Revision 37109
Added by Marek Horst about 9 years ago
workflow.xml | ||
---|---|---|
15 | 15 |
</property> |
16 | 16 |
</parameters> |
17 | 17 |
|
18 |
<start to="transformer"/> |
|
18 |
<start to="generate-schema"/> |
|
19 |
|
|
20 |
<action name="generate-schema"> |
|
21 |
<java> |
|
22 |
<job-tracker>${jobTracker}</job-tracker> |
|
23 |
<name-node>${nameNode}</name-node> |
|
24 |
<main-class>eu.dnetlib.iis.core.javamapreduce.hack.AvroSchemaGenerator</main-class> |
|
25 |
<arg>eu.dnetlib.iis.importer.auxiliary.schemas.DocumentContentUrl</arg> |
|
26 |
<arg>eu.dnetlib.iis.metadataextraction.schemas.DocumentText</arg> |
|
27 |
<capture-output /> |
|
28 |
</java> |
|
29 |
<ok to="transformer" /> |
|
30 |
<error to="fail" /> |
|
31 |
</action> |
|
32 |
|
|
19 | 33 |
<action name="transformer"> |
20 | 34 |
<pig> |
21 | 35 |
<job-tracker>${jobTracker}</job-tracker> |
... | ... | |
32 | 46 |
</configuration> |
33 | 47 |
<!-- Path to PIG script the workflow executes. --> |
34 | 48 |
<script>lib/scripts/transformer.pig</script> |
35 |
<param>schema_document_content_url=eu.dnetlib.iis.importer.auxiliary.schemas.DocumentContentUrl</param>
|
|
36 |
<param>schema_document_text=eu.dnetlib.iis.metadataextraction.schemas.DocumentText</param>
|
|
49 |
<param>schema_document_content_url=${wf:actionData('generate-schema')['eu.dnetlib.iis.importer.auxiliary.schemas.DocumentContentUrl']}</param>
|
|
50 |
<param>schema_document_text=${wf:actionData('generate-schema')['eu.dnetlib.iis.metadataextraction.schemas.DocumentText']}</param>
|
|
37 | 51 |
<param>input_document_content_url=${input_document_content_url}</param> |
38 | 52 |
<param>input_document_text=${input_document_text}</param> |
39 | 53 |
<param>output=${output}</param> |
Also available in: Unified diff
merging trunk changes with IIS-CDH-5.3.0 branch