|
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>Obtain data source params</DESCRIPTION>
|
|
15
|
<PARAMETERS>
|
|
16
|
<PARAM required="true" type="string" name="providerId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
|
17
|
</PARAMETERS>
|
|
18
|
<ARCS>
|
|
19
|
<ARC to="PREPARE_ENV_COLLECTION"/>
|
|
20
|
</ARCS>
|
|
21
|
</NODE>
|
|
22
|
|
|
23
|
<NODE name="PREPARE_ENV_COLLECTION" type="PrepareEnvCollectHadoopJobNode">
|
|
24
|
<DESCRIPTION>Set in the environment all the variable needed to the collection oozie job</DESCRIPTION>
|
|
25
|
<PARAMETERS>
|
|
26
|
<PARAM required="true" type="string" name="mdId" managedBy="system" category="MDSTORE_ID">$params.("harv_id")$</PARAM>
|
|
27
|
<PARAM required="true" type="string" name="collectionMode" managedBy="user" function="validValues(['REFRESH','INCREMENTAL'])"></PARAM>
|
|
28
|
<PARAM required="true" type="string" name="metadataEncoding" managedBy="user">XML</PARAM>
|
|
29
|
<PARAM managedBy="user" name="fromDateOverride" required="false" type="string"/>
|
|
30
|
</PARAMETERS>
|
|
31
|
<ARCS>
|
|
32
|
<ARC to="COLLECT_HADOOP"/>
|
|
33
|
</ARCS>
|
|
34
|
</NODE>
|
|
35
|
|
|
36
|
<NODE name="COLLECT_HADOOP" type="SubmitHadoopJob">
|
|
37
|
<DESCRIPTION>Start the Hadoop Job</DESCRIPTION>
|
|
38
|
<PARAMETERS>
|
|
39
|
<PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
|
|
40
|
<PARAM managedBy="user" name="cluster" required="true" type="string">DHP</PARAM>
|
|
41
|
<PARAM managedBy="system" name="envParams" required="true" type="string">
|
|
42
|
{
|
|
43
|
"apiDescription":"apiDescription",
|
|
44
|
"dataSourceInfo":"dataSourceInfo",
|
|
45
|
"identifierPath":"identifierPath",
|
|
46
|
"metadataEncoding":"metadataEncoding",
|
|
47
|
"timestamp":"timestamp",
|
|
48
|
"workflowId":"workflowId",
|
|
49
|
"mdStoreID":"mdId",
|
|
50
|
"collectionMode":"collectionMode",
|
|
51
|
"oozie.wf.application.path":"oozieWfPath"
|
|
52
|
}
|
|
53
|
</PARAM>
|
|
54
|
<PARAM managedBy="system" name="oozieReportActionsCsv" required="true" type="string">BeginRead,StartTransaction,CollectionWorker</PARAM>
|
|
55
|
</PARAMETERS>
|
|
56
|
<ARCS>
|
|
57
|
<ARC to="UPDATE_INFO"/>
|
|
58
|
</ARCS>
|
|
59
|
</NODE>
|
|
60
|
|
|
61
|
<NODE name="UPDATE_INFO" type="MDStoreToApiExtraFieldHadoop">
|
|
62
|
<DESCRIPTION>Update datasouce API extra fields</DESCRIPTION>
|
|
63
|
<PARAMETERS>
|
|
64
|
<PARAM required="true" type="string" name="mdId" managedBy="system">$params.("harv_id")$</PARAM>
|
|
65
|
<PARAM required="true" type="string" name="datasourceId" managedBy="system">$params.("dataprovider:id")$</PARAM>
|
|
66
|
<PARAM required="true" type="string" name="datasourceInterface" managedBy="system">$params.("dataprovider:interface")$</PARAM>
|
|
67
|
<PARAM required="true" type="string" name="extraFieldForTotal" managedBy="system">last_collection_total</PARAM>
|
|
68
|
<PARAM required="true" type="string" name="extraFieldForDate" managedBy="system">last_collection_date</PARAM>
|
|
69
|
<PARAM required="true" type="string" name="extraFieldForMdId" managedBy="system">last_collection_mdId</PARAM>
|
|
70
|
</PARAMETERS>
|
|
71
|
<ARCS>
|
|
72
|
<ARC to="success"/>
|
|
73
|
</ARCS>
|
|
74
|
</NODE>
|