Content deleted Content added
changed PMC test limit to 8000000, since PMCs greater than 7000000 have now been issued, see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7003341/ |
sync from sandbox; |
||
Line 11:
--============================<< H E L P E R F U N C T I O N S >>============================================
--[[--------------------------< W I K I D A T A _ A R T I C L E _ N A M E _ G E T >----------------------------
as an aid to internationalizing identifier-label wikilinks, gets identifier article names from wikidata.
returns :<lang code>:<article title> when <q> has an <article title> for <lang code>; nil else
for identifiers that do not have q, returns nil
for wikis that do not have mw.wikibase installed, returns nil
The call to mw.wikibase.getEntity() bumps the expensive parser function count
]]
local function wikidata_article_name_get (q)
if not is_set (q) or (q and not mw.wikibase) then -- when no q number or when a q number but mw.wikibase not installed on this wiki
return nil; -- abandon
end
local wd_article;
local this_wiki_code = cfg.this_wiki_code; -- wikipedia subdomain; 'en' for en.wikipedia.org
wd_article = mw.wikibase.getEntity (q):getSitelink (this_wiki_code .. 'wiki'); -- fetch article title from wd; nil when no title available at this wiki; bumps expensive parser function count
if wd_article then
wd_article = table.concat ({':', this_wiki_code, ':', wd_article}); -- interwiki-style link without brackets if taken from wd; leading colon required
end
return wd_article; -- article title from wd; nil else
end
--[[--------------------------< E X T E R N A L _ L I N K _ I D >----------------------------------------------
Line 33 ⟶ 65:
end
if not (cfg.use_identifier_redirects and is_set (options.redirect)) then -- redirect has priority so if enabled and available don't fetch from wikidata because expensive
wd_article = wikidata_article_name_get (options.q); -- if wikidata has an article title for this wiki, get it;
end
local label_link = (cfg.use_identifier_redirects and is_set (options.redirect) and options.redirect) or wd_article or options.link;
return table.concat ({
make_wikilink (
options.separator or ' ',
ext_link
Line 62 ⟶ 87:
local function internal_link_id(options)
local id = mw.ustring.gsub (options.id, '%d', cfg.date_names.local_digits); -- translate 'local' digits to Western 0-9
if not (cfg.use_identifier_redirects and is_set (options.redirect)) then -- redirect has priority so if enabled and available don't fetch from wikidata because expensive
wd_article = wikidata_article_name_get (options.q); -- if wikidata has an article title for this wiki, get it;
end
local label_link = (cfg.use_identifier_redirects and is_set (options.redirect) and options.redirect) or wd_article or options.link;
return table.concat (
{
make_wikilink (
options.separator or ' ', -- add the separator
make_wikilink (
table.concat (
Line 105 ⟶ 136:
end
return ''; -- |embargo= not set return empty string
end
--[=[-------------------------< I S _ V A L I D _ B I O R X I V _ D A T E >------------------------------------
returns true if:
2019-12-11T00:00Z <= biorxiv_date < today + 2 days
The dated form of biorxiv identifier has a start date of 2019-12-11. The unix timestamp for that date is {{#time:U|2019-12-11}} = 1576022400
biorxiv_date is the date provided in those |biorxiv= parameter values that are dated at time 00:00:00 UTC
today is the current date at time 00:00:00 UTC plus 48 hours
if today is 2015-01-01T00:00:00 then
adding 24 hours gives 2015-01-02T00:00:00 – one second more than today
adding 24 hours gives 2015-01-03T00:00:00 – one second more than tomorrow
This function does not work if it is fed month names for languages other than English. Wikimedia #time: parser
apparently doesn't understand non-Engish date month names. This function will always return false when the date
contains a non-English month name because good1 is false after the call to lang.formatDate(). To get around that
call this function with YYYY-MM-DD format dates.
