Nonbinary Wiki

Module:Citation/CS1/Identifiers: Difference between revisions

Module Discussion

Module:Citation/CS1/Identifiers (edit)

Revision as of 14:11, 9 January 2021

4,624 bytes added ,  3 years ago
sync from sandbox;
No edit summary
(sync from sandbox;)
Line 77: Line 77:


Formats a wiki-style external link
Formats a wiki-style external link
NB. Wikidata P1630 has a formatter URL with $1 placeholder for the ID which could be worked into our prefix/id/suffix
string, either overriding local definitions (auto-update) or as fallback for identifiers without local definitions.
But is expensive and could be risky if WD gets vandalized.
See Template_talk:Authority_control/Archive_8#Use_Wikidata_as_the_source_for_the_external_link


]]
]]
Line 93: Line 87:
if options.encode == true or options.encode == nil then
if options.encode == true or options.encode == nil then
url_string = mw.uri.encode( url_string );
url_string = mw.uri.encode (url_string);
end
end


Line 100: Line 94:
end
end


ext_link = mw.ustring.format ('[%s%s%s %s]', options.prefix, url_string, options.suffix or "", mw.text.nowiki(options.id));
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
if is_set (options.access) then
ext_link = substitute (cfg.presentation['ext-link-access-signal'], {cfg.presentation[options.access].class, cfg.presentation[options.access].title, ext_link}); -- add the free-to-read / paywall lock
ext_link = substitute (cfg.presentation['ext-link-access-signal'], {cfg.presentation[options.access].class, cfg.presentation[options.access].title, ext_link}); -- add the free-to-read / paywall lock
end
end
Line 116: Line 110:


Formats a wiki-style internal link
Formats a wiki-style internal link
TODO: Does not currently need to support options.access, options.encode, auto-linking and COinS (as in external_link_id),
but may be needed in the future for :m:Interwiki_map custom-prefixes like :arxiv:, :bibcode:, :DOI:, :hdl:, :ISSN:,
:JSTOR:, :Openlibrary:, :PMID:, :RFC:.


]]
]]


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


Line 141: Line 139:
--[[--------------------------< I S _ E M B A R G O E D >------------------------------------------------------
--[[--------------------------< I S _ E M B A R G O E D >------------------------------------------------------


Determines if a PMC identifier's online version is embargoed. Compares the date in |pmc-embargo-date= against today's date.  If embargo date is
Determines if a PMC identifier's online version is embargoed. Compares the date in |pmc-embargo-date= against
in the future, returns the content of |pmc-embargo-date=; otherwise, returns an empty string because the embargo has expired or because
today's date.  If embargo date is in the future, returns the content of |pmc-embargo-date=; otherwise, returns
|pmc-embargo-date= was not set in this cite.
an empty string because the embargo has expired or because |pmc-embargo-date= was not set in this cite.


]]
]]
Line 192: Line 190:
local lang_object = mw.getContentLanguage();
local lang_object = mw.getContentLanguage();


good1, biorxiv_ts = pcall (lang_object.formatDate, lang_object, 'U', biorxiv_date ); -- convert biorxiv_date value to Unix timestamp  
good1, biorxiv_ts = pcall (lang_object.formatDate, lang_object, 'U', biorxiv_date); -- convert biorxiv_date value to Unix timestamp  
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
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
Line 220: Line 218:
isxn_str = { isxn_str:byte(1, len) }; -- make a table of byte values '0' → 0x30 .. '9' → 0x39, 'X' → 0x58
isxn_str = { isxn_str:byte(1, len) }; -- make a table of byte values '0' → 0x30 .. '9' → 0x39, 'X' → 0x58
len = len + 1; -- adjust to be a loop counter
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
for i, v in ipairs (isxn_str) do -- loop through all of the bytes and calculate the checksum
if v == string.byte( "X" ) then -- if checkdigit is X (compares the byte value of 'X' which is 0x58)
if v == string.byte ("X" ) then -- if checkdigit is X (compares the byte value of 'X' which is 0x58)
temp = temp + 10*( len - i ); -- it represents 10 decimal
temp = temp + 10 * (len - i); -- it represents 10 decimal
else
else
temp = temp + tonumber( string.char(v) )*(len-i);
temp = temp + tonumber (string.char (v) )*(len-i);
end
end
end
end
Line 243: Line 241:
isxn_str = { isxn_str:byte(1, 13) }; -- make a table of byte values '0' → 0x30 .. '9' → 0x39
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
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
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
end
return temp % 10 == 0; -- sum modulo 10 is zero when ISBN-13/ISMN is correct
return temp % 10 == 0; -- sum modulo 10 is zero when ISBN-13/ISMN is correct
Line 315: Line 313:
]]
]]


