Project

General

Profile

« Previous | Next » 

Revision 36723

[maven-release-plugin] copy for tag dnet-openaireplus-profiles-1.0.4

View differences:

modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/deploy.info
1
{"type_source": "SVN", "goal": "package -U -T 4C source:jar", "url": "http://svn-public.driver.research-infrastructures.eu/driver/dnet40/modules/dnet-openaireplus-profiles/trunk/", "deploy_repository": "dnet4-snapshots", "version": "4", "mail": "sandro.labruzzo@isti.cnr.it,michele.artini@isti.cnr.it, claudio.atzori@isti.cnr.it, alessia.bardi@isti.cnr.it", "deploy_repository_url": "http://maven.research-infrastructures.eu/nexus/content/repositories/dnet4-snapshots", "name": "dnet-openaireplus-profiles"}
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/dedupGrouperJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="6b2d8db3-346f-4ddc-8591-39fd488c1191_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="dedupGrouperJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that closes the similarity mesh</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13
        	
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />		        	
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupGrouperMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put" />
22
				
23
				<!-- MISC -->				
24
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
25
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
26
				<PROPERTY key="mapreduce.map.speculative" value="false" />
27
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />	
28
			
29
				<PROPERTY key="mapred.reduce.tasks" value="0" />
30
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
31
				
32
		<!--  	Uncomment to override the default lib path -->			
33
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
34
        	</STATIC_CONFIGURATION>
35
        	<JOB_INTERFACE>
36
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
37
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
38
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
39
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />     
40
        	</JOB_INTERFACE>
41
        	<SCAN>
42
        		<FILTERS operator="MUST_PASS_ALL">
43
        			<FILTER type="prefix" param="entityTypeId" />
44
        		</FILTERS>
45
        		<FAMILIES>
46
        			<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
47
        			<FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
48
     				<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
49
        		</FAMILIES>
50
        	</SCAN>
51
        </HADOOP_JOB>
52
        <STATUS>
53
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
54
            <RUNNING_INSTANCES value="0"/>
55
            <CUMULATIVE_RUN value="0" />
56
        </STATUS>
57
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
58
    </BODY>
59
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/iisPreprocessingQuickJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="13beed98-81bf-4fbd-ab4f-de071177997c_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
    	<HADOOP_JOB name="iisPreprocessingQuickJob" type="oozie">
11
        	<DESCRIPTION>IIS preprocessing</DESCRIPTION>
12
            <STATIC_CONFIGURATION>
13
				<!-- Cluster wide -->
14
                <PROPERTY key="queueName" value="default"/>
15
				<PROPERTY key="user.name" value="dnet" />
16

  
17
				<!-- Runtime -->
18
                <PROPERTY key="workingDir" value="/tmp/integration/working_dir/preprocessing_quick_test"/>
19
                <PROPERTY key="oozie.wf.validate.ForkJoin" value="false"/>
20
                <PROPERTY key="input_referenceextraction_project" value="/user/marek.horst/share/referenceextraction/document_projects/2014-04-11"/>
21
                <PROPERTY key="input_referenceextraction_dataset" value="/user/marek.horst/share/referenceextraction/document_datasets/all/2014-04-11"/>
22
                <PROPERTY key="export_action_hbase_table_initialize" value="false"/>
23
            </STATIC_CONFIGURATION>
24
        	<JOB_INTERFACE>
25
 		       	<PARAM name="import_content_object_store_location" required="true" description="mdstore service location" />
26
        		<PARAM name="import_mdstore_service_location" required="true" description="mdstore service location" />
27
        		<PARAM name="import_dataset_mdstore_ids_csv" required="true" description="mdstore id for dataset records" />
28
        		<PARAM name="import_wos_mdstore_id" required="true" description="mdstore id for WoS records" />
29
        		<PARAM name="import_database_service_location" required="true" description="database service endpoint" />
30
        		<PARAM name="import_content_datacite_objectstores_csv" required="true" description="objecstore ids subject to dataset reference extraction" />
31
        		<PARAM name="import_content_wos_plaintext_objectstores_csv" required="true" description="objecstore ids for WoS plaintext" />
32
	      		<PARAM name="export_action_hbase_table_name" required="true" description="destination action manager table" />
33
        		<PARAM name="export_action_hbase_remote_zookeeper_quorum" required="true" description="ZK quorum" />
34
        		<PARAM name="export_action_hbase_remote_zookeeper_clientport" required="true" description="ZK port" />
35
        		<PARAM name="nameNode" required="true" description="hdfs name node" />
36
        		<PARAM name="jobTracker" required="true" description="job tracker name" />
37
        		<PARAM name="oozie.wf.application.path" required="true" description="oozie job application absolute path" />
38
       			<PARAM name="export_action_set_id_document_referencedProjects" required="true" description="target action set for project references" />
39
       			<PARAM name="export_action_set_id_document_referencedDatasets" required="true" description="target action set for dataset references" />
40
       			<PARAM name="export_action_set_id_entity_wos" required="true" description="target action set for WoS entities" />
41
     			<PARAM name="export_action_set_id_entity_dataset" required="true" description="target action set for dataset entities" />
42
        	</JOB_INTERFACE>
43
        </HADOOP_JOB>
44
        <STATUS>
45
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
46
            <RUNNING_INSTANCES value="0"/>
47
            <CUMULATIVE_RUN value="0" />
48
        </STATUS>
49
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
50
    </BODY>
51
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/dedupBuildRootsJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="895ce6a9-4131-4954-b9ed-949ff78f5448_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="dedupBuildRootsJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that build the roots and redirects the rels</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupBuildRootsMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupBuildRootsReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Writable" />				
27
				
28
				<!-- MISC -->				
29
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
30
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
31
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
32
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />				
33
			
34
				<PROPERTY key="mapred.reduce.tasks" value="500" />
35
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
36
	
37
		<!--  	Uncomment to override the default lib path -->			
38
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
39
        	</STATIC_CONFIGURATION>
40
        	<JOB_INTERFACE>
41
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
42
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
43
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
44
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />     
45
        	</JOB_INTERFACE>
46
        	<SCAN caching="10">
47
        		<FILTERS operator="MUST_PASS_ALL">
48
        			<FILTER type="prefix" param="entityTypeId" />
49
        		</FILTERS>
50
        		<FAMILIES/>        		
51
        	</SCAN>
52
        </HADOOP_JOB>
53
        <STATUS>
54
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
55
            <RUNNING_INSTANCES value="0"/>
56
            <CUMULATIVE_RUN value="0" />
57
        </STATUS>
58
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
59
    </BODY>
60
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/dedupPersonJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="29638605-235b-4cc1-9bf5-a5dd2fc84915_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="dedupPersonJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that scans a given entity type and creates the similarRel graph</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT  -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupPersonMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupPersonReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Writable" />				
27
				
28
				<!-- MISC -->				
29
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
30
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
31
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
32
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />				
33
			
34
				<PROPERTY key="mapred.reduce.tasks" value="1000" />
35
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
36
				
37
		<!--  	Uncomment to override the default lib path -->			
38
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
39
        	</STATIC_CONFIGURATION>
40
        	<JOB_INTERFACE>
41
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
42
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
43
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
44
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />     
45
        	</JOB_INTERFACE>
46
        	<SCAN>
47
        		<FILTERS operator="MUST_PASS_ALL">
48
        			<FILTER type="prefix" param="entityTypeId" />
49
        		</FILTERS>
50
        		<FAMILIES>
51
        			<FAMILY param="entityType" />
52
        		</FAMILIES>
53
        	</SCAN>
54
        </HADOOP_JOB>
55
        <STATUS>
56
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
57
            <RUNNING_INSTANCES value="0"/>
58
            <CUMULATIVE_RUN value="0" />
59
        </STATUS>
60
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
61
    </BODY>
62
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/iisPreprocessingJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="732656e3-5ac6-4344-9d1f-f5c805f53a06_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="iisPreprocessingJob" type="oozie">
11
 			<DESCRIPTION>IIS preprocessing</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13
				<!-- Cluster wide -->
14
                <PROPERTY key="queueName" value="default"/>
15
				<PROPERTY key="user.name" value="dnet.beta" />
16

