confirmed, superuser, Administrators
951
edits
Changes
Module:Citation/CS1/Identifiers (view source)
Revision as of 16:40, 12 March 2023
, 16:40, 12 March 20231 revision imported
--[[--------------------------< 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 >--------------------------------------
]]
]]
This function does not work if it is fed month names for languages other than English. Wikimedia #time: parser
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
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 lang.formatDate(). To get around that
contains a non-English month name because good1 is false after the call to lang_object.formatDate(). To get
call this function with YYYY-MM-DD format dates.
around that call this function with date parts and create a YYYY-MM-DD format date.
]=]
]=]
local function is_valid_biorxiv_date (biorxiv_date)
local function is_valid_biorxiv_date (y, m, d)
local biorxiv_date = table.concat ({y, m, d}, '-'); -- make ymd date
local good1, good2;
local good1, good2;
local biorxiv_ts, tomorrow_ts; -- to hold Unix timestamps representing the dates
local biorxiv_ts, tomorrow_ts; -- to hold Unix timestamps representing the dates
--[[--------------------------< N O R M A L I Z E _ L C C N >--------------------------------------------------
--[[--------------------------< N O R M A L I Z E _ L C C N >--------------------------------------------------
LCCN normalization (http://www.loc.gov/marc/lccn-namespace.html#normalization)
LCCN normalization (https://www.loc.gov/marc/lccn-namespace.html#normalization)
1. Remove all blanks.
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.
2. If there is a forward slash (/) in the string, remove it, and remove all characters to the right of the forward slash.
Returns a normalized LCCN for lccn() to validate. There is no error checking (step 3.b.1) performed in this function.
Returns a normalized LCCN for lccn() to validate. There is no error checking (step 3.b.1) performed in this function.
]]
]]
--[[--------------------------< A R X I V >--------------------------------------------------------------------
--[[--------------------------< A R X I V >--------------------------------------------------------------------
See: http://arxiv.org/help/arxiv_identifier
See: https://arxiv.org/help/arxiv_identifier
format and error check arXiv identifier. There are three valid forms of the identifier:
format and error check arXiv identifier. There are three valid forms of the identifier:
<date code> and <version> are as defined for 0704-1412
<date code> and <version> are as defined for 0704-1412
<number> is a five-digit number
<number> is a five-digit number
]]
]]
local handler = options.handler;
local handler = options.handler;
local year, month, version;
local year, month, version;
local err_cat = false; -- assume no error message
local err_msg = false; -- assume no error message
local text; -- output text
local text; -- output text
if ((not (90 < year or 8 > year)) or (1 > month or 12 < month)) or -- if invalid year or invalid month
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?
((91 == year and 7 > month) or (7 == year and 3 < month)) then -- if years ok, are starting and ending months ok?
err_msg = true; -- flag for error message
end
end
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)
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)?
((7 == year) and (4 > month)) then -- when year is 07, is month invalid (before April)?
err_msg = true; -- flag for error message
end
end
month = tonumber (month);
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)
if ((15 > year) or (1 > month or 12 < month)) then -- is year invalid or is month invalid? (doesn't test for future years)
err_msg = true; -- flag for error message
end
end
else
else
err_msg = true; -- not a recognized format; flag for error message
end
end
if err_cat then
if err_msg then
options.coins_list_t['ARXIV'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['ARXIV'] = nil; -- when error, unset so not included in COinS
end
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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handler.access});
if is_set (class) then
if is_set (class) then
text = table.concat ({text, ' [[//arxiv.org/archive/', class, ' ', class, ']]'}); -- external link within square brackets, not wikilink
text = table.concat ({text, ' [[//arxiv.org/archive/', class, ' ', class, ']]'}); -- external link within square brackets, not wikilink
else
else
set_message ('err_class_ignored');
end
end
end
end
return text;
return text;
end
end
Validates (sort of) and formats a bibcode ID.
Validates (sort of) and formats a bibcode ID.
