Project

General

Profile

« Previous | Next » 

Revision 60995

Added by Alessia Bardi over 3 years ago

Workflows that import data by-passing the aggregation system

View differences:

modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-scholexplorer.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="e03f256e-1e4d-4b3d-9c07-91faf5d25207_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2019-09-26T16:09:43+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Import ScholExplorer Links and entities</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="setInputPath" type="SetHdfsFile">
15 
                <DESCRIPTION>set the hdfs output path</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="user" name="hdfsPath" required="true" type="string">/usr/dnet/data/scholixDumpExport_2019_09_26</PARAM>
18 
                    <PARAM managedBy="system" name="hdfsPathParam" required="true" type="string">inputPath</PARAM>
19 
                </PARAMETERS>
20 
                <ARCS>
21 
                    <ARC to="prepareActionSets"/>
22 
                </ARCS>
23 
            </NODE>
24 
            <NODE name="prepareActionSets" type="PrepareActionSets">
25 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
26 
                <PARAMETERS>
27 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
28 
                        [
29 
                        {
30 
                        'set' : 'scholexplorer-dump',
31 
                        'jobProperty' : 'export_action_set_scholexplorer_dump',
32 
                        'enablingProperty' : 'active_scholexplorer_dump',
33 
                        'enabled' : 'true'
34 
                        }
35 
                        ]
36 
                    </PARAM>
37 
                </PARAMETERS>
38 
                <ARCS>
39 
                    <ARC to="extractOutputPath"/>
40 
                </ARCS>
41 
            </NODE>
42 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
43 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
44 
                <PARAMETERS>
45 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
46 
                </PARAMETERS>
47 
                <ARCS>
48 
                    <ARC to="importActionSet"/>
49 
                </ARCS>
50 
            </NODE>
51 
            <NODE isJoin="true" name="importActionSet" type="SubmitHadoopJob">
52 
                <DESCRIPTION>IIS main</DESCRIPTION>
53 
                <PARAMETERS>
54 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">importScholexplorerJob</PARAM>
55 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">DM</PARAM>
56 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
57 
                        {
58 
                        'mapred.input.dir':'inputPath',
59 
                        'mapred.output.dir':'outputPath'
60 
                        }
61 
                    </PARAM>
62 
                    <PARAM managedBy="user" name="simulation" required="true" type="boolean">false</PARAM>
63 
                </PARAMETERS>
64 
                <ARCS>
65 
                    <ARC to="updateActionSets"/>
66 
                </ARCS>
67 
            </NODE>
68 
            <NODE name="updateActionSets" type="UpdateActionSets">
69 
                <DESCRIPTION>update action sets</DESCRIPTION>
70 
                <PARAMETERS/>
71 
                <ARCS>
72 
                    <ARC to="success"/>
73 
                </ARCS>
74 
            </NODE>
75 
        </CONFIGURATION>
76 
        <STATUS>
77 
            <LAST_EXECUTION_ID>wf_20191022_141741_170</LAST_EXECUTION_ID>
78 
            <LAST_EXECUTION_DATE>2019-10-22T14:32:03+00:00</LAST_EXECUTION_DATE>
79 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
80 
            <LAST_EXECUTION_ERROR/>
81 
        </STATUS>
82 
    </BODY>
83 
</RESOURCE_PROFILE>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-doiboost-beta.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="e03f256e-1e4d-4b3d-9c07-91faf5d25208_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2021-04-19T13:28:18+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Import DOIboost</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="setMAGDumpPath" type="SetEnvParameter">
15 
                <DESCRIPTION>set the input path for MAG</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">MAGDumpPath</PARAM>
18 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/mag-2020-11-09</PARAM>
19 
                </PARAMETERS>
20 
                <ARCS>
21 
                    <ARC to="waitConfig"/>
22 
                </ARCS>
23 
            </NODE>
24 