  
17
				<!-- Runtime -->
18
				<PROPERTY key="workingDir" value="/tmp/integration/working_dir/preprocessing"  />
19
				<PROPERTY key="oozie.wf.application.path" value="/tmp/integration/apps/preprocessing"  />
20
				<PROPERTY key="oozie.wf.validate.ForkJoin" value="false"  />
21
				<PROPERTY key="export_action_hbase_table_initialize" value="false"/>
22
				<PROPERTY key="metadataextraction_default_cache_location" value="/cache/metadataextraction"/>
23
				<PROPERTY key="metadataextraction_excluded_checksums" value="1e5b574109da731f4918c7f91fc24864,bea4728578070c3d66774bf9454d41fe,da458477233b5561ae47042aa2a73086"/>
24
        	</STATIC_CONFIGURATION>
25
        	<JOB_INTERFACE>
26
 		       	<PARAM name="import_content_object_store_location" required="true" description="mdstore service location" />
27
        		<PARAM name="import_mdstore_service_location" required="true" description="mdstore service location" />
28
        		<PARAM name="import_dataset_mdstore_ids_csv" required="true" description="mdstore id for dataset records" />
29
        		<PARAM name="import_wos_mdstore_id" required="true" description="mdstore id for WoS records" />
30
        		<PARAM name="import_database_service_location" required="true" description="database service endpoint" />
31
        		<PARAM name="import_content_datacite_objectstores_csv" required="true" description="objecstore ids subject to dataset reference extraction" />
32
        		<PARAM name="import_content_wos_plaintext_objectstores_csv" required="true" description="objecstore ids for WoS plaintext" />
33
	      		<PARAM name="export_action_hbase_table_name" required="true" description="destination action manager table" />
34
        		<PARAM name="export_action_hbase_remote_zookeeper_quorum" required="true" description="ZK quorum" />
35
        		<PARAM name="export_action_hbase_remote_zookeeper_clientport" required="true" description="ZK port" />
36
        		<PARAM name="nameNode" required="true" description="hdfs name node" />
37
        		<PARAM name="jobTracker" required="true" description="job tracker name" />
38
        		<PARAM name="oozie.wf.application.path" required="true" description="oozie job application absolute path" />
39
       			<PARAM name="export_action_set_id_document_referencedProjects" required="true" description="target action set for project references" />
40
       			<PARAM name="export_action_set_id_document_referencedDatasets" required="true" description="target action set for dataset references" />
41
       			<PARAM name="export_action_set_id_entity_wos" required="true" description="target action set for WoS entities" />
42
     			<PARAM name="export_action_set_id_entity_dataset" required="true" description="target action set for dataset entities" />
43
        	</JOB_INTERFACE>
44
        </HADOOP_JOB>
45
        <STATUS>
46
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
47
            <RUNNING_INSTANCES value="0"/>
48
            <CUMULATIVE_RUN value="0" />
49
        </STATUS>
50
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
51
    </BODY>
52
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/dedupFindRootsJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="61f9270e-ffc1-4095-9f76-3852e4d227fb_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="dedupFindRootsJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that find the root of a similarity group</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupFindRootsMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put" />
22
				
23
				<!-- MISC -->				
24
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
25
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
26
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
27
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />				
28
			
29
				<PROPERTY key="mapred.reduce.tasks" value="0" />
30
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
31
				
32
		<!--  	Uncomment to override the default lib path -->			
33
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
34
        	</STATIC_CONFIGURATION>
35
        	<JOB_INTERFACE>
36
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
37
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
38
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
39
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />     
40
        	</JOB_INTERFACE>
41
        	<SCAN>
42
        		<FILTERS operator="MUST_PASS_ALL">
43
        			<FILTER type="prefix" param="entityTypeId" />
44
        		</FILTERS>
45
        		<FAMILIES>
46
        			<FAMILY param="entityType" />
47
        			<FAMILY value="resultResult_dedupSimilarity_isSimilarTo" />
48
        			<FAMILY value="personPerson_dedupSimilarity_isSimilarTo" />
49
     				<FAMILY value="organizationOrganization_dedupSimilarity_isSimilarTo" />
50
        		</FAMILIES>
51
        	</SCAN>
52
        </HADOOP_JOB>
53
        <STATUS>
54
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
55
            <RUNNING_INSTANCES value="0"/>
56
            <CUMULATIVE_RUN value="0" />
57
        </STATUS>
58
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
59
    </BODY>
60
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/iisMainJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="f6e4cbcd-b173-4f8d-9205-e64ba15f03ad_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="iisMainJob" type="oozie">
11
            <DESCRIPTION>IIS main workflow</DESCRIPTION>
12
            <STATIC_CONFIGURATION>
13
        	
14
				<!-- Cluster wide -->
15
                <PROPERTY key="queueName" value="default"/>
16
                <PROPERTY key="user.name" value="dnet.beta"/>
17

  
18
				<!-- Runtime -->
19
				<PROPERTY key="active_existence_filter" value="true"/>
20
                <PROPERTY key="import_hbase_approved_datasources_csv" value="$UNDEFINED$"/>
21
                <PROPERTY key="import_hbase_input_table" value="information_space-2014-11-05"/>
22
   				<PROPERTY key="metadataextraction_default_cache_location" value="/cache/metadataextraction"/>
23
                <PROPERTY key="export_action_hbase_table_initialize" value="true"/>
24
                <PROPERTY key="oozie.wf.validate.ForkJoin" value="false"/>
25
                <PROPERTY key="workingDir" value="/tmp/integration/working_dir/main"/>
26
				<PROPERTY key="metadataextraction_excluded_checksums" value="1e5b574109da731f4918c7f91fc24864,bea4728578070c3d66774bf9454d41fe,da458477233b5561ae47042aa2a73086"/>
27
            </STATIC_CONFIGURATION>
28
        	<JOB_INTERFACE>
29
        		<PARAM name="import_content_object_store_location" required="true" description="objectStore service endpoint" />
30
        		<PARAM name="import_content_objectstores_csv" required="true" description="csv list of the available object stores subject to processing" />
31
        		<PARAM name="import_mdstore_service_location" required="true" description="mdstore service location" />
32
        		<PARAM name="import_dataset_mdstore_ids_csv" required="true" description="mdstore ids for dataset records" />
33
	      		<PARAM name="export_action_hbase_table_name" required="true" description="destination action manager table" />
34
        		<PARAM name="export_action_hbase_remote_zookeeper_quorum" required="true" description="ZK quorum" />
35
        		<PARAM name="export_action_hbase_remote_zookeeper_clientport" required="true" description="ZK port" />
36
        		<PARAM name="nameNode" required="true" description="hdfs name node" />
37
        		<PARAM name="jobTracker" required="true" description="job tracker name" />
38
        		<PARAM name="oozie.wf.application.path" required="true" description="oozie job application absolute path" />
39
                <PARAM description="target action set for referenced projects" name="export_action_set_id_document_referencedProjects" required="true"/>
40
                <PARAM description="target action set for referenced datasets" name="export_action_set_id_document_referencedDatasets" required="true"/>
41
                <PARAM description="target action set for research initiative outcome" name="export_action_set_id_document_research_initiative" required="true"/>
42
                <PARAM description="target action set for document similarities" name="export_action_set_id_document_similarities_standard" required="true"/>
43
                <PARAM description="target action set for document statistics" name="export_action_set_id_document_statistics" required="true"/>
44
                <PARAM description="target action set for document classification" name="export_action_set_id_document_classes" required="true"/>
45
                <PARAM description="target action set for document citations" name="export_action_set_id_document_referencedDocuments" required="true"/>
46
                <PARAM description="target action set for dataset entities" name="export_action_set_id_entity_dataset" required="true"/>
47
       		
48
        		<!-- flags to enable/disable IIS modules -->
49
        		<PARAM name="active_metadataextraction_export" required="true" description="enable/disable the affiliation matching module" />
50
        		<PARAM name="active_citationmatching" required="true" description="enable/disable the citation matching module" />
51
        		<PARAM name="active_documentsclassification" required="true" description="enable/disable the document classification module" />
52
        		<PARAM name="active_documentssimilarity" required="true" description="enable/disable the document similarity module" />
53
        		<PARAM name="active_referenceextraction_dataset" required="true" description="enable/disable the dataset reference extraction module" />
54
        		<PARAM name="active_referenceextraction_project" required="true" description="enable/disable the project reference extracion module" />
55
        		<PARAM name="active_referenceextraction_researchinitiative" required="true" description="enable/disable the research initiative extraction module" />
56
        		<PARAM name="active_statistics" required="true" description="enable/disable the statistics module" />
57
        	</JOB_INTERFACE>
58
        </HADOOP_JOB>
59
        <STATUS>
60
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
61
            <RUNNING_INSTANCES value="0"/>
62
            <CUMULATIVE_RUN value="0"/>
63
        </STATUS>
64
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
65
    </BODY>
66
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/indexFeedJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="1c34963b-75b3-4440-9f42-72445a26c077_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="indexFeedJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that joins the entities on the hbase table and produces a sequence file containig the xml records</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.IndexFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.Text" />
22
				