local function arxiv (id, class)
local function arxiv (options)
local handler = cfg.id_handlers['ARXIV'];
local id = options.id;
local class = options.Class; -- TODO: lowercase?
local handler = options.handler;
local year, month, version;
local year, month, version;
local err_cat = false; -- assume no error message
local err_cat = false; -- assume no error message
Line 323: Line 323:
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 9107-0703 format with or without version
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 9107-0703 format with or without version
year, month = id:match("^%a[%a%.%-]+/([90]%d)([01]%d)%d%d%d[v%d]*$");
year, month = id:match("^%a[%a%.%-]+/([90]%d)([01]%d)%d%d%d[v%d]*$");
year = tonumber(year);
year = tonumber (year);
month = tonumber(month);
month = tonumber (month);
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?
Line 332: Line 332:
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 with or without version
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 with or without version
year, month = id:match("^(%d%d)([01]%d)%.%d%d%d%d[v%d]*$");
year, month = id:match("^(%d%d)([01]%d)%.%d%d%d%d[v%d]*$");
year = tonumber(year);
year = tonumber (year);
month = tonumber(month);
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)
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)?
Line 341: Line 341:
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 with or without version
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 with or without version
year, month = id:match("^(%d%d)([01]%d)%.%d%d%d%d%d[v%d]*$");
year, month = id:match("^(%d%d)([01]%d)%.%d%d%d%d%d[v%d]*$");
year = tonumber(year);
year = tonumber (year);
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_cat = true; -- flag for error message
err_cat = true; -- flag for error message
Line 386: Line 386:
]]
]]


local function bibcode (id, access)
local function bibcode (options)
local handler = cfg.id_handlers['BIBCODE'];
local id = options.id;
local access = options.access;
local handler = options.handler;
local err_type;
local err_type;
local year;
local year;
Line 402: Line 404:
err_type = cfg.err_msg_supl.value; -- so value error
err_type = cfg.err_msg_supl.value; -- so value error
else
else
local next_year = tonumber(os.date ('%Y')) + 1; -- get the current year as a number and add one for next year
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
year = tonumber (year); -- convert year portion of bibcode to a number
if (1000 > year) or (year > next_year) then
if (1000 > year) or (year > next_year) then
Line 414: Line 416:


if is_set (err_type) then -- if there was an error detected
if is_set (err_type) then -- if there was an error detected
text = text .. ' ' .. set_message( 'err_bad_bibcode', {err_type});
text = text .. ' ' .. set_message ('err_bad_bibcode', {err_type});
end
end
return text;
return text;
Line 435: Line 437:
]]
]]


local function biorxiv(id)
local function biorxiv (options)
local handler = cfg.id_handlers['BIORXIV'];
local id = options.id;
local handler = options.handler;
local err_cat = true; -- flag; assume that there will be an error
local err_cat = true; -- flag; assume that there will be an error
Line 461: Line 464:
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}) .. (err_cat and (' ' .. set_message ('err_bad_biorxiv')) or '');
end
end


