value as PMIDs approach local handler = cfg.id_handlers['PMID']; local err_cat = ; -- presume that PMID is valid if id:match("[^%d]") then -- if PMID has anything but digits err_cat = ' ' .. set_error('bad_pmid'); -- set an error message else -- PMID is only digits local id_num = tonumber(id); -- convert id to a number for range testing-- if 1 > id_num or test_limit < id_num then -- if PMID is outside test limit boundaries if 1 > id_num or handler.id_limit < id_num then -- if PMID is outside test limit boundaries err_cat = ' ' .. set_error('bad_pmid'); -- set an error message end end return external_link_id .. err_cat;end
----------------------------< S 2 C I D >--------------------------------------------------------------------
Format an s2cid, do simple error checking
S2CIDs are sequential numbers beginning at 1 and counting up. This code checks the s2cid to see that it is onlydigits and is less than test_limit; the value in local variable test_limit will need to be updated periodicallyas more S2CIDs are issued.
local function s2cid (id, access)-- local test_limit = cfg.id_limits.s2cid; -- update this value as S2CIDs approach local handler = cfg.id_handlers['S2CID']; local err_cat = ; -- presume that S2CID is valid local id_num; local text; id_num = id:match ('^[1-9]%d*$'); -- id must be all digits; must not begin with 0; no open access flag
if is_set (id_num) then -- id_num has a value so test it id_num = tonumber(id_num); -- convert id_num to a number for range testing-- if test_limit < id_num then -- if S2CID is outside test limit boundaries if handler.id_limit < id_num then -- if S2CID is outside test limit boundaries err_cat = ' ' .. set_error('bad_s2cid'); -- set an error message end
else -- when id format incorrect err_cat = ' ' .. set_error('bad_s2cid'); -- set an error message end
text = external_link_id .. err_cat;
return text;end
----------------------------< S B N >------------------------------------------------------------------------
9-digit form of isbn10; uses same check-digit validation when sbn is prefixed with an additional '0' to make 10 digits
local function sbn (id) local check; local err_type = ;
if nil ~= id:match("[^%s-0-9X]") then return false, cfg.err_msg_supl.char; -- fail if sbn contains anything but digits, hyphens, or the uppercase X end
id=id:gsub("[%s-]", ""); -- strip spaces and hyphens from the sbn
if 9 ~= id:len then return false, cfg.err_msg_supl.length; -- fail if incorrect length end
if id:match("^%d*X?$")
return is_valid_isxn('0' .. id, 10), cfg.err_msg_supl.check; -- prefix sbn with '0' and validate as isbn10end
----------------------------< S S R N >----------------------------------------------------------------------
Format an ssrn, do simple error checking
SSRNs are sequential numbers beginning at 100? and counting up. This code checks the ssrn to see that it isonly digits and is greater than 99 and less than test_limit; the value in local variable test_limit will needto be updated periodically as more SSRNs are issued.
local function ssrn (id)-- local test_limit = cfg.id_limits.ssrn; -- update this value as SSRNs approach local handler = cfg.id_handlers['SSRN']; local err_cat = ; -- presume that SSRN is valid local id_num; local text; id_num = id:match ('^%d+$'); -- id must be all digits
if is_set (id_num) then -- id_num has a value so test it id_num = tonumber(id_num); -- convert id_num to a number for range testing-- if 100 > id_num or test_limit < id_num then -- if SSRN is outside test limit boundaries if 100 > id_num or handler.id_limit < id_num then -- if SSRN is outside test limit boundaries err_cat = ' ' .. set_error('bad_ssrn'); -- set an error message end else -- when id format incorrect err_cat = ' ' .. set_error('bad_ssrn'); -- set an error message end text = external_link_id .. err_cat;
return text;end
----------------------------< U S E N E T _ I D >------------------------------------------------------------
Validate and format a usenet message id. Simple error checking, looks for 'id-left@id-right' not enclosed in'<' and/or '>' angle brackets.
