Seeds of the Word

Module:Citation/CS1/Identifiers: Difference between revisions

Browse history interactively
← Older edit
Content deleted Content added
m 1 revision imported
m 1 revision imported
 
(5 intermediate revisions by 2 users not shown)
Line 1:
 
--[[--------------------------< F O R W A R D D E C L A R A T I O N S >--------------------------------------
]]
Line 87 ⟶ 86:
if options.encode == true or options.encode == nil then
url_string = mw.uri.encode (url_string, 'PATH');
end
 
Line 180 ⟶ 179:
This function does not work if it is fed month names for languages other than English. Wikimedia #time: parser
apparently doesn't understand non-English 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 langlang_object.formatDate(). To get around that
around that call this function with date parts and create a YYYY-MM-DD format datesdate.
 
]=]
 
local function is_valid_biorxiv_date (biorxiv_datey, m, d)
local biorxiv_date = table.concat ({y, m, d}, '-'); -- make ymd date
local good1, good2;
local biorxiv_ts, tomorrow_ts; -- to hold Unix timestamps representing the dates
Line 242:
isxn_str = { isxn_str:byte(1, 13) }; -- 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 == 0; -- sum modulo 10 is zero when ISBN-13/ISMN is correct
Line 250:
--[[--------------------------< N O R M A L I Z E _ L C C N >--------------------------------------------------
 
LCCN normalization (httphttps://www.loc.gov/marc/lccn-namespace.html#normalization)
1. Remove all blanks.
2. If there is a forward slash (/) in the string, remove it, and remove all characters to the right of the forward slash.
Line 260:
 
Returns a normalized LCCN for lccn() to validate. There is no error checking (step 3.b.1) performed in this function.
 
]]
 
Line 286 ⟶ 287:
--[[--------------------------< A R X I V >--------------------------------------------------------------------
 
See: httphttps://arxiv.org/help/arxiv_identifier
 
format and error check arXiv identifier. There are three valid forms of the identifier:
Line 311 ⟶ 312:
<date code> and <version> are as defined for 0704-1412
<number> is a five-digit number
 
]]
 
Line 318 ⟶ 320:
local handler = options.handler;
local year, month, version;
local err_caterr_msg = false; -- assume no error message
local text; -- output text
Line 327 ⟶ 329:
if ((not (90 < year or 8 > year)) or (1 > month or 12 < month)) or -- if invalid year or invalid month
((91 == year and 7 > month) or (7 == year and 3 < month)) then -- if years ok, are starting and ending months ok?
err_caterr_msg = true; -- flag for error message
end
 
Line 336 ⟶ 338:
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 == year) and (4 > month)) then -- when year is 07, is month invalid (before April)?
err_caterr_msg = true; -- flag for error message
end
 
Line 344 ⟶ 346:
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_caterr_msg = true; -- flag for error message
end
 
else
err_caterr_msg = true; -- not a recognized format; flag for error message
end
 
if err_msg then
err_cat = err_cat and table.concat ({' ', set_message ('err_bad_arxiv')}) or ''; -- set error message if flag is true
options.coins_list_t['ARXIV'] = nil; -- when error, unset so not included in COinS
end
local err_msg_t = {};
if err_msg then
set_message ('err_bad_arxiv');
end
 
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handler.access}) .. err_cat;
 
if is_set (class) then
Line 360 ⟶ 369:
text = table.concat ({text, ' [[//arxiv.org/archive/', class, ' ', class, ']]'}); -- external link within square brackets, not wikilink
else
text = table.concat ({text, ' ', set_message ('err_class_ignored')});
end
end
 
return text;
end
 
Line 372 ⟶ 381:
Validates (sort of) and formats a bibcode ID.
 
