Project

General

Profile

1
<workflow-app xmlns="uri:oozie:workflow:0.4" name="transformers_ingest_citations">
2
	
3
	<parameters>
4
		<property>
5
            <name>input</name>
6
            <description>input containing document metadata records</description>
7
        </property>
8
        <property>
9
            <name>output</name>
10
            <description>extracted citations</description>
11
        </property>
12
	</parameters>
13
	 
14
    <start to="generate-schema"/>
15
    
16
    <action name="generate-schema">
17
	    <java>
18
    		<job-tracker>${jobTracker}</job-tracker>
19
        	<name-node>${nameNode}</name-node>
20
	        <main-class>eu.dnetlib.iis.core.javamapreduce.hack.AvroSchemaGenerator</main-class>
21
	        <arg>eu.dnetlib.iis.citationmatching.direct.schemas.DocumentMetadata</arg>
22
	        <arg>eu.dnetlib.iis.citationmatching.direct.schemas.Citation</arg>
23
	        <capture-output />
24
	    </java>
25
	    <ok to="transformer" />
26
	    <error to="fail" />
27
	</action>
28
    
29
    <action name="transformer">
30
        <pig>
31
            <job-tracker>${jobTracker}</job-tracker>
32
            <name-node>${nameNode}</name-node>
33
			<!-- The data generated by this node is deleted in this section -->
34
			<prepare>
35
				<delete path="${nameNode}${output}" />
36
			</prepare>
37
            <configuration>
38
                <property>
39
                    <name>mapred.job.queue.name</name>
40
                    <value>${queueName}</value>
41
                </property>
42
            </configuration>
43
            <!-- Path to PIG script the workflow executes. -->
44
            <script>lib/scripts/transformer/transformer.pig</script>
45
            
46
            <param>input=${input}</param>
47
            <param>schema_input=${wf:actionData('generate-schema')['eu.dnetlib.iis.citationmatching.direct.schemas.DocumentMetadata']}</param>
48
            
49
            <param>output=${output}</param>
50
            <param>schema_output=${wf:actionData('generate-schema')['eu.dnetlib.iis.citationmatching.direct.schemas.Citation']}</param>
51
        </pig>
52
        <ok to="end"/>
53
        <error to="fail"/>
54
    </action>
55
    <kill name="fail">
56
		<message>Unfortunately, the workflow failed -- error message:
57
			[${wf:errorMessage(wf:lastErrorNode())}]</message>
58
    </kill>
59
    <end name="end"/>
60
</workflow-app>
    (1-1/1)