            <NODE isStart="true" name="setInputPathMAG" type="SetEnvParameter">
25 
                <DESCRIPTION>set the input path for MAG</DESCRIPTION>
26 
                <PARAMETERS>
27 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathMAG</PARAM>
28 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/mag</PARAM>
29 
                </PARAMETERS>
30 
                <ARCS>
31 
                    <ARC to="waitConfig"/>
32 
                </ARCS>
33 
            </NODE>
34 
            <NODE isStart="true" name="setInputPathCrossref" type="SetEnvParameter">
35 
                <DESCRIPTION>set the input path for Crossref</DESCRIPTION>
36 
                <PARAMETERS>
37 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathCrossref</PARAM>
38 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/crossref</PARAM>
39 
                </PARAMETERS>
40 
                <ARCS>
41 
                    <ARC to="waitConfig"/>
42 
                </ARCS>
43 
            </NODE>
44 
            <NODE isStart="true" name="setCrossrefTimestamp" type="SetEnvParameter">
45 
                <DESCRIPTION>set the timestamp for the Crossref incremental harvesting</DESCRIPTION>
46 
                <PARAMETERS>
47 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">crossrefTimestamp</PARAM>
48 
                    <PARAM managedBy="user" name="parameterValue" required="false" type="string">1607614921429</PARAM>
49 
                </PARAMETERS>
50 
                <ARCS>
51 
                    <ARC to="waitConfig"/>
52 
                </ARCS>
53 
            </NODE>
54 
            <NODE isStart="true" name="setInputPathUnpayWall" type="SetEnvParameter">
55 
                <DESCRIPTION>set the input path for UnpayWall</DESCRIPTION>
56 
                <PARAMETERS>
57 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathUnpayWall</PARAM>
58 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/unpayWall</PARAM>
59 
                </PARAMETERS>
60 
                <ARCS>
61 
                    <ARC to="waitConfig"/>
62 
                </ARCS>
63 
            </NODE>
64 
            <NODE isStart="true" name="setInputPathOrcid" type="SetEnvParameter">
65 
                <DESCRIPTION>set the input path for ORCID</DESCRIPTION>
66 
                <PARAMETERS>
67 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathOrcid</PARAM>
68 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/orcid_activities_2020/last_orcid_dataset</PARAM>
69 
                </PARAMETERS>
70 
                <ARCS>
71 
                    <ARC to="waitConfig"/>
72 
                </ARCS>
73 
            </NODE>
74 
            <NODE isStart="true" name="setWorkingPathOrcid" type="SetEnvParameter">
75 
                <DESCRIPTION>set the working path for ORCID</DESCRIPTION>
76 
                <PARAMETERS>
77 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">workingPathOrcid</PARAM>
78 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/orcid</PARAM>
79 
                </PARAMETERS>
80 
                <ARCS>
81 
                    <ARC to="waitConfig"/>
82 
                </ARCS>
83 
            </NODE>
84 
            <NODE isStart="true" name="setHostedByMapPath" type="SetEnvParameter">
85 
                <DESCRIPTION>set the hostedBy map path</DESCRIPTION>
86 
                <PARAMETERS>
87 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">hostedByMapPath</PARAM>
88 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/hostedBy/hbMap.gz</PARAM>
89 
                </PARAMETERS>
90 
                <ARCS>
91 
                    <ARC to="waitConfig"/>
92 
                </ARCS>
93 
            </NODE>
94 
            <NODE isStart="true" name="setResumeFrom" type="SetEnvParameter">
95 
                <DESCRIPTION>set the oozie workflow name from which the execution will be resumed</DESCRIPTION>
96 
                <PARAMETERS>
97 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">resumeFrom</PARAM>
98 
                    <PARAM function="validValues(['ConvertCrossrefToOAF','ResetMagWorkingPath', 'PreprocessMag', 'PreprocessUW', 'PreprocessORCID', 'CreateDOIBoost', 'GenerateActionSet'])" managedBy="user" name="parameterValue" required="false" type="string">PreprocessORCID</PARAM>
99 
                </PARAMETERS>
100 
                <ARCS>
101 
                    <ARC to="waitConfig"/>
102 
                </ARCS>
103 
            </NODE>
104 
            <NODE isJoin="true" name="waitConfig">
105 
                <DESCRIPTION>wait configurations</DESCRIPTION>
106 
                <PARAMETERS/>
107 
                <ARCS>
108 
                    <ARC to="prepareActionSets"/>