Format for bibcodes is specified here: http://adsabs.harvard.edu/abs_doc/help_pages/data.html#bibcodes
Format for bibcodes is specified here: https://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
But, this: 2015arXiv151206696F is apparently valid so apparently, the only things that really matter are length, 19 characters
local handler = options.handler;
local handler = options.handler;
local err_type;
local err_type;
local err_msg = '';
local year;
local year;
if is_set (err_type) then -- if there was an error detected
if is_set (err_type) then -- if there was an error detected
set_message ('err_bad_bibcode', {err_type});
options.coins_list_t['BIBCODE'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['BIBCODE'] = nil; -- when error, unset so not included in COinS
end
end
return text;
return text;
end
end
local id = options.id;
local id = options.id;
local handler = options.handler;
local handler = options.handler;
local err_cat = true; -- flag; assume that there will be an error
local err_msg = true; -- flag; assume that there will be an error
local patterns = {
local patterns = {
if m then -- m is nil when id is the six-digit form
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)
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
break; -- date fail; break out early so we don't unset the error message
end
end
end
end
err_msg = nil; -- we found a match so unset the error message
break; -- and done
break; -- and done
end
end
end -- err_cat remains set here when no match
end -- err_cat remains set here when no match
if err_cat then
if err_msg then
options.coins_list_t['BIORXIV'] = nil; -- when error, unset so not included in COinS
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
end
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator,
prefix = handler.prefix, id = id, separator = handler.separator,
encode = handler.encode, access = handler.access}) .. (err_cat and (' ' .. set_message ('err_bad_biorxiv')) or '');
encode = handler.encode, access = handler.access});
end
end
The description of the structure of this identifier can be found at Help_talk:Citation_Style_1/Archive_26#CiteSeerX_id_structure
The description of the structure of this identifier can be found at Help_talk:Citation_Style_1/Archive_26#CiteSeerX_id_structure
]]
]]
local handler = options.handler;
local handler = options.handler;
local matched;
local matched;
local text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
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,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode,
matched = id:match ("^10%.1%.1%.[1-9]%d?%d?%d?%.[1-9]%d?%d?%d?$");
matched = id:match ("^10%.1%.1%.[1-9]%d?%d?%d?%.[1-9]%d?%d?%d?$");
if not matched then
if not matched then
set_message ('err_bad_citeseerx' );
options.coins_list_t['CITESEERX'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['CITESEERX'] = nil; -- when error, unset so not included in COinS
end
end
return text;
return text;
end
end
and terminal punctuation may not be technically correct but it appears, that in practice these characters are rarely
and terminal punctuation may not be technically correct but it appears, that in practice these characters are rarely
if ever used in DOI names.
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
]]
]]
local ignore_invalid = options.accept;
local ignore_invalid = options.accept;
local handler = options.handler;
local handler = options.handler;
local err_cat;
local err_flag;
local text;
local text;
end
end
local registrant = id:match ('^10%.([^/]+)/[^%s–]-[^%.,]$'); -- registrant set when DOI has the proper basic form
local registrant = mw.ustring.match (id, '^10%.([^/]+)/[^%s–]-[^%.,]$'); -- registrant set when DOI has the proper basic form
local registrant_err_patterns = { -- these patterns are for code ranges that are not supported
local registrant_err_patterns = { -- these patterns are for code ranges that are not supported
'^[^1-3]%d%d%d%d%.%d%d*$', -- 5 digits with subcode (0xxxx, 40000+); accepts: 10000–39999
'^[^1-3]%d%d%d%d%.%d%d*$', -- 5 digits with subcode (0xxxx, 40000+); accepts: 10000–39999
'^[^1-9]%d%d%d$', -- 4 digits without subcode (0xxx); accepts: 1000–9999
'^[^1-9]%d%d%d$', -- 4 digits without subcode (0xxx); accepts: 1000–9999
'^%d%d%d%d%d%d+', -- 6 or more digits
'^%d%d%d%d%d%d+', -- 6 or more digits
'^%d%d?%d?$', -- less than 4 digits without subcode (with subcode is legitimate)
'^%d%d?%d?$', -- less than 4 digits without subcode (3 digits with subcode is legitimate)
'^%d%d?%.[%d%.]+', -- 1 or 2 digits with subcode
'^5555$', -- test registrant will never resolve
'^5555$', -- test registrant will never resolve
'[^%d%.]', -- any character that isn't a digit or a dot
'[^%d%.]', -- any character that isn't a digit or a dot
for i, pattern in ipairs (registrant_err_patterns) do -- spin through error patterns
for i, pattern in ipairs (registrant_err_patterns) do -- spin through error patterns
if registrant:match (pattern) then -- to validate registrant codes
if registrant:match (pattern) then -- to validate registrant codes
err_flag = set_message ('err_bad_doi'); -- when found, mark this DOI as bad
break; -- and done
break; -- and done
end
end
end
end
else
else
err_flag = set_message ('err_bad_doi'); -- invalid directory or malformed
end
end
else
else
end
end
if err_cat then
if err_flag then
options.coins_list_t['DOI'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['DOI'] = nil; -- when error, unset so not included in COinS
end
end
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
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,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access,
auto_link = not (err_cat 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
auto_link = not (err_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 '');
}) .. (inactive or '');
return text .. (err_cat and err_cat or ''); -- parentheses required
return text;
end
end
if ever used in HDLs.