Line 472: Line 475:
]]
]]


local function citeseerx (id)
local function citeseerx (options)
local handler = cfg.id_handlers['CITESEERX'];
local id = options.id;
local handler = options.handler;
local matched;
local matched;
Line 482: Line 486:
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
text = text .. ' ' .. set_message( 'err_bad_citeseerx' );
text = text .. ' ' .. set_message ('err_bad_citeseerx' );
end
end
return text;
return text;
Line 505: Line 509:
]]
]]


local function doi (id, inactive, access, ignore_invalid)
local function doi (options)
local id = options.id;
local inactive = options.DoiBroken
local access = options.access;
local ignore_invalid = options.accept;
local handler = options.handler;
local err_cat;
local err_cat;
local handler = cfg.id_handlers['DOI'];
 
local text;
local text;
if is_set (inactive) then
if is_set (inactive) then
Line 526: Line 534:
end
end
if is_set(inactive_year) and is_set (inactive_month) then
if is_set (inactive_year) and is_set (inactive_month) then
set_message ('maint_doi_inactive_dated', {inactive_year, inactive_month, ' '});
set_message ('maint_doi_inactive_dated', {inactive_year, inactive_month, ' '});
elseif is_set(inactive_year) then
elseif is_set (inactive_year) then
set_message ('maint_doi_inactive_dated', {inactive_year, '', ''});
set_message ('maint_doi_inactive_dated', {inactive_year, '', ''});
else
else
Line 546: Line 554:
'^%d%d?%d?$', -- less than 4 digits without subcode (with subcode is legitimate)
'^%d%d?%d?$', -- less than 4 digits without subcode (with subcode is legitimate)
'^5555$', -- test registrant will never resolve
'^5555$', -- test registrant will never resolve
'%s', -- any space character in registrant
'[^%d%.]', -- any character that isn't a digit or a dot
}
}


Line 594: Line 602:
]]
]]


local function hdl(id, access)
local function hdl (options)
local handler = cfg.id_handlers['HDL'];
local id = options.id;
local access = options.access;
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 http://www.handle.net/proxy_servlet.html
'noredirect',
'noredirect',
Line 630: Line 640:


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
text = text .. ' ' .. set_message( 'err_bad_hdl' );
text = text .. ' ' .. set_message ('err_bad_hdl' );
end
end
return text;
return text;
Line 642: Line 652:
]]
]]


local function isbn (isbn_str, ignore_invalid)
local function isbn (options)
local handler = cfg.id_handlers['ISBN'];
local isbn_str = options.id;
local ignore_invalid = options.accept;
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 ISBN = internal_link_id ({link = handler.link, label = handler.label, redirect = handler.redirect,
local ISBN = internal_link_id ({link = handler.link, label = handler.label, redirect = handler.redirect,
Line 673: Line 686:
return return_result (false, cfg.err_msg_supl.form);
return return_result (false, cfg.err_msg_supl.form);
end
end
return return_result (is_valid_isxn(id, 10), cfg.err_msg_supl.check);
return return_result (is_valid_isxn (id, 10), cfg.err_msg_supl.check);
else
else
if id:match ('^%d+$') == nil then
if id:match ('^%d+$') == nil then
Line 703: Line 716:
]]
]]


local function asin (id, domain)
local function asin (options)
local id = options.id;
local domain = options.ASINTLD;
local err_cat = ""
local err_cat = ""


Line 710: Line 726:
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 isbn (id) then -- see if ASIN value is or validates as ISBN-10
if isbn (options) then -- see if ASIN value is or validates as ISBN-10
if not id:find ('^630') then -- 630xxxxxxx is (apparently) not a valid isbn prefix but is used by amazon as a numeric identifier
if not id:find ('^630') then -- 630xxxxxxx is (apparently) not a valid isbn prefix but is used by amazon as a numeric identifier
set_message ('maint_asin'); -- begins with something other than 630 so possibly an isbn
set_message ('maint_asin'); -- begins with something other than 630 so possibly an isbn
Line 721: Line 737:
end
end
end
end
if not is_set(domain) then  
if not is_set (domain) then  
domain = "com";
domain = "com";
elseif in_array (domain, {'jp', 'uk'}) then -- Japan, United Kingdom
elseif in_array (domain, {'jp', 'uk'}) then -- Japan, United Kingdom
Line 728: Line 744:
domain = "com." .. domain;
domain = "com." .. domain;
end
end
local handler = cfg.id_handlers['ASIN'];
 
