|
1
|
<workflow-app xmlns="uri:oozie:workflow:0.2" name="test-union_input_2">
|
|
2
|
<start to="producer"/>
|
|
3
|
<action name="producer">
|
|
4
|
<java>
|
|
5
|
<job-tracker>${jobTracker}</job-tracker>
|
|
6
|
<name-node>${nameNode}</name-node>
|
|
7
|
<!-- The data generated by this node is deleted in this section -->
|
|
8
|
<prepare>
|
|
9
|
<delete path="${nameNode}${workingDir}/producer" />
|
|
10
|
<mkdir path="${nameNode}${workingDir}/producer" />
|
|
11
|
</prepare>
|
|
12
|
<configuration>
|
|
13
|
<property>
|
|
14
|
<name>mapred.job.queue.name</name>
|
|
15
|
<value>${queueName}</value>
|
|
16
|
</property>
|
|
17
|
</configuration>
|
|
18
|
<!-- This is simple wrapper for the Java code -->
|
|
19
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
20
|
<!-- The business Java code that gets to be executed -->
|
|
21
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
22
|
<!-- Specification of the output ports -->
|
|
23
|
<arg>-C{metadata_orig,
|
|
24
|
eu.dnetlib.iis.importer.schemas.DocumentMetadata,
|
|
25
|
eu/dnetlib/iis/collapsers/union/data/metadata_orig.json}</arg>
|
|
26
|
<arg>-C{metadata_cermine,
|
|
27
|
eu.dnetlib.iis.importer.schemas.DocumentMetadata,
|
|
28
|
eu/dnetlib/iis/collapsers/union/data/metadata_cermine.json}</arg>
|
|
29
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
|
30
|
<arg>-Ometadata_orig=${workingDir}/producer/metadata_orig</arg>
|
|
31
|
<arg>-Ometadata_cermine=${workingDir}/producer/metadata_cermine</arg>
|
|
32
|
</java>
|
|
33
|
<ok to="union"/>
|
|
34
|
<error to="fail"/>
|
|
35
|
</action>
|
|
36
|
<action name="union">
|
|
37
|
<sub-workflow>
|
|
38
|
<app-path>${wf:appPath()}/union</app-path>
|
|
39
|
<configuration>
|
|
40
|
<property>
|
|
41
|
<name>jobTracker</name>
|
|
42
|
<value>${jobTracker}</value>
|
|
43
|
</property>
|
|
44
|
<property>
|
|
45
|
<name>nameNode</name>
|
|
46
|
<value>${nameNode}</value>
|
|
47
|
</property>
|
|
48
|
<property>
|
|
49
|
<name>queueName</name>
|
|
50
|
<value>${queueName}</value>
|
|
51
|
</property>
|
|
52
|
<!-- Working directory of the subworkflow -->
|
|
53
|
<property>
|
|
54
|
<name>workingDir</name>
|
|
55
|
<value>${workingDir}/union/working_dir</value>
|
|
56
|
</property>
|
|
57
|
<property>
|
|
58
|
<name>schema_input</name>
|
|
59
|
<value>eu.dnetlib.iis.importer.schemas.DocumentMetadata</value>
|
|
60
|
</property>
|
|
61
|
<property>
|
|
62
|
<name>schema_output</name>
|
|
63
|
<value>eu.dnetlib.iis.collapsers.schemas.DocumentMetadataEnvelope</value>
|
|
64
|
</property>
|
|
65
|
<property>
|
|
66
|
<name>origin_1</name>
|
|
67
|
<value>orig</value>
|
|
68
|
</property>
|
|
69
|
<property>
|
|
70
|
<name>origin_2</name>
|
|
71
|
<value>cermine</value>
|
|
72
|
</property>
|
|
73
|
<!-- Input ports. -->
|
|
74
|
<property>
|
|
75
|
<name>input_1</name>
|
|
76
|
<value>${workingDir}/producer/metadata_orig</value>
|
|
77
|
</property>
|
|
78
|
<property>
|
|
79
|
<name>input_2</name>
|
|
80
|
<value>${workingDir}/producer/metadata_cermine</value>
|
|
81
|
</property>
|
|
82
|
<!-- Output port bound to given path -->
|
|
83
|
<property>
|
|
84
|
<name>output</name>
|
|
85
|
<value>${workingDir}/collapser_union/output</value>
|
|
86
|
</property>
|
|
87
|
</configuration>
|
|
88
|
</sub-workflow>
|
|
89
|
<ok to="consumer"/>
|
|
90
|
<error to="fail"/>
|
|
91
|
</action>
|
|
92
|
<action name="consumer">
|
|
93
|
<java>
|
|
94
|
<job-tracker>${jobTracker}</job-tracker>
|
|
95
|
<name-node>${nameNode}</name-node>
|
|
96
|
<configuration>
|
|
97
|
<property>
|
|
98
|
<name>mapred.job.queue.name</name>
|
|
99
|
<value>${queueName}</value>
|
|
100
|
</property>
|
|
101
|
</configuration>
|
|
102
|
<!-- This is simple wrapper for the Java code -->
|
|
103
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
104
|
<!-- The business Java code that gets to be executed -->
|
|
105
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>
|
|
106
|
<!-- Specification of the input ports -->
|
|
107
|
<arg>-C{output,
|
|
108
|
eu.dnetlib.iis.collapsers.schemas.DocumentMetadataEnvelope,
|
|
109
|
eu/dnetlib/iis/collapsers/union/data/metadata_output_2.json}</arg>
|
|
110
|
<!-- All input and output ports have to be bound to paths in HDFS -->
|
|
111
|
<arg>-Ioutput=${workingDir}/collapser_union/output</arg>
|
|
112
|
</java>
|
|
113
|
<ok to="end" />
|
|
114
|
<error to="fail" />
|
|
115
|
</action>
|
|
116
|
<kill name="fail">
|
|
117
|
<message>Unfortunately, the workflow failed -- error message:
|
|
118
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
119
|
</kill>
|
|
120
|
<end name="end"/>
|
|
121
|
</workflow-app>
|
|
122
|
<!--property>
|
|
123
|
<name>origin_1</name>
|
|
124
|
<value>orig</value>
|
|
125
|
</property>
|
|
126
|
<property>
|
|
127
|
<name>origin_2</name>
|
|
128
|
<value>cermine</value>
|
|
129
|
</property-->
|
|
130
|
<!-- Input ports. -->
|
|
131
|
<!--property>
|
|
132
|
<name>input_port_1</name>
|
|
133
|
<value>${workingDir}/producer/metadata_orig</value>
|
|
134
|
</property>
|
|
135
|
<property>
|
|
136
|
<name>input_port_2</name>
|
|
137
|
<value>${workingDir}/producer/metadata_cermine</value>
|
|
138
|
</property-->
|