Metacat

-- use default values initially

CREATE TABLE doi_registration AS

select 

text('testing') as ezid_account,

doc.doctype,

id.guid, 

'https://cn.dataone.org/cn/v1/resolve/' || regexp_replace(regexp_replace(id.guid, '/', '%2F', 'g'), ':', '%3A', 'g') as url, 

text('Unknown') as title, 

text(null) as creator,

text(null) as publisher,

to_char(doc.date_created, 'YYYY') as pub_date

from identifier id, xml_documents doc

where guid like 'doi%'

and id.docid = doc.docid

and id.rev = doc.rev

UNION ALL

select 

text('testing') as ezid_account,

doc.doctype,

id.guid, 

'https://cn.dataone.org/cn/v1/resolve/' || regexp_replace(regexp_replace(id.guid, '/', '%2F', 'g'), ':', '%3A', 'g') as url, 

text('Unknown') as title, 

text(null) as creator,

text(null) as publisher,

to_char(doc.date_created, 'YYYY') as pub_date

from identifier id, xml_revisions doc

where guid like 'doi%'

and id.docid = doc.docid

and id.rev = doc.rev;

--update defaults

update doi_registration

set title = 'Data file'

where doctype = 'BIN';

update doi_registration

set title = 'Legacy EML file'

where doctype like '%eml%2.0.0beta%';

-- update title using node information

update doi_registration doi

set title = child.nodedata

from identifier id, xml_documents docs, xml_nodes nodes, xml_nodes parent, xml_nodes child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'title'

and parent.nodename = 'dataset';

-- update pubDate using nodes

update doi_registration doi

set pub_date = child.nodedata

from identifier id, xml_documents docs, xml_nodes nodes, xml_nodes parent, xml_nodes child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'pubDate'

and parent.nodename = 'dataset';

-- update publisher using nodes

update doi_registration doi

set publisher = child.nodedata

from identifier id, xml_documents docs, xml_nodes nodes, xml_nodes parent, xml_nodes child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'publisher'

and parent.nodename = 'dataset';

-- update organization using nodes

update doi_registration doi

set creator = child.nodedata

from identifier id, xml_documents docs, xml_nodes nodes, xml_nodes parent, xml_nodes child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'organizationName'

and parent.nodename = 'creator'

and nodes.nodeid = (select min(thisnode.nodeid) from xml_nodes thisnode where thisnode.parentnodeid = parent.nodeid and nodetype = 'ELEMENT');

-- update creator using nodes

-- creator/individualName/surName

-- creator/individualName/givenName

update doi_registration doi

set creator = child.nodedata

from identifier id, 

xml_documents docs, 

xml_nodes nodes, 

xml_nodes parent, 

xml_nodes child, 

xml_nodes grandparent

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and parent.parentnodeid = grandparent.nodeid

and nodes.nodename = 'surName'

and parent.nodename = 'individualName'

and grandparent.nodename = 'creator'

and parent.nodeid = (select min(thisnode.nodeid) from xml_nodes thisnode where thisnode.parentnodeid = grandparent.nodeid and nodetype = 'ELEMENT');

-- add the first name if we have it

update doi_registration doi

set creator = child.nodedata ||' '|| creator

from identifier id, 

xml_documents docs, 

xml_nodes nodes, 

xml_nodes parent, 

xml_nodes child, 

xml_nodes grandparent

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and parent.parentnodeid = grandparent.nodeid

and nodes.nodename = 'givenName'

and parent.nodename = 'individualName'

and grandparent.nodename = 'creator'

and parent.nodeid = (select min(thisnode.nodeid) from xml_nodes thisnode where thisnode.parentnodeid = grandparent.nodeid and nodetype = 'ELEMENT');

-- update title using revisions

update doi_registration doi

set title = child.nodedata

from identifier id, xml_revisions docs, xml_nodes_revisions nodes, xml_nodes_revisions parent, xml_nodes_revisions child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'title'

and parent.nodename = 'dataset';