if ever used in HDLs.
Query string parameters are named here: http://www.handle.net/proxy_servlet.html. query strings are not displayed
Query string parameters are named here: https://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
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.
have to detect the query string so that it isn't URL-encoded with the rest of the identifier.
local access = options.access;
local access = options.access;
local handler = options.handler;
local handler = options.handler;
local query_params = { -- list of known query parameters from http://www.handle.net/proxy_servlet.html
local query_params = { -- list of known query parameters from https://www.handle.net/proxy_servlet.html
'noredirect',
'noredirect',
'ignore_aliases',
'ignore_aliases',
if nil == id:match("^[^%s–]-/[^%s–]-[^%.,]$") then -- HDL must contain a forward slash, must not contain spaces, endashes, and must not end with period or comma
if nil == id:match("^[^%s–]-/[^%s–]-[^%.,]$") then -- HDL must contain a forward slash, must not contain spaces, endashes, and must not end with period or comma
set_message ('err_bad_hdl' );
options.coins_list_t['HDL'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['HDL'] = nil; -- when error, unset so not included in COinS
end
end
return text;
return text;
end
end
if not check then -- and there is an error
if not check then -- and there is an error
options.coins_list_t['ISBN'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['ISBN'] = nil; -- when error, unset so not included in COinS
set_message ('err_bad_isbn', err_type); -- set an error message
return ISBN; -- return id text
end
end
end
end
return ISBN;
return ISBN; -- return id text
end
end
local domain = options.ASINTLD;
local domain = options.ASINTLD;
local err_cat = ''
local 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]$") then
if not id:match("^[%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u][%d%u]$") then
err_flag = set_message ('err_bad_asin'); -- ASIN is not a mix of 10 uppercase alpha and numeric characters
else
else
if id:match("^%d%d%d%d%d%d%d%d%d[%dX]$") then -- if 10-digit numeric (or 9 digits with terminal X)
if id:match("^%d%d%d%d%d%d%d%d%d[%dX]$") then -- if 10-digit numeric (or 9 digits with terminal X)
if is_valid_isxn (id, 10) then -- see if ASIN value is or validates as ISBN-10
if is_valid_isxn (id, 10) then -- see if ASIN value is or validates as ISBN-10
if not id:find ('^63[01]') then -- 630xxxxxxx and 631xxxxxxx are (apparently) not a valid isbn prefixes but are used by amazon as a numeric identifier
if not id:find ('^63[01]') then -- 630xxxxxxx and 631xxxxxxx are (apparently) not a valid isbn prefixes but are used by amazon as a numeric identifier
err_flag = set_message ('err_bad_asin'); -- ASIN has ISBN-10 form but begins with something other than 630/1 so probably an isbn
end
end
elseif not is_set (err_cat) then
elseif not is_set (err_flag) then
err_flag = set_message ('err_bad_asin'); -- ASIN is not ISBN-10
end
end
elseif not id:match("^%u[%d%u]+$") then
elseif not id:match("^%u[%d%u]+$") then
err_flag = set_message ('err_bad_asin'); -- asin doesn't begin with uppercase alpha
end
end
end
end
domain = "com." .. domain;
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)
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
end
local handler = options.handler;
local handler = options.handler;
if not is_set (err_cat) then
if not is_set (err_flag) then
options.coins_list_t['ASIN'] = handler.prefix .. domain .. "/dp/" .. id; -- experiment for asin coins
options.coins_list_t['ASIN'] = handler.prefix .. domain .. "/dp/" .. id; -- asin for coins
else
else
options.coins_list_t['ASIN'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['ASIN'] = nil; -- when error, unset so not included in COinS
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
return external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
prefix = handler.prefix .. domain .. "/dp/",
prefix = handler.prefix .. domain .. "/dp/",
id = id, encode = handler.encode, separator = handler.separator}) .. err_cat;
id = id, encode = handler.encode, separator = handler.separator})
end
end
Determines whether an ISMN string is valid. Similar to ISBN-13, ISMN is 13 digits beginning 979-0-... and uses the
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 http://www.ismn-international.org/download/Web_ISMN_Users_Manual_2008-6.pdf
same check digit calculations. See https://www.ismn-international.org/download/Web_ISMN_Users_Manual_2008-6.pdf
section 2, pages 9–12.