109 
                </ARCS>
110 
            </NODE>
111 
            <NODE name="prepareActionSets" type="PrepareActionSets">
112 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
113 
                <PARAMETERS>
114 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
115 
                        [
116 
                        {
117 
                        'set' : 'doiboost',
118 
                        'jobProperty' : 'export_action_set_doiboost',
119 
                        'enablingProperty' : 'active_doiboost',
120 
                        'enabled' : 'true'
121 
                        }
122 
                        ]
123 
                    </PARAM>
124 
                </PARAMETERS>
125 
                <ARCS>
126 
                    <ARC to="extractOutputPath"/>
127 
                </ARCS>
128 
            </NODE>
129 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
130 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
131 
                <PARAMETERS>
132 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
133 
                </PARAMETERS>
134 
                <ARCS>
135 
                    <ARC to="updateDOIBoost"/>
136 
                </ARCS>
137 
            </NODE>
138 
            <NODE name="updateDOIBoost" type="SubmitHadoopJob">
139 
                <DESCRIPTION>prepare a new version of DOIBoost</DESCRIPTION>
140 
                <PARAMETERS>
141 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
142 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
143 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
144 
                        {
145 
                        'crossrefTimestamp' : 'crossrefTimestamp',
146 
                        'hostedByMapPath' : 'hostedByMapPath',
147 
                        'MAGDumpPath' :'MAGDumpPath',
148 
                        'inputPathMAG' : 'inputPathMAG',
149 
                        'inputPathCrossref' : 'inputPathCrossref',
150 
                        'inputPathUnpayWall' : 'inputPathUnpayWall',
151 
                        'inputPathOrcid' : 'inputPathOrcid',
152 
                        'outputPath' : 'outputPath',
153 
                        'workingPathOrcid':'workingPathOrcid',
154 
                        'resumeFrom' : 'resumeFrom'
155 
                        }
156 
                    </PARAM>
157 
                    <PARAM managedBy="system" name="params" required="true" type="string">
158 
                        {
159 
                        'oozie.wf.application.path' : ' /lib/dnet/STABLE_IDS/actionmanager/doiboost/oozie_app',
160 
                        'workingPath' : '/data/doiboost/process',
161 
                        'sparkExecutorIntersectionMemory' : '9G',
162 
                        'sparkExecutorMemory' : '6G',
163 
                        'esServer' : 'ip-90-147-167-25.ct1.garrservices.it',
164 
                        'esIndex' : 'crossref'
165 
                        }
166 
                    </PARAM>
167 
                    <PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
168 
                </PARAMETERS>
169 
                <ARCS>
170 
                    <ARC to="updateActionSets"/>
171 
                </ARCS>
172 
            </NODE>
173 
            <NODE name="updateActionSets" type="UpdateActionSets">
174 
                <DESCRIPTION>update action sets</DESCRIPTION>
175 
                <PARAMETERS/>
176 
                <ARCS>
177 
                    <ARC to="success"/>
178 
                </ARCS>
179 
            </NODE>
180 
        </CONFIGURATION>
181 
        <STATUS>
182 
            <LAST_EXECUTION_ID>wf_20210416_122501_878</LAST_EXECUTION_ID>
183 
            <LAST_EXECUTION_DATE>2021-04-16T13:58:50+00:00</LAST_EXECUTION_DATE>
184 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
185 
            <LAST_EXECUTION_ERROR/>
186 
        </STATUS>
187 
    </BODY>
188 
</RESOURCE_PROFILE>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-ror-beta.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="7cf3cfed-fbfb-46ca-b4da-aa43beb58f19_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2021-05-06T09:48:33+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Update ROR actionset</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import Infospace</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="setInputPath" type="SetEnvParameter">
15 
                <DESCRIPTION>Set the base path containing the no_doi_dataset folder</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPath</PARAM>
18 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/ror/ror-data-2021-04-06.json</PARAM>
19 
