1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="b6addffa-bae1-42db-811c-ae20a5c514a6_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
|
4
|
<RESOURCE_TYPE value="WorkflowDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2006-05-04T18:13:51.0Z"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<WORKFLOW_NAME menuSection="InfoSpace Provision">OpenAIRE Provision</WORKFLOW_NAME>
|
11
|
<WORKFLOW_DESCRIPTION>OpenAIRE Data Provision</WORKFLOW_DESCRIPTION>
|
12
|
<WORKFLOW_INFO/>
|
13
|
<WORKFLOW_FAMILY>Data Provision</WORKFLOW_FAMILY>
|
14
|
<WORKFLOW_PRIORITY>35</WORKFLOW_PRIORITY>
|
15
|
<!-- <ADMIN_EMAIL>alessia.bardi@isti.cnr.it,claudio.atzori@isti.cnr.it</ADMIN_EMAIL> -->
|
16
|
<CONFIGURATION status="EXECUTABLE" start="MANUAL">
|
17
|
|
18
|
<PARAMETERS>
|
19
|
<PARAM name="table" description="HBase table to be dropped" required="true" managedBy="user" type="string"></PARAM>
|
20
|
<PARAM name="cluster" description="Hadoop cluster logical name" required="true" managedBy="user" type="string" function="validValues(['DM','IIS'])"></PARAM>
|
21
|
<PARAM name="reuseRegionInfo" description="Reuse table splits" required="true" managedBy="user" type="boolean">true</PARAM>
|
22
|
|
23
|
<PARAM name="tableConf" description="HBase table configuration param name" required="true" managedBy="system" type="string">hbase.table.conf</PARAM>
|
24
|
<PARAM name="tableCols" description="HBase table schema param name" required="true" managedBy="system" type="string">hbase.table.cols</PARAM>
|
25
|
|
26
|
<PARAM name="db" description="relational db name" required="true" type="string" managedBy="user" />
|
27
|
|
28
|
<PARAM name="mappingDS" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
29
|
<PARAM name="mappingDS_ORG" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
30
|
<PARAM name="mappingORG" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
31
|
<PARAM name="mappingPERSON" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
32
|
<PARAM name="mappingPRJ_ORG" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
33
|
<PARAM name="mappingPRJ_PERSON" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
34
|
<PARAM name="mappingPRJ" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
35
|
<PARAM name="mappingCLAIM_RELS" description="xslt mapping function" required="true" type="string" managedBy="user" function="obtainValues('dbmf2hbaseMappings', {})" />
|
36
|
|
37
|
|
38
|
<PARAM name="mappingPublication" description="xslt mapping for publications" required="true" type="string" managedBy="user" function="obtainValues('oaf2hbaseMappings', {})" />
|
39
|
<PARAM name="reusePublication" description="reuse publications on HDFS?" required="true" type="boolean" managedBy="user" />
|
40
|
<PARAM name="hdfsPathPublication" description="hdfs path for publications" required="true" type="string" managedBy="user" />
|
41
|
|
42
|