local function usenet_id (id) local handler = cfg.id_handlers['USENETID'];
local text = external_link_id if not id:match('^.+@.+$') or not id:match('^[^<].*[^>]$')then -- doesn't have '@' or has one or first or last character is '< or '>' text = text .. ' ' .. set_error('bad_usenet_id') -- add an error message if the message id is invalid end return textend
--
local function zbl (id) local handler = cfg.id_handlers['ZBL']; local err_cat = ; if id:match('^%d%d%d%d%d%d%d%d$') then -- is this identifier using temporary format? add_maint_cat ('zbl'); -- yes, add maint cat elseif not id:match('^%d?%d?%d?%d%.%d%d%d%d%d$') then -- not temporary, is it normal format? err_cat = ' ' .. set_error('bad_zbl'); -- no, set an error message end return external_link_id .. err_cat;end
--
--
local function build_id_list(id_list, options) local new_list, handler = ;
local function fallback(k) return end; for k, v in pairs(id_list) do -- k is uc identifier name as index to cfg.id_handlers; e.g. cfg.id_handlers['ISBN'], v is a table -- fallback to read-only cfg handler = setmetatable(fallback(k));
if handler.mode
'internal' then table.insert(new_list,); elseif handler.mode ~= 'manual' then error(cfg.messages['unknown_ID_mode']); elseif k
'ASIN' then table.insert(new_list,); elseif k
'BIORXIV' then table.insert(new_list,); elseif k
'DOI' then table.insert(new_list,); elseif k
'HDL' then table.insert(new_list,); elseif k
'ISMN' then table.insert(new_list,); elseif k
'JFM' then table.insert(new_list,); elseif k
'MR' then table.insert(new_list,); elseif k
'OL' or k
'PMC' then table.insert(new_list,); elseif k
'S2CID' then table.insert(new_list,); elseif k
'SSRN' then table.insert(new_list,); elseif k
'ZBL' then table.insert(new_list,); else error(cfg.messages['unknown_manual_ID']); end end local function comp(a, b) -- used in following table.sort return a[1]:lower < b[1]:lower; end table.sort(new_list, comp); for k, v in ipairs(new_list) do new_list[k] = v[2]; end return new_list;end
----------------------------< E X T R A C T _ I D S >------------------------------------------------------------
Populates ID table from arguments using configuration settings. Loops through cfg.id_handlers and searches args forany of the parameters listed in each cfg.id_handlers['...'].parameters. If found, adds the parameter and value tothe identifier list. Emits redundant error message is more than one alias exists in args
local function extract_ids(args) local id_list = ; -- list of identifiers found in args for k, v in pairs(cfg.id_handlers) do -- k is uc identifier name as index to cfg.id_handlers; e.g. cfg.id_handlers['ISBN'], v is a table v = select_one(args, v.parameters, 'redundant_parameters'); -- v.parameters is a table of aliases for k; here we pick one from args if present if is_set(v) then id_list[k] = v; end -- if found in args, add identifier to our list end return id_list;end
----------------------------< E X T R A C T _ I D _ A C C E S S _ L E V E L S >--------------------------------------
Fetches custom id access levels from arguments using configuration settings.Parameters which have a predefined access level (e.g. arxiv) do not use thisfunction as they are directly rendered as free without using an additional parameter.
access-level values must match the case used in cfg.keywords_lists['id-access'] (lowercase unless there is some special reason for something else)
local function extract_id_access_levels(args, id_list) local id_accesses_list = ; for k, v in pairs(cfg.id_handlers) do local access_param = v.custom_access; -- name of identifier's access-level parameter if is_set(access_param) then local access_level = args[access_param]; -- get the assigned value if there is one if is_set (access_level) then if not in_array (access_level, cfg.keywords_lists['id-access']) then -- exact match required table.insert(z.message_tail,); access_level = nil; -- invalid so unset end if not is_set(id_list[k]) then -- identifer access-level must have a matching identifier table.insert(z.message_tail,); -- param name is uppercase in cfg.id_handlers (k); lowercase for error message end id_accesses_list[k] = cfg.keywords_xlate[access_level]; -- get translated keyword end end end return id_accesses_list;end
----------------------------< S E T _ S E L E C T E D _ M O D U L E S >--------------------------------------
Sets local cfg table and imported functions table to same (live or sandbox) as that used by the other modules.
local function set_selected_modules (cfg_table_ptr, utilities_page_ptr) cfg = cfg_table_ptr;
is_set = utilities_page_ptr.is_set; -- import functions from select Module:Citation/CS1/Utilities module in_array = utilities_page_ptr.in_array; set_error = utilities_page_ptr.set_error; select_one = utilities_page_ptr.select_one; add_maint_cat = utilities_page_ptr.add_maint_cat; substitute = utilities_page_ptr.substitute; make_wikilink = utilities_page_ptr.make_wikilink;
z = utilities_page_ptr.z; -- table of tables in Module:Citation/CS1/Utilitiesend
----------------------------< E X P O R T E D F U N C T I O N S >------------------------------------------
return