]=]
local function is_valid_biorxiv_date (biorxiv_date)
local good1, good2;
local biorxiv_ts, tomorrow_ts; -- to hold unix time stamps representing the dates
local lang_object = mw.getContentLanguage();
good1, biorxiv_ts = pcall (lang_object.formatDate, lang_object, 'U', biorxiv_date ); -- convert biorxiv_date value to unix timesatmp
good2, tomorrow_ts = pcall (lang_object.formatDate, lang_object, 'U', 'today + 2 days' ); -- today midnight + 2 days is one second more than all day tomorrow
if good1 and good2 then -- lang.formatDate() returns a timestamp in the local script which which tonumber() may not understand
biorxiv_ts = tonumber (biorxiv_ts) or lang_object:parseFormattedNumber (biorxiv_ts); -- convert to numbers for the comparison;
tomorrow_ts = tonumber (tomorrow_ts) or lang_object:parseFormattedNumber (tomorrow_ts);
else
return false; -- one or both failed to convert to unix time stamp
end
return ((1576022400 <= biorxiv_ts) and (biorxiv_ts < tomorrow_ts)) -- 2012-12-11T00:00Z <= biorxiv_date < tomorrow's date
end
Line 120 ⟶ 190:
local function is_valid_isxn (isxn_str, len)
local temp = 0;
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
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)
temp = temp + 10*( len - i ); -- it represents 10 decimal
else
temp = temp + tonumber( string.char(v) )*(len-i);
end
end
return temp % 11 == 0; -- returns true if calculation result is zero
end
Line 254 ⟶ 324:
err_cat = err_cat and table.concat ({' ', set_error ('bad_arxiv')}) or ''; -- set error message if flag is true
text = external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=handler.access}) .. err_cat;
Line 292 ⟶ 362:
local year;
local text = external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix, id=id, separator=handler.separator, encode=handler.encode,
access=access});
if 19 ~= id:len() then
err_type =
else
year = id:match ("^(%d%d%d%d)[%a][%w&%.][%w&%.][%w&%.][%w.]+[%a%.]$") --
if not year then -- if nil then no pattern match
err_type =
else
local next_year = tonumber(os.date ('%Y'))+1; -- get the current year as a number and add one for next year
year = tonumber (year); -- convert year portion of bibcode to a number
if (1000 > year) or (year > next_year) then
err_type =
end
if id:find('&%.') then
err_type =
end
end
Line 323 ⟶ 393:
--[[--------------------------< B I O R X I V >-----------------------------------------------------------------
Format bioRxiv id and do simple error checking.
6 digits. After 2019-12-11, biorXiv ids retained the six-digit identifier but prefixed that with a yyyy.mm.dd.
date and suffixed with an optional version identifier.
The bioRxiv id is the string of characters:
https://doi.org/10.1101/078733 -> 10.1101/078733
or a date followed by a six-digit number followed by an optional version indicator 'v' and one or more digits:
https://www.biorxiv.org/content/10.1101/2019.12.11.123456v2 -> 10.1101/2019.12.11.123456v2
see https://www.biorxiv.org/about-biorxiv
]]
Line 331 ⟶ 408:
local function biorxiv(id)
local handler = cfg.id_handlers['BIORXIV'];
local err_cat = ' ' .. set_error( 'bad_biorxiv');
local patterns = {
'^10.1101/%d%d%d%d%d%d$', -- simple 6-digit identifier (before 2019-12-11)
'^10.1101/(20[1-9]%d)%.([01]%d)%.([0-3]%d)%.%d%d%d%d%d%dv%d+$', -- y.m.d. date + 6-digit identifier + version (after 2019-12-11)
'^10.1101/(20[1-9]%d)%.([01]%d)%.([0-3]%d)%.%d%d%d%d%d%d$', -- y.m.d. date + 6-digit identifier (after 2019-12-11)
}
for _, pattern in ipairs (patterns) do -- spin through the patterns looking for a match
if id:match (pattern) then
local y, m, d = id:match (pattern); -- found a match, attempt to get year, month and date from the identifier
if m then -- m is nil when id is the six-digit form
if not is_valid_biorxiv_date (y .. '-' .. m .. '-' .. d) then -- validate the encoded date; TODO: don't ignore leapyear and actual month lengths ({{#time:}} is a poor date validator)
break; -- break out early so we don't unset the error message
end
end
err_cat = ''; -- we found a match so unset the error message
break; -- and done
end
end -- err_cat remains set here when no match
return external_link_id ({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;
Line 354 ⟶ 447:
local matched;
local text = external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix, id=id, separator=handler.separator, encode=handler.encode,
access=handler.access});
Line 414 ⟶ 507:
end
text = external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=access}) .. (inactive or '')
Line 427 ⟶ 520:
'^%d%d?%d?$', -- less than 4 digits without subcode (with subcode is legitimate)
'^5555$', -- test registrant will never resolve
'%s', -- any space character in registrant
}
Line 458 ⟶ 552:
terminal punctuation may not be technically correct but it appears, that in practice these characters are rarely
if ever used in HDLs.