<PARAM name="mappingDataset" description="xslt mapping for datasets" required="true" type="string" managedBy="user" function="obtainValues('odf2hbaseMappings', {})" />
|
43
|
<PARAM name="reuseDataset" description="reuse datasets on HDFS?" required="true" type="boolean" managedBy="user" />
|
44
|
<PARAM name="hdfsPathDataset" description="hdfs path for datasets" required="true" type="string" managedBy="user" />
|
45
|
|
46
|
</PARAMETERS>
|
47
|
<WORKFLOW>
|
48
|
|
49
|
<NODE name="resetHbase" type="LaunchWorkflowTemplate" isStart="true">
|
50
|
<DESCRIPTION>Reset HBase table</DESCRIPTION>
|
51
|
<PARAMETERS>
|
52
|
<PARAM name="wfTemplateId" value="75345aba-c069-43f4-90aa-e13688d9845e_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ==" />
|
53
|
<PARAM name="wfTemplateParams">
|
54
|
<MAP>
|
55
|
<ENTRY key="table" ref="table"/>
|
56
|
<ENTRY key="cluster" ref="cluster"/>
|
57
|
<ENTRY key="reuseRegionInfo" ref="reuseRegionInfo"/>
|
58
|
</MAP>
|
59
|
</PARAM>
|
60
|
</PARAMETERS>
|
61
|
|
62
|
<ARCS>
|
63
|
<ARC to="db2hbase" />
|
64
|
</ARCS>
|
65
|
</NODE>
|
66
|
|
67
|
<NODE name="db2hbase" type="LaunchWorkflowTemplate">
|
68
|
<DESCRIPTION>load db to hbase</DESCRIPTION>
|
69
|
<PARAMETERS>
|
70
|
<PARAM name="wfTemplateId" value="3ffe4cd8-73f1-4566-85fc-26ac32e8af56_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ==" />
|
71
|
<PARAM name="wfTemplateParams">
|
72
|
<MAP>
|
73
|
<ENTRY key="db" ref="db" />
|
74
|
<ENTRY key="hbaseTable" ref="table" />
|
75
|
<ENTRY key="cluster" ref="cluster" />
|
76
|
|
77
|
<ENTRY key="sqlQueryDS" value="/eu/dnetlib/msro/workflows/hbase/queryDatasources.sql" />
|
78
|
<ENTRY key="mappingDS" ref="mappingDS" />
|
79
|
|
80
|
<ENTRY key="sqlQueryDS_ORG" value="/eu/dnetlib/msro/workflows/hbase/queryDatasourceOrganization.sql" />
|
81
|
<ENTRY key="mappingDS_ORG" ref="mappingDS_ORG" />
|
82
|
|
83
|
<ENTRY key="sqlQueryORG" value="/eu/dnetlib/msro/workflows/hbase/queryOrganizations.sql"/>
|
84
|
<ENTRY key="mappingORG" ref="mappingORG" />
|
85
|
|
86
|
<ENTRY key="sqlQueryPERSON" value="/eu/dnetlib/msro/workflows/hbase/queryPersons.sql"/>
|
87
|
<ENTRY key="mappingPERSON" ref="mappingPERSON" />
|
88
|
|
89
|
<ENTRY key="sqlQueryPRJ_ORG" value="/eu/dnetlib/msro/workflows/hbase/queryProjectOrganization.sql"/>
|
90
|
<ENTRY key="mappingPRJ_ORG" ref="mappingPRJ_ORG" />
|
91
|
|
92
|
<ENTRY key="sqlQueryPRJ_PERSON" value="/eu/dnetlib/msro/workflows/hbase/queryProjectContactperson.sql"/>
|
93
|
<ENTRY key="mappingPRJ_PERSON" ref="mappingPRJ_PERSON" />
|
94
|
|
95
|
<ENTRY key="sqlQueryPRJ" value="/eu/dnetlib/msro/workflows/hbase/queryProjects.sql"/>
|
96
|
<ENTRY key="mappingPRJ" ref="mappingPRJ" />
|
97
|
|
98
|
<ENTRY key="sqlQueryCLAIM_RELS" value="/eu/dnetlib/msro/workflows/hbase/queryClaimsRel.sql"/>
|
99
|
<ENTRY key="mappingCLAIM_RELS" ref="mappingCLAIM_RELS" />
|
100
|
</MAP>
|
101
|
</PARAM>
|
102
|
</PARAMETERS>
|
103
|
|
104
|
<ARCS>
|
105
|
<ARC to="oaf2hbase" />
|
106
|
</ARCS>
|
107
|
|
108
|
</NODE>
|
109
|
|
110
|