section 2, pages 9–12.
if false == valid_ismn then
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
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
set_message ('err_bad_ismn'); -- create an error message if the ISMN is invalid
end
end
like this:
like this:
|issn=0819 4327 gives: [http://www.worldcat.org/issn/0819 4327 0819 4327] -- can't have spaces in an external link
|issn=0819 4327 gives: [https://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
This code now prevents that by inserting a hyphen at the ISSN midpoint. It also validates the ISSN for length
if false == valid_issn then
if false == valid_issn then
options.coins_list_t['ISSN'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['ISSN'] = nil; -- when error, unset so not included in COinS
set_message ('err_bad_issn', (options.hkey == 'EISSN') and 'e' or ''); -- create an error message if the ISSN is invalid
end
end
end
end
return text
return text;
end
end
local handler = options.handler;
local handler = options.handler;
local id_num;
local id_num;
id_num = id:match ('^[Jj][Ff][Mm](.*)$'); -- identifier with jfm prefix; extract identifier
id_num = id:match ('^[Jj][Ff][Mm](.*)$'); -- identifier with jfm prefix; extract identifier
id = id_num; -- jfm matches pattern
id = id_num; -- jfm matches pattern
else
else
set_message ('err_bad_jfm' ); -- set an error message
options.coins_list_t['JFM'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['JFM'] = 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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode});
end
end
local access = options.access;
local access = options.access;
local handler = options.handler;
local handler = options.handler;
if id:find ('[Jj][Ss][Tt][Oo][Rr]') or id:find ('^https?://') or id:find ('%s') then
if id:find ('[Jj][Ss][Tt][Oo][Rr]') or id:find ('^https?://') or id:find ('%s') then
set_message ('err_bad_jstor'); -- set an error message
options.coins_list_t['JSTOR'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['JSTOR'] = 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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access});
end
end
Format LCCN link and do simple error checking. LCCN is a character string 8-12 characters long. The length of
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.
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
length = 8 then all digits
length = 8 then all digits
local lccn = options.id;
local lccn = options.id;
local handler = options.handler;
local handler = options.handler;
local err_cat = ''; -- presume that LCCN is valid
local err_flag; -- presume that LCCN is valid
local id = lccn; -- local copy of the LCCN
local id = lccn; -- local copy of the LCCN
if 8 == len then
if 8 == len then
if id:match("[^%d]") then -- if LCCN has anything but digits (nil if only digits)
if id:match("[^%d]") then -- if LCCN has anything but digits (nil if only digits)
err_flag = set_message ('err_bad_lccn'); -- set an error message
end
end
elseif 9 == len then -- LCCN should be adddddddd
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?
if nil == id:match("%l%d%d%d%d%d%d%d%d") then -- does it match our pattern?
