1
|
<NODE name="SET_INFO" isStart="true" type="SetProviderInfo">
|
2
|
<DESCRIPTION>Set information about current provider</DESCRIPTION>
|
3
|
<PARAMETERS>
|
4
|
<PARAM required="true" type="string" name="providerId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
5
|
<PARAM required="true" type="string" name="providerName" managedBy="system">$params.("dataprovider:name")$</PARAM>
|
6
|
<PARAM required="true" type="string" name="api" managedBy="system">$params.("dataprovider:interface")$</PARAM>
|
7
|
</PARAMETERS>
|
8
|
<ARCS>
|
9
|
<ARC to="obtainParams" />
|
10
|
</ARCS>
|
11
|
</NODE>
|
12
|
|
13
|
<NODE name="obtainParams" type="ObtainOpenaireDataSourceParams">
|
14
|
<DESCRIPTION>Fetch records from MDStore</DESCRIPTION>
|
15
|
<PARAMETERS>
|
16
|
<PARAM required="true" type="string" name="providerId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
17
|
</PARAMETERS>
|
18
|
<ARCS>
|
19
|
<ARC to="START_READING"/>
|
20
|
</ARCS>
|
21
|
</NODE>
|
22
|
|
23
|
<NODE name="START_READING" type="StartReadingMDStore">
|
24
|
<DESCRIPTION>Start reading Hadoop MD Store</DESCRIPTION>
|
25
|
<PARAMETERS>
|
26
|
<PARAM required="true" type="string" name="mdId" managedBy="system">$params.("harv_id")$</PARAM>
|
27
|
</PARAMETERS>
|
28
|
<ARCS>
|
29
|
<ARC to="PREPARE_STORE_VERSION"/>
|
30
|
</ARCS>
|
31
|
</NODE>
|
32
|
|
33
|
<NODE name="PREPARE_STORE_VERSION" type="PrepareMDStoreVersion">
|
34
|
<DESCRIPTION>Prepare a new MdStore Version</DESCRIPTION>
|
35
|
<PARAMETERS>
|
36
|
<PARAM required="true" type="string" name="mdId" managedBy="system">$params.("tran_id")$</PARAM>
|
37
|
</PARAMETERS>
|
38
|
<ARCS>
|
39
|
<ARC to="PREPARE_ENV_TRANSFORM"/>
|
40
|
</ARCS>
|
41
|
</NODE>
|
42
|
|
43
|
<NODE name="PREPARE_ENV_TRANSFORM" type="PrepareEnvTransformHadoopJobNode">
|
44
|
<DESCRIPTION>Put in the environment all the variable needed to the transform oozie job </DESCRIPTION>
|
45
|
<PARAMETERS>
|
46
|
<PARAM managedBy="user" name="hdfsBasePath" required="true" type="string"></PARAM>
|
47
|
</PARAMETERS>
|
48
|
<ARCS>
|
49
|
<ARC to="TRANSFORM_HADOOP"/>
|
50
|
</ARCS>
|
51
|
</NODE>
|
52
|
|
53
|
|
54
|
<NODE name="TRANSFORM_HADOOP" type="SubmitDnetHadoopJobNode">
|
55
|
<DESCRIPTION>Transform original records</DESCRIPTION>
|
56
|
<PARAMETERS>
|
57
|
<PARAM required="true" type="string" name="ruleId" category="TRANSFORMATION_RULE_ID" managedBy="user" function="listProfiles('TransformationRuleDSResourceType', '//TITLE')"></PARAM>
|
58
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">dnetHadoopCollection</PARAM>
|
59
|
<PARAM managedBy="system" name="cluster" required="true" type="string">DHP</PARAM>
|
60
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
61
|
{
|
62
|
"hdfsNativePath":"hdfsNativePath",
|
63
|
"hdfsTransformedPath":"hdfsTransformedPath",
|
64
|
"timestamp":"timestamp",
|
65
|
"workflowId":"workflowId"
|
66
|
}
|
67
|
</PARAM>
|
68
|
</PARAMETERS>
|
69
|
<ARCS>
|
70
|
<ARC to="COMMIT_VERSION"/>
|
71
|
</ARCS>
|
72
|
</NODE>
|
73
|
|
74
|
<NODE name="COMMIT_VERSION" type="CommitMDStoreVersion">
|
75
|
<DESCRIPTION>Commit the mdstore version</DESCRIPTION>
|
76
|
<PARAMETERS/>
|
77
|
<ARCS>
|
78
|
<ARC to="END_READING"/>
|
79
|
</ARCS>
|
80
|
</NODE>
|
81
|
|
82
|
<NODE name="END_READING" type="EndReadingMDStore">
|
83
|
<DESCRIPTION>End reading Hadoop MD Store</DESCRIPTION>
|
84
|
<PARAMETERS/>
|
85
|
<ARCS>
|
86
|
<ARC to="UPDATE_INFO"/>
|
87
|
</ARCS>
|
88
|
</NODE>
|
89
|
|
90
|
<NODE name="UPDATE_INFO" type="MDStoreToApiExtraFieldHadoop">
|
91
|
<DESCRIPTION>Update datasouce API extra fields</DESCRIPTION>
|
92
|
<PARAMETERS>
|
93
|
<PARAM required="true" type="string" name="mdId" managedBy="system">$params.("tran_id")$</PARAM>
|
94
|
<PARAM required="true" type="string" name="datasourceId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
95
|
<PARAM required="true" type="string" name="datasourceInterface" managedBy="system">$params.("dataprovider:interface")$</PARAM>
|
96
|
<PARAM required="true" type="string" name="extraFieldForTotal" managedBy="system">last_aggregation_total</PARAM>
|
97
|
<PARAM required="true" type="string" name="extraFieldForDate" managedBy="system">last_aggregation_date</PARAM>
|
98
|
<PARAM required="true" type="string" name="extraFieldForMdId" managedBy="system">last_aggregation_mdId</PARAM>
|
99
|
</PARAMETERS>
|
100
|
<ARCS>
|
101
|
<ARC to="success"/>
|
102
|
</ARCS>
|
103
|
</NODE>
|
104
|
|