Format for bibcodes is specified here: httphttps://adsabs.harvard.edu/abs_doc/help_pages/data.html#bibcodes
 
But, this: 2015arXiv151206696F is apparently valid so apparently, the only things that really matter are length, 19 characters
Line 391 ⟶ 400:
local handler = options.handler;
local err_type;
local err_msg = '';
local year;
 
Line 404 ⟶ 414:
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
Line 416 ⟶ 426:
 
if is_set (err_type) then -- if there was an error detected
text = text .. ' ' .. set_message ('err_bad_bibcode', {err_type});
options.coins_list_t['BIBCODE'] = nil; -- when error, unset so not included in COinS
 
end
 
return text;
end
Line 440 ⟶ 453:
local id = options.id;
local handler = options.handler;
local err_caterr_msg = true; -- flag; assume that there will be an error
local patterns = {
Line 453 ⟶ 466:
 
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 leap-year and actual month lengths ({{#time:}} is a poor date validator)
break; -- date fail; break out early so we don't unset the error message
end
end
err_caterr_msg = nil; -- we found a match so unset the error message
break; -- and done
end
end -- err_cat remains set here when no match
 
if err_msg then
options.coins_list_t['BIORXIV'] = nil; -- when error, unset so not included in COinS
set_message ('err_bad_biorxiv'); -- and set the error message
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator,
encode = handler.encode, access = handler.access}) .. (err_cat and (' ' .. set_message ('err_bad_biorxiv')) or '');
end
 
Line 473 ⟶ 491:
 
The description of the structure of this identifier can be found at Help_talk:Citation_Style_1/Archive_26#CiteSeerX_id_structure
 
]]
 
Line 479 ⟶ 498:
local handler = options.handler;
local matched;
 
local text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode,
Line 486 ⟶ 505:
matched = id:match ("^10%.1%.1%.[1-9]%d?%d?%d?%.[1-9]%d?%d?%d?$");
if not matched then
text = text .. ' ' .. set_message ('err_bad_citeseerx' );
options.coins_list_t['CITESEERX'] = nil; -- when error, unset so not included in COinS
end
 
return text;
end
Line 506 ⟶ 527:
and terminal punctuation may not be technically correct but it appears, that in practice these characters are rarely
if ever used in DOI names.
 
https://www.doi.org/doi_handbook/2_Numbering.html -- 2.2 Syntax of a DOI name
https://www.doi.org/doi_handbook/2_Numbering.html#2.2.2 -- 2.2.2 DOI prefix
 
]]
Line 515 ⟶ 539:
local ignore_invalid = options.accept;
local handler = options.handler;
local err_caterr_flag;
 
local text;
Line 544 ⟶ 568:
end
 
local registrant = id:mw.ustring.match (id, '^10%.([^/]+)/[^%s–]-[^%.,]$'); -- registrant set when DOI has the proper basic form
 
local registrant_err_patterns = { -- these patterns are for code ranges that are not supported
'^[^1-3]%d%d%d%d%.%d%d*$', -- 5 digits with subcode (0xxxx, 40000+); accepts: 10000–39999
'^[^1-45]%d%d%d%d$', -- 5 digits without subcode (0xxxx, 4000060000+); accepts: 10000–4999910000–59999
'^[^1-9]%d%d%d%.%d%d*$', -- 4 digits with subcode (0xxx); accepts: 1000–9999
'^[^1-9]%d%d%d$', -- 4 digits without subcode (0xxx); accepts: 1000–9999
'^%d%d%d%d%d%d+', -- 6 or more digits
'^%d%d?%d?$', -- less than 4 digits without subcode (3 digits with subcode is legitimate)
'^%d%d?%.[%d%.]+', -- 1 or 2 digits with subcode
'^5555$', -- test registrant will never resolve
'[^%d%.]', -- any character that isn't a digit or a dot
Line 558 ⟶ 583:
 
if not ignore_invalid then
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
err_caterr_flag = ' ' .. set_message ('err_bad_doi'); -- when found, mark this DOI as bad
break; -- and done
end
end
else
err_caterr_flag = ' ' .. set_message ('err_bad_doi'); -- invalid directory or malformed
end
else
Line 572 ⟶ 597:
end
 
if err_flag then
options.coins_list_t['DOI'] = nil; -- when error, unset so not included in COinS
end
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access,
auto_link = not (err_caterr_flag or is_set (inactive) or ignore_invalid) and 'doi' or nil -- do not auto-link when |doi-broken-date= has a value or when there is a DOI error or (to play it safe, after all, auto-linking is not essential) when invalid DOIs are ignored
}) .. (inactive or '');
 
return text;
return text .. (err_cat and err_cat or ''); -- parentheses required
end
 
Line 596 ⟶ 625:
if ever used in HDLs.
 