-- update pubDate using revisions

update doi_registration doi

set pub_date = child.nodedata

from identifier id, xml_revisions docs, xml_nodes_revisions nodes, xml_nodes_revisions parent, xml_nodes_revisions child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'pubDate'

and parent.nodename = 'dataset';

-- update publisher using revisions

update doi_registration doi

set publisher = child.nodedata

from identifier id, xml_revisions docs, xml_nodes_revisions nodes, xml_nodes_revisions parent, xml_nodes_revisions child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'publisher'

and parent.nodename = 'dataset';

-- update organization using nodes

update doi_registration doi

set creator = child.nodedata

from identifier id, xml_revisions docs, xml_nodes_revisions nodes, xml_nodes_revisions parent, xml_nodes_revisions child

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and nodes.nodename = 'organizationName'

and parent.nodename = 'creator'

and nodes.nodeid = (select min(thisnode.nodeid) from xml_nodes_revisions thisnode where thisnode.parentnodeid = parent.nodeid and nodetype = 'ELEMENT');

-- update creator using revisions

update doi_registration doi

set creator = child.nodedata

from identifier id, 

xml_revisions docs, 

xml_nodes_revisions nodes, 

xml_nodes_revisions parent, 

xml_nodes_revisions child, 

xml_nodes_revisions grandparent

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and parent.parentnodeid = grandparent.nodeid

and nodes.nodename = 'surName'

and parent.nodename = 'individualName'

and grandparent.nodename = 'creator'

and parent.nodeid = (select min(thisnode.nodeid) from xml_nodes_revisions thisnode where thisnode.parentnodeid = grandparent.nodeid and nodetype = 'ELEMENT');

-- add the first name if we have it

update doi_registration doi

set creator = child.nodedata ||' '|| creator

from identifier id, 

xml_revisions docs, 

xml_nodes_revisions nodes, 

xml_nodes_revisions parent, 

xml_nodes_revisions child, 

xml_nodes_revisions grandparent

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev = docs.rev

and docs.rootnodeid = nodes.rootnodeid

and nodes.nodeid = child.parentnodeid

and nodes.parentnodeid = parent.nodeid

and parent.parentnodeid = grandparent.nodeid

and nodes.nodename = 'givenName'

and parent.nodename = 'individualName'

and grandparent.nodename = 'creator'

and parent.nodeid = (select min(thisnode.nodeid) from xml_nodes_revisions thisnode where thisnode.parentnodeid = grandparent.nodeid and nodetype = 'ELEMENT');

-- use xml_documents for defaults that are still missing 

update doi_registration doi

set creator = doc.user_owner

from xml_documents doc, identifier id

where doi.guid = id.guid

and id.docid = doc.docid

and id.rev = doc.rev

and doi.creator is null;

update doi_registration doi

set creator = doc.user_owner

from xml_revisions doc, identifier id

where doi.guid = id.guid

and id.docid = doc.docid

and id.rev = doc.rev

and doi.creator is null;

-- set publisher

update doi_registration

set publisher = creator

where publisher is null or publisher = '';

-- clean up

update doi_registration

set title = replace(title, E'\n', ' ');

update doi_registration

set title = regexp_replace(title, E'\\s+', ' ', 'g');

update doi_registration

set title = regexp_replace(title, E'^\\s+', '', 'g');

update doi_registration

set ezid_account = 'KNB'

where guid like 'doi:10.5063%';

--16988

update doi_registration

set ezid_account = 'PISCO'

where guid like 'doi:10.6085%';

--98078

update doi_registration

set ezid_account = 'LTER'

where guid like 'doi:10.6073%';