23
				<!-- JOB GLOBAL -->		
24
                <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />
25
                <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text"/>
26
				
27
				<!-- MISC -->					
28
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
29
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
30
				<PROPERTY key="mapred.reduce.tasks" value="0" />
31
				<PROPERTY key="mapred.fairscheduler.pool" value="solr"/>
32
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
33
				
34
		<!--  	Uncomment to override the default lib path -->			
35
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
36
        	</STATIC_CONFIGURATION>
37
        	<JOB_INTERFACE>
38
        		<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
39
        		<PARAM name="mapred.output.dir" required="true" description="destination path on hdfs for rotten index xml records" />
40
        		
41
        		<PARAM name="index.solr.url" required="false" description="url used to instantiate the solr client" /> 
42
       			<PARAM name="index.solr.collection" required="true" description="target solr collection to be fed" />
43

  
44
       			<PARAM name="id" required="true" description="index DS id" />
45
				<PARAM name="index.shutdown.wait.time" required="true" description="wait time before shut down the solr client pool" />
46
       			<PARAM name="index.buffer.flush.threshold" required="true" description="indexing buffer flush threshold" />
47
       			<PARAM name="index.feed.timestamp" required="true" description="timestamp used as ds_version" />
48
				<PARAM name="index.solr.sim.mode" required="true" description="boolean value, allows to run this job in simulation mode" />
49
				<PARAM name="index.xslt" required="true" description="record transformer created by the MSRO service" />
50
        	</JOB_INTERFACE>
51
        	<SCAN>
52
        		<FILTERS />
53
        		<FAMILIES />
54
        	</SCAN>
55
        </HADOOP_JOB>
56
        <STATUS>
57
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
58
            <RUNNING_INSTANCES value="0"/>
59
            <CUMULATIVE_RUN value="0" />
60
        </STATUS>
61
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
62
    </BODY>
63
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/oaiFeedJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="03d7af20-63bb-4790-a052-6cdbc1e05fce_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2015-02-09T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="oaiFeedJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that feeds the OAI store</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.NullOutputFormat" />	
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.oai.OaiFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.NullWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.io.NullWritable" />				
22
				
23
				<!-- JOB GLOBAL -->		
24
<!--                 <PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.NullWritable" /> -->
25
<!--                 <PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.NullWritable"/> -->
26
				
27
				<!-- MISC -->					
28
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
29
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
30
				<PROPERTY key="mapred.reduce.tasks" value="0" />
31
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
32
				
33
				<!--  	Uncomment to override the default lib path -->			
34
				<PROPERTY key="job.lib" value="/user/dnet/lib/dnet-mapreduce-jobs-assembly-0.0.6.3-SNAPSHOT.jar"/> 
35
        	</STATIC_CONFIGURATION>
36
        	<JOB_INTERFACE>
37
        		<PARAM name="mapred.input.dir" required="true" description="source sequence file on hdfs" />
38
				<PARAM name="services.publisher.oai.collection" required="true" description="target mongodb collection" />
39
				<PARAM name="oaiConfiguration" required="true" description="configuration bean used to guide the OAI feeding" />
40
				<PARAM name="oai.feed.date" required="true" description="timestamp" />				
41
				<PARAM name="services.publisher.oai.host" required="true" description="mongodb host" />
42
				<PARAM name="services.publisher.oai.port" required="true" description="mongodb port" />
43
				<PARAM name="services.publisher.oai.db" required="true" description="mongodb database name" />	
44
				<PARAM name="services.publisher.oai.skipDuplicates" required="true" description="skip duplicated records." />	
45
				<PARAM name="services.publisher.oai.duplicateXPath" required="true" description="records with this xpath are identified as duplicates" />																
46
        	</JOB_INTERFACE>
47
        	<SCAN>
48
        		<FILTERS />
49
        		<FAMILIES />
50
        	</SCAN>
51
        </HADOOP_JOB>
52
        <STATUS>
53
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
54
            <RUNNING_INSTANCES value="0"/>
55
            <CUMULATIVE_RUN value="0" />
56
        </STATUS>
57
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
58
    </BODY>
59
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/mdStoreHdfsImportJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="5cd115e7-9650-4263-9c0d-a3fbf6d9549d_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="mdStoreHdfsImportJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that maps xml metada records from a sequence file into an hbase table</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13
        	
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />
17
				
18
				<!-- MAPPER -->        	
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dataimport.ImportRecordsMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Put" />
22
			
23
				<!-- MISC -->
24
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />
25
				<PROPERTY key="mapreduce.map.speculative" value="false" />
26
				<PROPERTY key="mapred.reduce.tasks" value="0" />
27
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
28
				
29
		<!--  	Uncomment to override the default lib path -->			
30
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
31
				
32
        	</STATIC_CONFIGURATION>
33
        	<JOB_INTERFACE>
34
        		<PARAM name="mapred.input.dir" required="true" description="input sequence file" />
35
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
36
        		<PARAM name="hbase.import.xslt" required="true" description="mapping" />
37
        	</JOB_INTERFACE>
38
        </HADOOP_JOB>
39
        <STATUS>
40
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
41
            <RUNNING_INSTANCES value="0"/>
42
            <CUMULATIVE_RUN value="0" />
43
        </STATUS>
44
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
45
    </BODY>
46
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/prepareIndexDataJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="ed8c0a4e-7cf2-49df-bfed-fcfab0699ade_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="prepareIndexDataJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that joins the entities on the hbase table and produces a sequence file containig the xml records</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.index.PrepareFeedMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.index.PrepareFeedReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.io.Text" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Text" />				
27
				
28
				<!-- MISC -->
29
				<PROPERTY key="mapred.output.compress" value="false" />						
30
				<PROPERTY key="mapred.compress.map.output" value="true" />	
31
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
32
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
33
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
34
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />
35
							
36
				<PROPERTY key="mapred.reduce.tasks" value="500" />
37
				<PROPERTY key="dfs.blocksize" value="32M" />
38
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
39
				
40
		<!--  	Uncomment to override the default lib path -->			
41
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
42
        	</STATIC_CONFIGURATION>
43
        	<JOB_INTERFACE>
44
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
45
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
46
        		<PARAM name="mapred.output.dir" required="true" description="target sequence file on hdfs" /> 
47
        		
48
        		<PARAM name="index.entity.links" required="true" description="entity joiner configuration" /> 
49
        		<PARAM name="contextmap" required="true" description="context map (ContextDSResources)" />        		
50
        	</JOB_INTERFACE>
51
        	<SCAN>
52
        		<FILTERS />
53
        		<FAMILIES />
54
        	</SCAN>
55
        </HADOOP_JOB>
56
        <STATUS>
57
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
58
            <RUNNING_INSTANCES value="0"/>
59
            <CUMULATIVE_RUN value="0" />
60
        </STATUS>
61
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
62
    </BODY>
63
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/copyTableJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="cd176186-58ba-44b7-ae7e-25994c0c95bb_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="copytable" type="admin">
11
 			<DESCRIPTION>copies a table on the same cluster or on a remote one, mimics http://hbase.apache.org/book/ops_mgt.html#copytable</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- MISC -->				
15
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
16
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
17
			
18
				<PROPERTY key="mapred.reduce.tasks" value="0" />
19
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
20
				
21
		<!--  	Uncomment to override the default lib path -->			
22
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
23
        	</STATIC_CONFIGURATION>
24
        	<JOB_INTERFACE>
25
        		<PARAM name="new.name" required="true" description="target hbase table" />
26
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
27
        		<PARAM name="peer.adr" required="false" description="target hbase quorum" />
28
        	</JOB_INTERFACE>
29
        	<SCAN caching="100">
30
          		<FILTERS />
31
        		<FAMILIES />      		
32
        	</SCAN>
33
        </HADOOP_JOB>
34
        <STATUS>
35
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
36
            <RUNNING_INSTANCES value="0"/>
37
            <CUMULATIVE_RUN value="0" />
38
        </STATUS>
39
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
40
    </BODY>
41
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/dedupCandidateScanJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="675f1436-205a-4b19-8b6b-35e1c17fb125_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="dedupCandidateScanJob" type="mapreduce">
11
 			<DESCRIPTION>map reduce job that scans a given entity type and creates the similarRel graph</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13

  
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat" />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />				
17
        	
18
        		<!-- MAPPER -->
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text" />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />
22
				
23
				<!-- REDUCER -->
24
				<PROPERTY key="mapreduce.reduce.class" value="eu.dnetlib.data.mapreduce.hbase.dedup.DedupReducer" />
25
				<PROPERTY key="mapred.output.key.class" value="org.apache.hadoop.hbase.io.ImmutableBytesWritable" />				
26
				<PROPERTY key="mapred.output.value.class" value="org.apache.hadoop.io.Writable" />				
27
				
