Module:Authority control/auxiliary: Difference between revisions

From The Goon Show Depository
m (1 revision imported)
No edit summary
 
(5 intermediate revisions by 3 users not shown)
Line 1: Line 1:
require('strict')
-- require('strict')
local p = {}
local p = {}
--[[======================================================]]
--[[======================================================]]
--[[            Format validation functions              ]]
--[[            Format validation functions              ]]
--[[======================================================]]
--[[======================================================]]
p.botanistV = function(id)
function p.botanistV(id)
return mw.ustring.match(id,"^[%u%l%d%. '-]+$")
return mw.ustring.match(id,"^[%u%l%d%. '-]+$")
end
end


p.validateIsni = function(id) --Validate ISNI (and ORCID) and retuns it as a 16 characters string or returns false if it's invalid. See http://support.orcid.org/knowledgebase/articles/116780-structure-of-the-orcid-identifier
-- NCDA: NOID Check Digit Algorithm; see [[wikipedia:Check digit#NCDA]]
local ncda -- leave this as a local since NCDA is commonly used among ARK identifiers and could be useful for validating other identifiers later
do -- initialize these constants only once but scope them in a block so local namespace doesn't get cluttered with these
local r29s = [[0123456789bcdfghjkmnpqrstvwxz]] -- radix 29 "betanumeric" digit string
local r29n = r29s:len()
local r29v2d, r29d2v = {}, {}
for i = 1, r29n do
local v, d = i-1, r29s:sub(i, i)
r29v2d[v], r29d2v[d] = d, v
end
function ncda(sid)
local n, sum = sid:len(), 0
for i = 1, n do
sum = sum + i * (r29d2v[sid:sub(i, i)] or 0)
end
return r29v2d[sum % r29n]
end
end
function p.validateBNF(id)
local FRBNF = id:sub(1, -2)
return FRBNF:match('^%d%d%d%d%d%d%d%d$') ~= nil and ncda('cb'..FRBNF) == id:sub(-1) and id
end
 
function p.validateIsni(id) --Validate ISNI (and ORCID) and returns it as a 16 characters string or returns false if it's invalid. See http://support.orcid.org/knowledgebase/articles/116780-structure-of-the-orcid-identifier
id = id:gsub( '[ %-]', '' ):upper()
id = id:gsub( '[ %-]', '' ):upper()
if not id:match( '^%d%d%d%d%d%d%d%d%d%d%d%d%d%d%d[%dX]$' ) then
if not id:match( '^%d%d%d%d%d%d%d%d%d%d%d%d%d%d%d[%dX]$' ) then
Line 32: Line 55:
end
end


local function splitLccn(id)
function p.orcidV(id)
if id:match( '^%l%l?%l?%d%d%d%d%d%d%d%d%d?%d?$' ) then
id = id:gsub( '^(%l+)(%d+)(%d%d%d%d%d%d)$', '%1/%2/%3' )
end
if id:match( '^%l%l?%l?/%d%d%d?%d?/%d+$' ) then
return mw.text.split( id, '/' )
end
return false
end
 
p.lccnV = function(id)
local function append(str, c, length)
while str:len() < length do
str = c..str
end
return str
end
local parts = splitLccn(id) --e.g. n78039510
if not parts then
return false
end
local lccnType = parts[1] ~= 'sh' and 'names' or 'subjects'
return lccnType .. '/' .. parts[1] .. parts[2] .. append( parts[3], '0', 6 )
end
 
p.orcidV = function(id)
id = p.validateIsni(id)
id = p.validateIsni(id)
if not id then
if not id then
Line 65: Line 63:
end
end


p.tlsV = function(id)
function p.tlsV(id)
id = id:gsub(' +', '_')
id = id:gsub(' +', '_')
local idlen = mw.ustring.len(id)
local idlen = mw.ustring.len(id)
Line 76: Line 74:
end
end
return id
return id
end
p.worldcatidV = function(id)
if not id:match( '^viaf%-%d+$' ) and
  not id:match( '^lccn%-n[a-z]?[0-9%-]+$' ) and
  not id:match( '^n[cps]%-.+$' ) then
return false
end
return mw.uri.encode(id, 'PATH')
end
p.zbmathV = function(id)
local ps = {'%l[%l%-]*', '%.%l[%l%-]*', '%.%d*'}
return id:match( '^'..ps[1]..'$' ) -- prefix with no capture options
  or id:match( '^'..ps[1]..ps[2]..'$' ) -- prefix with first capture option
  or id:match( '^'..ps[1]..ps[3]..'$' ) -- prefix with second capture option
  or id:match( '^'..ps[1]..ps[2]..ps[3]..'$' ) -- prefix and both capture options
end
end