Query string parameters are named here: httphttps://www.handle.net/proxy_servlet.html. query strings are not displayed
but since '?' is an allowed character in an HDL, '?' followed by one of the query parameters is the only way we
have to detect the query string so that it isn't URL-encoded with the rest of the identifier.
Line 606 ⟶ 635:
local access = options.access;
local handler = options.handler;
local query_params = { -- list of known query parameters from httphttps://www.handle.net/proxy_servlet.html
'noredirect',
'ignore_aliases',
Line 640 ⟶ 669:
 
if nil == id:match("^[^%s–]-/[^%s–]-[^%.,]$") then -- HDL must contain a forward slash, must not contain spaces, endashes, and must not end with period or comma
text = text .. ' ' .. set_message ('err_bad_hdl' );
options.coins_list_t['HDL'] = nil; -- when error, unset so not included in COinS
end
 
return text;
end
Line 664 ⟶ 695:
else -- here when not ignoring
if not check then -- and there is an error
options.coins_list_t['ISBN'] = nil; -- when error, unset so not included in COinS
return ISBN .. set_message ('err_bad_isbn', {err_type}, false, ' '); -- display an error message
set_message ('err_bad_isbn', err_type); -- set an error message
return ISBN; -- return id text
end
end
return ISBN; -- return id text
end
 
Line 686 ⟶ 719:
return return_result (false, cfg.err_msg_supl.form);
end
returnif return_resultnot (is_valid_isxn (id, 10), cfg.err_msg_supl.check);then -- test isbn-10 for numerical validity
return return_result (false, cfg.err_msg_supl.check); -- fail if isbn-10 is not numerically valid
end
if id:find ('^63[01]') then -- 630xxxxxxx and 631xxxxxxx are (apparently) not valid isbn group ids but are used by amazon as numeric identifiers (asin)
return return_result (false, cfg.err_msg_supl.group); -- fail if isbn-10 begins with 630/1
end
return return_result (true, cfg.err_msg_supl.check); -- pass if isbn-10 is numerically valid
else
if id:match ('^%d+$') == nil then
Line 709 ⟶ 748:
Error message if not 10 characters, if not ISBN-10, if mixed and first character is a digit.
 
|asin=630....... isand |asin=631....... are (apparently) not a legitimate ISBN though it checksums as one; these do not cause this
do not cause this function to emit the maint_asin message
 
This function is positioned here because it calls isbn()
Line 720 ⟶ 759:
local domain = options.ASINTLD;
local err_cat = ""err_flag;
 
if not id:match("^[%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u]$") then
err_caterr_flag = ' ' .. set_message ('err_bad_asin'); -- ASIN is not a mix of 10 uppercase alpha and numeric characters
else
if id:match("^%d%d%d%d%d%d%d%d%d[%dX]$") then -- if 10-digit numeric (or 9 digits with terminal X)
if isbnis_valid_isxn (optionsid, 10) then -- see if ASIN value is or validates as ISBN-10
if not id:find ('^63063[01]') then -- 630xxxxxxx isand 631xxxxxxx are (apparently) not a valid isbn prefixprefixes but isare used by amazon as a numeric identifier
err_flag = set_message ('maint_asinerr_bad_asin'); -- ASIN has ISBN-10 form but begins with something other than 630/1 so possiblyprobably an isbn
end
elseif not is_set (err_caterr_flag) then
err_caterr_flag = ' ' .. set_message ('err_bad_asin'); -- ASIN is not ISBN-10
end
elseif not id:match("^%u[%d%u]+$") then
err_caterr_flag = ' ' .. set_message ('err_bad_asin'); -- asin doesn't begin with uppercase alpha
end
end
if (not is_set (domain)) or in_array (domain, {'us'}) then -- default: United States
domain = "com";
elseif in_array (domain, {'jp', 'uk'}) then -- Japan, United Kingdom
domain = "co." .. domain;
elseif in_array (domain, {'au', 'br', 'mxz.cn'}) then -- Australia, Brazil, MexicoChina
domain = "cn";
elseif in_array (domain, {'au', 'br', 'mx', 'sg', 'tr'}) then -- Australia, Brazil, Mexico, Singapore, Turkey
domain = "com." .. domain;
elseif not in_array (domain, {'ae', 'ca', 'cn', 'de', 'es', 'fr', 'in', 'it', 'nl', 'pl', 'sa', 'se', 'co.jp', 'co.uk', 'com', 'com.au', 'com.br', 'com.mx', 'com.sg', 'com.tr'}) then -- Arabic Emirates, Canada, China, Germany, Spain, France, Indonesia, Italy, Netherlands, Poland, Saudi Arabia, Sweden (as of 2021-03 Austria (.at), Liechtenstein (.li) and Switzerland (.ch) still redirect to the German site (.de) with special settings, so don't maintain local ASINs for them)
err_flag = set_message ('err_bad_asin_tld'); -- unsupported asin-tld value
end
 