local handler = options.handler;
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/",
Line 743: Line 760:
]]
]]


local function ismn (id)
local function ismn (options)
local handler = cfg.id_handlers['ISMN'];
local id = options.id;
local handler = options.handler;
local text;
local text;
local valid_ismn = true;
local valid_ismn = true;
Line 752: Line 770:
id = id:gsub ('[%s-]', ''); -- remove hyphens and white space
id = id:gsub ('[%s-]', ''); -- remove hyphens and white space


if 13 ~= id:len() or id:match( "^9790%d*$" ) == nil then -- ISMN must be 13 digits and begin with 9790
if 13 ~= id:len() or id:match ("^9790%d*$" ) == nil then -- ISMN must be 13 digits and begin with 9790
valid_ismn = false;
valid_ismn = false;
else
else
Line 769: Line 787:


if false == valid_ismn then
if false == valid_ismn then
text = text .. ' ' .. set_message( 'err_bad_ismn' ) -- add an error message if the ISMN is invalid
text = text .. ' ' .. set_message ('err_bad_ismn' ) -- add an error message if the ISMN is invalid
end  
end  
Line 791: Line 809:
]]
]]


local function issn (id, e, ignore_invalid)
local function issn (options)
local id = options.id;
local handler = options.handler;
local ignore_invalid = options.accept;
 
local issn_copy = id; -- save a copy of unadulterated ISSN; use this version for display if ISSN does not validate
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 text;
local valid_issn = true;
local valid_issn = true;
handler = cfg.id_handlers[e and 'EISSN' or 'ISSN'];


id = id:gsub ('[%s-]', ''); -- remove hyphens and whitespace
id = id:gsub ('[%s-]', ''); -- remove hyphens and whitespace


if 8 ~= id:len() or nil == id:match( "^%d*X?$" ) then -- validate the ISSN: 8 digits long, containing only 0-9 or X in the last position
if 8 ~= id:len() or nil == id:match ("^%d*X?$" ) then -- validate the ISSN: 8 digits long, containing only 0-9 or X in the last position
valid_issn = false; -- wrong length or improper character
valid_issn = false; -- wrong length or improper character
else
else
valid_issn = is_valid_isxn(id, 8); -- validate ISSN
valid_issn = is_valid_isxn (id, 8); -- validate ISSN
end
end


if true == valid_issn then
if true == valid_issn then
id = string.sub( id, 1, 4 ) .. "-" .. string.sub( id, 5 ); -- if valid, display correctly formatted version
id = string.sub (id, 1, 4 ) .. "-" .. string.sub (id, 5 ); -- if valid, display correctly formatted version
else
else
id = issn_copy; -- if not valid, show the invalid ISSN with error message
id = issn_copy; -- if not valid, show the invalid ISSN with error message
Line 820: Line 839:
else
else
if false == valid_issn then
if false == valid_issn then
text = text .. ' ' .. set_message ('err_bad_issn', e and 'e' or ''); -- add an error message if the ISSN is invalid
text = text .. ' ' .. set_message ('err_bad_issn', (options.hkey == 'EISSN') and 'e' or ''); -- add an error message if the ISSN is invalid
end  
end  
end
end
Line 834: Line 853:
]]
]]


local function jfm (id)
local function jfm (options)
local handler = cfg.id_handlers['JFM'];
local id = options.id;
local handler = options.handler;
local id_num;
local id_num;
local err_cat = '';
local err_cat = '';
Line 843: Line 863:
if is_set (id_num) then
if is_set (id_num) then
set_message ('maint_jfm_format');
set_message ('maint_jfm_format');
else -- plain number without mr prefix
else -- plain number without JFM prefix
id_num = id; -- if here id does not have prefix
id_num = id; -- if here id does not have prefix
end
end
Line 850: Line 870:
id = id_num; -- jfm matches pattern
id = id_num; -- jfm matches pattern
else
else
err_cat = ' ' .. set_message( 'err_bad_jfm' ); -- set an error message
err_cat = ' ' .. set_message ('err_bad_jfm' ); -- set an 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, encode = handler.encode}) .. err_cat;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode}) .. err_cat;
end
--[[--------------------------< J S T O R >--------------------------------------------------------------------
Format a JSTOR with some error checking
]]
local function jstor (options)
local id = options.id;
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
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
end