Line 98: Line 79:
--[[                Custom link functions                ]]
--[[                Custom link functions                ]]
--[[======================================================]]
--[[======================================================]]
p.ISILlink = function(id,label)
function p.ISILlink(id,label)
if not id:match('^%D%D?%D?%D?%-.+$') then
if not id:match('^%D%D?%D?%D?%-.+$') then
return false
return false
end
end
for _,prefix in ipairs({'AT','AU','BE','CA','CH','DE','FI','FR','IT','KR','NZ','US','ZDB'}) do
for _, prefix in ipairs({'AT', 'AU', 'BE', 'CA', 'CH', 'DE', 'FI', 'FR', 'IT', 'KR', 'NZ', 'US', 'ZDB'}) do
if id:match('^'..prefix..'%-') then
if id:match('^'..prefix..'%-') then
return '<span class="uid">[https://w3id.org/isil/'..id..' ' .. (label or 'ISIL') .. ']</span>'
return '[https://w3id.org/isil/'..id..' ' .. (label or 'ISIL') .. ']'
end
end
end
end
return '[[International Standard Identifier for Libraries and Related Organizations|ISIL]]\n**<span class="uid">' .. id .. '</span>'
return '[[International Standard Identifier for Libraries and Related Organizations|ISIL]]: ' .. id
end
end


p.uscgLink = function(id)
function p.uscgLink(id)
local id2 = id:match( '^[1-7]%-%d%d?%d?%d?%d?$' ) or id:match( '^[1-7]%-%d%d?%d?%d?%d?%.%d*[1-9]$' )
local id2 = id:match( '^[1-7]%-%d%d?%d?%d?%d?$' ) or id:match( '^[1-7]%-%d%d?%d?%d?%d?%.%d*[1-9]$' )
if id2 then
if id2 then
return '<span class="uid">[https://www.navcen.uscg.gov/pdf/lightlists/LightList%20V'..mw.ustring.sub(id2,1,1)..'.pdf '..id2..']</span>'
return '[https://www.navcen.uscg.gov/pdf/lightlists/LightList%20V'..mw.ustring.sub(id2,1,1)..'.pdf '..id2..']'
else
else
return false
return false
Line 119: Line 100:
end
end