err_flag = set_message ('err_bad_lccn'); -- set an error message
end
end
elseif 10 == len then -- LCCN should be aadddddddd or dddddddddd
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 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
if nil == id:match("^%l%l%d%d%d%d%d%d%d%d") then -- ... see if it matches our pattern
err_flag = set_message ('err_bad_lccn'); -- no match, set an error message
end
end
end
end
elseif 11 == len then -- LCCN should be aaadddddddd or adddddddddd
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
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_flag = set_message ('err_bad_lccn'); -- no match, set an error message
end
end
elseif 12 == len then -- LCCN should be aadddddddddd
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
if not id:match("^%l%l%d%d%d%d%d%d%d%d%d%d") then -- see if it matches our pattern
err_flag = set_message ('err_bad_lccn'); -- no match, set an error message
end
end
else
else
err_flag = set_message ('err_bad_lccn'); -- wrong length, set an error message
end
end
if not is_set (err_cat) and nil ~= lccn:find ('%s') then
if not is_set (err_flag) and nil ~= lccn:find ('%s') then
err_flag = set_message ('err_bad_lccn'); -- lccn contains a space, set an error message
end
end
if is_set (err_cat) then
if is_set (err_flag) then
options.coins_list_t['LCCN'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['LCCN'] = 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,
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;
prefix = handler.prefix, id = lccn, separator = handler.separator, encode = handler.encode});
end
end
local id_num;
local id_num;
local id_len;
local id_len;
id_num = id:match ('^[Mm][Rr](%d+)$'); -- identifier with mr prefix
id_num = id:match ('^[Mm][Rr](%d+)$'); -- identifier with mr prefix
id = string.rep ('0', 7-id_len) .. id_num; -- zero-fill leading digits
id = string.rep ('0', 7-id_len) .. id_num; -- zero-fill leading digits
else
else
set_message ('err_bad_mr'); -- set an error message
options.coins_list_t['MR'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['MR'] = 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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode});
end
end
local handler = options.handler;
local handler = options.handler;
local number;
local number;
if id:match('^ocm%d%d%d%d%d%d%d%d$') then -- ocm prefix and 8 digits; 001 field (12 characters)
if id:match('^ocm%d%d%d%d%d%d%d%d$') then -- ocm prefix and 8 digits; 001 field (12 characters)
number = id:match('ocm(%d+)'); -- get the number
number = id:match('ocm(%d+)'); -- get the number
id = number; -- exclude prefix, if any, from external link
id = number; -- exclude prefix, if any, from external link
else
else
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
options.coins_list_t['OCLC'] = 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_msg;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode});
end
end
local handler = options.handler;
local handler = options.handler;
local ident, code = id:gsub('^OL', ''):match("^(%d+([AMW]))$"); -- strip optional OL prefix followed immediately by digits followed by 'A', 'M', or 'W';
local ident, code = id:gsub('^OL', ''):match("^(%d+([AMW]))$"); -- strip optional OL prefix followed immediately by digits followed by 'A', 'M', or 'W';
local error_msg = '';
local err_flag;
local prefix = { -- these are appended to the handler.prefix according to code
local prefix = { -- these are appended to the handler.prefix according to code
['A']='authors/OL',
['A']='authors/OL',
code = 'X'; -- no code or id completely invalid
code = 'X'; -- no code or id completely invalid
ident = id; -- copy id to ident so that we display the flawed identifier
ident = id; -- copy id to ident so that we display the flawed identifier
err_flag = set_message ('err_bad_ol');
end
end
if not is_set (error_msg) then
if not is_set (err_flag) then
options.coins_list_t['OL'] = handler.prefix .. prefix[code] .. ident; -- experiment for ol coins
options.coins_list_t['OL'] = handler.prefix .. prefix[code] .. ident; -- experiment for ol coins
else
else
prefix = handler.prefix .. prefix[code],
prefix = handler.prefix .. prefix[code],
id = ident, separator = handler.separator, encode = handler.encode,
id = ident, separator = handler.separator, encode = handler.encode,
access = access}) .. error_msg;
access = access});
end
end
local access = options.access;
local access = options.access;
local handler = options.handler;
local handler = options.handler;
if id:match("[^%d]") then -- if OSTI has anything but digits
if id:match("[^%d]") then -- if OSTI has anything but digits
set_message ('err_bad_osti'); -- set an error message
options.coins_list_t['OSTI'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['OSTI'] = nil; -- when error, unset so not included in COinS
else -- OSTI is only digits
else -- OSTI is only digits
local id_num = tonumber (id); -- convert id to a number for range testing
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
if 1018 > id_num or handler.id_limit < id_num then -- if OSTI is outside test limit boundaries
set_message ('err_bad_osti'); -- set an error message
options.coins_list_t['OSTI'] = nil; -- when error, unset so not included in COinS
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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access});
end
end
local embargo = options.Embargo; -- TODO: lowercase?