Line 872: Line 913:
]]
]]


local function lccn(lccn)
local function lccn (options)
local handler = cfg.id_handlers['LCCN'];
local lccn = options.id;
local handler = options.handler;
local err_cat = ''; -- presume that LCCN is valid
local err_cat = ''; -- presume that LCCN is valid
local id = lccn; -- local copy of the LCCN
local id = lccn; -- local copy of the LCCN
Line 921: Line 963:
]]
]]


local function mr (id)
local function mr (options)
local handler = cfg.id_handlers['MR'];
local id = options.id;
local handler = options.handler;
local id_num;
local id_num;
local id_len;
local id_len;
Line 937: Line 980:
id_len = id_num and id_num:len() or 0;
id_len = id_num and id_num:len() or 0;
if (7 >= id_len) and (0 ~= id_len) then
if (7 >= id_len) and (0 ~= id_len) then
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
err_cat = ' ' .. set_message ('err_bad_mr'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_mr'); -- set an error message
Line 954: Line 997:
]]
]]


local function oclc (id)
local function oclc (options)
local handler = cfg.id_handlers['OCLC'];
local id = options.id;
local handler = options.handler;
local number;
local number;
local err_msg = ''; -- empty string for concatenation
local err_msg = ''; -- empty string for concatenation
Line 996: Line 1,040:
]]
]]


local function openlibrary(id, access)
local function openlibrary (options)
local handler = cfg.id_handlers['OL'];
local id = options.id;
local access = options.access;
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 error_msg = '';
Line 1,017: Line 1,063:
id = ident, separator = handler.separator, encode = handler.encode,
id = ident, separator = handler.separator, encode = handler.encode,
access = access}) .. error_msg;
access = access}) .. error_msg;
end
--[[--------------------------< O S T I >----------------------------------------------------------------------
Format OSTI and do simple error checking. OSTIs are sequential numbers beginning at 1 and counting up.  This
code checks the OSTI to see that it contains only digits and is less than test_limit specified in the configuration;
the value in test_limit will need to be updated periodically as more OSTIs are issued.
NB. 1018 is the lowest OSTI number found in the wild (so far) and resolving OK on the OSTI site
]]
local function osti (options)
local id = options.id;
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
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
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
end


Line 1,037: Line 1,113:
]]
]]


local function pmc (id, embargo)
local function pmc (options)
local handler = cfg.id_handlers['PMC'];
local id = options.id;
local embargo = options.Embargo; -- TODO: lowercase?
local handler = options.handler;
local err_cat;
local err_cat;
local id_num;
local id_num;
Line 1,052: Line 1,130:


if is_set (id_num) then -- id_num has a value so test it
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
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_cat = ' ' .. set_message ('err_bad_pmc'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_pmc'); -- set an error message
Line 1,088: Line 1,166:
]]
]]


local function pmid(id)
local function pmid (options)
local handler = cfg.id_handlers['PMID'];
local id = options.id;
local handler = options.handler;
local err_cat = ''; -- presume that PMID is valid
local err_cat = ''; -- presume that PMID is valid
Line 1,095: Line 1,174:
err_cat = ' ' .. set_message ('err_bad_pmid'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_pmid'); -- set an error message
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
err_cat = ' ' .. set_message ('err_bad_pmid'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_pmid'); -- set an error message
Line 1,103: Line 1,182:
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}) .. err_cat;
end
--[[--------------------------< R F C >------------------------------------------------------------------------
Format RFC and do simple error checking. RFCs are sequential numbers beginning at 1 and counting up.  This
code checks the RFC to see that it contains only digits and is less than test_limit specified in the configuration;
the value in test_limit will need to be updated periodically as more RFCs are issued.
An index of all RFCs is here: https://tools.ietf.org/rfc/
]]
local function rfc (options)
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
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
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
end