28
				<!-- MISC -->				
29
				<PROPERTY key="mapred.compress.map.output" value="true" />	
30
				<PROPERTY key="mapred.reduce.tasks.speculative.execution" value="false" />	
31
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false" />	
32
				<PROPERTY key="mapreduce.map.speculative" value="false" />		
33
				<PROPERTY key="mapreduce.reduce.speculative" value="false" />				
34
			
35
				<PROPERTY key="mapred.reduce.tasks" value="1000" />
36
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
37
				
38
		<!--  	Uncomment to override the default lib path -->			
39
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
40
        	</STATIC_CONFIGURATION>
41
        	<JOB_INTERFACE>
42
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
43
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
44
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
45
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />
46
        		<PARAM name="dedup.pace.conf" required="true" description="dedup pace configuration" />
47
        		<PARAM name="dedup.wf.conf" required="true" description="dedup workflow configuration" />        		
48
        	</JOB_INTERFACE>
49
        	<SCAN>
50
        		<FILTERS operator="MUST_PASS_ALL">
51
        			<FILTER type="prefix" param="entityTypeId" />
52
        		</FILTERS>
53
        		<FAMILIES>
54
        			<FAMILY param="entityType" />
55
           			<FAMILY value="resultResult_dedup_merges" />
56
        			<FAMILY value="personPerson_dedup_merges" />
57
     				<FAMILY value="organizationOrganization_dedup_merges" />        			
58
        		</FAMILIES>
59
        	</SCAN>
60
        </HADOOP_JOB>
61
        <STATUS>
62
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
63
            <RUNNING_INSTANCES value="0"/>
64
            <CUMULATIVE_RUN value="0" />
65
        </STATUS>
66
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
67
    </BODY>
68
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/sqoopStatsUpdateJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="bf0ab07b-36bf-4164-ab73-342bfb11e51a_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="StatsExportJob" type="oozie">
11
            <DESCRIPTION>Job for importing data from HBASE to the relational Stats Database</DESCRIPTION>
12
            <STATIC_CONFIGURATION>
13
			
14
				<!-- Cluster wide -->
15
                <PROPERTY key="queueName" value="default"/>
16
                <PROPERTY key="user.name" value="dnet"/> <!-- username = sqoop?? -->
17
                <PROPERTY key="workingDir" value="/user/dnet/lib/stats/working_dir"/>
18
                <PROPERTY key="numReducers" value="1"/>
19
                
20
				<PROPERTY key="oozie.wf.application.path" value="hdfs://nmis-hadoop-cluster/user/eri.katsari/stats/oozie_app"/><!-- edit this property! -->
21
				<PROPERTY key="Stats_db_Url" value="jdbc:postgresql://node1.t.openaire.research-infrastructures.eu:5432/stats"/><!-- complete the jdbc url with the actual value! -->
22
				<PROPERTY key="Stats_db_User" value="sqoop"/>
23
				<PROPERTY key="Stats_db_Pass" value="sqoop"/>
24
				<PROPERTY key="Stats_db_Driver" value="org.postgresql.Driver"/>
25
				<PROPERTY key="Stats_db_table_map" value="datasourceLanguage=datasource_languages,datasource=datasource,project=project,result=result,organization=organization,datasourceOrganization=datasource_organizations,datasourceTopic=datasource_topics,projectOrganization=project_organizations,resultClaim=result_claims,resultClassification=result_classifications,resultConcept=result_concepts,resultLanguage=result_languages,resultOrganization=result_organizations,resultResult=result_results,resultProject=project_results,resultTopic=result_topics,category=category,context=context,claim=claim,concept=concept,datasourceLanguage=datasource_languages,resultLanguage=result_languages,resultDatasource=result_datasources"/>
26
				<PROPERTY key="Stats_sqoop_RecsPerStatement" value="10000"/>
27
				<PROPERTY key="Stats_sqoop_StatementPerTrans" value="1000000"/>
28
				<PROPERTY key="Stats_sqoop_ReducersCount" value="4"/>
29
				<PROPERTY key="Stats_output_Path" value="/tmp/stats/"/>
30
				<PROPERTY key="Stats_null_String_Field" value="null"/>
31
				<PROPERTY key="Stats_null_Numeric_Field" value="null"/>
32
				<PROPERTY key="Stats_enclosing_Character" value="#"/>
33
				<PROPERTY key="Stats_delim_Character" value="!"/>
34
				<PROPERTY key="out1" value="datasource"/>
35
				<PROPERTY key="out2" value="project"/>
36
				<PROPERTY key="out3" value="organization"/>
37
				<PROPERTY key="out4" value="datasourceOrganization"/>
38
				<PROPERTY key="out5" value="datasourceTopic"/>
39
				<PROPERTY key="out6" value="datasourceLanguage"/>
40
				<PROPERTY key="out7" value="projectOrganization"/>
41
				<PROPERTY key="out8" value="resultClaim"/>
42
				<PROPERTY key="out9" value="resultClassification"/>
43
				<PROPERTY key="out10" value="resultConcept"/>
44
				<PROPERTY key="out11" value="resultLanguage"/>
45
				<PROPERTY key="out12" value="resultOrganization"/>
46
				<PROPERTY key="out13" value="resultResult"/>
47
				<PROPERTY key="out14" value="resultProject"/>
48
				<PROPERTY key="out15" value="category"/>
49
				<PROPERTY key="out16" value="resultTopic"/>
50
				<PROPERTY key="out17" value="resultDatasource"/>
51
				<PROPERTY key="out18" value="result"/>
52
				<PROPERTY key="out19" value="claim"/>
53
				<PROPERTY key="out20" value="concept"/>
54
            </STATIC_CONFIGURATION>
55
            <JOB_INTERFACE>
56
                <PARAM name="nameNode" required="true" description="hdfs name node"/>
57
                <PARAM name="jobTracker" required="true" description="job tracker name"/>
58
                <PARAM name="Stats_Hbase_Source_Table" required="true" description="Hbase Table with Protobuffs."/>
59
                <PARAM name="Stats_indexConf" required="true" description="Index Entity Links configuration."/>
60
                <PARAM name="isLookupEndpoint" required="true" description="IS lookup service endpoint"/>
61
            </JOB_INTERFACE>
62
        </HADOOP_JOB>
63
        <STATUS>
64
            <LAST_SUBMISSION_DATE value="2014-11-14T19:57:25+00:00"/>
65
            <RUNNING_INSTANCES value="0"/>
66
            <CUMULATIVE_RUN value="75"/>
67
        </STATUS>
68
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
69
    </BODY>
70
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/HadoopJobConfigurationDSResources/HadoopJobConfigurationDSResourceType/resetDedupJob.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="bc4f377a-af07-403d-a019-af60aa557652_SGFkb29wSm9iQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL0hhZG9vcEpvYkNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="HadoopJobConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="HadoopJobConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2001-12-31T12:00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <HADOOP_JOB name="resetDedupJob" type="mapreduce">
11
 			<DESCRIPTION>map only job that reset the dedup jobs</DESCRIPTION>
12
        	<STATIC_CONFIGURATION>
13
        	
14
				<!-- I/O FORMAT -->
15
				<PROPERTY key="mapreduce.inputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableInputFormat"  />
16
				<PROPERTY key="mapreduce.outputformat.class" value="org.apache.hadoop.hbase.mapreduce.TableOutputFormat" />		
17

  
18
				<!-- MAPPER -->        	
19
				<PROPERTY key="mapreduce.map.class" value="eu.dnetlib.data.mapreduce.hbase.reset.HBaseResetMapper" />
20
				<PROPERTY key="mapred.mapoutput.key.class" value="org.apache.hadoop.io.Text"  />
21
				<PROPERTY key="mapred.mapoutput.value.class" value="org.apache.hadoop.hbase.client.Mutation"  />
22

  
23
				<!-- MISC -->
24
				<PROPERTY key="mapred.map.tasks.speculative.execution" value="false"  />
25
				<PROPERTY key="mapreduce.map.speculative" value="false"  />
26
				<PROPERTY key="mapred.reduce.tasks" value="0"  />
27
				<!-- <PROPERTY key="user.name" value="dnet" /> -->
28
				
29
		<!--  	Uncomment to override the default lib path -->			
30
		<!--	<PROPERTY key="job.lib" value="/user/dnet/dnet-mapreduce-jobs-0.0.2-SNAPSHOT-jar-with-dependencies.jar"/> -->
31
				
