Project

General

Profile

1 60316 claudio.at
<NODE name="SET_INFO" isStart="true" type="SetProviderInfo">
2 57299 sandro.lab
	<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 60316 claudio.at
		<ARC to="PREPARE_ENV_COLLECTION"/>
20 57299 sandro.lab
	</ARCS>
21
</NODE>
22 60316 claudio.at
23 57299 sandro.lab
<NODE name="PREPARE_ENV_COLLECTION" type="PrepareEnvCollectHadoopJobNode">
24 60316 claudio.at
    <DESCRIPTION>Set in the environment all the variable needed to the collection oozie job</DESCRIPTION>
25 57299 sandro.lab
    <PARAMETERS>
26 60318 claudio.at
        <PARAM required="true" type="string" name="mdId" managedBy="system" category="MDSTORE_ID">$params.("harv_id")$</PARAM>
27 60316 claudio.at
        <PARAM required="true" type="string" name="collectionMode" managedBy="user" function="validValues(['REFRESH','INCREMENTAL'])"></PARAM>
28 57299 sandro.lab
    </PARAMETERS>
29
    <ARCS>
30
        <ARC to="COLLECT_HADOOP"/>
31
    </ARCS>
32
</NODE>
33 60316 claudio.at
34
<NODE name="COLLECT_HADOOP" type="SubmitHadoopJob">
35 57299 sandro.lab
    <DESCRIPTION>Start the Hadoop Job</DESCRIPTION>
36
    <PARAMETERS>
37 60316 claudio.at
        <PARAM managedBy="system" name="hadoopJob" required="true" type="string">executeOozieJob</PARAM>
38 57299 sandro.lab
        <PARAM managedBy="user" name="cluster" required="true" type="string">DHP</PARAM>
39
        <PARAM managedBy="system" name="envParams" required="true" type="string">
40
            {
41 60316 claudio.at
                "mdId":"mdId",
42
                "collectionMode":"collectionMode",
43 57299 sandro.lab
                "apiDescription":"apiDescription",
44
                "sequenceFilePath":"sequenceFilePath",
45 60316 claudio.at
                "dataSourceInfo":"dataSourceInfo",
46 57299 sandro.lab
                "timestamp":"timestamp",
47
                "identifierPath":"identifierPath",
48 60318 claudio.at
                "workflowId":"workflowId",
49
                "oozie.wf.application.path":"oozieWfPath"
50 57299 sandro.lab
            }
51
        </PARAM>
52
    </PARAMETERS>
53
    <ARCS>
54 60316 claudio.at
        <ARC to="UPDATE_INFO"/>
55 57299 sandro.lab
    </ARCS>
56
</NODE>
57
58
<NODE name="UPDATE_INFO" type="MDStoreToApiExtraFieldHadoop">
59
	<DESCRIPTION>Update datasouce API extra fields</DESCRIPTION>
60
	<PARAMETERS>
61
		<PARAM required="true" type="string" name="mdId" managedBy="system">$params.("harv_id")$</PARAM>
62
		<PARAM required="true" type="string" name="datasourceId" managedBy="system">$params.("dataprovider:id")$</PARAM>
63
		<PARAM required="true" type="string" name="datasourceInterface" managedBy="system">$params.("dataprovider:interface")$</PARAM>
64
		<PARAM required="true" type="string" name="extraFieldForTotal" managedBy="system">last_collection_total</PARAM>
65
		<PARAM required="true" type="string" name="extraFieldForDate" managedBy="system">last_collection_date</PARAM>
66
		<PARAM required="true" type="string" name="extraFieldForMdId" managedBy="system">last_collection_mdId</PARAM>
67
	</PARAMETERS>
68
	<ARCS>
69
		<ARC to="success"/>
70
	</ARCS>
71
</NODE>