Metacat

-- use default values initially

text('testing') as ezid_account,

doc.doctype,

id.guid,

text(null) as publisher,

where guid like 'doi%'

and id.rev = doc.rev

select

text('testing') as ezid_account,

doc.doctype,

id.guid,

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 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

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';

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';

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 parent.nodename = 'publisher'

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

update doi_registration doi

set publisher = 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 = 'publisher'

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

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 = 'publisher'

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

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');

-- creator/individualName/surName

-- creator/individualName/givenName

update doi_registration doi

xml_documents docs,

xml_nodes nodes,

xml_nodes parent,

xml_nodes child,

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'

update doi_registration doi

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 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 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 parent.nodename = 'publisher'

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

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,

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 = 'publisher'

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

--include the first name (careful not to rerun this part)

update doi_registration doi

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 = 'publisher'

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

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 doi_registration doi

xml_revisions docs,

xml_nodes_revisions nodes,

xml_nodes_revisions parent,

xml_nodes_revisions child,

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'

update doi_registration doi

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 = '';

update doi_registration

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

update doi_registration

--16988

update doi_registration

--98078

update doi_registration

--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.rev = doc.rev;

-- 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.rev = rev.rev;

--update resourceType

update doi_registration

update doi_registration

update doi_registration doi

set title = 'Data file - ' || doc.docname

from identifier id, xml_documents doc

where doi.guid = id.guid

and id.docid = doc.docid

and id.rev = doc.rev

-- update the data file titles from revisions

update doi_registration doi

set title = 'Data file - ' || doc.docname

from identifier id, xml_revisions doc

where doi.guid = id.guid

and id.docid = doc.docid

and id.rev = doc.rev

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 obsoletedby

update doi_registration doi

set obsoletedBy = newer.guid

from identifier id, identifier newer

where doi.guid = id.guid

and id.docid = newer.docid

and newer.rev = (select min(next.rev) from identifier next where next.docid = id.docid and next.rev > id.rev);

-- update the obsolets

update doi_registration doi

set obsoletes = older.guid

from identifier id, identifier older

where doi.guid = id.guid

and id.docid = older.docid

and older.rev = (select max(prev.rev) from identifier prev where prev.docid = id.docid and prev.rev < id.rev);

/**

select doi.guid, older.guid as obsoletes

from doi_registration doi, identifier id, identifier older

where doi.guid = id.guid

and id.docid = older.docid

and older.rev = (select max(prev.rev) from identifier prev where prev.docid = id.docid and prev.rev < id.rev)

and doi.guid like 'doi:10.5063/AA/ABS.4%';

**/

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 id.guid in (select guid from doi_registration);

INSERT INTO ecogrid_docids

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

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

where 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 = 'url'

and parent.nodename = 'online'

and child.nodedata like 'ecogrid%'

and child.nodetype = 'TEXT'

and id.guid in (select guid from doi_registration);

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 doi_registration doi

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

from ecogrid_docids eco

where doi.guid = eco.guid;

from doi_registration

where resourceMapId is null;

update doi_registration

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

where resourceMapId is not null;

update doi_registration doi

set creator = user_owner

from identifier id, xml_documents docs

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev= docs.rev

update doi_registration doi

set publisher = user_owner

from identifier id, xml_documents docs

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev= docs.rev

update doi_registration doi

set creator = user_owner

from identifier id, xml_revisions docs

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev= docs.rev

update doi_registration doi

set publisher = user_owner

from identifier id, xml_revisions docs

where doi.guid = id.guid

and id.docid = docs.docid

and id.rev= docs.rev

update doi_registration

set publisher = null

where publisher = 'document';

case when givenName is null then

	sn

else

update doi_registration

case when givenName is null then

	sn

else

select creator, count(*) as cnt

from doi_registration

where creator like 'uid=%'

group by creator

order by cnt desc;

update doi_registration doi

set creator = meta.creator,

publisher = meta.publisher

from doi_registration meta, identifier id, ecogrid_docids eco

where doi.guid = id.guid

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

and eco.guid = meta.guid;

update doi_registration

set creator = 'Kellogg Biological Station'

where (creator is null or trim(creator) = '')

and guid like '%knb-lter-kbs%';

update doi_registration

set publisher = 'Kellogg Biological Station'

where (publisher is null or trim(publisher) = '')

and guid like '%knb-lter-kbs%';