local embargo = options.Embargo; -- TODO: lowercase?
local handler = options.handler;
local handler = options.handler;
local err_cat;
local err_flag;
local id_num;
local id_num;
local text;
local text;
id_num = tonumber (id_num); -- convert id_num to a number for range testing
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
if 1 > id_num or handler.id_limit < id_num then -- if PMC is outside test limit boundaries
err_flag = set_message ('err_bad_pmc'); -- set an error message
else
else
id = tostring (id_num); -- make sure id is a string
id = tostring (id_num); -- make sure id is a string
end
end
else -- when id format incorrect
else -- when id format incorrect
err_flag = set_message ('err_bad_pmc'); -- set an error message
end
end
handler.separator,
handler.separator,
id,
id,
});
});
else
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
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,
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handler.access,
auto_link = not err_cat and 'pmc' or nil -- do not auto-link when PMC has error
auto_link = not err_flag and 'pmc' or nil -- do not auto-link when PMC has error
}) .. (err_cat and err_cat or ''); -- parentheses required
});
end
end
if err_cat then
if err_flag then
options.coins_list_t['PMC'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['PMC'] = nil; -- when error, unset so not included in COinS
end
end
local id = options.id;
local id = options.id;
local handler = options.handler;
local handler = options.handler;
if id:match("[^%d]") then -- if PMID has anything but digits
if id:match("[^%d]") then -- if PMID has anything but digits
set_message ('err_bad_pmid'); -- set an error message
options.coins_list_t['PMID'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['PMID'] = nil; -- when error, unset so not included in COinS
else -- PMID is only digits
else -- PMID is only digits
local id_num = tonumber (id); -- convert id to a number for range testing
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
if 1 > id_num or handler.id_limit < id_num then -- if PMID is outside test limit boundaries
set_message ('err_bad_pmid'); -- set an error message
options.coins_list_t['PMID'] = nil; -- when error, unset so not included in COinS
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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode});
end
end
local id = options.id;
local id = options.id;
local handler = options.handler;
local handler = options.handler;
if id:match("[^%d]") then -- if RFC has anything but digits
if id:match("[^%d]") then -- if RFC has anything but digits
set_message ('err_bad_rfc'); -- set an error message
options.coins_list_t['RFC'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['RFC'] = nil; -- when error, unset so not included in COinS
else -- RFC is only digits
else -- RFC is only digits
local id_num = tonumber (id); -- convert id to a number for range testing
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
if 1 > id_num or handler.id_limit < id_num then -- if RFC is outside test limit boundaries
set_message ('err_bad_rfc'); -- set an error message
options.coins_list_t['RFC'] = nil; -- when error, unset so not included in COinS
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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = handler.access});
end
end
local access = options.access;
local access = options.access;
local handler = options.handler;
local handler = options.handler;
local id_num;
local id_num;
local text;
local text;
id_num = tonumber (id_num); -- convert id_num to a number for range testing
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
if handler.id_limit < id_num then -- if S2CID is outside test limit boundaries
set_message ('err_bad_s2cid'); -- set an error message
options.coins_list_t['S2CID'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['S2CID'] = nil; -- when error, unset so not included in COinS
end
end
else -- when id format incorrect
else -- when id format incorrect
set_message ('err_bad_s2cid'); -- set an error message
options.coins_list_t['S2CID'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['S2CID'] = nil; -- when error, unset so not included in COinS
end
end
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access});
return text;
return text;
local ignore_invalid = options.accept;
local ignore_invalid = options.accept;
local handler = options.handler;
local handler = options.handler;
local function return_result (check, err_type) -- local function to handle the various returns
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,
local SBN = internal_link_id ({link = handler.link, label = handler.label, redirect = handler.redirect,
prefix = handler.prefix, id = id, separator = handler.separator});
prefix = handler.prefix, id = id, separator = handler.separator});
if not ignore_invalid then -- if not ignoring SBN errors
if not ignore_invalid then -- if not ignoring SBN errors
if not check then
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
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
set_message ('err_bad_sbn', {err_type}); -- display an error message
return SBN;
end
end
else