                </PARAMETERS>
20 
                <ARCS>
21 
                    <ARC to="prepareActionSets"/>
22 
                </ARCS>
23 
            </NODE>
24 
            <NODE name="prepareActionSets" type="PrepareActionSets">
25 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
26 
                <PARAMETERS>
27 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
28 
                        [
29 
                        {
30 
                        'set' : 'ror',
31 
                        'jobProperty' : 'export_action_set_ror',
32 
                        'enablingProperty' : 'active_ror',
33 
                        'enabled' : 'true'
34 
                        }
35 
                        ]
36 
                    </PARAM>
37 
                </PARAMETERS>
38 
                <ARCS>
39 
                    <ARC to="extractOutputPath"/>
40 
                </ARCS>
41 
            </NODE>
42 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
43 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
44 
                <PARAMETERS>
45 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
46 
                </PARAMETERS>
47 
                <ARCS>
48 
                    <ARC to="rorUpdate"/>
49 
                </ARCS>
50 
            </NODE>
51 
            <NODE name="rorUpdate" type="SubmitHadoopJob">
52 
                <DESCRIPTION>update the ROR actionset</DESCRIPTION>
53 
                <PARAMETERS>
54 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
55 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
56 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
57 
                        {
58 
                        'rorJsonInputPath' : 'inputPath',
59 
                        'rorActionSetPath': 'outputPath'
60 
                        }
61 
                    </PARAM>
62 
                    <PARAM managedBy="system" name="params" required="true" type="string">
63 
                        {
64 
                        'oozie.wf.application.path' : '/lib/dnet/STABLE_IDS/actionmanager/ror/oozie_app',
65 
                        'workingDir': '/tmp/import_ror_actionset'
66 
                        }
67 
                    </PARAM>
68 
                    <PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
69 
                </PARAMETERS>
70 
                <ARCS>
71 
                    <ARC to="updateActionSets"/>
72 
                </ARCS>
73 
            </NODE>
74 
            <NODE name="updateActionSets" type="UpdateActionSets">
75 
                <DESCRIPTION>update action sets</DESCRIPTION>
76 
                <PARAMETERS/>
77 
                <ARCS>
78 
                    <ARC to="success"/>
79 
                </ARCS>
80 
            </NODE>
81 
        </CONFIGURATION>
82 
        <STATUS>
83 
            <LAST_EXECUTION_ID>wf_20210506_135405_174</LAST_EXECUTION_ID>
84 
            <LAST_EXECUTION_DATE>2021-05-06T13:55:05+00:00</LAST_EXECUTION_DATE>
85 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
86 
            <LAST_EXECUTION_ERROR/>
87 
        </STATUS>
88 
    </BODY>
89 
</RESOURCE_PROFILE>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-h2020-classification-beta.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="4bb067d5-a2f2-42b9-844c-4e1d8d71b80f_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2020-12-01T14:33:42+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Import H2020classification</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import H2020classification</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="prepareActionSets" type="PrepareActionSets">
15 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
18 
                        [
19 
                        {
20 
                        'set' : 'h2020classification',
21 
                        'jobProperty' : 'export_action_set_h2020classification',
22 
                        'enablingProperty' : 'active_h2020classification',
23 
                        'enabled' : 'true'
24 
                        }
25 
                        ]
26 
                    </PARAM>
27 
                </PARAMETERS>
28 
                <ARCS>
29 
                    <ARC to="extractOutputPath"/>
30 
                </ARCS>
31 
            </NODE>
32 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
33 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
34 
                <PARAMETERS>
35 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
36 
                </PARAMETERS>
37 
                <ARCS>
38 
                    <ARC to="h2020ClassificationUpdate"/>
39 
                </ARCS>
40 