local handler = options.handler;
 
if not is_set (err_flag) then
options.coins_list_t['ASIN'] = handler.prefix .. domain .. "/dp/" .. id; -- asin for coins
else
options.coins_list_t['ASIN'] = nil; -- when error, unset so not included in COinS
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix .. domain .. "/dp/",
id = id, encode = handler.encode, separator = handler.separator}) .. err_cat;
end
 
Line 755 ⟶ 804:
 
Determines whether an ISMN string is valid. Similar to ISBN-13, ISMN is 13 digits beginning 979-0-... and uses the
same check digit calculations. See httphttps://www.ismn-international.org/download/Web_ISMN_Users_Manual_2008-6.pdf
section 2, pages 9–12.
 
ismn value not made part of COinS metadata because we don't have a url or isn't a COinS-defined identifier (rft.xxx)
or an identifier registered at info-uri.info (info:)
 
]]
Line 787 ⟶ 839:
 
if false == valid_ismn then
options.coins_list_t['ISMN'] = nil; -- when error, unset so not included in COinS; not really necessary here because ismn not made part of COinS
text = text .. ' ' .. set_message ('err_bad_ismn' ) -- add an error message if the ISMN is invalid
set_message ('err_bad_ismn'); -- create an error message if the ISMN is invalid
end
Line 800 ⟶ 853:
like this:
 
|issn=0819 4327 gives: [httphttps://www.worldcat.org/issn/0819 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 length
Line 839 ⟶ 892:
else
if false == valid_issn then
options.coins_list_t['ISSN'] = nil; -- when error, unset so not included in COinS
text = text .. ' ' .. set_message ('err_bad_issn', (options.hkey == 'EISSN') and 'e' or ''); -- add an error message if the ISSN is invalid
set_message ('err_bad_issn', (options.hkey == 'EISSN') and 'e' or ''); -- create an error message if the ISSN is invalid
end
end
return text;
end
 
Line 857 ⟶ 911:
local handler = options.handler;
local id_num;
 
local err_cat = '';
id_num = id:match ('^[Jj][Ff][Mm](.*)$'); -- identifier with jfm prefix; extract identifier
 
Line 870 ⟶ 923:
id = id_num; -- jfm matches pattern
else
err_cat = ' ' .. set_message ('err_bad_jfm' ); -- set an error message
options.coins_list_t['JFM'] = nil; -- when error, unset so not included in COinS
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode}) .. err_cat;
end
 
Line 888 ⟶ 942:
local access = options.access;
local handler = options.handler;
local err_msg = '';
 
if id:find ('[Jj][Ss][Tt][Oo][Rr]') or id:find ('^https?://') or id:find ('%s') then
err_msg = ' ' .. set_message ('err_bad_jstor'); -- set an error message
options.coins_list_t['JSTOR'] = nil; -- when error, unset so not included in COinS
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access}) .. err_msg;
end
 