32
        	</STATIC_CONFIGURATION>
33
        	<JOB_INTERFACE>
34
        		<PARAM name="hbase.mapred.inputtable" required="true" description="source hbase table" />
35
        		<PARAM name="hbase.mapred.outputtable" required="true" description="target hbase table" />
36
        		<PARAM name="hbase.mapreduce.inputtable" required="true" description="source hbase table" />
37
        		<PARAM name="hbase.mapreduce.outputtable" required="true" description="target hbase table" />        		
38
        	</JOB_INTERFACE>
39
        </HADOOP_JOB>
40
        <STATUS>
41
            <LAST_SUBMISSION_DATE value="2001-12-31T12:00:00"/>
42
            <RUNNING_INSTANCES value="0"/>
43
            <CUMULATIVE_RUN value="0" />
44
        </STATUS>
45
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
46
    </BODY>
47
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/OAIPublisherConfigurationDSResources/OAIPublisherConfigurationDSResourceType/OAIPublisherConfiguration-OpenAire.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="53b2865b-53a3-444c-a3a5-d09bb7be9374_T0FJUHVibGlzaGVyQ29uZmlndXJhdGlvbkRTUmVzb3VyY2VzL09BSVB1Ymxpc2hlckNvbmZpZ3VyYXRpb25EU1Jlc291cmNlVHlwZQ=="/>
4
        <RESOURCE_TYPE value="OAIPublisherConfigurationDSResourceType"/>
5
        <RESOURCE_KIND value="OAIPublisherConfigurationDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2014-10-23T12:33:34+00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <CONFIGURATION>
11
            <IDSCHEME>oai</IDSCHEME>
12
            <IDNAMESPACE>dnet</IDNAMESPACE>
13
            <CURRENTDB>oaistore</CURRENTDB>
14
            <OAISETS>
15
                <OAISET enabled="true">
16
                    <spec>OpenAccess</spec>
17
                    <name>Set of Open Access results</name>
18
                    <description>OPEN ACCESS publications and datasets</description>
19
                    <query>(license = "OPEN")</query>
20
                </OAISET>
21
                <OAISET enabled="true">
22
                    <spec>publications</spec>
23
                    <name>Publications</name>
24
                    <description>Set of all publications aggregated in OpenAIREPlus</description>
25
                    <query>resulttypeid="publication"</query>
26
                </OAISET>
27
                <OAISET enabled="true">
28
                    <spec>organizations</spec>
29
                    <name>Organizations</name>
30
                    <description>Set of all organizations aggregated in OpenAIREPlus</description>
31
                    <query>oaftype="organization"</query>
32
                </OAISET>
33
                <OAISET enabled="true">
34
                    <spec>datasets</spec>
35
                    <name>datasets</name>
36
                    <description>Set of all datasets aggregated in OpenAIREPlus</description>
37
                    <query>resulttypeid="dataset"</query>
38
                </OAISET>
39
                <OAISET enabled="true">
40
                    <spec>projects</spec>
41
                    <name>projects</name>
42
                    <description>Set of all projects: includes FP7, WellcomeTrust and, soon, H2020 funded projects</description>
43
                    <query>oaftype exact "project"</query>
44
                </OAISET>
45
                <OAISET enabled="true">
46
                    <spec>FP7Projects</spec>
47
                    <name>FP7 projects</name>
48
                    <description>Projects funded by EC FP7</description>
49
                    <query>(oaftype exact "project" and funder exact "corda_______::FP7")</query>
50
                </OAISET>
51
                <OAISET enabled="true">
52
                    <spec>WTProjects</spec>
53
                    <name>Wellcome Trust projects</name>
54
                    <description>Projects funded by Wellcome Trust</description>
55
                    <query>(oaftype exact "project" and funder exact "wt::WT")</query>
56
                </OAISET>
57
            </OAISETS>
58
            <METADATAFORMATS>
59
                <METADATAFORMAT exportable="true" metadataPrefix="oaf">
60
                    <NAMESPACE>http://namespace.openaire.eu/oaf</NAMESPACE>
61
                    <SCHEMA>http://www.openaire.eu/schema/0.2/oaf-0.2.xsd</SCHEMA>
62
                    <SOURCE_METADATA_FORMAT interpretation="openaire" layout="index" name="oaf"/>
63
                    <TRANSFORMATION_RULE/>
64
                    <BASE_QUERY>(oaftype &lt;&gt; "person")</BASE_QUERY>
65
                </METADATAFORMAT>
66
                <METADATAFORMAT exportable="true" metadataPrefix="oai_dc">
67
                    <NAMESPACE>http://www.openarchives.org/OAI/2.0/oai_dc/</NAMESPACE>
68
                    <SCHEMA>http://www.openarchives.org/OAI/2.0/oai_dc.xsd</SCHEMA>
69
                    <SOURCE_METADATA_FORMAT interpretation="openaire" layout="index" name="oaf"/>
70
                    <TRANSFORMATION_RULE>oaf2dc_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU=</TRANSFORMATION_RULE>
71
                    <BASE_QUERY>oaftype="result"</BASE_QUERY>
72
                </METADATAFORMAT>
73
            </METADATAFORMATS>
74
            <INDICES>
75
                <INDEX name="objIdentifier" repeatable="false">
76
                    <SOURCE interpretation="openaire" layout="index" name="oaf" path="//*[local-name() ='objIdentifier']"/>
77
                </INDEX>
78
                <INDEX name="set" repeatable="true">
79
                    <SOURCE interpretation="openaire" layout="index" name="oaf" path="//collectedfrom/@name"/>
80
                </INDEX>
81
                <INDEX name="license" repeatable="false">
82
                    <SOURCE interpretation="openaire" layout="index" name="oaf" path="//bestlicense/@classid"/>
83
                </INDEX>
84
                <INDEX name="oaftype" repeatable="false">
85
                    <SOURCE interpretation="openaire" layout="index" name="oaf" path="local-name(//*[local-name()='entity']/*)"/>
86
                </INDEX>
87
                <INDEX name="resulttypeid" repeatable="false">
88
                    <SOURCE interpretation="openaire" layout="index" name="oaf" path="//*[local-name()='entity']/*[local-name()='result']/resulttype/@classid"/>
89
                </INDEX>
90
                <INDEX name="funder" repeatable="false">
91
                    <SOURCE interpretation="openaire" layout="index" name="oaf" path="//*[local-name()='entity']/*[local-name()='project']/fundingtree//funding_level_0/id"/>
92
                </INDEX>
93
            </INDICES>
94
        </CONFIGURATION>
95
        <STATUS>
96
            <LAST_UPDATE value="2001-12-31T12:00:00"/>
97
        </STATUS>
98
        <SECURITY_PARAMETERS>SECURITY_PARAMETERS</SECURITY_PARAMETERS>
99
    </BODY>
100
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/dc2oaf_cleaning_OPENAIREplus_compliant_phaidra.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="d152b8be-723d-42f5-8a13-e9de7d66a2b8_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
4
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
5
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2013-11-27T00:21:15+00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <CONFIGURATION>
11
            <IMPORTED/>
12
            <SCRIPT>
13
                <TITLE>dc_cleaning_OPENAIREplus_compliant_phaidra</TITLE>
14
                <CODE><![CDATA[
15
declare_script "dc_cleaning_OpenAIREplus_compliant_phaidra";
16
declare_ns oaf = "http://namespace.openaire.eu/oaf";
17
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
18
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
19
declare_ns dc = "http://purl.org/dc/elements/1.1/";
20
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
21
$var0 = "''";
22
$var1 = "'corda_______::'";
23
$varDummy = "''";
24
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
25
static $varRepoid = xpath:"//dri:repositoryId";
26
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//OFFICIAL_NAME"]);
27
dri:objIdentifier = xpath:"//dri:objIdentifier";
28
dri:repositoryId = $varRepoid;
29
//dri:repositoryId = xpath:"//dri:repositoryId";
30
dri:recordIdentifier = RegExpr(xpath:"//dri:recordIdentifier", $var0, "s/^(.*)(::)/$2/");
31
apply xpath:"//dc:creator" if xpath:"string-length(.) > 0" dc:creator = xpath:"normalize-space(.)"; else $varDummy = "''";
32
apply xpath:"//dc:title" if xpath:"string-length(.) > 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
33
apply xpath:"//dc:subject" if xpath:"string-length(.) > 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
34
