Module:Citation/CS1/Identifiers: Difference between revisions

    (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: Line 11:


    --============================<< H E L P E R  F U N C T I O N S >>============================================
    --============================<< 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 >----------------------------------------------
    --[[--------------------------< E X T E R N A L _ L I N K _ I D >----------------------------------------------
    Line 33: Line 65:
    end
    end


    -- this_wiki_code = mw.language.getContentLanguage():getCode(); -- get this wikipedia's language code (subdomain)
    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;
    if string.match (mw.site.server, 'wikidata') then
    this_wiki_code = mw.getCurrentFrame():preprocess('{{int:lang}}'); -- on wikidata so use interface language setting instead
    end
    end
    if is_set (options.q) and mw.wikibase then -- wikibase test here avoids script errors in third party wikis that aren't using mw.wikibase extension
    local label_link = (cfg.use_identifier_redirects and is_set (options.redirect) and options.redirect) or wd_article or options.link;
    wd_article = mw.wikibase.getEntity (options.q):getSitelink (this_wiki_code .. 'wiki'); -- fetch article title from wd
    if wd_article then
    wd_article = table.concat ({':', this_wiki_code, ':', wd_article}); -- make interwiki link if taken from wd; leading colon required
    end
    end


    return table.concat ({
    return table.concat ({
    make_wikilink (wd_article or options.link, options.label), -- wikidata link or locally specified
    make_wikilink (label_link, options.label), -- redirect, wikidata link, or locally specified link (in that order)
    options.separator or '&nbsp;',
    options.separator or '&nbsp;',
    ext_link
    ext_link
    Line 62: Line 87:
    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
    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 (
    return table.concat (
    {
    {
    make_wikilink (options.link, options.label),
    make_wikilink (label_link, options.label), -- wiki link the identifier label
    options.separator or '&nbsp;',
    options.separator or '&nbsp;', -- add the separator
    make_wikilink (
    make_wikilink (
    table.concat (
    table.concat (
    Line 105: Line 136:
    end
    end
    return ''; -- |embargo= not set return empty string
    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
    end


    Line 120: Line 190:
    local function is_valid_isxn (isxn_str, len)
    local function is_valid_isxn (isxn_str, len)
    local temp = 0;
    local temp = 0;
    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
    return temp % 11 == 0; -- returns true if calculation result is zero
    return temp % 11 == 0; -- returns true if calculation result is zero
    end
    end


    Line 254: Line 324:
    err_cat = err_cat and table.concat ({' ', set_error ('bad_arxiv')}) or ''; -- set error message if flag is true
    err_cat = err_cat and table.concat ({' ', set_error ('bad_arxiv')}) or ''; -- set error message if flag is true
    text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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}) .. err_cat;


    Line 292: Line 362:
    local year;
    local year;


    local text = external_link_id({link=handler.link, label=handler.label, q = handler.q,
    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,
    access=access});
    access=access});
    if 19 ~= id:len() then
    if 19 ~= id:len() then
    err_type = 'length';
    err_type = cfg.err_msg_supl.length;
    else
    else
    year = id:match ("^(%d%d%d%d)[%a][%w&%.][%w&%.][%w&%.][%w.]+[%a%.]$") --  
    year = id:match ("^(%d%d%d%d)[%a][%w&%.][%w&%.][%w&%.][%w.]+[%a%.]$") --  
    if not year then -- if nil then no pattern match
    if not year then -- if nil then no pattern match
    err_type = '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
    err_type = 'year'; -- year out of bounds
    err_type = cfg.err_msg_supl.year; -- year out of bounds
    end
    end
    if id:find('&%.') then
    if id:find('&%.') then
    err_type = 'journal'; -- journal abbreviation must not have '&.' (if it does its missing a letter)
    err_type = cfg.err_msg_supl.journal; -- journal abbreviation must not have '&.' (if it does its missing a letter)
    end
    end
    end
    end
    Line 323: Line 393:
    --[[--------------------------< B I O R X I V >-----------------------------------------------------------------
    --[[--------------------------< B I O R X I V >-----------------------------------------------------------------


    Format bioRxiv id and do simple error checking.  BiorXiv ids are exactly 6 digits.
    Format bioRxiv id and do simple error checking.  Before 2019-12-11, biorXiv ids were 10.1101/ followed by exactly
    The bioRxiv id is the number following the last slash in the bioRxiv-issued DOI:
    6 digits.  After 2019-12-11, biorXiv ids retained the six-digit identifier but prefixed that with a yyyy.mm.dd.
    https://doi.org/10.1101/078733 -> 078733
    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: Line 408:
    local function biorxiv(id)
    local function biorxiv(id)
    local handler = cfg.id_handlers['BIORXIV'];
    local handler = cfg.id_handlers['BIORXIV'];
    local err_cat = ''; -- presume that bioRxiv id is valid
    local err_cat = ' ' .. set_error( 'bad_biorxiv'); -- assume that there will be an error
    if nil == id:match("^%d%d%d%d%d%d$") then -- if bioRxiv id has anything but six digits
    local patterns = {
    err_cat = ' ' .. set_error( 'bad_biorxiv'); -- set an error message
    '^10.1101/%d%d%d%d%d%d$', -- simple 6-digit identifier (before 2019-12-11)
    end
    '^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)
    }
    return external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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,
    prefix=handler.prefix,id=id,separator=handler.separator,
    encode=handler.encode, access=handler.access}) .. err_cat;
    encode=handler.encode, access=handler.access}) .. err_cat;
    Line 354: Line 447:
    local matched;
    local matched;
    local text = external_link_id({link=handler.link, label=handler.label, q = handler.q,
    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,
    access=handler.access});
    access=handler.access});
    Line 414: Line 507:
    end
    end


    text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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}) .. (inactive or '')
    prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=access}) .. (inactive or '')