else
local id = options.id;
local id = options.id;
local handler = options.handler;
local handler = options.handler;
local id_num;
local id_num;
local text;
local text;
id_num = tonumber (id_num); -- convert id_num to a number for range testing
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
if 100 > id_num or handler.id_limit < id_num then -- if SSRN is outside test limit boundaries
set_message ('err_bad_ssrn'); -- set an error message
options.coins_list_t['SSRN'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['SSRN'] = nil; -- when error, unset so not included in COinS
end
end
else -- when id format incorrect
else -- when id format incorrect
set_message ('err_bad_ssrn'); -- set an error message
options.coins_list_t['SSRN'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['SSRN'] = nil; -- when error, unset so not included in COinS
end
end
text = external_link_id ({link = handler.link, label = handler.label, q = handler.q, redirect = handler.redirect,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = options.access});
return text;
return text;
if not id:match('^.+@.+$') or not id:match('^[^<].*[^>]$') then -- doesn't have '@' or has one or first or last character is '< or '>'
if not id:match('^.+@.+$') or not id:match('^[^<].*[^>]$') then -- doesn't have '@' or has one or first or last character is '< or '>'
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
options.coins_list_t['USENETID'] = nil; -- when error, unset so not included in COinS
end
end
return text
return text;
end
end
local id = options.id;
local id = options.id;
local handler = options.handler;
local handler = options.handler;
if id:match('^%d%d%d%d%d%d%d%d$') then -- is this identifier using temporary format?
if id:match('^%d%d%d%d%d%d%d%d$') then -- is this identifier using temporary format?
set_message ('maint_zbl'); -- yes, add maint cat
set_message ('maint_zbl'); -- yes, add maint cat
elseif not id:match('^%d?%d?%d?%d%.%d%d%d%d%d$') then -- not temporary, is it normal format?
elseif not id:match('^%d?%d?%d?%d%.%d%d%d%d%d$') then -- not temporary, is it normal format?
set_message ('err_bad_zbl'); -- no, set an error message
options.coins_list_t['ZBL'] = nil; -- when error, unset so not included in COinS
options.coins_list_t['ZBL'] = 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,
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;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode});
end
end
if is_set (access_level) then
if is_set (access_level) then
if not in_array (access_level, cfg.keywords_lists['id-access']) then -- exact match required
if not in_array (access_level, cfg.keywords_lists['id-access']) then -- exact match required
set_message ('err_invalid_param_val', {access_param, access_level});
access_level = nil; -- invalid so unset
access_level = nil; -- invalid so unset
end
end
if not is_set (id_list[k]) then -- identifier access-level must have a matching identifier
if not is_set (id_list[k]) then -- identifier access-level must have a matching identifier
set_message ('err_param_access_requires_param', {k:lower()}); -- parameter name is uppercase in cfg.id_handlers (k); lowercase for error message
end
end
id_accesses_list[k] = cfg.keywords_xlate[access_level]; -- get translated keyword
id_accesses_list[k] = cfg.keywords_xlate[access_level]; -- get translated keyword
options_t.handler = cfg.id_handlers[hkey];
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
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
options_t.coins_list_t[hkey] = v; -- id value without accept-as-written markup for metadata
if options_t.handler.access and not in_array (options_t.handler.access, cfg.keywords_lists['id-access']) then
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
if func_map[hkey] then
table.insert (ID_list_t, {hkey, func_map[hkey] (options_t)}); -- call the function and add the results to the output sequence table
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
else
error (cfg.messages['unknown_ID_key'] .. ' ' .. hkey); -- here when func_map doesn't have a function for hkey
error (cfg.messages['unknown_ID_key'] .. hkey); -- here when func_map doesn't have a function for hkey
end
end
end
end
for _, v in ipairs (ID_support_t) do
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
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
end
]]
]]
local function identifier_lists_get (args, options_t, ID_support_t)
local function identifier_lists_get (args_t, options_t, ID_support_t)
local ID_list_coins_t = extract_ids (args); -- get a table of identifiers and their values for use locally and for use in COinS
local ID_list_coins_t = extract_ids (args_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
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
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
auto_link_urls = auto_link_urls, -- table of identifier URLs to be used when auto-linking |title=
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
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;
is_embargoed = is_embargoed;
set_selected_modules = set_selected_modules;
set_selected_modules = set_selected_modules;
}
}