apply xpath:"//dc:publisher" if xpath:"string-length(.) > 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
35
apply xpath:"//dc:source" if xpath:"string-length(.) > 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
36
dc:contributor = xpath:"//dc:contributor";
37
dc:description = xpath:"//dc:description";
38
//if xpath:"//*[local-name() = 'setSpec' and text()='ec_fundedresources']" dc:relation = skipRecord(); else $varDummy = "''";
39
$varHttpTest = "''";
40
if xpath:"//dc:identifier[starts-with(., 'http')]" $varHttpTest = "true"; else dc:identifier = skipRecord();
41
apply xpath:"//dc:identifier" if xpath:"starts-with(normalize-space(.), 'http')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
42
static dr:dateOfCollection = getValue(CURRENTDATE, []);
43
// dc:type = xpath:"//dc:type";
44
dc:language = Convert(xpath:"//dc:language", Languages);
45
//if xpath:"//dc:rights[text()='info:eu-repo/semantics/openAccess']" dc:publisher = xpath:"//dc:publisher"; else dc:publisher = skipRecord();
46
dc:date = xpath:"//dc:date";
47
oaf:dateAccepted = Convert(xpath:"descendant-or-self::dc:date", DateISO8601, "yyyy-MM-dd", "min()");
48
apply xpath:"//dc:date" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/date')" oaf:embargoenddate = RegExpr(xpath:"normalize-space(.)", $var0, "s/^(.*info:eu-repo\/date\/embargoEnd\/)//gmi"); else $var0 = "''";
49
apply xpath:"//dc:relation" if xpath:"string-length(substring-after(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7/')) = 6 and number(substring-after(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7/')) &gt; 100000" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = skipRecord();
50
//apply xpath:"//dc:relation" if xpath:"string-length(translate(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7','')) = 6" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
51
//apply xpath:"//dc:relation" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7')" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
52
//comment-js-09-10-2012 apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" dc:rights = empty; else dc:rights = xpath:"normalize-space(.)";
53
// static oaf:datasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
54
//
55
//
56
//
57
static $varDsType = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//CONFIGURATION/DATASOURCE_TYPE"]);
58
//if xpath:"$varDsType='aggregator'" oaf:hostingDatasourceid = xpath:"//prov:baseURL"; else oaf:hostingDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
59
oaf:collectedDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
60
//
61
apply xpath:"//dc:type" if xpath:"." dr:CobjCategory = Convert(xpath:"normalize-space(.)", TextTypologies); else dc:type = xpath:".";
62
apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" oaf:accessrights = Convert(xpath:"normalize-space(.)", AccessRights); else dc:rights = xpath:".";
63
if xpath:"//dc:rights[starts-with(normalize-space(.), 'info:eu-repo/semantics')]" $var0 = "''"; else oaf:accessrights = "OPEN";
64
//if xpath:"count(//dc:rights) = 0" oaf:accessrights = "OPEN"; else $var0 = "''";
65
// oaf:accessrights = Convert(xpath:"normalize-space(//dc:rights)", AccessRights);
66
//
67
static oaf:collectedFrom = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
68
static oaf:hostedBy = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
69
//
70
$varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '(10[.][0-9]{4,}[^\s"/<>]*/[^\s"<>]+)');
71
oaf:identifier = set(xpath:"$varId//value", @identifierType = "doi";);
72
oaf:datasourceprefix = xpath:"//oaf:datasourceprefix";
73
end
74
]]></CODE>
75
            </SCRIPT>
76
        </CONFIGURATION>
77
        <STATUS/>
78
        <SECURITY_PARAMETERS/>
79
    </BODY>
80
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/dc2oaf_cleaning_OPENAIREplus_pensoft.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="7a0f18a4-7015-45ec-a9e5-1c7368889d7f_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
4
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
5
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2013-02-28T12:14:22+01:00"/>
8
    </HEADER>
9
    <BODY>
10
        <CONFIGURATION>
11
            <IMPORTED/>
12
            <SCRIPT>
13
                <TITLE>dc_cleaning_OPENAIREplus_pensoft</TITLE>
14
                <CODE><![CDATA[
15
declare_script "dc_cleaning_OpenAIREplus_pensoft";
16
declare_ns oaf = "http://namespace.openaire.eu/oaf";
17
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
18
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
19
declare_ns dc = "http://purl.org/dc/elements/1.1/";
20
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
21
$var0 = "''";
22
$var1 = "'corda_______::'";
23
static $varHostedByPrefix = "'openaire____::issn'";
24
$varDummy = "''";
25
$varHostedByName = RegExpr(xpath:"//dc:source", $var0, "s/(\s+)(\d+)(.*)$//gmi");
26
$varHostedByIdTemp = RegExpr(xpath:"//dc:relation[last()]", $var0, "s/^(.*)(\/)|(-)//gmi");
27
$varHostedById = xpath:"concat($varHostedByPrefix, $varHostedByIdTemp)";
28
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
29
static $varRepoid = xpath:"//dri:repositoryId";
30
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//OFFICIAL_NAME"]);
31
dri:objIdentifier = xpath:"//dri:objIdentifier";
32
dri:repositoryId = $varRepoid;
33
dri:recordIdentifier = RegExpr(xpath:"//dri:recordIdentifier", $var0, "s/^(.*)(::)/$2/");
34
apply xpath:"//dc:creator" if xpath:"string-length(.) > 0" dc:creator = xpath:"normalize-space(.)"; else $varDummy = "''";
35
apply xpath:"//dc:title" if xpath:"string-length(.) > 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
36
apply xpath:"//dc:subject" if xpath:"string-length(.) > 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
37
apply xpath:"//dc:publisher" if xpath:"string-length(.) > 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
38
apply xpath:"//dc:source" if xpath:"string-length(.) > 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
39
dc:contributor = xpath:"//dc:contributor";
40
dc:description = xpath:"//dc:description";
41
$varHttpTest = "''";
42
if xpath:"//dc:identifier[starts-with(., 'http')]" $varHttpTest = "true"; else dc:identifier = skipRecord();
43
apply xpath:"//dc:identifier" if xpath:"starts-with(normalize-space(.), 'http')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
44
static dr:dateOfCollection = getValue(CURRENTDATE, []);
45
// dc:type = xpath:"//dc:type";
46
dc:language = Convert(xpath:"//dc:language", Languages);
47
dc:date = xpath:"//dc:date";
48
oaf:dateAccepted = Convert(xpath:"descendant-or-self::dc:date", DateISO8601, "yyyy-MM-dd", "min()");
49
apply xpath:"//dc:date" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/date')" oaf:embargoenddate = RegExpr(xpath:"normalize-space(.)", $var0, "s/^(.*info:eu-repo\/date\/embargoEnd\/)//gmi"); else $var0 = "''";
50
apply xpath:"//dc:relation" if xpath:"string-length(substring-after(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7/')) = 6" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
51
//comment-js-09-10-2012 apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" dc:rights = empty; else dc:rights = xpath:"normalize-space(.)";
52
//
53
static $varDsType = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//CONFIGURATION/DATASOURCE_TYPE"]);
54
//if xpath:"$varDsType='aggregator'" oaf:hostingDatasourceid = xpath:"//prov:baseURL"; else oaf:hostingDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
55
// oaf:collectedDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
56
//
57
apply xpath:"//dc:type" if xpath:"." dr:CobjCategory = Convert(xpath:"normalize-space(.)", TextTypologies); else dc:type = xpath:".";
58
apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" oaf:accessrights = Convert(xpath:"normalize-space(.)", AccessRights); else dc:rights = xpath:".";
59
if xpath:"//dc:rights[starts-with(normalize-space(.), 'info:eu-repo/semantics')]" $var0 = "''"; else oaf:accessrights = "OPEN";
60
//if xpath:"count(//dc:rights) = 0" oaf:accessrights = "OPEN"; else $var0 = "''";
61
// oaf:accessrights = Convert(xpath:"normalize-space(//dc:rights)", AccessRights);
62
//
63
static oaf:collectedFrom = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
64
static oaf:hostedBy = set("''", @name = $varHostedByName; , @id = $varHostedById;);
65
//
66
$varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '(10[.][0-9]{4,}[^\s"/<>]*/[^\s"<>]+)');
67
oaf:identifier = set(xpath:"$varId//value", @identifierType = "doi";);
68
oaf:datasourceprefix = xpath:"//oaf:datasourceprefix";
69
end
70
]]></CODE>
71
            </SCRIPT>
72