<NODE name="oaf2hbase" type="LaunchWorkflowTemplate">
|
111
|
<DESCRIPTION>OAF to HBase</DESCRIPTION>
|
112
|
<PARAMETERS>
|
113
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ==" />
|
114
|
<PARAM name="wfTemplateParams">
|
115
|
<MAP>
|
116
|
<ENTRY key="hbaseTable" ref="table"/>
|
117
|
<ENTRY key="cluster" ref="cluster"/>
|
118
|
<ENTRY key="mdFormat" value="OAF"/>
|
119
|
<ENTRY key="mapping" ref="mappingPublication"/>
|
120
|
<ENTRY key="reuseMdRecords" ref="reusePublication"/>
|
121
|
<ENTRY key="hdfsPath" ref="hdfsPathPublication"/>
|
122
|
</MAP>
|
123
|
</PARAM>
|
124
|
</PARAMETERS>
|
125
|
<ARCS>
|
126
|
<ARC to="odf2hbase" />
|
127
|
</ARCS>
|
128
|
</NODE>
|
129
|
|
130
|
|
131
|
<NODE name="odf2hbase" type="LaunchWorkflowTemplate">
|
132
|
<DESCRIPTION>OAF to HBase</DESCRIPTION>
|
133
|
<PARAMETERS>
|
134
|
<PARAM name="wfTemplateId" value="5b05a65a-4eeb-4862-bc55-b35c7ec3baf0_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ==" />
|
135
|
<PARAM name="wfTemplateParams">
|
136
|
<MAP>
|
137
|
<ENTRY key="hbaseTable" ref="table"/>
|
138
|
<ENTRY key="cluster" ref="cluster"/>
|
139
|
<ENTRY key="mdFormat" value="ODF"/>
|
140
|
<ENTRY key="mapping" ref="mappingDataset"/>
|
141
|
<ENTRY key="reuseMdRecords" ref="reuseDataset"/>
|
142
|
<ENTRY key="hdfsPath" ref="hdfsPathDataset"/>
|
143
|
</MAP>
|
144
|
</PARAM>
|
145
|
</PARAMETERS>
|
146
|
<ARCS>
|
147
|
<ARC to="actions2hbase" />
|
148
|
</ARCS>
|
149
|
</NODE>
|
150
|
|
151
|
<NODE name="actions2hbase">
|
152
|
<DESCRIPTION>actions to hbase</DESCRIPTION>
|
153
|
<PARAMETERS/>
|
154
|
|
155
|
<ARCS>
|
156
|
<ARC to="dedupPerson" />
|
157
|
</ARCS>
|
158
|
</NODE>
|
159
|
<NODE name="dedupPerson">
|
160
|
<DESCRIPTION>dedup person</DESCRIPTION>
|
161
|
<PARAMETERS/>
|
162
|
|
163
|
<ARCS>
|
164
|
<ARC to="closeMesh" />
|
165
|
</ARCS>
|
166
|
</NODE>
|
167
|
<NODE name="closeMesh">
|
168
|
<DESCRIPTION>close similarity mesh</DESCRIPTION>
|
169
|
<PARAMETERS/>
|
170
|
|
171
|
<ARCS>
|
172
|
<ARC to="applyBlacklist" />
|
173
|
</ARCS>
|
174
|
</NODE>
|
175
|
<NODE name="applyBlacklist">
|
176
|
<DESCRIPTION>apply blacklist</DESCRIPTION>
|
177
|
<PARAMETERS/>
|
178
|
|
179
|
<ARCS>
|
180
|
<ARC to="updateProvision" />
|
181
|
</ARCS>
|
182
|
</NODE>
|
183
|
<NODE name="updateProvision">
|
184
|
<DESCRIPTION>update provision</DESCRIPTION>
|
185
|
<PARAMETERS/>
|
186
|
|
187
|
<ARCS>
|
188
|
<ARC to="preparePrepublic" />
|
189
|
</ARCS>
|
190
|
</NODE>
|
191
|
<NODE name="preparePrepublic">
|
192
|
<DESCRIPTION>prepare pre-public infospace</DESCRIPTION>
|
193
|
<PARAMETERS/>
|
194
|
|
195
|
<ARCS>
|
196
|
<ARC to="success" />
|
197
|
</ARCS>
|
198
|
</NODE>
|
199
|
|
200
|
</WORKFLOW>
|
201
|
|
202
|
</CONFIGURATION>
|
203
|
|
204
|
<NOTIFICATIONS/>
|
205
|
|
206
|
<SCHEDULING enabled="false">
|
207
|
<CRON>29 5 22 ? * *</CRON>
|
208
|
<MININTERVAL>10080</MININTERVAL>
|
209
|
</SCHEDULING>
|
210
|
|
211
|
<STATUS/>
|
212
|
</BODY>
|
213
|
</RESOURCE_PROFILE>
|