p.ccg = function(id)
function p.ccg(id)
id = id:match('^[NAIP]?[1-9]%d*$') or id:match('^[NAIP]?[1-9]%d*%.%d+$')
id = id:match('^[NAIP]?[1-9]%d*$') or id:match('^[NAIP]?[1-9]%d*%.%d+$')
if not id then return false end
if not id then return false end
Line 128: Line 109:
local v
local v
if prefix=='N' then
if prefix=='N' then
local int = interval({1,7.5,14.4,100,121,173,211,235,269.99,326,396,450.1,471.7,499,n=suffix})
local int = interval({1, 7.5, 14.4, 100, 121, 173, 211, 235, 269.99, 326, 396, 450.1, 471.7, 499, n=suffix})
if int=='1-2' then v = 1
if int=='1-2' then v = 1
elseif int=='2-3' then v = 75
elseif int=='2-3' then v = 75
Line 146: Line 127:
link = link .. 'publications/list-lights/newfoundland/n' .. v .. '-en'
link = link .. 'publications/list-lights/newfoundland/n' .. v .. '-en'
elseif prefix=='A' then
elseif prefix=='A' then
local int = interval({5,114.5,145,163,268,271,301.5,327,686.5,704.85,883.2,942,1085,1169.1,1584.5,1773,1823.55,2190,2369,2389,n=suffix})
local int = interval({5, 114.5, 145, 163, 268, 271, 301.5, 327, 686.5, 704.85, 883.2, 942, 1085, 1169.1, 1584.5, 1773, 1823.55, 2190, 2369, 2389, n=suffix})
if int=='1-2' then v = 5
if int=='1-2' then v = 5
elseif int=='2-3' then v = 1145
elseif int=='2-3' then v = 1145
elseif int=='3-4' then v = 145
elseif int=='3-4' then v = 145
elseif int=='4-5' then v = 162
elseif int=='4-5' then v = 162
elseif int=='5-6' then v = 271
elseif int=='5-6' then v = 268
elseif int=='6-7' then v = 3015
elseif int=='6-7' then v = 271
elseif int=='7-8' then v = 327
elseif int=='7-8' then v = 3015
elseif int=='8-9' then v = 6865
elseif int=='8-9' then v = 327
elseif int=='9-10' then v = 7048
elseif int=='9-10' then v = 6865
elseif int=='10-11' then v = 883
elseif int=='10-11' then v = 7048
elseif int=='11-12' then v = 942
elseif int=='11-12' then v = 883
elseif int=='12-13' then v = 1085
elseif int=='12-13' then v = 942
elseif int=='13-14' then v = 11691
elseif int=='13-14' then v = 1085
elseif int=='14-15' then v = 15845
elseif int=='14-15' then v = 11691
elseif int=='15-16' then v = 1773
elseif int=='15-16' then v = 15845
elseif int=='16-17' then v = 182355
elseif int=='16-17' then v = 1773
elseif int=='17-18' then v = 2190
elseif int=='17-18' then v = 182355
elseif int=='18-19' then v = 2369
elseif int=='18-19' then v = 2190
elseif int=='19-20' then v = 2389
elseif int=='19-20' then v = 2369
elseif int=='20-21' then v = 2389
end
end
link = link .. 'publications/list-lights/atl/a' .. v .. '-en'
link = link .. 'publications/list-lights/atl/a' .. v .. '-en'
elseif prefix=='I' then
elseif prefix=='I' then
local int = interval({0.05,401.1,403.4,551.06,552,624,708,731.2,768,814,983,1046,1059.6,1082,1162,1204.7,1233.3,1328,1330,1346.2,1377.8,1408,1410,1420,1445,1470,1520,1534,1540.6,1554,1557.7,1558.8,1563.1,1625.5,1671.7,1716.96,2545,n=suffix})
local int = interval({0.05, 401.1, 403.4, 551.06, 552, 624, 708, 731.2, 768, 814, 983, 1046, 1059.6, 1082, 1162, 1204.7, 1233.3, 1328, 1330, 1346.2, 1377.8, 1408, 1410, 1420, 1445, 1470, 1520, 1534, 1540.6, 1554, 1557.7, 1558.8, 1563.1, 1625.5, 1671.7, 1716.96, 2545, n=suffix})
if int=='1-2' then v = '01'
if int=='1-2' then v = '01'
elseif int=='2-3' then v = 4011
elseif int=='2-3' then v = 4011
Line 214: Line 196:
link = link .. 'list-lights'
link = link .. 'list-lights'
end
end
return '[[CCG (identifier)|CCG]] <span class="uid">[' .. link .. ' ' .. id .. ']</span>'
return '[[CCG (identifier)|CCG]]: [' .. link .. ' ' .. id .. ']'
end
end


return p
return p

Latest revision as of 07:12, 27 August 2024

-- require('strict')
local p = {}
--[[======================================================]]
--[[            Format validation functions               ]]
--[[======================================================]]
function p.botanistV(id)
	return mw.ustring.match(id,"^[%u%l%d%. '-]+$")
end