Line 903 ⟶ 957:
Format LCCN link and do simple error checking. LCCN is a character string 8-12 characters long. The length of
the LCCN dictates the character type of the first 1-3 characters; the rightmost eight are always digits.
https://oclc-research.github.io/infoURI-Frozen/info-uri.info/info:lccn/reg.html
http://info-uri.info/registry/OAIHandler?verb=GetRecord&metadataPrefix=reg&identifier=info:lccn/
 
length = 8 then all digits
Line 916 ⟶ 970:
local lccn = options.id;
local handler = options.handler;
local err_cat = ''err_flag; -- presume that LCCN is valid
local id = lccn; -- local copy of the LCCN
 
Line 924 ⟶ 978:
if 8 == len then
if id:match("[^%d]") then -- if LCCN has anything but digits (nil if only digits)
err_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- set an error message
end
elseif 9 == len then -- LCCN should be adddddddd
if nil == id:match("%l%d%d%d%d%d%d%d%d") then -- does it match our pattern?
err_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- set an error message
end
elseif 10 == 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 == id:match("^%l%l%d%d%d%d%d%d%d%d") then -- ... see if it matches our pattern
err_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- no match, set an error message
end
end
elseif 11 == 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_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- no match, set an error message
end
elseif 12 == len then -- LCCN should be aadddddddddd
if not id:match("^%l%l%d%d%d%d%d%d%d%d%d%d") then -- see if it matches our pattern
err_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- no match, set an error message
end
else
err_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- wrong length, set an error message
end
 
if not is_set (err_caterr_flag) and nil ~= lccn:find ('%s') then
err_caterr_flag = ' ' .. set_message ('err_bad_lccn'); -- lccn contains a space, set an error message
end
 
if is_set (err_flag) then
options.coins_list_t['LCCN'] = nil; -- when error, unset so not included in COinS
end
 
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = lccn, separator = handler.separator, encode = handler.encode}) .. err_cat;
end
 
Line 968 ⟶ 1,026:
local id_num;
local id_len;
 
local err_cat = '';
id_num = id:match ('^[Mm][Rr](%d+)$'); -- identifier with mr prefix
 
Line 982 ⟶ 1,039:
id = string.rep ('0', 7-id_len) .. id_num; -- zero-fill leading digits
else
err_cat = ' ' .. set_message ('err_bad_mr'); -- set an error message
options.coins_list_t['MR'] = nil; -- when error, unset so not included in COinS
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode}) .. err_cat;
end
 
Line 1,001 ⟶ 1,059:
local handler = options.handler;
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
Line 1,024 ⟶ 1,081:
id = number; -- exclude prefix, if any, from external link
else
err_msg = ' ' .. set_message ('err_bad_oclc') -- add an error message if the id is malformed
options.coins_list_t['OCLC'] = nil; -- when error, unset so not included in COinS
end
local text =return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode}) .. err_msg;
 
return text;
end
 
Line 1,045 ⟶ 1,101:
local handler = options.handler;
local ident, code = id:gsub('^OL', ''):match("^(%d+([AMW]))$"); -- strip optional OL prefix followed immediately by digits followed by 'A', 'M', or 'W';
local error_msg = ''err_flag;
local prefix = { -- these are appended to the handler.prefix according to code
['A']='authors/OL',
Line 1,056 ⟶ 1,112:
code = 'X'; -- no code or id completely invalid
ident = id; -- copy id to ident so that we display the flawed identifier
error_msgerr_flag = ' ' .. set_message ('err_bad_ol');
end
 
if not is_set (err_flag) then
options.coins_list_t['OL'] = handler.prefix .. prefix[code] .. ident; -- experiment for ol coins
else
options.coins_list_t['OL'] = nil; -- when error, unset so not included in COinS
end
 
Line 1,062 ⟶ 1,124:
prefix = handler.prefix .. prefix[code],
id = ident, separator = handler.separator, encode = handler.encode,
access = access}) .. error_msg;
end
 
Line 1,080 ⟶ 1,142:
local access = options.access;
local handler = options.handler;
 
local err_cat = ''; -- presume that OSTI is valid
if id:match("[^%d]") then -- if OSTI has anything but digits
err_cat = ' ' .. set_message ('err_bad_osti'); -- set an error message
options.coins_list_t['OSTI'] = nil; -- when error, unset so not included in COinS
else -- OSTI is only digits
local id_num = tonumber (id); -- convert id to a number for range testing
if 1018 > id_num or handler.id_limit < id_num then -- if OSTI is outside test limit boundaries
err_cat = ' ' .. set_message ('err_bad_osti'); -- set an error message
options.coins_list_t['OSTI'] = nil; -- when error, unset so not included in COinS
end
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access}) .. err_cat;
end
 