    Line 427: Line 520:
    '^%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
    }
    }
    Line 458: Line 552:
    terminal punctuation may not be technically correct but it appears, that in practice these characters are rarely
    terminal punctuation may not be technically correct but it appears, that in practice these characters are rarely
    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
    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: Line 561:
    local function hdl(id, access)
    local function hdl(id, access)
    local handler = cfg.id_handlers['HDL'];
    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 text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
    local hdl, suffix, param = id:match ('(.-)(%?(%a+).+)$'); -- look for query string
    prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=access})
    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–]-[^%.,]$") then -- hdl must contain a fwd slash, must not contain spaces, endashes, and must not end with period or comma
    if nil == id:match("^[^%s–]-/[^%s–]-[^%.,]$") then -- hdl must contain a fwd slash, must not contain spaces, endashes, and must not end with period or comma
    Line 482: Line 609:
    local function isbn( isbn_str )
    local function isbn( isbn_str )
    if nil ~= isbn_str:match("[^%s-0-9X]") then
    if nil ~= isbn_str:match("[^%s-0-9X]") then
    return false, 'invalid character'; -- fail if isbn_str contains anything but digits, hyphens, or the uppercase X
    return false, cfg.err_msg_supl.char; -- fail if isbn_str contains anything but digits, hyphens, or the uppercase X
    end
    end
    isbn_str = isbn_str:gsub( "-", "" ):gsub( " ", "" ); -- remove hyphens and spaces
    isbn_str = isbn_str:gsub( "-", "" ):gsub( " ", "" ); -- remove hyphens and spaces
    Line 488: Line 615:
       
       
    if len ~= 10 and len ~= 13 then
    if len ~= 10 and len ~= 13 then
    return false, 'length'; -- fail if incorrect length
    return false, cfg.err_msg_supl.length; -- fail if incorrect length
    end
    end


    if len == 10 then
    if len == 10 then
    if isbn_str:match( "^%d*X?$" ) == nil then -- fail if isbn_str has 'X' anywhere but last position
    if isbn_str:match( "^%d*X?$" ) == nil then -- fail if isbn_str has 'X' anywhere but last position
    return false, 'invalid form';
    return false, cfg.err_msg_supl.form;
    end
    end
    return is_valid_isxn(isbn_str, 10), 'checksum';
    return is_valid_isxn(isbn_str, 10), cfg.err_msg_supl.check;
    else
    else
    if isbn_str:match( "^%d+$" ) == nil then
    if isbn_str:match( "^%d+$" ) == nil then
    return false, 'invalid character'; -- fail if isbn13 is not all digits
    return false, cfg.err_msg_supl.char; -- fail if isbn13 is not all digits
    end
    end
    if isbn_str:match( "^97[89]%d*$" ) == nil then
    if isbn_str:match( "^97[89]%d*$" ) == nil then
    return false, 'invalid prefix'; -- fail when isbn13 does not begin with 978 or 979
    return false, cfg.err_msg_supl.prefix; -- fail when isbn13 does not begin with 978 or 979
    end
    end
    if isbn_str:match ('^9790') then
    if isbn_str:match ('^9790') then
    return false, 'invalid group id'; -- group identifier '0' is reserved to ismn
    return false, cfg.err_msg_supl.group; -- group identifier '0' is reserved to ismn
    end
    end
    return is_valid_isxn_13 (isbn_str), 'checksum';
    return is_valid_isxn_13 (isbn_str), cfg.err_msg_supl.check;
    end
    end
    end
    end
    Line 546: Line 673:
    end
    end
    local handler = cfg.id_handlers['ASIN'];
    local handler = cfg.id_handlers['ASIN'];
    return external_link_id({link=handler.link,
    return external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
    label=handler.label, q = handler.q, 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}) .. err_cat;
    end
    end
    Line 575: Line 702:
    end
    end