            </NODE>
41 
            <NODE name="h2020ClassificationUpdate" type="SubmitHadoopJob">
42 
                <DESCRIPTION>prepare updates for the H2020 Classification</DESCRIPTION>
43 
                <PARAMETERS>
44 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
45 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
46 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
47 
                        {
48 
                        'outputPath': 'outputPath'
49 
                        }
50 
                    </PARAM>
51 
                    <PARAM managedBy="system" name="params" required="true" type="string">
52 
                        {
53 
                        'oozie.wf.application.path' : ' /lib/dnet/BETA/actionmanager/project/oozie_app',
54 
                        'workingDir' : '/tmp/beta_provision/working_dir/h2020classification',
55 
                        'projectFileURL' : 'https://cordis.europa.eu/data/cordis-h2020projects.csv',
56 
                        'programmeFileURL' : 'http://cordis.europa.eu/data/reference/cordisref-H2020programmes.csv',
57 
                        'topicFileURL':'http://cordis.europa.eu/data/reference/cordisref-H2020topics.xlsx',
58 
                        'postgresURL':'jdbc:postgresql://beta.services.openaire.eu:5432/dnet_openaireplus',
59 
                        'postgresUser':'dnet',
60 
                        'postgresPassword':'dnetPwd'
61 
                        }
62 
                    </PARAM>
63 
                    <PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
64 
                </PARAMETERS>
65 
                <ARCS>
66 
                    <ARC to="updateActionSets"/>
67 
                </ARCS>
68 
            </NODE>
69 
            <NODE name="updateActionSets" type="UpdateActionSets">
70 
                <DESCRIPTION>update action sets</DESCRIPTION>
71 
                <PARAMETERS/>
72 
                <ARCS>
73 
                    <ARC to="success"/>
74 
                </ARCS>
75 
            </NODE>
76 
        </CONFIGURATION>
77 
        <STATUS>
78 
            <LAST_EXECUTION_ID>wf_20201005_124642_515</LAST_EXECUTION_ID>
79 
            <LAST_EXECUTION_DATE>2020-10-05T12:57:35+00:00</LAST_EXECUTION_DATE>
80 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
81 
            <LAST_EXECUTION_ERROR/>
82 
        </STATUS>
83 
    </BODY>
84 
</RESOURCE_PROFILE>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-orcid-nodoi-beta.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="7cf3cfed-fbfb-46ca-b4da-aa43beb58f18_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2021-04-22T09:03:40+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Update ORCID (no DOI) actionset</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import Infospace</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="setInputPath" type="SetEnvParameter">
15 
                <DESCRIPTION>Set the base path containing the no_doi_dataset folder</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPath</PARAM>
18 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/orcid_activities_2020</PARAM>
19 
                </PARAMETERS>
20 
                <ARCS>
21 
                    <ARC to="prepareActionSets"/>
22 
                </ARCS>
23 
            </NODE>
24 
            <NODE name="prepareActionSets" type="PrepareActionSets">
25 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
26 
                <PARAMETERS>
27 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
28 
                        [
29 
                        {
30 
                        'set' : 'orcidworks-no-doi',
31 
                        'jobProperty' : 'export_action_set_orcidworks_no_doi',
32 
                        'enablingProperty' : 'active_orcidworks_no_doi',
33 
                        'enabled' : 'true'
34 
                        }
35 
                        ]
36 
                    </PARAM>
37 
                </PARAMETERS>
38 
                <ARCS>
39 
                    <ARC to="extractOutputPath"/>
40 
                </ARCS>
41 
            </NODE>
42 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
43 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
44 
                <PARAMETERS>
45 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
46 
                </PARAMETERS>
47 
                <ARCS>
48 
                    <ARC to="orcidNoDoiUpdate"/>
49 
                </ARCS>
50 
            </NODE>
51 
            <NODE name="orcidNoDoiUpdate" type="SubmitHadoopJob">
52 
                <DESCRIPTION>prepare updates for the Orcid No Doi</DESCRIPTION>
53 
                <PARAMETERS>
