1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="ingest_pmc_citations">
|
2
|
|
3
|
<parameters>
|
4
|
<property>
|
5
|
<name>input_document_nlm</name>
|
6
|
<description>input document text with NLMs</description>
|
7
|
</property>
|
8
|
<property>
|
9
|
<name>input_dedup_map</name>
|
10
|
<description>mapping between original OpenAire IDs and deduplicated documents' IDs</description>
|
11
|
</property>
|
12
|
<property>
|
13
|
<name>input_pmid_to_oaid</name>
|
14
|
<description>pmid to oaid mappings</description>
|
15
|
</property>
|
16
|
<property>
|
17
|
<name>input_doi_to_oaid</name>
|
18
|
<description>doi to oaid mappings</description>
|
19
|
</property>
|
20
|
<property>
|
21
|
<name>output_citation</name>
|
22
|
<description>extracted citations</description>
|
23
|
</property>
|
24
|
<property>
|
25
|
<name>reduce_tasks</name>
|
26
|
<value>2</value>
|
27
|
</property>
|
28
|
</parameters>
|
29
|
|
30
|
<start to="ingest"/>
|
31
|
|
32
|
<action name="ingest">
|
33
|
<java>
|
34
|
<job-tracker>${jobTracker}</job-tracker>
|
35
|
<name-node>${nameNode}</name-node>
|
36
|
<!-- The data generated by this node is deleted in this section -->
|
37
|
<prepare>
|
38
|
<delete path="${nameNode}${output_citation}"/>
|
39
|
</prepare>
|
40
|
<configuration>
|
41
|
<property>
|
42
|
<name>mapred.job.queue.name</name>
|
43
|
<value>${queueName}</value>
|
44
|
</property>
|
45
|
</configuration>
|
46
|
<main-class>eu.dnetlib.iis.ingest.pmc.citations.ResolvedCitationsImporter</main-class>
|
47
|
<java-opts>-Dmapred.reduce.tasks=${reduce_tasks}</java-opts>
|
48
|
<arg>${nameNode}${input_document_nlm}</arg>
|
49
|
<arg>${nameNode}${input_dedup_map}</arg>
|
50
|
<arg>${nameNode}${input_pmid_to_oaid}</arg>
|
51
|
<arg>${nameNode}${input_doi_to_oaid}</arg>
|
52
|
<arg>${nameNode}${output_citation}</arg>
|
53
|
</java>
|
54
|
<ok to="end"/>
|
55
|
<error to="fail"/>
|
56
|
</action>
|
57
|
|
58
|
<kill name="fail">
|
59
|
<message>Unfortunately, the process failed -- error message: [${wf:errorMessage(wf:lastErrorNode())}]</message>
|
60
|
</kill>
|
61
|
<end name="end"/>
|
62
|
</workflow-app>
|