1
|
<?xml version="1.0"?>
|
2
|
<!-- Note that documentation placed in comments in this file uses the
|
3
|
"markdown" syntax (along with its way of dividing text into sections). -->
|
4
|
<workflow-app xmlns="uri:oozie:workflow:0.3" name="test-referenceextraction_researchinitiative_main_sampletest">
|
5
|
<start to="producer" />
|
6
|
|
7
|
<action name="producer">
|
8
|
<java>
|
9
|
<job-tracker>${jobTracker}</job-tracker>
|
10
|
<name-node>${nameNode}</name-node>
|
11
|
<!-- The data generated by this node is deleted in this section -->
|
12
|
<prepare>
|
13
|
<delete path="${nameNode}${workingDir}/producer" />
|
14
|
<mkdir path="${nameNode}${workingDir}/producer" />
|
15
|
</prepare>
|
16
|
<configuration>
|
17
|
<property>
|
18
|
<name>mapred.job.queue.name</name>
|
19
|
<value>${queueName}</value>
|
20
|
</property>
|
21
|
</configuration>
|
22
|
<!-- This is simple wrapper for the Java code -->
|
23
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
24
|
<!-- The business Java code that gets to be executed -->
|
25
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
26
|
<!-- Specification of the output ports -->
|
27
|
<arg>-C{document_text,
|
28
|
eu.dnetlib.iis.metadataextraction.schemas.DocumentText,
|
29
|
eu/dnetlib/iis/referenceextraction/researchinitiative/data/empty.json}</arg>
|
30
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
31
|
<arg>-Odocument_text=${workingDir}/producer/document_text</arg>
|
32
|
</java>
|
33
|
<ok to="referenceextraction_researchinitiative" />
|
34
|
<error to="fail" />
|
35
|
</action>
|
36
|
|
37
|
<action name="referenceextraction_researchinitiative">
|
38
|
<sub-workflow>
|
39
|
<app-path>${wf:appPath()}/referenceextraction_researchinitiative</app-path>
|
40
|
<configuration>
|
41
|
<property>
|
42
|
<name>jobTracker</name>
|
43
|
<value>${jobTracker}</value>
|
44
|
</property>
|
45
|
<property>
|
46
|
<name>nameNode</name>
|
47
|
<value>${nameNode}</value>
|
48
|
</property>
|
49
|
<property>
|
50
|
<name>queueName</name>
|
51
|
<value>${queueName}</value>
|
52
|
</property>
|
53
|
<!-- Working directory of the subworkflow -->
|
54
|
<property>
|
55
|
<name>workingDir</name>
|
56
|
<value>${workingDir}/referenceextraction_researchinitiative/working_dir</value>
|
57
|
</property>
|
58
|
<property>
|
59
|
<name>input_document_text</name>
|
60
|
<value>${workingDir}/producer/document_text</value>
|
61
|
</property>
|
62
|
<property>
|
63
|
<name>output_document_to_research_initiative</name>
|
64
|
<value>${workingDir}/referenceextraction_researchinitiative/document_to_research_initiative</value>
|
65
|
</property>
|
66
|
</configuration>
|
67
|
</sub-workflow>
|
68
|
<ok to="consumer"/>
|
69
|
<error to="fail"/>
|
70
|
</action>
|
71
|
|
72
|
<action name="consumer">
|
73
|
<java>
|
74
|
<job-tracker>${jobTracker}</job-tracker>
|
75
|
<name-node>${nameNode}</name-node>
|
76
|
<configuration>
|
77
|
<property>
|
78
|
<name>mapred.job.queue.name</name>
|
79
|
<value>${queueName}</value>
|
80
|
</property>
|
81
|
</configuration>
|
82
|
<!-- This is simple wrapper for the Java code -->
|
83
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
84
|
<!-- The business Java code that gets to be executed -->
|
85
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>
|
86
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
87
|
<arg>-C{document_to_dataset,
|
88
|
eu.dnetlib.iis.referenceextraction.researchinitiative.schemas.DocumentToConceptId,
|
89
|
eu/dnetlib/iis/referenceextraction/researchinitiative/data/empty.json}</arg>
|
90
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
91
|
<arg>-Idocument_to_dataset=${workingDir}/referenceextraction_researchinitiative/document_to_research_initiative</arg>
|
92
|
</java>
|
93
|
<ok to="end" />
|
94
|
<error to="fail" />
|
95
|
</action>
|
96
|
|
97
|
<kill name="fail">
|
98
|
<message>Unfortunately, the process failed -- error message:
|
99
|
[${wf:errorMessage(wf:lastErrorNode())}]
|
100
|
</message>
|
101
|
</kill>
|
102
|
|
103
|
<end name="end"/>
|
104
|
</workflow-app>
|