Line 1,117 ⟶ 1,180:
local embargo = options.Embargo; -- TODO: lowercase?
local handler = options.handler;
local err_caterr_flag;
local id_num;
local text;
Line 1,132 ⟶ 1,195:
id_num = tonumber (id_num); -- convert id_num to a number for range testing
if 1 > id_num or handler.id_limit < id_num then -- if PMC is outside test limit boundaries
err_caterr_flag = ' ' .. set_message ('err_bad_pmc'); -- set an error message
else
id = tostring (id_num); -- make sure id is a string
end
else -- when id format incorrect
err_caterr_flag = ' ' .. set_message ('err_bad_pmc'); -- set an error message
end
Line 1,146 ⟶ 1,209:
handler.separator,
id,
(err_cat and err_cat or '') -- parens required
});
else
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect, -- no embargo date or embargo has expired, ok to link to article
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handler.access,
auto_link = not err_caterr_flag and 'pmc' or nil -- do not auto-link when PMC has error
});
}) .. (err_cat and err_cat or ''); -- parentheses required
end
 
if err_flag then
options.coins_list_t['PMC'] = nil; -- when error, unset so not included in COinS
end
 
return text;
end
Line 1,169 ⟶ 1,236:
local id = options.id;
local handler = options.handler;
 
local err_cat = ''; -- presume that PMID is valid
if id:match("[^%d]") then -- if PMID has anything but digits
err_cat = ' ' .. set_message ('err_bad_pmid'); -- set an error message
options.coins_list_t['PMID'] = nil; -- when error, unset so not included in COinS
else -- PMID is only digits
local id_num = tonumber (id); -- convert id to a number for range testing
if 1 > id_num or handler.id_limit < id_num then -- if PMID is outside test limit boundaries
err_cat = ' ' .. set_message ('err_bad_pmid'); -- set an error message
options.coins_list_t['PMID'] = nil; -- when error, unset so not included in COinS
end
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode}) .. err_cat;
end
 
Line 1,198 ⟶ 1,266:
local id = options.id;
local handler = options.handler;
 
local err_cat = ''; -- presume that RFC is valid
if id:match("[^%d]") then -- if RFC has anything but digits
err_cat = ' ' .. set_message ('err_bad_rfc'); -- set an error message
options.coins_list_t['RFC'] = nil; -- when error, unset so not included in COinS
else -- RFC is only digits
local id_num = tonumber (id); -- convert id to a number for range testing
if 1 > id_num or handler.id_limit < id_num then -- if RFC is outside test limit boundaries
err_cat = ' ' .. set_message ('err_bad_rfc'); -- set an error message
options.coins_list_t['RFC'] = nil; -- when error, unset so not included in COinS
end
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handler.access}) .. err_cat;
end
 
Line 1,228 ⟶ 1,297:
local access = options.access;
local handler = options.handler;
local err_cat = ''; -- presume that S2CID is valid
local id_num;
local text;
Line 1,237 ⟶ 1,305:
id_num = tonumber (id_num); -- convert id_num to a number for range testing
if handler.id_limit < id_num then -- if S2CID is outside test limit boundaries
err_cat = ' ' .. set_message ('err_bad_s2cid'); -- set an error message
options.coins_list_t['S2CID'] = nil; -- when error, unset so not included in COinS
end
 
else -- when id format incorrect
err_cat = ' ' .. set_message ('err_bad_s2cid'); -- set an error message
options.coins_list_t['S2CID'] = nil; -- when error, unset so not included in COinS
end
 
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access}) .. err_cat;
 
return text;
Line 1,254 ⟶ 1,323:
 
9-digit form of ISBN-10; uses same check-digit validation when SBN is prefixed with an additional '0' to make 10 digits
 
sbn value not made part of COinS metadata because we don't have a url or isn't a COinS-defined identifier (rft.xxx)
or an identifier registered at info-uri.info (info:)
 