54 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
55 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
56 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
57 
                        {
58 
                        'workingPath' : 'inputPath',
59 
                        'outputPath': 'outputPath'
60 
                        }
61 
                    </PARAM>
62 
                    <PARAM managedBy="system" name="params" required="true" type="string">
63 
                        {
64 
                        'oozie.wf.application.path' : '/lib/dnet/STABLE_IDS/actionmanager/orcidnodoi_actionset/oozie_app',
65 
                        'spark2GenNoDoiDatasetMaxExecutors' : '200',
66 
                        'spark2GenNoDoiDatasetExecutorMemory' : '2G'
67 
                        }
68 
                    </PARAM>
69 
                    <PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
70 
                </PARAMETERS>
71 
                <ARCS>
72 
                    <ARC to="updateActionSets"/>
73 
                </ARCS>
74 
            </NODE>
75 
            <NODE name="updateActionSets" type="UpdateActionSets">
76 
                <DESCRIPTION>update action sets</DESCRIPTION>
77 
                <PARAMETERS/>
78 
                <ARCS>
79 
                    <ARC to="success"/>
80 
                </ARCS>
81 
            </NODE>
82 
        </CONFIGURATION>
83 
        <STATUS>
84 
            <LAST_EXECUTION_ID>wf_20210505_073553_347</LAST_EXECUTION_ID>
85 
            <LAST_EXECUTION_DATE>2021-05-05T07:43:23+00:00</LAST_EXECUTION_DATE>
86 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
87 
            <LAST_EXECUTION_ERROR/>
88 
        </STATUS>
89 
    </BODY>
90 
</RESOURCE_PROFILE>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-doiboost-prod.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="e03f256e-1e4d-4b3d-9c07-91faf5d25208_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2021-05-07T13:09:35+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Import DOIboost</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="setInputPathMAG" type="SetEnvParameter">
15 
                <DESCRIPTION>set the input path for MAG</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathMAG</PARAM>
18 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/mag</PARAM>
19 
                </PARAMETERS>
20 
                <ARCS>
21 
                    <ARC to="waitConfig"/>
22 
                </ARCS>
23 
            </NODE>
24 
            <NODE isStart="true" name="setInputPathCrossref" type="SetEnvParameter">
25 
                <DESCRIPTION>set the input path for Crossref</DESCRIPTION>
26 
                <PARAMETERS>
27 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathCrossref</PARAM>
28 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/crossref</PARAM>
29 
                </PARAMETERS>
30 
                <ARCS>
31 
                    <ARC to="waitConfig"/>
32 
                </ARCS>
33 
            </NODE>
34 
            <NODE isStart="true" name="setInputPathUnpayWall" type="SetEnvParameter">
35 
                <DESCRIPTION>set the input path for UnpayWall</DESCRIPTION>
36 
                <PARAMETERS>
37 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">inputPathUnpayWall</PARAM>
38 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/unpayWall</PARAM>
39 
                </PARAMETERS>
40 
                <ARCS>
41 
                    <ARC to="waitConfig"/>
42 
                </ARCS>
43 
            </NODE>
44 
            <NODE isStart="true" name="setWorkingPathOrcid" type="SetEnvParameter">
45 
                <DESCRIPTION>set the working path for ORCID</DESCRIPTION>
46 
                <PARAMETERS>
47 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">workingPathOrcid</PARAM>
48 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/orcid</PARAM>
49 
                </PARAMETERS>
50 
                <ARCS>
51 
                    <ARC to="waitConfig"/>
52 
                </ARCS>
53 
            </NODE>
54 
            <NODE isStart="true" name="setHostedByMapPath" type="SetEnvParameter">
55 
                <DESCRIPTION>set the hostedBy map path</DESCRIPTION>
56 
                <PARAMETERS>
57 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">hostedByMapPath</PARAM>
58 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/doiboost/input/hostedBy/hbMap.gz</PARAM>
59 
                </PARAMETERS>
60 
                <ARCS>
61 
                    <ARC to="waitConfig"/>
62 
                </ARCS>
63 
            </NODE>
64 