Query string parameters are named here: http://www.handle.net/proxy_servlet.html. query strings are not displayed
but since '?' is anallowed character in an hdl, '?' followed by one of the query parameters is the only way we
have to detect the query string so that it isn't url encoded with the rest of the identifier.
]]
Line 463 ⟶ 561:
local function hdl(id, access)
local handler = cfg.id_handlers['HDL'];
local query_params = { -- list of known query parameters from http://www.handle.net/proxy_servlet.html
'noredirect',
'ignore_aliases',
'auth',
'cert',
'index',
'type',
'urlappend',
'locatt',
'action',
}
local hdl, suffix, param = id:match ('(.-)(%?(%a+).+)$'); -- look for query string
local found;
if hdl then -- when there are query strings, this is the handle identifier portion
for _, q in ipairs (query_params) do -- spin through the list of query parameters
if param:match ('^' .. q) then -- if the query string begins with one of the parameters
found = true; -- announce a find
break; -- and stop looking
end
end
end
if found then
id = hdl; -- found so replace id with the handle portion; this will be url encoded, suffix will not
else
suffix = ''; -- make sure suffix is empty string for concatenation else
end
local text = external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
prefix=handler.prefix, id=id, suffix=suffix, separator=handler.separator, encode=handler.encode, access=access})
if nil == id:match("^[^%s–]-/[^%s–]-[^%.,]$") then -- hdl must contain a fwd slash, must not contain spaces, endashes, and must not end with period or comma
Line 482 ⟶ 609:
local function isbn( isbn_str )
if nil ~= isbn_str:match("[^%s-0-9X]") then
return false,
end
isbn_str = isbn_str:gsub( "-", "" ):gsub( " ", "" ); -- remove hyphens and spaces
Line 488 ⟶ 615:
if len ~= 10 and len ~= 13 then
return false,
end
if len == 10 then
if isbn_str:match( "^%d*X?$" ) == nil then -- fail if isbn_str has 'X' anywhere but last position
return false,
end
return is_valid_isxn(isbn_str, 10),
else
if isbn_str:match( "^%d+$" ) == nil then
return false,
end
if isbn_str:match( "^97[89]%d*$" ) == nil then
return false,
end
if isbn_str:match ('^9790') then
return false,
end
return is_valid_isxn_13 (isbn_str),
end
end
Line 546 ⟶ 673:
end
local handler = cfg.id_handlers['ASIN'];
return external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
id=id, encode=handler.encode, separator = handler.separator}) .. err_cat;
end
Line 575 ⟶ 702:
end
-- text = internal_link_id ({link
-- prefix=handler.prefix, id=
text = table.concat (
Line 634 ⟶ 761:
end
text = external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode})
if false == valid_issn then
text = text .. ' ' .. set_error( 'bad_issn', e and 'e' or '' ) -- add an error message if the issn is invalid
Line 670 ⟶ 797:
end
return external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode}) .. err_cat;
end
Line 727 ⟶ 854:
end
return external_link_id ({link
prefix=handler.prefix,id=lccn,separator=handler.separator, encode=handler.encode}) .. err_cat;
end
Line 759 ⟶ 886:
end
return external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode}) .. err_cat;
end
Line 800 ⟶ 927:
end
local text = external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix, id=id, separator=handler.separator, encode=handler.encode}) .. err_msg;
Line 821 ⟶ 948:
if not is_set (ident) then -- if malformed return an error
return external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix .. 'OL',
id=id, separator=handler.separator, encode = handler.encode,
Line 830 ⟶ 957:
if ( code == "A" ) then
return external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix .. 'authors/OL',
id=id, separator=handler.separator, encode = handler.encode,
Line 837 ⟶ 964:
if ( code == "M" ) then
return external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix .. 'books/OL',
