source: TI07-MOLES/trunk/DIF/XQuery/DIF2MOLES.xquery @ 3143

(: Version for one DIF per dgMetadata instance :)
(: Note algoritm for creating non-pre-existing organisations :)
(: dgPersons are not created as one can't tell automatically which are people and which are orgs, and orgs are simpler :)  
import module namespace voclib='http://ndg.nerc.ac.uk/xquery/lib/vocab' at 'xmldb:exist:///db/xqueryLib/Vocabs/vocab_xquery_lib.xquery';
import module namespace inputParse='http://ndg.nerc.ac.uk/xquery/lib/inputParse' at 'xmldb:exist:///db/xqueryLib/Utilities/inputParse_xquery_lib.xquery';
declare default element namespace 'http://ndg.nerc.ac.uk/moles';
declare namespace dif='http://gcmd.gsfc.nasa.gov/Aboutus/xml/dif/';
declare namespace xsi='http://www.w3.org/2001/XMLSchema-instance'; 

(: Replacable parameters :)
declare variable $targetCollection as xs:string {'/db/discovery'};
declare variable $input_repository as xs:string {'Input_Repository_Code'};
declare variable $input_repository_local as xs:string {'Input_Repository_LocalID'};
declare variable $input_entry_id as xs:string {'ndg.noc.soton.ac.uk__DIF__NOCSDAT110'};
declare variable $localIdentifier as xs:string {'ndg.noc.soton.ac.uk__DIF__NOCSDAT110'};

declare variable $output_local_id as xs:string {'Output_LocalID'};
(: Keep
declare variable $targetCollection as xs:string {'TargetCollection'};
declare variable $input_repository as xs:string {'Input_Repository_Code'};
declare variable $input_repository_local as xs:string {'Input_Repository_LocalID'};
declare variable $input_entry_id as xs:string {'Input_Entry_ID'};
declare variable $output_local_id as xs:string {'Output_LocalID'};
:)
for $DIF in collection($targetCollection)/dif:DIF[dif:Entry_ID=$input_entry_id]
return
element dgMetadata {
element dgMetadataRecord {
element dgMetadataID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier {$output_local_id}
else
element localIdentifier {data($DIF/dif:Entry_ID)}
},
element dgMetadataDescription {
element metadataDescriptionID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier {concat('generated_desc-', $output_local_id)}
else
element localIdentifier {concat('generated_desc-', escape-uri(string($DIF/dif:Entry_ID), true()))}
},
element metadataDescriptionLastUpdated {current-date()}                 ,
element abstract {
element abstractText {string($DIF/dif:Summary)}
}
,
for $descOnline in $DIF/dif:Related_URL
return
element descriptionSection {
element descriptionOnlineReference {
element dgSimpleLink {escape-uri(data($descOnline/dif:URL), true())}},
if (exists($descOnline/dif:URL_Content_Type)) then 
element dgReferenceName {data($descOnline/dif:URL_Content_Type)}
else ()
}
},
element name {string($DIF/dif:Entry_Title)},
element abbreviation {string($DIF/dif:Entry_Title)},
element dgDataEntity {
element dgDataSetType {''},
element dgDataSummary {
for $parameter in $DIF/dif:Parameters
return
element dgParameterSummary {
element dgParameterValue {
element dgValueDataParameter {
element Value {''},
element dgStandardUnit {
element dgValidTerm {'dummy'},
element dgValidTermID {
element ParentListID {$voclib:unknown_vocab_id},
element TermID {escape-uri('dummy unit', true())}
}
}
}
},
element dgStdParameterMeasured {
if ($parameter/dif:Category!='') then element dgValidTerm {string($parameter/dif:Category)}
else  element dgValidTerm {'unknown'},
element dgValidTermID {
element ParentListID {concat($voclib:gcmd_science_valids_categories, '/current')},
element TermID {escape-uri($parameter/dif:Category, true())}
},
element dgValidSubterm {
if ($parameter/dif:Topic!='') then element dgValidTerm {string($parameter/dif:Topic)}
else  element dgValidTerm {'unknown'},
element dgValidTermID {
element ParentListID {concat($voclib:gcmd_science_valids_topics, '/current')},
element TermID {escape-uri($parameter/dif:Topic, true())}
},
element dgValidSubterm {
if ($parameter/dif:Term!='') then element dgValidTerm {string($parameter/dif:Term)}
else  element dgValidTerm {'unknown'},
element dgValidTermID {
element ParentListID {concat($voclib:gcmd_science_valids_terms, '/current')},
element TermID {escape-uri($parameter/dif:Term, true())}
},
if (exists($parameter/dif:Variable) and $parameter/dif:Variable!='') then