Line 1,116: Line 1,224:
]]
]]


local function s2cid (id, access)
local function s2cid (options)
local handler = cfg.id_handlers['S2CID'];
local id = options.id;
local access = options.access;
local handler = options.handler;
local err_cat = ''; -- presume that S2CID is valid
local err_cat = ''; -- presume that S2CID is valid
local id_num;
local id_num;
Line 1,125: Line 1,235:


  if is_set (id_num) then -- id_num has a value so test it
  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
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
err_cat = ' ' .. set_message ('bad_s2cid_err'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_s2cid'); -- set an error message
end
end


else -- when id format incorrect
else -- when id format incorrect
err_cat = ' ' .. set_message ('bad_s2cid_err'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_s2cid'); -- set an error message
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:gsub ('%.%a%a', ''), separator = handler.separator, encode = handler.encode, access = access}) .. err_cat;
prefix = handler.prefix, id = id, separator = handler.separator, encode = handler.encode, access = access}) .. err_cat;


return text;
return text;
Line 1,147: Line 1,257:
]]
]]


local function sbn (id, ignore_invalid)
local function sbn (options)
local handler = cfg.id_handlers['SBN'];
local id = options.id;
local ignore_invalid = options.accept;
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,
Line 1,190: Line 1,302:
]]
]]


local function ssrn (id)
local function ssrn (options)
local handler = cfg.id_handlers['SSRN'];
local id = options.id;
local handler = options.handler;
local err_cat = ''; -- presume that SSRN is valid
local err_cat = ''; -- presume that SSRN is valid
local id_num;
local id_num;
Line 1,199: Line 1,312:


if is_set (id_num) then -- id_num has a value so test it
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
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
err_cat = ' ' .. set_message ('err_bad_ssrn'); -- set an error message
err_cat = ' ' .. set_message ('err_bad_ssrn'); -- set an error message
Line 1,221: Line 1,334:
]]
]]


local function usenet_id (id)
local function usenet_id (options)
local handler = cfg.id_handlers['USENETID'];
local id = options.id;
local handler = options.handler;


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,
Line 1,245: Line 1,359:
]]
]]


local function zbl (id)
local function zbl (options)
local handler = cfg.id_handlers['ZBL'];
local id = options.id;
local handler = options.handler;
local err_cat = '';
local err_cat = '';
Line 1,261: Line 1,376:


--============================<< I N T E R F A C E  F U N C T I O N S >>==========================================
--============================<< I N T E R F A C E  F U N C T I O N S >>==========================================
--[[--------------------------< B U I L D _ I D _ L I S T >--------------------------------------------------------
Takes a table of IDs created by extract_ids() and turns it into a table of formatted ID outputs.
inputs:
id_list – table of identifiers built by extract_ids()
options – table of various template parameter values used to modify some manually handled identifiers
]]
local function build_id_list( id_list, options )
local new_list, handler = {};
local accept;
local function fallback(k) return { __index = function(t,i) return cfg.id_handlers[k][i] end } end;
for k, v in pairs( id_list ) do
v, accept = has_accept_as_written (v); -- remove and note accept-as-written markup if present
-- k is uppercase 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( { ['id'] = v, ['access'] = options.IdAccessLevels[k] }, fallback(k) );
if handler.mode == 'external' then
table.insert( new_list, {handler.label, external_link_id( handler ) } );
elseif handler.mode == 'internal' then
table.insert( new_list, {handler.label, internal_link_id( handler ) } );
elseif handler.mode ~= 'manual' then
error( cfg.messages['unknown_ID_mode'] );
elseif k == 'ARXIV' then
table.insert( new_list, {handler.label, arxiv( v, options.Class ) } );
elseif k == 'ASIN' then
table.insert( new_list, {handler.label, asin( v, options.ASINTLD ) } );
elseif k == 'BIBCODE' then
table.insert( new_list, {handler.label, bibcode( v, handler.access ) } );
elseif k == 'BIORXIV' then
table.insert( new_list, {handler.label, biorxiv( v ) } );
elseif k == 'CITESEERX' then
table.insert( new_list, {handler.label, citeseerx( v ) } );
elseif k == 'DOI' then
table.insert( new_list, {handler.label, doi( v, options.DoiBroken, handler.access, accept) } );
elseif k == 'EISSN' then
table.insert( new_list, {handler.label, issn( v, true, accept) } ); -- true distinguishes EISSN from ISSN
elseif k == 'HDL' then
table.insert( new_list, {handler.label, hdl( v, handler.access ) } );
elseif k == 'ISBN' then
table.insert( new_list, {handler.label, isbn (v, (accept or options.IgnoreISBN)) } );
elseif k == 'ISMN' then
table.insert( new_list, {handler.label, ismn( v ) } );
elseif k == 'ISSN' then
table.insert( new_list, {handler.label, issn( v, false, accept) } );
elseif k == 'JFM' then
table.insert( new_list, {handler.label, jfm( v ) } );
elseif k == 'LCCN' then
table.insert( new_list, {handler.label, lccn( v ) } );
elseif k == 'MR' then
table.insert( new_list, {handler.label, mr( v ) } );
elseif k == 'OCLC' then
table.insert( new_list, {handler.label, oclc( v ) } );
elseif k == 'OL' or k == 'OLA' then
table.insert( new_list, {handler.label, openlibrary( v, handler.access ) } );
elseif k == 'PMC' then
table.insert( new_list, {handler.label, pmc( v, options.Embargo ) } );
elseif k == 'PMID' then
table.insert( new_list, {handler.label, pmid( v ) } );
elseif k == 'S2CID' then
table.insert( new_list, {handler.label, s2cid( v, handler.access ) } );
elseif k == 'SBN' then
table.insert( new_list, {handler.label, sbn (v, accept) } );
elseif k == 'SSRN' then
table.insert( new_list, {handler.label, ssrn( v ) } );
elseif k == 'USENETID' then
table.insert( new_list, {handler.label, usenet_id( v ) } );
elseif k == 'ZBL' then
table.insert( new_list, {handler.label, zbl( v ) } );
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 >------------------------------------------------------------
--[[--------------------------< E X T R A C T _ I D S >------------------------------------------------------------
Line 1,363: Line 1,385:
]]
]]


local function extract_ids( args )
local function extract_ids (args)
local id_list = {}; -- list of identifiers found in args
local id_list = {}; -- list of identifiers found in args
for k, v in pairs( cfg.id_handlers ) do -- k is uppercase identifier name as index to cfg.id_handlers; e.g. cfg.id_handlers['ISBN'], v is a table
for k, v in pairs (cfg.id_handlers) do -- k is uppercase identifier name as index to cfg.id_handlers; e.g. cfg.id_handlers['ISBN'], v is a table
v = select_one( args, v.parameters, 'err_redundant_parameters' ); -- v.parameters is a table of aliases for k; here we pick one from args if present
v = select_one (args, v.parameters, 'err_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
if is_set (v) then id_list[k] = v; end -- if found in args, add identifier to our list
end
end
return id_list;
return id_list;
Line 1,375: Line 1,397:
--[[--------------------------< E X T R A C T _ I D _ A C C E S S _ L E V E L S >--------------------------------------
--[[--------------------------< 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.
Fetches custom id access levels from arguments using configuration settings. Parameters which have a predefined access
Parameters which have a predefined access level (e.g. arxiv) do not use this
level (e.g. arxiv) do not use this function as they are directly rendered as free without using an additional parameter.
function as they are directly rendered as free without using an additional parameter.
 