-- NCDA: NOID Check Digit Algorithm; see [[wikipedia:Check digit#NCDA]]
local ncda -- leave this as a local since NCDA is commonly used among ARK identifiers and could be useful for validating other identifiers later
do -- initialize these constants only once but scope them in a block so local namespace doesn't get cluttered with these
	local r29s = [[0123456789bcdfghjkmnpqrstvwxz]] -- radix 29 "betanumeric" digit string
	local r29n = r29s:len()
	local r29v2d, r29d2v = {}, {}
	for i = 1, r29n do
		local v, d = i-1, r29s:sub(i, i)
		r29v2d[v], r29d2v[d] = d, v
	end
	function ncda(sid)
		local n, sum = sid:len(), 0
		for i = 1, n do
			sum = sum + i * (r29d2v[sid:sub(i, i)] or 0)
		end
		return r29v2d[sum % r29n]
	end
end
function p.validateBNF(id)
	local FRBNF = id:sub(1, -2)
	return FRBNF:match('^%d%d%d%d%d%d%d%d$') ~= nil and ncda('cb'..FRBNF) == id:sub(-1) and id
end

function p.validateIsni(id) --Validate ISNI (and ORCID) and returns it as a 16 characters string or returns false if it's invalid. See http://support.orcid.org/knowledgebase/articles/116780-structure-of-the-orcid-identifier
	id = id:gsub( '[ %-]', '' ):upper()
	if not id:match( '^%d%d%d%d%d%d%d%d%d%d%d%d%d%d%d[%dX]$' ) then
		return false
	end
	local total = 0
	for i = 1, 15 do
		local digit = id:byte( i ) - 48 --Get integer value
		total = (total + digit) * 2
	end
	local remainder = total % 11
	local result = (12 - remainder) % 11
	local checkdigit
	if result == 10 then
		checkdigit = 'X'
	else
		checkdigit=tostring( result )
	end
	if checkdigit ~= string.char( id:byte( 16 ) ) then
		return false
	end
	return id
end

function p.orcidV(id)
	id = p.validateIsni(id)
	if not id then
		return false
	end
	return id:sub( 1, 4 )..'-'..id:sub( 5, 8 )..'-'..id:sub( 9, 12 )..'-'..id:sub( 13, 16 )
end

function p.tlsV(id)
	id = id:gsub(' +', '_')
	local idlen = mw.ustring.len(id)
	if idlen < 4 or idlen > 90 then
		return false
	end
	local regex = '^%u'..string.rep("[%w_',%.%-%(%)%*%/–&]", idlen - 1)..'$'
	if not mw.ustring.match(id,regex ) then
		return false
	end
	return id
end

--[[======================================================]]
--[[                Custom link functions                 ]]
--[[======================================================]]
function p.ISILlink(id,label)
	if not id:match('^%D%D?%D?%D?%-.+$') then
		return false
	end
	for _, prefix in ipairs({'AT', 'AU', 'BE', 'CA', 'CH', 'DE', 'FI', 'FR', 'IT', 'KR', 'NZ', 'US', 'ZDB'}) do
		if id:match('^'..prefix..'%-') then
			return '[https://w3id.org/isil/'..id..' ' .. (label or 'ISIL') .. ']'
		end
	end
	return '[[International Standard Identifier for Libraries and Related Organizations|ISIL]]: ' .. id
end

function p.uscgLink(id)
	local id2 = id:match( '^[1-7]%-%d%d?%d?%d?%d?$' ) or id:match( '^[1-7]%-%d%d?%d?%d?%d?%.%d*[1-9]$' )
	if id2 then
		return '[https://www.navcen.uscg.gov/pdf/lightlists/LightList%20V'..mw.ustring.sub(id2,1,1)..'.pdf '..id2..']'
	else
		return false
	end
end