element dgValidSubterm {
element dgValidTerm {string($parameter/dif:Variable)},
element dgValidTermID {
element ParentListID {concat($voclib:gcmd_science_valids_variables, '/current')},
element TermID {escape-uri($parameter/dif:Variable, true())}
},
if (exists($parameter/dif:Detailed_Variable) and $parameter/dif:Detailed_Variable!='') then
element dgValidSubterm {
element dgValidTerm {string($parameter/dif:Detailed_Variable)},
element dgValidTermID {
element ParentListID {$voclib:unknown_vocab_id},
element TermID {escape-uri($parameter/dif:Detailed_Variable, true())}
},
element ListLevel {4}}
else (),
element ListLevel {3}
}
else (),
element ListLevel {2}
},
element ListLevel {1}
},
element ListLevel {0}
},
element ParameterName {
concat (string($parameter/dif:Category), ' > ', string($parameter/dif:Topic), ' > ', string($parameter/dif:Term)),
if (exists($parameter/dif:Variable) and $parameter/dif:Variable!='') then
concat(' > ', string($parameter/dif:Variable))
else (),
if (exists($parameter/dif:Detailed_Variable) and $parameter/dif:Detailed_Variable!='') then
concat(' > ', string($parameter/dif:Detailed_Variable))
else ()
},
element ParameterAbbreviation {
concat (string($parameter/dif:Category), ' > ', string($parameter/dif:Topic), ' > ', string($parameter/dif:Term)),
if (exists($parameter/dif:Variable) and $parameter/dif:Variable!='') then
concat(' > ', string($parameter/dif:Variable))
else (),
if (exists($parameter/dif:Detailed_Variable) and $parameter/dif:Detailed_Variable!='') then
concat(' > ', string($parameter/dif:Detailed_Variable))
else ()
}
},
if (exists($DIF/dif:Spatial_Coverage) or exists($DIF/dif:Paleo_Temporal_Coverage) or exists($DIF/dif:Location) or exists($DIF/dif:Temporal_Coverage)) then
element dgDataCoverage {
if (exists($DIF/dif:Spatial_Coverage)  or exists($DIF/dif:Location))  then
element dgSpatialCoverage {
for $boundingbox in $DIF/dif:Spatial_Coverage[exists(dif:Northernmost_Latitude)
and exists(dif:Southernmost_Latitude)
and exists(dif:Easternmost_Longitude)
and exists(dif:Westernmost_Longitude)]
return
element BoundingBox {
element LimitNorth {data(inputParse:fix-coord($boundingbox/dif:Northernmost_Latitude))},
element LimitSouth {data(inputParse:fix-coord($boundingbox/dif:Southernmost_Latitude))},
element LimitWest {data(inputParse:fix-coord($boundingbox/dif:Westernmost_Longitude))},
element LimitEast {data(inputParse:fix-coord($boundingbox/dif:Easternmost_Longitude))}
},
for $location in $DIF/dif:Location
return
element dgArea {
element dgValidTerm {string($location)},
element dgValidTermID {
element ParentListID {concat($voclib:gcmd_location_valids, '/current')},
element TermID {escape-uri($location, true())}
}
}
}
else (),
if (exists($DIF/dif:Temporal_Coverage) or exists($DIF/dif:Paleo_Temporal_Coverage) or exists($DIF/dif:Chronostratigraphic_Unit)) then
element dgTemporalCoverage {
for $temporalcoverage in $DIF/dif:Temporal_Coverage[exists(dif:Start_Date)]
return
element DateRange {
element DateRangeStart {string($temporalcoverage/dif:Start_Date)},
element DateRangeEnd {string($temporalcoverage/dif:Stop_Date)}
},
for $paleotemporalcoverage in $DIF/dif:Paleo_Temporal_Coverage[exists(dif:Paleo_Start_Date)]
return
element DateRange {
element DateRangeStart {string($paleotemporalcoverage/dif:Paleo_Start_Date)},
element DateRangeEnd {string($paleotemporalcoverage/dif:Paleo_Stop_Date)}
},
for $chronostratigraphic in $DIF/dif:Chronostratigraphic_Unit
return
element dgChronostratigraphicTerm {
element dgValidTerm {string($chronostratigraphic)},
element dgValidTermID {
element ParentListID {concat($voclib:gcmd_chronostratigraphic_valids, '/current')},
element TermID {escape-uri($chronostratigraphic, true())}
}
}
}
else ()
}
else ()
},
element dgDataRoles {
if (exists($DIF/dif:Data_Set_Citation/dif:Dataset_Creator)) then
element dgDataCreator {
element dgMetadataID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier {concat('generated_creator-', $output_local_id)}
else
element localIdentifier {concat('generated_creator-', escape-uri(string($DIF/dif:Entry_ID), true()))}
},
element roleName {'Data Creator'},
element abbreviation {'Creator'},
for $creatorID in $DIF/dif:Data_Set_Citation/dif:Dataset_Creator
