1
|
<RESOURCE_PROFILE>
|
2
|
<HEADER>
|
3
|
<RESOURCE_IDENTIFIER value="12f1db9b-3c3e-4fd6-a425-932afd6ef3be_V29ya2Zsb3dUZW1wbGF0ZURTUmVzb3VyY2VzL1dvcmtmbG93VGVtcGxhdGVEU1Jlc291cmNlVHlwZQ=="/>
|
4
|
<RESOURCE_TYPE value="WorkflowTemplateDSResourceType"/>
|
5
|
<RESOURCE_KIND value="WorkflowTemplateDSResources"/>
|
6
|
<RESOURCE_URI value=""/>
|
7
|
<DATE_OF_CREATION value="2016-06-25T10:24:32+00:00"/>
|
8
|
</HEADER>
|
9
|
<BODY>
|
10
|
<CONFIGURATION>
|
11
|
|
12
|
<PARAMETERS>
|
13
|
<PARAM name="cluster" description="Hadoop cluster name" required="true" type="string" />
|
14
|
<PARAM name="inputRecordsPath" description="path on HDFS pointing to the sequence file with the input xml records" required="true" type="string" />
|
15
|
|
16
|
<PARAM name="oaiDBName" description="db name" required="true" type="string" />
|
17
|
<PARAM name="oaiCollectionName" description="collection name" required="true" type="string" default="oaf-index-openaire" />
|
18
|
<PARAM name="skipDuplicates" description="skip duplicates" required="true" type="boolean" />
|
19
|
<PARAM name="duplicateXPath" description="xpath to identify duplicated records" required="true" type="string" default="//*[local-name()='entity']//*[local-name()='datainfo']/*[local-name()='deletedbyinference'][./text() = 'true']"/>
|
20
|
|
21
|
<PARAM name="fieldNames" description="list of field names to index" required="true" type="string" default="set,datestamp;license,oaftype;set,oaftype;oaftype,funder;resulttypeid,oaftype;oaftype,funding;resulttypeid,funder;resulttypeid,funding" />
|
22
|
<PARAM name="configuredOnly" description="rebuild only the indicies in the OAI configuration profile" required="true" type="boolean" default="false" />
|
23
|
</PARAMETERS>
|
24
|
|
25
|
<WORKFLOW>
|
26
|
|
27
|
<NODE name="prepareOAI" type="PrepareOaiJob" isStart="true">
|
28
|
<DESCRIPTION>Prepare params for OAI store feeding</DESCRIPTION>
|
29
|
<PARAMETERS>
|
30
|
<PARAM name="oaiConfigurationParam" value="oaiConfiguration" />
|
31
|
<PARAM name="oaiFeedDateParam" value="oai.feed.date" />
|
32
|
</PARAMETERS>
|
33
|
<ARCS>
|
34
|
<ARC to="OAIDropStore"/>
|
35
|
</ARCS>
|
36
|
</NODE>
|
37
|
|
38
|
<NODE name="OAIDropStore" type="OAIDropStore">
|
39
|
<DESCRIPTION>Drop the store: upserts are too expensive!</DESCRIPTION>
|
40
|
<PARAMETERS>
|
41
|
<PARAM name="dbName" ref="oaiDBName" />
|
42
|
<PARAM name="collectionName" ref="oaiCollectionName" />
|
43
|
</PARAMETERS>
|
44
|
<ARCS>
|
45
|
<ARC to="OAICreateStore"/>
|
46
|
</ARCS>
|
47
|
</NODE>
|
48
|
|
49
|
<NODE name="OAICreateStore" type="OAICreateStore">
|
50
|
<DESCRIPTION/>
|
51
|
<PARAMETERS>
|
52
|
<PARAM name="dbName" ref="oaiDBName" />
|
53
|
<PARAM name="collectionName" ref="oaiCollectionName" />
|
54
|
</PARAMETERS>
|
55
|
<ARCS>
|
56
|
<ARC to="feedOAI"/>
|
57
|
</ARCS>
|
58
|
</NODE>
|
59
|
|
60
|