]]
Line 1,261 ⟶ 1,333:
local ignore_invalid = options.accept;
local handler = options.handler;
local function return_result (check, err_type) -- local function to handle the various returns
local SBN = internal_link_id ({link = handler.link, label = handler.label, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator});
if not ignore_invalid then -- if not ignoring SBN errors
if not check then
options.coins_list_t['SBN'] = nil; -- when error, unset so not included in COinS; not really necessary here because sbn not made part of COinS
return SBN .. set_message ('err_bad_sbn', {err_type}, false, ' '); -- display an error message
set_message ('err_bad_sbn', {err_type}); -- display an error message
return SBN;
end
else
Line 1,278 ⟶ 1,352:
end
 
local ident = id:gsub ('[%s-]', ''); -- remove hyphens and whitespace; they interfere with the rest of the tests
 
if 9 ~= ident:len() then
Line 1,305 ⟶ 1,379:
local id = options.id;
local handler = options.handler;
local err_cat = ''; -- presume that SSRN is valid
local id_num;
local text;
Line 1,314 ⟶ 1,387:
id_num = tonumber (id_num); -- convert id_num to a number for range testing
if 100 > id_num or handler.id_limit < id_num then -- if SSRN is outside test limit boundaries
err_cat = ' ' .. set_message ('err_bad_ssrn'); -- set an error message
options.coins_list_t['SSRN'] = nil; -- when error, unset so not included in COinS
end
else -- when id format incorrect
err_cat = ' ' .. set_message ('err_bad_ssrn'); -- set an error message
options.coins_list_t['SSRN'] = nil; -- when error, unset so not included in COinS
end
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handleroptions.access}) .. err_cat;
 
return text;
Line 1,342 ⟶ 1,417:
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_message ('err_bad_usenet_id') -- add an error message if the message id is invalid
options.coins_list_t['USENETID'] = nil; -- when error, unset so not included in COinS
end
return text;
end
 
Line 1,362 ⟶ 1,438:
local id = options.id;
local handler = options.handler;
 
local err_cat = '';
if id:match('^%d%d%d%d%d%d%d%d$') then -- is this identifier using temporary format?
set_message ('maint_zbl'); -- yes, add maint cat
elseif not id:match('^%d?%d?%d?%d%.%d%d%d%d%d$') then -- not temporary, is it normal format?
err_cat = ' ' .. set_message ('err_bad_zbl'); -- no, set an error message
options.coins_list_t['ZBL'] = nil; -- when error, unset so not included in COinS
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode}) .. err_cat;
end
 
Line 1,414 ⟶ 1,490:
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, { set_message ('invalid_param_valerr_invalid_param_val', {access_param, access_level}, true) } );
access_level = nil; -- invalid so unset
end
if not is_set (id_list[k]) then -- identifier access-level must have a matching identifier
table.insert (z.message_tail, { set_message ('err_param_access_requires_param', {k:lower()}, true) } ); -- parameter 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
Line 1,432 ⟶ 1,508:
render the identifiers into a sorted sequence table
 
<id_list_tID_list_coins_t> is a table of k/v pairs where k is same as key in cfg.id_handlers and v is the assigned value
<options_t> is a table of is a table of various k/v option pairs provided in the call to new_build_id_list();
modified by this function and passed to all identifier rendering functions
<access_levels_t> is a table of k/v pairs where k is same as key in cfg.id_handlers and v is the assigned value (if valid)
 
returns a sequence table of sorted (by hkey - 'handler' key) rendered identifier strings
 