function p.ccg(id)
	id = id:match('^[NAIP]?[1-9]%d*$') or id:match('^[NAIP]?[1-9]%d*%.%d+$')
	if not id then return false end
	local link = 'https://www.notmar.gc.ca/'
	local prefix = string.sub(id, 1, 1) -- get first character of id
	local suffix = string.sub(id, 2) -- remove first character of id
	local interval = require('Module:Interval')._main
	local v
	if prefix=='N' then
		local int = interval({1, 7.5, 14.4, 100, 121, 173, 211, 235, 269.99, 326, 396, 450.1, 471.7, 499, n=suffix})
		if int=='1-2' then v = 1
		elseif int=='2-3' then v = 75
		elseif int=='3-4' then v = 144
		elseif int=='4-5' then v = 100
		elseif int=='5-6' then v = 121
		elseif int=='6-7' then v = 173
		elseif int=='7-8' then v = 211
		elseif int=='8-9' then v = 235
		elseif int=='9-10' then v = 26999
		elseif int=='10-11' then v = 326
		elseif int=='11-12' then v = 396
		elseif int=='12-13' then v = 4501
		elseif int=='13-14' then v = 4717
		elseif int=='14-15' then v = 499
		end
		link = link .. 'publications/list-lights/newfoundland/n' .. v .. '-en'
	elseif prefix=='A' then
		local int = interval({5, 114.5, 145, 163, 268, 271, 301.5, 327, 686.5, 704.85, 883.2, 942, 1085, 1169.1, 1584.5, 1773, 1823.55, 2190, 2369, 2389, n=suffix})
		if int=='1-2' then v = 5
		elseif int=='2-3' then v = 1145
		elseif int=='3-4' then v = 145
		elseif int=='4-5' then v = 162
		elseif int=='5-6' then v = 268
		elseif int=='6-7' then v = 271
		elseif int=='7-8' then v = 3015
		elseif int=='8-9' then v = 327
		elseif int=='9-10' then v = 6865
		elseif int=='10-11' then v = 7048
		elseif int=='11-12' then v = 883
		elseif int=='12-13' then v = 942
		elseif int=='13-14' then v = 1085
		elseif int=='14-15' then v = 11691
		elseif int=='15-16' then v = 15845
		elseif int=='16-17' then v = 1773
		elseif int=='17-18' then v = 182355
		elseif int=='18-19' then v = 2190
		elseif int=='19-20' then v = 2369
		elseif int=='20-21' then v = 2389
		end
		link = link .. 'publications/list-lights/atl/a' .. v .. '-en'
	elseif prefix=='I' then
		local int = interval({0.05, 401.1, 403.4, 551.06, 552, 624, 708, 731.2, 768, 814, 983, 1046, 1059.6, 1082, 1162, 1204.7, 1233.3, 1328, 1330, 1346.2, 1377.8, 1408, 1410, 1420, 1445, 1470, 1520, 1534, 1540.6, 1554, 1557.7, 1558.8, 1563.1, 1625.5, 1671.7, 1716.96, 2545, n=suffix})
		if int=='1-2' then v = '01'
		elseif int=='2-3' then v = 4011
		elseif int=='3-4' then v = 4034
		elseif int=='4-5' then v = 55106
		elseif int=='5-6' then v = 552
		elseif int=='6-7' then v = 624
		elseif int=='7-8' then v = 708
		elseif int=='8-9' then v = 7312
		elseif int=='9-10' then v = 768
		elseif int=='10-11' then v = 814
		elseif int=='11-12' then v = 983
		elseif int=='12-13' then v = 1046
		elseif int=='13-14' then v = 10596
		elseif int=='14-15' then v = 1082
		elseif int=='15-16' then v = 1162
		elseif int=='16-17' then v = 12047
		elseif int=='17-18' then v = 12333
		elseif int=='18-19' then v = 1328
		elseif int=='19-20' then v = 1330
		elseif int=='20-21' then v = 13462
		elseif int=='21-22' then v = 13778
		elseif int=='22-23' then v = 1408
		elseif int=='23-24' then v = 1410
		elseif int=='24-25' then v = 1420
		elseif int=='25-26' then v = 1445
		elseif int=='26-27' then v = 1470
		elseif int=='27-28' then v = 1520
		elseif int=='28-29' then v = 1534
		elseif int=='29-30' then v = 15406
		elseif int=='30-31' then v = 1554
		elseif int=='31-32' then v = 15577
		elseif int=='32-33' then v = 15588
		elseif int=='33-34' then v = 1562
		elseif int=='34-35' then v = 16255
		elseif int=='35-36' then v = 16717
		elseif int=='36-37' then v = 171696
		elseif int=='37-38' then v = 2545
		end
		link = link .. 'publications/list-lights/inland-waters/i' .. v .. '-en'
	elseif prefix=='P' then
		link = link .. 'publications/list-lights/pac/p'
	else
		link = link .. 'list-lights'
	end
	return '[[CCG (identifier)|CCG]]: [' .. link .. ' ' .. id .. ']'
end

return p