--49482

/**

 * Additional modifications, July 5th, 2012

 */

ALTER TABLE doi_registration

ADD COLUMN resourceType text,

ADD COLUMN objectFormat text,

ADD COLUMN obsoletedBy text,

ADD COLUMN obsoletes text,

ADD COLUMN resourceMapId text,

ADD COLUMN resourceMapLocation text,

ADD COLUMN access text;

-- update access values

update doi_registration

set access = 'protected';

update doi_registration doi

set access = 'public'

from xml_access a

where doi.guid = a.guid

and a.principal_name = 'public'

and a.permission >= '4'

and a.perm_type = 'allow';

-- update the objectFormat from xml_documents

update doi_registration doi

set objectFormat = doc.doctype

from identifier id, xml_documents doc

where doi.guid = id.guid

and id.docid = doc.docid;

-- update the objectFormat from xml_revisions

update doi_registration doi

set objectFormat = rev.doctype

from identifier id, xml_revisions rev

where doi.guid = id.guid

and id.docid = rev.docid;

--update resourceType

update doi_registration

set resourceType = 'Dataset/METADATA';

update doi_registration

set resourceType = 'Dataset/DATA'

where objectFormat = 'BIN';

-- update the objectFormat from SM table (will be subset)

update doi_registration doi

set objectFormat = sm.object_format

from systemMetadata sm

where doi.guid = sm.guid;

--16938

--update revision history from SM

update doi_registration doi

set obsoletes = sm.obsoletes,

obsoletedBy = sm.obsoleted_by

from systemMetadata sm

where doi.guid = sm.guid;

--update resourceMap pointer TBD

-- NOTE: some of these maps might not actually exist

update doi_registration doi

set resourceMapId = 'resourceMap_'||id.docid||'.'||id.rev

from identifier id

where doi.guid = id.guid

and id.docid = rev.docid

and objectFormat like 'eml%';

--update the resourcemapid for described data

CREATE TABLE ecogrid_docids AS

select distinct id.docid, id.rev, id.guid, substring(xpi.nodedata from 'ecogrid://knb/(.*)$') as data_docid

from xml_path_index xpi, identifier id, xml_documents xmld

where xpi.docid = xmld.docid 

and xmld.docid = id.docid

and xmld.rev = id.rev

and xpi.path like 'dataset/%/physical/distribution/online/url' 

and xpi.nodedata like 'ecogrid%';

-- TODO: include revisions for described data

-- NOTE: some of thee maps might not actually exist

update doi_registration doi

set resourceMapId = 'resourceMap_'||eco.docid||'.'||eco.rev

from identifier id, ecogrid_docids eco

where doi.guid = id.guid

and id.docid||'.'||id.rev = eco.data_docid;

--update the resource map location

update doi_registration

set resourceMapLocation = 'https://cn.dataone.org/cn/v1/resolve/' || resourceMapId

where resourceMapId is not null;

--update creator usinig LDAP lookup

CREATE TABLE ecoinfo_dn (dn text, givenName text, sn text);

COPY ecoinfo_dn FROM '/tmp/ecoinfo_dn.csv' WITH CSV HEADER;

update doi_registration

set creator = givenName|| ' ' || sn

from ecoinfo_dn

where creator = dn;

update doi_registration

set publisher = givenName|| ' ' || sn

from ecoinfo_dn

where publisher = dn;

select creator, count(*) as cnt

from doi_registration

where creator like 'uid=%'

group by creator

order by cnt desc;

-- copy to the external file

COPY 

(select ezid_account, 

guid as dc_identifier, 

url as datacite_url, 

title as dc_title, 

creator as dc_creator, 

publisher as dc_publisher, 

pub_date as datacite_publicationYear,

resourceType,

objectFormat,

obsoletedBy,

obsoletes,

resourceMapId,

resourceMapLocation,

access

from doi_registration 

order by dc_identifier) 

TO '/tmp/doi_registration.csv'

WITH CSV HEADER;

--164548

--drop table doi_registration;

--drop table ecogrid_docids;

--drop table ecoinfo_dn;