    -- text = internal_link_id({link = handler.link, label = handler.label, -- use this (or external version) when there is some place to link to
    -- text = internal_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect, -- use this (or external version) when there is some place to link to
    -- prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode})
    -- prefix=handler.prefix, id=id_copy, separator=handler.separator, encode=handler.encode})
       
       
    text = table.concat (
    text = table.concat (
    Line 634: Line 761:
    end
    end
    text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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})
     
    if false == valid_issn then
    if false == valid_issn then
    text = text .. ' ' .. set_error( 'bad_issn', e and 'e' or '' ) -- add an error message if the issn is invalid
    text = text .. ' ' .. set_error( 'bad_issn', e and 'e' or '' ) -- add an error message if the issn is invalid
    Line 670: Line 797:
    end
    end
    return external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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
    end
    Line 727: Line 854:
    end
    end


    return external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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}) .. err_cat;
    end
    end
    Line 759: Line 886:
    end
    end
    return external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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
    end
    Line 800: Line 927:
    end
    end
    local text = external_link_id({link=handler.link, label=handler.label, q = handler.q,
    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}) .. err_msg;
    prefix=handler.prefix, id=id, separator=handler.separator, encode=handler.encode}) .. err_msg;


    Line 821: Line 948:


    if not is_set (ident) then -- if malformed return an error
    if not is_set (ident) then -- if malformed return an error
    return external_link_id({link=handler.link, label=handler.label, q = handler.q,
    return external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
    prefix=handler.prefix .. 'OL',
    prefix=handler.prefix .. 'OL',
    id=id, separator=handler.separator, encode = handler.encode,
    id=id, separator=handler.separator, encode = handler.encode,
    Line 830: Line 957:
    if ( code == "A" ) then
    if ( code == "A" ) then
    return external_link_id({link=handler.link, label=handler.label, q = handler.q,
    return external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
    prefix=handler.prefix .. 'authors/OL',
    prefix=handler.prefix .. 'authors/OL',
    id=id, separator=handler.separator, encode = handler.encode,
    id=id, separator=handler.separator, encode = handler.encode,
    Line 837: Line 964:
    if ( code == "M" ) then
    if ( code == "M" ) then
    return external_link_id({link=handler.link, label=handler.label, q = handler.q,
    return external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
    prefix=handler.prefix .. 'books/OL',
    prefix=handler.prefix .. 'books/OL',
    id=id, separator=handler.separator, encode = handler.encode,
    id=id, separator=handler.separator, encode = handler.encode,
    Line 844: Line 971:


    if ( code == "W" ) then
    if ( code == "W" ) then
    return external_link_id({link=handler.link, label=handler.label, q = handler.q,
    return external_link_id ({link=handler.link, label=handler.label, q=handler.q, redirect=handler.redirect,
    prefix=handler.prefix .. 'works/OL',
    prefix=handler.prefix .. 'works/OL',
    id=id, separator=handler.separator, encode = handler.encode,
    id=id, separator=handler.separator, encode = handler.encode,
    Line 870: Line 997:


    local function pmc(id, embargo)
    local function pmc(id, embargo)
    local test_limit = 8000000; -- update this value as PMCs approach
    local test_limit = cfg.id_limits.pmc; -- update this value as PMCs approach
    local handler = cfg.id_handlers['PMC'];
    local handler = cfg.id_handlers['PMC'];
    local err_cat = ''; -- presume that PMC is valid
    local err_cat = ''; -- presume that PMC is valid
    Line 904: Line 1,031:
    });
    });
    else
    else
    text = external_link_id({link = handler.link, label = handler.label, q = handler.q, -- 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}) .. err_cat;
    prefix=handler.prefix,id=id,separator=handler.separator, encode=handler.encode, access=handler.access}) .. err_cat;
    end
    end
    Line 920: Line 1,047:


    local function pmid(id)
    local function pmid(id)
    local test_limit = 33000000; -- update this value as PMIDs approach
    local test_limit = cfg.id_limits.pmid; -- update this value as PMIDs approach
    local handler = cfg.id_handlers['PMID'];
    local handler = cfg.id_handlers['PMID'];
    local err_cat = ''; -- presume that PMID is valid
    local err_cat = ''; -- presume that PMID is valid
    Line 933: Line 1,060:
    end
    end
    return external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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
    --[[--------------------------< 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
    end


    Line 949: Line 1,140:


    local function ssrn (id)
    local function ssrn (id)
    local test_limit = 3500000; -- update this value as SSRNs approach
    local test_limit = cfg.id_limits.ssrn; -- update this value as SSRNs approach
    local handler = cfg.id_handlers['SSRN'];
    local handler = cfg.id_handlers['SSRN'];
    local err_cat = ''; -- presume that SSRN is valid
    local err_cat = ''; -- presume that SSRN is valid
    Line 966: Line 1,157:
    end
    end
    text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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}) .. err_cat;


    Line 983: Line 1,174:
    local handler = cfg.id_handlers['USENETID'];
    local handler = cfg.id_handlers['USENETID'];


    local text = external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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})
       
       
    Line 1,014: Line 1,205:
    end
    end
    return external_link_id({link = handler.link, label = handler.label, q = handler.q,
    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
    end
    Line 1,095: Line 1,286:
    elseif k == 'PMID' then
    elseif k == 'PMID' then
    table.insert( new_list, {handler.label, pmid( v ) } );
    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
    elseif k == 'SSRN' then
    table.insert( new_list, {handler.label, ssrn( v ) } );
    table.insert( new_list, {handler.label, ssrn( v ) } );
    Retrieved from "https://nonbinary.wiki/wiki/Module:Citation/CS1/Identifiers"