id=id, separator=handler.separator, encode = handler.encode,
Line 844 ⟶ 971:
if ( code == "W" ) then
return external_link_id ({link=handler.link, label=handler.label, q
prefix=handler.prefix .. 'works/OL',
id=id, separator=handler.separator, encode = handler.encode,
Line 870 ⟶ 997:
local function pmc(id, embargo)
local test_limit =
local handler = cfg.id_handlers['PMC'];
local err_cat = ''; -- presume that PMC is valid
Line 904 ⟶ 1,031:
});
else
text = external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=handler.access}) .. err_cat;
end
Line 920 ⟶ 1,047:
local function pmid(id)
local test_limit =
local handler = cfg.id_handlers['PMID'];
local err_cat = ''; -- presume that PMID is valid
Line 933 ⟶ 1,060:
end
return external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode}) .. err_cat;
end
--[[--------------------------< S 2 C I D >--------------------------------------------------------------------
Format an s2cid, do simple error checking
S2CIDs are sequential numbers beginning at 1 and counting up. This code checks the s2cid to see that it is only
digits and is less than test_limit; the value in local variable test_limit will need to be updated periodically
as more S2CIDs are issued.
]]
local function s2cid (id, access)
local test_limit = cfg.id_limits.s2cid; -- update this value as S2CIDs approach
local handler = cfg.id_handlers['S2CID'];
local err_cat = ''; -- presume that S2CID is valid
local id_num;
local text;
id_num = id:match ('^[1-9]%d+$'); -- id must be all digits; must not begin with 0; no open access flag
if is_set (id_num) then -- id_num has a value so test it
id_num = tonumber(id_num); -- convert id_num to a number for range testing
if test_limit < id_num then -- if S2CID is outside test limit boundaries
err_cat = ' ' .. set_error( 'bad_s2cid' ); -- set an error message
end
else -- when id format incorrect
err_cat = ' ' .. set_error( 'bad_s2cid' ); -- set an error message
end
text = external_link_id ({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;
return text;
end
--[[--------------------------< S B N >------------------------------------------------------------------------
9-digit form of isbn10; uses same check-digit validation when sbn is prefixed with an additional '0' to make 10 digits
]]
local function sbn (id)
local check;
local err_type = '';
if nil ~= id:match("[^%s-0-9X]") then
return false, cfg.err_msg_supl.char; -- fail if sbn contains anything but digits, hyphens, or the uppercase X
end
id=id:gsub( "[%s-]", "" ); -- strip spaces and hyphens from the sbn
if 9 ~= id:len() then
return false, cfg.err_msg_supl.length; -- fail if incorrect length
end
if id:match( "^%d*X?$" ) == nil then -- fail if sbn has 'X' anywhere but last position
return false, cfg.err_msg_supl.form;
end
return is_valid_isxn('0' .. id, 10), cfg.err_msg_supl.check; -- prefix sbn with '0' and validate as isbn10
end
Line 949 ⟶ 1,140:
local function ssrn (id)
local test_limit =
local handler = cfg.id_handlers['SSRN'];
local err_cat = ''; -- presume that SSRN is valid
Line 966 ⟶ 1,157:
end
text = external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=handler.access}) .. err_cat;
Line 983 ⟶ 1,174:
local handler = cfg.id_handlers['USENETID'];
local text = external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode})
Line 1,014 ⟶ 1,205:
end
return external_link_id ({link
prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode}) .. err_cat;
end
Line 1,095 ⟶ 1,286:
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
local SBN = internal_link_id (handler);
local check; -- boolean validation result
local err_type = '';
check, err_type = sbn (v);
if not check then
SBN = SBN .. set_error( 'bad_sbn', {err_type}, false, " ", "" ); -- display an error message
end
table.insert( new_list, {handler.label, SBN } );
elseif k == 'SSRN' then
table.insert( new_list, {handler.label, ssrn( v ) } );
| |||