returns a table of k/v pairs where k is same as the identifier's key in cfg.id_handlers and v is the assigned (valid) keyword


access-level values must match the case used in cfg.keywords_lists['id-access'] (lowercase unless there is some special reason for something else)
access-level values must match the case used in cfg.keywords_lists['id-access'] (lowercase unless there is some special reason for something else)
Line 1,383: Line 1,406:
]]
]]


local function extract_id_access_levels( args, id_list )
local function extract_id_access_levels (args, id_list)
local id_accesses_list = {};
local id_accesses_list = {};
for k, v in pairs( cfg.id_handlers ) do
for k, v in pairs (cfg.id_handlers) do
local access_param = v.custom_access; -- name of identifier's access-level parameter
local access_param = v.custom_access; -- name of identifier's access-level parameter
if is_set(access_param) then
if is_set (access_param) then
local access_level = args[access_param]; -- get the assigned value if there is one
local access_level = args[access_param]; -- get the assigned value if there is one
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
table.insert( z.message_tail, { set_message( 'invalid_param_val', {access_param, access_level}, true ) } );
table.insert (z.message_tail, { set_message ('invalid_param_val', {access_param, access_level}, true) } );
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
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
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
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
Line 1,402: Line 1,425:
end
end
return id_accesses_list;
return id_accesses_list;
end
--[[--------------------------< B U I L D _ I D _ L I S T >----------------------------------------------------
render the identifiers into a sorted sequence table
<id_list_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) rendered identifier strings
]]
local function build_id_list (id_list_t, options_t, access_levels_t)
local new_list_t = {};
local accept;
local func_map = { --function map points to functions associated with hkey identifier
['ARXIV'] = arxiv,
['ASIN'] = asin,
['BIBCODE'] = bibcode,
['BIORXIV'] = biorxiv,
['CITESEERX'] = citeseerx,
['DOI'] = doi,
['EISSN'] = issn,
['HDL'] = hdl,
['ISBN'] = isbn,
['ISMN'] = ismn,
['ISSN'] = issn,
['JFM'] = jfm,
['JSTOR'] = jstor,
['LCCN'] = lccn,
['MR'] = mr,
['OCLC'] = oclc,
['OL'] = openlibrary,
['OSTI'] = osti,
['PMC'] = pmc,
['PMID'] = pmid,
['RFC']  = rfc,
['S2CID'] = s2cid,
['SBN'] = sbn,
['SSRN'] = ssrn,
['USENETID'] = usenet_id,
['ZBL'] = zbl,
}
for hkey, v in pairs (id_list_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
options_t.hkey = hkey; -- ~/Configuration handler key
options_t.id = v; -- add that identifier value to the options table
options_t.accept = accept; -- add the accept boolean flag
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];
if func_map[hkey] then
table.insert (new_list_t, {hkey, func_map[hkey] (options_t)}); -- call the function and add the results to the output sequence table
-- 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
else
error (cfg.messages['unknown_ID_key'] .. ' ' .. hkey); -- here when func_map doesn't have a function for hkey
end
end
local function comp (a, b) -- used by following table.sort()
return a[1]:lower() < b[1]:lower(); -- sort by hkey
end
table.sort (new_list_t, comp); -- sequence table of tables sort
for k, v in ipairs (new_list_t) do -- convert sequence table of tables to simple sequence table of strings
new_list_t[k] = v[2];
end
return new_list_t;
end
--[[--------------------------< I D E N T I F I E R _ L I S T S _ G E T >--------------------------------------
Creates two identifier lists: a k/v table of identifiers and their values to be used locally and for use in the
COinS metadata, and a sequence table of the rendered identifier strings that will be included in the rendered
citation.
]]
local function identifier_lists_get (args, options_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_access_levels_t = extract_id_access_levels (args, 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
end
end


Line 1,432: Line 1,550:
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=
build_id_list = build_id_list,
identifier_lists_get = identifier_lists_get, -- experiment to replace individual calls to build_id_list, extract_ids, extract_id_access_levels
extract_ids = extract_ids,
extract_id_access_levels = extract_id_access_levels,
is_embargoed = is_embargoed;
is_embargoed = is_embargoed;
set_selected_modules = set_selected_modules;
set_selected_modules = set_selected_modules;
}
}
Retrieved from "https://nonbinary.wiki/wiki/Special:MobileDiff/17953"