            <NODE isStart="true" name="setResumeFrom" type="SetEnvParameter">
65 
                <DESCRIPTION>set the oozie workflow name from which the execution will be resumed</DESCRIPTION>
66 
                <PARAMETERS>
67 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">resumeFrom</PARAM>
68 
                    <PARAM function="validValues(['ConvertCrossrefToOAF','PreprocessMag','PreprocessUW', 'ProcessORCID', 'CreateDOIBoost', 'GenerateActionSet'])" managedBy="user" name="parameterValue" required="false" type="string">ConvertCrossrefToOAF</PARAM>
69 
                </PARAMETERS>
70 
                <ARCS>
71 
                    <ARC to="waitConfig"/>
72 
                </ARCS>
73 
            </NODE>
74 
            <NODE isJoin="true" name="waitConfig">
75 
                <DESCRIPTION>wait configurations</DESCRIPTION>
76 
                <PARAMETERS/>
77 
                <ARCS>
78 
                    <ARC to="prepareActionSets"/>
79 
                </ARCS>
80 
            </NODE>
81 
            <NODE name="prepareActionSets" type="PrepareActionSets">
82 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
83 
                <PARAMETERS>
84 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
85 
                        [
86 
                        {
87 
                        'set' : 'doiboost',
88 
                        'jobProperty' : 'export_action_set_doiboost',
89 
                        'enablingProperty' : 'active_doiboost',
90 
                        'enabled' : 'true'
91 
                        }
92 
                        ]
93 
                    </PARAM>
94 
                </PARAMETERS>
95 
                <ARCS>
96 
                    <ARC to="extractOutputPath"/>
97 
                </ARCS>
98 
            </NODE>
99 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
100 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
101 
                <PARAMETERS>
102 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
103 
                </PARAMETERS>
104 
                <ARCS>
105 
                    <ARC to="updateDOIBoost"/>
106 
                </ARCS>
107 
            </NODE>
108 
            <NODE name="updateDOIBoost" type="SubmitHadoopJob">
109 
                <DESCRIPTION>prepare a new version of DOIBoost</DESCRIPTION>
110 
                <PARAMETERS>
111 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
112 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
113 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
114 
                        {
115 
                        'hostedByMapPath' : 'hostedByMapPath',
116 
                        'inputPathMAG' : 'inputPathMAG',
117 
                        'inputPathCrossref' : 'inputPathCrossref',
118 
                        'inputPathUnpayWall' : 'inputPathUnpayWall',
119 
                        'outputPath' : 'outputPath',
120 
                        'workingPathOrcid':'workingPathOrcid',
121 
                        'resumeFrom' : 'resumeFrom'
122 
                        }
123 
                    </PARAM>
124 
                    <PARAM managedBy="system" name="params" required="true" type="string">
125 
                        {
126 
                        'oozie.wf.application.path' : ' /lib/dnet/PROD/doiboost/process/oozie_app',
127 
                        'workingPath' : '/data/doiboost/process_p',
128 
                        'sparkExecutorIntersectionMemory' : '9G',
129 
                        'sparkExecutorMemory' : '6G'
130 
                        }
131 
                    </PARAM>
132 
                    <PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
133 
                </PARAMETERS>
134 
                <ARCS>
135 
                    <ARC to="updateActionSets"/>
136 
                </ARCS>
137 
            </NODE>
138 
            <NODE name="updateActionSets" type="UpdateActionSets">
139 
                <DESCRIPTION>update action sets</DESCRIPTION>
140 
                <PARAMETERS/>
141 
                <ARCS>
142 
                    <ARC to="success"/>
143 
                </ARCS>
144 
            </NODE>
145 
        </CONFIGURATION>
146 
        <STATUS>
147 
            <LAST_EXECUTION_ID>wf_20210512_082537_192</LAST_EXECUTION_ID>
148 
            <LAST_EXECUTION_DATE>2021-05-12T10:19:27+00:00</LAST_EXECUTION_DATE>
149 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
150 
            <LAST_EXECUTION_ERROR/>
151 
        </STATUS>
152 
    </BODY>
153 
</RESOURCE_PROFILE>
modules/dnet-openaireplus-workflows/trunk/src/main/resources/eu/dnetlib/test/profiles/openaireplus/workflows/import/import-datacite-beta.xml
1 
<RESOURCE_PROFILE>