]]
 
local function build_id_list (id_list_tID_list_coins_t, options_t, access_levels_t)
local new_list_tID_list_t = {};
local accept;
local func_map = { --function map points to functions associated with hkey identifier
Line 1,473 ⟶ 1,549:
}
 
for hkey, v in pairs (id_list_tID_list_coins_t) do
v, accept = has_accept_as_written (v); -- remove accept-as-written markup if present; accept is boolean true when markup removed; false else
-- every function gets the options table with value v and accept boolean
Line 1,481 ⟶ 1,557:
options_t.access = access_levels_t[hkey]; -- add the access level for those that have an |<identifier-access= parameter
options_t.handler = cfg.id_handlers[hkey];
options_t.coins_list_t = ID_list_coins_t; -- pointer to ID_list_coins_t; for |asin= and |ol=; also to keep erroneous values out of the citation's metadata
if func_map[hkey] then
options_t.coins_list_t[hkey] = v; -- id value without accept-as-written markup for metadata
table.insert (new_list_t, {hkey, func_map[hkey] (options_t)}); -- call the function and add the results to the output sequence table
 
if options_t.handler.access and not in_array (options_t.handler.access, cfg.keywords_lists['id-access']) then
-- TODO: also retrieve identifier validity status, OL A/M/W/X type and ASIN TLD info from the corresponding function call for improved metadata generation in COinS() in ~/COinS
error (cfg.messages['unknown_ID_access'] .. options_t.handler.access); -- here when handler access key set to a value not listed in list of allowed id access keywords
end
 
if func_map[hkey] then
local id_text = func_map[hkey] (options_t); -- call the function to get identifier text and any error message
table.insert (ID_list_t, {hkey, id_text}); -- add identifier text to the output sequence table
else
error (cfg.messages['unknown_ID_key'] .. ' ' .. hkey); -- here when func_map doesn't have a function for hkey
end
end
Line 1,495 ⟶ 1,576:
end
 
table.sort (new_list_tID_list_t, comp); -- sequence table of tables sort
for k, v in ipairs (new_list_tID_list_t) do -- convert sequence table of tables to simple sequence table of strings
ID_list_t[k] = v[2]; -- v[2] is the identifier rendering from the call to the various functions in func_map{}
new_list_t[k] = v[2];
end
return new_list_tID_list_t;
end
 
 
--[[--------------------------< O P T I O N S _ C H E C K >----------------------------------------------------
 
check that certain option parameters have their associated identifier parameters with values
 
<ID_list_coins_t> is a table of k/v pairs where k is same as key in cfg.id_handlers and v is the assigned value
<ID_support_t> is a sequence table of tables created in citation0() where each subtable has four elements:
[1] is the support parameter's assigned value; empty string if not set
[2] is a text string same as key in cfg.id_handlers
[3] is cfg.error_conditions key used to create error message
[4] is original ID support parameter name used to create error message
returns nothing; on error emits an appropriate error message
 
]]
 
local function options_check (ID_list_coins_t, ID_support_t)
for _, v in ipairs (ID_support_t) do
if is_set (v[1]) and not ID_list_coins_t[v[2]] then -- when support parameter has a value but matching identifier parameter is missing or empty
set_message (v[3], (v[4])); -- emit the appropriate error message
end
end
end
 
Line 1,512 ⟶ 1,617:
]]
 
local function identifier_lists_get (argsargs_t, options_t, ID_support_t)
local ID_list_coins_t = extract_ids (argsargs_t); -- get a table of identifiers and their values for use locally and for use in COinS
options_check (ID_list_coins_t, ID_support_t); -- ID support parameters must have matching identifier parameters
local ID_access_levels_t = extract_id_access_levels (args, ID_list_coins_t); -- get a table of identifier access levels
local ID_access_levels_t = extract_id_access_levels (args_t, ID_list_coins_t); -- get a table of identifier access levels
local ID_list_t = build_id_list (ID_list_coins_t, options_t, ID_access_levels_t); -- get a sequence table of rendered identifier strings
 
-- TODO: add code to retrieve identifier validity status, OL A/M/W/X type and ASIN TLD info from ID_list_t and add this to ID_list_coins_t for improved metadata generation in COinS() in ~/COinS
 
return ID_list_t, ID_list_coins_t; -- return the tables
Line 1,550 ⟶ 1,654:
auto_link_urls = auto_link_urls, -- table of identifier URLs to be used when auto-linking |title=
identifier_lists_get = identifier_lists_get, -- experiment to replace individual calls to build_id_list(), extract_ids, extract_id_access_levels
is_embargoed = is_embargoed;
set_selected_modules = set_selected_modules;
Retrieved from "https://en.seminaverbi.bibleget.io/wiki/Module:Citation/CS1/Identifiers"