<NODE name="feedOAI" type="SubmitHadoopJob">
|
61
|
<DESCRIPTION>M/O oai feeding</DESCRIPTION>
|
62
|
<PARAMETERS>
|
63
|
<PARAM name="cluster" ref="cluster" />
|
64
|
<PARAM name="hadoopJob" value="oaiFeedJob" />
|
65
|
<PARAM name="jobParams">
|
66
|
<MAP>
|
67
|
<ENTRY key="mapred.input.dir" ref="inputRecordsPath"/>
|
68
|
<ENTRY key="services.publisher.oai.collection" ref="oaiCollectionName"/>
|
69
|
<ENTRY key="services.publisher.oai.db" ref="oaiDBName"/>
|
70
|
<ENTRY key="services.publisher.oai.skipDuplicates" ref="skipDuplicates"/>
|
71
|
<ENTRY key="services.publisher.oai.duplicateXPath" ref="duplicateXPath"/>
|
72
|
<ENTRY key="oaiConfiguration" env="oaiConfiguration"/>
|
73
|
<ENTRY key="oai.feed.date" env="oai.feed.date"/>
|
74
|
<ENTRY key="services.publisher.oai.host" property="services.publisher.oai.host"/>
|
75
|
<ENTRY key="services.publisher.oai.port" property="services.publisher.oai.port"/>
|
76
|
</MAP>
|
77
|
</PARAM>
|
78
|
</PARAMETERS>
|
79
|
<ARCS>
|
80
|
<ARC to="CompoundIndexes"/>
|
81
|
</ARCS>
|
82
|
</NODE>
|
83
|
|
84
|
<NODE name="CompoundIndexes" type="OAICreateIndex">
|
85
|
<DESCRIPTION>Create composite indexes for the OAI store</DESCRIPTION>
|
86
|
<PARAMETERS>
|
87
|
<PARAM name="dbName" ref="oaiDBName" />
|
88
|
<PARAM name="collectionName" ref="oaiCollectionName"/>
|
89
|
<PARAM name="fieldNames" ref="fieldNames" />
|
90
|
</PARAMETERS>
|
91
|
<ARCS>
|
92
|
<ARC to="ConfigIndexes"/>
|
93
|
</ARCS>
|
94
|
</NODE>
|
95
|
|
96
|
<NODE name="ConfigIndexes" type="OAIEnsureIndexes">
|
97
|
<DESCRIPTION>Ensure an index exists on fields as specified in the configuration profile</DESCRIPTION>
|
98
|
<PARAMETERS>
|
99
|
<PARAM name="dbName" ref="oaiDBName" />
|
100
|
<PARAM name="collectionName" ref="oaiCollectionName"/>
|
101
|
</PARAMETERS>
|
102
|
<ARCS>
|
103
|
<ARC to="SetsCount"/>
|
104
|
</ARCS>
|
105
|
</NODE>
|
106
|
|
107
|
<NODE name="SetsCount" type="OAISetsCountUpdate">
|
108
|
<DESCRIPTION>Count records in each OAI set, for each exported metadata format</DESCRIPTION>
|
109
|
<PARAMETERS>
|
110
|
<PARAM name="dbName" ref="oaiDBName" />
|
111
|
<PARAM name="collectionName" ref="oaiCollectionName"/>
|
112
|
<PARAM name="configuredOnly" ref="configuredOnly" />
|
113
|
</PARAMETERS>
|
114
|
<ARCS>
|
115
|
<ARC to="ConfigSets"/>
|
116
|
</ARCS>
|
117
|
</NODE>
|
118
|
|
119
|
<NODE name="ConfigSets" type="RefreshSetsFromConfig">
|
120
|
<DESCRIPTION>Reads the current OAI configuration and upsert OAI sets accordingly (counts are updated as well)</DESCRIPTION>
|
121
|
<PARAMETERS>
|
122
|
<PARAM name="dbName" ref="oaiDBName" />
|
123
|
<PARAM name="collectionName" ref="oaiCollectionName"/>
|
124
|
</PARAMETERS>
|
125
|
<ARCS>
|
126
|
<ARC to="success"/>
|
127
|
</ARCS>
|
128
|
</NODE>
|
129
|
|
130
|
</WORKFLOW>
|
131
|
|
132
|
</CONFIGURATION>
|
133
|
|
134
|
</BODY>
|
135
|
</RESOURCE_PROFILE>
|