        </CONFIGURATION>
73
        <STATUS/>
74
        <SECURITY_PARAMETERS/>
75
    </BODY>
76
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/wos_rules.xml
1
<?xml version="1.0" encoding="UTF-8"?>
2
<RESOURCE_PROFILE>
3
    <HEADER>
4
        <RESOURCE_IDENTIFIER value="5ac71ab7-a250-4d3d-9d5e-8e12d983ff4a_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
5
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
6
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
7
        <RESOURCE_URI value=""/>
8
        <DATE_OF_CREATION value="2013-11-11T11:11:11+01:00"/>
9
    </HEADER>
10
    <BODY>
11
        <CONFIGURATION>
12
            <IMPORTED/>
13
            <SCRIPT>
14
                <TITLE>oaf_transformation_wos</TITLE>
15
                <CODE><![CDATA[
16
declare_script "oaf_transformation_wos";
17
declare_ns oaf = "http://namespace.openaire.eu/oaf";
18
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
19
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
20
declare_ns dc = "http://purl.org/dc/elements/1.1/";
21
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
22
// +++ header section +++
23
dri:objIdentifier = xpath:"//dri:objIdentifier";
24
dri:recordIdentifier = xpath:"//csvRecord/column[@name='UT']";
25
dri:dateOfCollection = xpath:"//dri:dateOfCollection";
26
// dri:repositoryId
27
oaf:datasourceprefix = "webcrawl____";
28
// +++ metadata section +++
29
dc:language = Convert(xpath:"//csvRecord/column[@name='LA']", Languages);
30
dc:title = xpath:"//csvRecord/column[@name='TI']";
31
%myTemplate = split(xpath:"//csvRecord/column[@name='AF']/text()", "dc:creator", ";");
32
// dc:creator = xpath:"//csvRecord/column[@name='AF']/text()";
33
$varIssn = xpath:"//csvRecord/column[@name='SN']";
34
oaf:journal = set(xpath:"//csvRecord/column[@name='SO']", @issn = $varIssn;);
35
dr:CobjCategory = Convert(xpath:"//csvRecord/column[@name='DT']", TextTypologies);
36
// subjects must be ignored
37
// dc:subject = xpath:"//csvRecord/column[@name='ID']";
38
// abstract must be ignored
39
// dc:description = xpath:"//csvRecord/column[@name='AB']";
40
dc:publisher = xpath:"//csvRecord/column[@name='PU']";
41
oaf:dateAccepted = Convert(xpath:"descendant-or-self::column[@name='PY']", DateISO8601, "yyyy-MM-dd", "min()");
42
$varDoi = xpath:"concat('http://dx.doi.org/', normalize-space(//csvRecord/column[@name='DI']))";
43
dc:identifier = $varDoi;
44
if xpath:"//csvRecord/column[@name='OA'][text()='Y']" oaf:accessrights = "OPEN"; else oaf:accessrights = "CLOSED";
45
oaf:identifier = set(xpath:"//csvRecord/column[@name='DI']", @identifierType = "doi";);
46
oaf:hostedBy = set("''", @name="Unknown Repository";, @id="openaire____::1256f046-bf1f-4afc-8b47-d0b147148b18";);
47
if xpath:"//csvRecord/column[@name='DI'][starts-with(., '10.')]" oaf:collectedFrom = set("''", @name="Crossref";, @id="openaire____::crossref";); else oaf:collectedFrom = set("''", @name="Web Crawl";, @id="openaire____::webcrawl";);
48
end
49
]]></CODE>
50
            </SCRIPT>
51
        </CONFIGURATION>
52
        <STATUS/>
53
        <SECURITY_PARAMETERS/>
54
    </BODY>
55
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/dc2oaf_cleaning_OPENAIREplus.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="d8bf1473-4af6-4c18-8b27-fb59759908ed_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
4
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
5
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2013-02-28T12:14:22+01:00"/>
8
    </HEADER>
9
    <BODY>
10
        <CONFIGURATION>
11
            <IMPORTED/>
12
            <SCRIPT>
13
                <TITLE>dc_cleaning_OPENAREplus</TITLE>
14
                <CODE><![CDATA[
15
declare_script "dc_cleaning_OpenAIREplus";
16
declare_ns oaf = "http://namespace.openaire.eu/oaf";
17
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
18
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
19
declare_ns dc = "http://purl.org/dc/elements/1.1/";
20
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
21
$var0 = "''";
22
$var1 = "'corda_______::'";
23
$varDummy = "''";
24
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
25
static $varRepoid = xpath:"//dri:repositoryId";
26
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//OFFICIAL_NAME"]);
27
dri:objIdentifier = xpath:"//dri:objIdentifier";
28
dri:repositoryId = $varRepoid;
29
//dri:repositoryId = xpath:"//dri:repositoryId";
30
dri:recordIdentifier = RegExpr(xpath:"//dri:recordIdentifier", $var0, "s/^(.*)(::)/$2/");
31
apply xpath:"//dc:creator" if xpath:"string-length(.) > 0" dc:creator = xpath:"normalize-space(.)"; else $varDummy = "''";
32
apply xpath:"//dc:title" if xpath:"string-length(.) > 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
33
apply xpath:"//dc:subject" if xpath:"string-length(.) > 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
34
apply xpath:"//dc:publisher" if xpath:"string-length(.) > 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
35
apply xpath:"//dc:source" if xpath:"string-length(.) > 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
36
dc:contributor = xpath:"//dc:contributor";
37
dc:description = xpath:"//dc:description";
38
$varHttpTest = "''";
39
if xpath:"//dc:identifier[starts-with(., 'http')]" $varHttpTest = "true"; else dc:identifier = skipRecord();
40
apply xpath:"//dc:identifier" if xpath:"starts-with(normalize-space(.), 'http')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
41
static dr:dateOfCollection = getValue(CURRENTDATE, []);
42
// dc:type = xpath:"//dc:type";
43
dc:language = Convert(xpath:"//dc:language", Languages);
44
//if xpath:"//dc:rights[text()='info:eu-repo/semantics/openAccess']" dc:publisher = xpath:"//dc:publisher"; else dc:publisher = skipRecord();
45
dc:date = xpath:"//dc:date";
46
oaf:dateAccepted = Convert(xpath:"descendant-or-self::dc:date", DateISO8601, "yyyy-MM-dd", "min()");
47
apply xpath:"//dc:date" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/date')" oaf:embargoenddate = RegExpr(xpath:"normalize-space(.)", $var0, "s/^(.*info:eu-repo\/date\/embargoEnd\/)//gmi"); else $var0 = "''";
48
apply xpath:"//dc:relation" if xpath:"string-length(translate(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7','')) = 5" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
49
//comment-js-09-10-2012 apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" dc:rights = empty; else dc:rights = xpath:"normalize-space(.)";
50
// static oaf:datasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
51
//
52
//
53
//
54
static $varDsType = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//CONFIGURATION/DATASOURCE_TYPE"]);
55
//if xpath:"$varDsType='aggregator'" oaf:hostingDatasourceid = xpath:"//prov:baseURL"; else oaf:hostingDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
56
oaf:collectedDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
57
//
58
apply xpath:"//dc:type" if xpath:"." dr:CobjCategory = Convert(xpath:"normalize-space(.)", TextTypologies); else dc:type = xpath:".";
59
apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" oaf:accessrights = Convert(xpath:"normalize-space(.)", AccessRights); else dc:rights = xpath:".";
60
oaf:accessrights = "OPEN";
61
// oaf:accessrights = Convert(xpath:"normalize-space(//dc:rights)", AccessRights);
62
//
63
static oaf:collectedFrom = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
64
static oaf:hostedBy = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
65
//
66
$varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '(10[.][0-9]{4,}[^\s"/<>]*/[^\s"<>]+)');
67
oaf:identifier = set(xpath:"$varId//value", @identifierType = "doi";);
68
oaf:datasourceprefix = xpath:"//oaf:datasourceprefix";
69
end
70
]]></CODE>
71
            </SCRIPT>
72
        </CONFIGURATION>
73
        <STATUS/>
74
        <SECURITY_PARAMETERS/>
75
    </BODY>
76
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/dc2oaf_cleaning_OPENAIREplus_compliant_dlib.xml
1
<?xml version="1.0" encoding="UTF-8"?>
2
<RESOURCE_PROFILE>
3
    <HEADER>
4
        <RESOURCE_IDENTIFIER value="99a3948a-ed80-4fdd-83c2-ea021b0fc069_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
5
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
6
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
7
        <RESOURCE_URI value=""/>
8
        <DATE_OF_CREATION value="2015-02-23T15:32:50+00:00"/>
9
    </HEADER>
10
    <BODY>
11
        <CONFIGURATION>
12
            <IMPORTED/>
13
            <SCRIPT>
14
                <TITLE>dc_cleaning_OPENAIREplus_compliant_dlib</TITLE>
15
                <CODE>declare_script "dc_cleaning_OpenAIREplus_compliant_dlib";
16