2 
    <HEADER>
3 
        <RESOURCE_IDENTIFIER value="50593ba8-bb4b-413a-b852-b2c7e4a56093_V29ya2Zsb3dEU1Jlc291cmNlcy9Xb3JrZmxvd0RTUmVzb3VyY2VUeXBl"/>
4 
        <RESOURCE_TYPE value="WorkflowDSResourceType"/>
5 
        <RESOURCE_KIND value="WorkflowDSResources"/>
6 
        <RESOURCE_URI value=""/>
7 
        <DATE_OF_CREATION value="2021-04-21T09:11:15+00:00"/>
8 
    </HEADER>
9 
    <BODY>
10 
        <WORKFLOW_NAME>Import Datacite ActionSet</WORKFLOW_NAME>
11 
        <WORKFLOW_TYPE>Import InfoSpace</WORKFLOW_TYPE>
12 
        <WORKFLOW_PRIORITY>30</WORKFLOW_PRIORITY>
13 
        <CONFIGURATION start="manual">
14 
            <NODE isStart="true" name="setDatacitePath" type="SetEnvParameter">
15 
                <DESCRIPTION>set the input path for Datacite content</DESCRIPTION>
16 
                <PARAMETERS>
17 
                    <PARAM managedBy="system" name="parameterName" required="true" type="string">datacitePath</PARAM>
18 
                    <PARAM managedBy="user" name="parameterValue" required="true" type="string">/data/datacite/datacite_oaf</PARAM>
19 
                </PARAMETERS>
20 
                <ARCS>
21 
                    <ARC to="prepareActionSets"/>
22 
                </ARCS>
23 
            </NODE>
24 
            <NODE name="prepareActionSets" type="PrepareActionSets">
25 
                <DESCRIPTION>prepare action sets</DESCRIPTION>
26 
                <PARAMETERS>
27 
                    <PARAM managedBy="system" name="sets" required="true" type="string">
28 
                        [
29 
                        {
30 
                        'set' : 'datacite',
31 
                        'jobProperty' : 'export_action_set_datacite',
32 
                        'enablingProperty' : 'active_datacite',
33 
                        'enabled' : 'true'
34 
                        }
35 
                        ]
36 
                    </PARAM>
37 
                </PARAMETERS>
38 
                <ARCS>
39 
                    <ARC to="extractOutputPath"/>
40 
                </ARCS>
41 
            </NODE>
42 
            <NODE name="extractOutputPath" type="ExtractOutputPath">
43 
                <DESCRIPTION>extract the hdfs output path generated in the previous node</DESCRIPTION>
44 
                <PARAMETERS>
45 
                    <PARAM managedBy="system" name="hdfsOutputPathParam" required="true" type="string">outputPath</PARAM>
46 
                </PARAMETERS>
47 
                <ARCS>
48 
                    <ARC to="generateDataciteActionSet"/>
49 
                </ARCS>
50 
            </NODE>
51 
            <NODE name="generateDataciteActionSet" type="SubmitHadoopJob">
52 
                <DESCRIPTION>prepare a new version of Datacite ActionSet</DESCRIPTION>
53 
                <PARAMETERS>
54 
                    <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
55 
                    <PARAM managedBy="system" name="cluster" required="true" type="string">IIS</PARAM>
56 
                    <PARAM managedBy="system" name="envParams" required="true" type="string">
57 
                        {
58 
                        'sourcePath' : 'datacitePath',
59 
                        'outputPath' : 'outputPath'
60 
                        }
61 
                    </PARAM>
62 
                    <PARAM managedBy="system" name="params" required="true" type="string">
63 
                        {
64 
                        'oozie.wf.application.path' : '/lib/dnet/STABLE_IDS/actionmanager/datacite_actionset/oozie_app',
65 
                        'sparkExecutorMemory' : '7G'
66 
                        }
67 
                    </PARAM>
68 
                    <PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">build-report</PARAM>
69 
                </PARAMETERS>
70 
                <ARCS>
71 
                    <ARC to="updateActionSets"/>
72 
                </ARCS>
73 
            </NODE>
74 
            <NODE name="updateActionSets" type="UpdateActionSets">
75 
                <DESCRIPTION>update action sets</DESCRIPTION>
76 
                <PARAMETERS/>
77 
                <ARCS>
78 
                    <ARC to="success"/>
79 
                </ARCS>
80 
            </NODE>
81 
        </CONFIGURATION>
82 
        <STATUS>
83 
            <LAST_EXECUTION_ID>wf_20210421_091120_270</LAST_EXECUTION_ID>
84 
            <LAST_EXECUTION_DATE>2021-04-21T09:17:25+00:00</LAST_EXECUTION_DATE>
85 
            <LAST_EXECUTION_STATUS>SUCCESS</LAST_EXECUTION_STATUS>
86 
            <LAST_EXECUTION_ERROR/>
87 
        </STATUS>
88 
    </BODY>
89 
</RESOURCE_PROFILE>

Also available in: Unified diff