return
element dgRoleHolder {
element dgMetadataID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier  {escape-uri(concat('generated_orgcit-', string($creatorID), '-', $output_local_id), true())}
else
element localIdentifier  {escape-uri(concat('generated_orgcit-', string($creatorID), '-', data($DIF/dif:Entry_ID)), true())}
},
element startDate {current-date()}
}
}
else if (exists($DIF/dif:Originating_Center)) then
element dgDataCreator {
element dgMetadataID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier  {escape-uri(concat('generated_creator-', $output_local_id), true())}
else
element localIdentifier  {escape-uri(concat('generated_creator-', data($DIF/dif:Entry_ID)), true())}
},
element roleName {'Data Creator'},
element abbreviation {'Creator'},
for $creatorID in $DIF/dif:Originating_Center
return
element dgRoleHolder {
element dgOrganisationID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier  {escape-uri(concat('generated_orgcit-', string($creatorID), '-', $output_local_id), true())}
else
element localIdentifier  {escape-uri(concat('generated_orgcit-', string($creatorID), '-', data($DIF/dif:Entry_ID)), true())}
},
element startDate {current-date()}
}
}
else (),
element dgDataCurator {
element dgMetadataID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier  {escape-uri(concat('generated_curator-', $output_local_id), true())}
else
element localIdentifier  {escape-uri(concat('generated_curator-', data($DIF/dif:Entry_ID)), true())}
},
element roleName {'Data Curator'},
element abbreviation {'Curator'},
element dgRoleHolder {
element dgOrganisationID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
element localIdentifier {$input_repository_local}
},
element startDate {current-date()}
}
}
}
},
element dgStructuredKeyword {
element dgValidTerm {'d2b converted record'},
element dgValidTermID {
element ParentListID {$voclib:unknown_vocab_id},
element TermID {'d2b'}
}
},
for $structuredKeywords in $DIF/dif:Keyword
return 
element dgStructuredKeyword {
element dgValidTerm {string($structuredKeywords)},
element dgValidTermID {
if ($structuredKeywords='MDIP' or $structuredKeywords='NERC' or $structuredKeywords='NERC_DDC' or $structuredKeywords='DDDP') then
element ParentListID {concat($voclib:ndg_data_provider_vocab, '/current')}
else
element ParentListID {$voclib:unknown_vocab_id}
,
element TermID {escape-uri($structuredKeywords, true())}
}
},
for $structuredKeywords in $DIF/dif:ISO_Topic_Category
return 
element dgStructuredKeyword {
element dgValidTerm {string($structuredKeywords)},
element dgValidTermID {
element ParentListID {concat($voclib:iso_topic_list, '/current')},
element TermID {escape-uri($structuredKeywords, true())}
}
},
if (exists($DIF/dif:DIF_Creation_Date) or exists($DIF/dif:Last_DIF_Revision_Date)) then 
element dgMetadataProvenance {
if (exists($DIF/dif:DIF_Creation_Date)) then
element RecordCreation {
element CreatedDate {
if (string($DIF/dif:DIF_Creation_Date) castable as xs:date) then 
string($DIF/dif:DIF_Creation_Date) cast as xs:date
else (current-date())
},
element CreatedBy {$input_repository}
}
else 
element RecordCreation {
element CreatedDate {current-date()},
element CreatedBy {'MOLES Import'}
},
if (exists($DIF/dif:Last_DIF_Revision_Date)) then
element RecordUpdate {
element UpdateDate {
if (string($DIF/dif:Last_DIF_Revision_Date) castable as xs:dateTime) then 
string($DIF/dif:Last_DIF_Revision_Date) cast as xs:dateTime
else (current-dateTime())
},
element UpdatedBy {$input_repository}
}
else ()
}
else ()
} (: </dgMetadataRecord>:),
for $creator in distinct-values($DIF/(dif:Data_Set_Citation/dif:Dataset_Creator | dif:Originating_Center))
return
element dgOrganisation {
element dgMetadataID {
element schemeIdentifier {'NDG-B0'},
element repositoryIdentifier {$input_repository},
if ($output_local_id != 'Output_LocalID') then
element localIdentifier  {escape-uri(concat('generated_orgcit-', string($creator), '-', $output_local_id), true())}
else
element localIdentifier  {escape-uri(concat('generated_orgcit-', string($creator), '-', data($DIF/dif:Entry_ID)), true())}
},
element name {string($creator)},
element abbreviation {string($creator)},
element contactDetails {''}
}
} (:    </dgMetadata> :)