declare_ns oaf = "http://namespace.openaire.eu/oaf";
17
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
18
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
19
declare_ns dc = "http://purl.org/dc/elements/1.1/";
20
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
21
$var0 = "''";
22
$var1 = "'corda_______::'";
23
$varDummy = "''";
24
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
25
static $varRepoid = xpath:"//dri:repositoryId";
26
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//OFFICIAL_NAME"]);
27
dri:objIdentifier = xpath:"//dri:objIdentifier";
28
dri:repositoryId = $varRepoid;
29
//dri:repositoryId = xpath:"//dri:repositoryId";
30
dri:recordIdentifier = xpath:"//dri:recordIdentifier";
31
if xpath:"count(//dlib-meta0.1//creator) = 0" dc:creator = skipRecord(); else $varDummy = "''";
32
//apply xpath:"//dc:creator" if xpath:"string-length(.) &gt; 0" dc:creator = xpath:"normalize-space(.)"; else $varDummy = "''";
33
apply xpath:"//dlib-meta0.1//creator" if xpath:"string-length(.) &gt; 0" dc:creator = Convert(xpath:".", Person); else $varDummy = "''";
34
apply xpath:"//dlib-meta0.1/title" if xpath:"string-length(.) &gt; 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
35
apply xpath:"//dc:subject" if xpath:"string-length(.) &gt; 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
36
apply xpath:"//dlib-meta0.1//publisher" if xpath:"string-length(.) &gt; 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
37
apply xpath:"//dc:source" if xpath:"string-length(.) &gt; 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
38
dc:contributor = xpath:"//dc:contributor";
39
dc:description = xpath:"//dc:description";
40
$varHttpTest = "''";
41
if xpath:"//dlib-meta0.1//identifier[starts-with(., 'http')]" $varHttpTest = "true"; else dc:identifier = skipRecord();
42
apply xpath:"//dlib-meta0.1//identifier" if xpath:"starts-with(normalize-space(.), 'http')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
43
static dr:dateOfCollection = getValue(CURRENTDATE, []);
44
// dc:type = xpath:"//dc:type";
45
dc:language = Convert(xpath:"//dlib-meta0.1/language", Languages);
46
//dc:language = "eng";
47
//if xpath:"//dc:rights[text()='info:eu-repo/semantics/openAccess']" dc:publisher = xpath:"//dc:publisher"; else dc:publisher = skipRecord();
48
dc:date = xpath:"//dlib-meta0.1/date[@date-type='publication']";
49
oaf:dateAccepted = xpath:"concat(replace(//date[@date-type='publication'], '\D', ''), '-01', '-01')";
50
//  Convert(xpath:"//date[@date-type='publication']", DateISO8601, "yyyy-MM-dd", "min()");
51
//
52
static $varDsType = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//CONFIGURATION/DATASOURCE_TYPE"]);
53
//if xpath:"$varDsType='aggregator'" oaf:hostingDatasourceid = xpath:"//prov:baseURL"; else oaf:hostingDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
54
oaf:collectedDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
55
//
56
apply xpath:"//dlib-meta0.1//type[@resource-type='work']" if xpath:"." dr:CobjCategory = Convert(xpath:"normalize-space(.)", TextTypologies); else dc:type = xpath:".";
57
dc:rights = xpath:"//dlib-meta0.1/rights";
58
oaf:accessrights = "OPEN";
59
//if xpath:"count(//dc:rights) = 0" oaf:accessrights = "OPEN"; else $var0 = "''";
60
// oaf:accessrights = Convert(xpath:"normalize-space(//dc:rights)", AccessRights);
61
//
62
static oaf:collectedFrom = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
63
static oaf:hostedBy = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
64
//
65
// $varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '(10[.][0-9]{4,}[^\s"/&lt;&gt;]*/[^\s"&lt;&gt;]+)');
66
oaf:identifier = set(xpath:"//dlib-meta0.1/identifier[@uri-type='DOI']", @identifierType = "doi";);
67
oaf:datasourceprefix = xpath:"//oaf:datasourceprefix";
68
oaf:journal = set(xpath:"//dlib-meta0.1//serial-name", @eissn = xpath:"//dlib-meta0.1//issn";);
69
end</CODE>
70
            </SCRIPT>
71
        </CONFIGURATION>
72
        <STATUS/>
73
        <SECURITY_PARAMETERS/>
74
    </BODY>
75
</RESOURCE_PROFILE>
modules/dnet-openaireplus-profiles/tags/dnet-openaireplus-profiles-1.0.4/trunk/src/main/resources/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/dc2oaf_cleaning_OPENAIREplus_compliant_creator_comma.xml
1
<RESOURCE_PROFILE>
2
    <HEADER>
3
        <RESOURCE_IDENTIFIER value="bc13c7ee-869f-4b76-81a8-2ee6ffcb81fa_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
4
        <RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
5
        <RESOURCE_KIND value="TransformationRuleDSResources"/>
6
        <RESOURCE_URI value=""/>
7
        <DATE_OF_CREATION value="2014-04-09T08:05:59+00:00"/>
8
    </HEADER>
9
    <BODY>
10
        <CONFIGURATION>
11
            <IMPORTED/>
12
            <SCRIPT>
13
                <TITLE>dc_cleaning_OPENAIREplus_compliant_creator_comma</TITLE>
14
                <CODE><![CDATA[
15
declare_script "dc_cleaning_OpenAIREplus_compliant_creator_comma";
16
declare_ns oaf = "http://namespace.openaire.eu/oaf";
17
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
18
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
19
declare_ns dc = "http://purl.org/dc/elements/1.1/";
20
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
21
$var0 = "''";
22
$var1 = "'corda_______::'";
23
$varDummy = "''";
24
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
25
static $varRepoid = xpath:"//dri:repositoryId";
26
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//OFFICIAL_NAME"]);
27
dri:objIdentifier = xpath:"//dri:objIdentifier";
28
dri:repositoryId = $varRepoid;
29
//dri:repositoryId = xpath:"//dri:repositoryId";
30
dri:recordIdentifier = RegExpr(xpath:"//dri:recordIdentifier", $var0, "s/^(.*)(::)/$2/");
31
if xpath:"count(//dc:creator) = 0" dc:creator = skipRecord(); else $varDummy = "''";
32
apply xpath:"//dc:creator" if xpath:"string-length(.) > 0" dc:creator = xpath:"concat(substring-before(normalize-space(.), ' '), ', ', substring-after(normalize-space(.), ' '))"; else $varDummy = "''";
33
apply xpath:"//dc:contributor" if xpath:"string-length(.) > 0" dc:contributor = xpath:"concat(substring-before(normalize-space(.), ' '), ', ', substring-after(normalize-space(.), ' '))"; else $varDummy = "''";
34
apply xpath:"//dc:title" if xpath:"string-length(.) > 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
35
apply xpath:"//dc:subject" if xpath:"string-length(.) > 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
36
apply xpath:"//dc:publisher" if xpath:"string-length(.) > 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
37
apply xpath:"//dc:source" if xpath:"string-length(.) > 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
38
dc:description = xpath:"//dc:description";
39
$varHttpTest = "''";
40
if xpath:"//dc:identifier[starts-with(., 'http')]" $varHttpTest = "true"; else dc:identifier = skipRecord();
41
apply xpath:"//dc:identifier" if xpath:"starts-with(normalize-space(.), 'http')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
42
static dr:dateOfCollection = getValue(CURRENTDATE, []);
43
// dc:type = xpath:"//dc:type";
44
dc:language = Convert(xpath:"//dc:language", Languages);
45
//dc:language = "eng";
46
//if xpath:"//dc:rights[text()='info:eu-repo/semantics/openAccess']" dc:publisher = xpath:"//dc:publisher"; else dc:publisher = skipRecord();
47
dc:date = xpath:"//dc:date";
48
oaf:dateAccepted = Convert(xpath:"descendant-or-self::dc:date", DateISO8601, "yyyy-MM-dd", "min()");
49
apply xpath:"//dc:date" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/date')" oaf:embargoenddate = RegExpr(xpath:"normalize-space(.)", $var0, "s/^(.*info:eu-repo\/date\/embargoEnd\/)//gmi"); else $var0 = "''";
50
apply xpath:"//dc:relation" if xpath:"string-length(substring-after(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7/')) = 6 and not(contains(normalize-space(.), '/12345'))" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
51
//apply xpath:"//dc:relation" if xpath:"string-length(translate(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7','')) = 6" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
52
//apply xpath:"//dc:relation" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/grantAgreement/EC/FP7')" oaf:projectid = RegExpr(xpath:"normalize-space(.)", $var1, "s/^(.*info:eu-repo\/grantAgreement\/EC\/FP7\/)//gmi"); else dc:relation = xpath:"normalize-space(.)";
53
//comment-js-09-10-2012 apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" dc:rights = empty; else dc:rights = xpath:"normalize-space(.)";
... This diff was truncated because it exceeds the maximum size that can be displayed.

Also available in: Unified diff