|
1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="test-mainworkflows_integration_primary_processing">
|
|
2
|
|
|
3
|
<global>
|
|
4
|
<job-tracker>${jobTracker}</job-tracker>
|
|
5
|
<name-node>${nameNode}</name-node>
|
|
6
|
<configuration>
|
|
7
|
<property>
|
|
8
|
<name>mapred.job.queue.name</name>
|
|
9
|
<value>${queueName}</value>
|
|
10
|
</property>
|
|
11
|
</configuration>
|
|
12
|
</global>
|
|
13
|
|
|
14
|
<start to="generate-schema" />
|
|
15
|
|
|
16
|
<action name="generate-schema">
|
|
17
|
<java>
|
|
18
|
<main-class>eu.dnetlib.iis.core.javamapreduce.hack.AvroSchemaGenerator</main-class>
|
|
19
|
<arg>eu.dnetlib.iis.mainworkflows.schemas.DocumentContentClasspath</arg>
|
|
20
|
<arg>eu.dnetlib.iis.metadataextraction.schemas.DocumentText</arg>
|
|
21
|
<arg>org.apache.avro.Schema.Type.NULL</arg>
|
|
22
|
<capture-output />
|
|
23
|
</java>
|
|
24
|
<ok to="producer" />
|
|
25
|
<error to="fail" />
|
|
26
|
</action>
|
|
27
|
|
|
28
|
<action name="producer">
|
|
29
|
<java>
|
|
30
|
<!-- The data generated by this node is deleted in this section -->
|
|
31
|
<prepare>
|
|
32
|
<delete path="${nameNode}${workingDir}/producer" />
|
|
33
|
<mkdir path="${nameNode}${workingDir}/producer" />
|
|
34
|
</prepare>
|
|
35
|
<!-- This is simple wrapper for the Java code -->
|
|
36
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
37
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
|
38
|
<arg>-C{document_metadata,
|
|
39
|
eu.dnetlib.iis.importer.schemas.DocumentMetadata,
|
|
40
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/document_metadata.json}</arg>
|
|
41
|
<arg>-C{document_to_project,
|
|
42
|
eu.dnetlib.iis.importer.schemas.DocumentToProject,
|
|
43
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
44
|
<arg>-C{document_content_classpath,
|
|
45
|
eu.dnetlib.iis.mainworkflows.schemas.DocumentContentClasspath,
|
|
46
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/document_content_classpath.json}</arg>
|
|
47
|
<arg>-C{person,
|
|
48
|
eu.dnetlib.iis.importer.schemas.Person,
|
|
49
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/person.json}</arg>
|
|
50
|
<arg>-C{project,
|
|
51
|
eu.dnetlib.iis.importer.schemas.Project,
|
|
52
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/project.json}</arg>
|
|
53
|
<arg>-C{project_concept,
|
|
54
|
eu.dnetlib.iis.importer.schemas.Concept,
|
|
55
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/concept.json}</arg>
|
|
56
|
<arg>-C{dataset,
|
|
57
|
eu.dnetlib.iis.importer.schemas.DataSetReference,
|
|
58
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/dataset.json}</arg>
|
|
59
|
<arg>-C{extracted_document_metadata,
|
|
60
|
eu.dnetlib.iis.metadataextraction.schemas.ExtractedDocumentMetadata,
|
|
61
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/metadataextraction/extracted_document_metadata.json}</arg>
|
|
62
|
<arg>-C{document_text_wos,
|
|
63
|
eu.dnetlib.iis.metadataextraction.schemas.DocumentText,
|
|
64
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
65
|
<arg>-C{citation,
|
|
66
|
eu.dnetlib.iis.ingest.pmc.citations.schemas.Citation,
|
|
67
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/meta/citation_pmc.json}</arg>
|
|
68
|
<arg>-C{document_text_classpath,
|
|
69
|
eu.dnetlib.iis.mainworkflows.schemas.DocumentContentClasspath,
|
|
70
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/input/metadataextraction/document_text_classpath.json}</arg>
|
|
71
|
<arg>-Odocument_metadata=${workingDir}/producer/document_metadata</arg>
|
|
72
|
<arg>-Odocument_to_project=${workingDir}/producer/document_to_project</arg>
|
|
73
|
<arg>-Odocument_content_classpath=${workingDir}/producer/document_content_classpath</arg>
|
|
74
|
<arg>-Operson=${workingDir}/producer/person</arg>
|
|
75
|
<arg>-Oproject=${workingDir}/producer/project</arg>
|
|
76
|
<arg>-Oproject_concept=${workingDir}/producer/project_concept</arg>
|
|
77
|
<arg>-Odataset=${workingDir}/producer/dataset</arg>
|
|
78
|
<arg>-Oextracted_document_metadata=${workingDir}/producer/extracted_document_metadata</arg>
|
|
79
|
<arg>-Odocument_text_wos=${workingDir}/producer/document_text_wos</arg>
|
|
80
|
<arg>-Ocitation=${workingDir}/producer/citation</arg>
|
|
81
|
<arg>-Odocument_text_classpath=${workingDir}/producer/document_text_classpath</arg>
|
|
82
|
</java>
|
|
83
|
<ok to="document_text_producer"/>
|
|
84
|
<error to="fail"/>
|
|
85
|
</action>
|
|
86
|
|
|
87
|
<action name="document_text_producer">
|
|
88
|
<map-reduce>
|
|
89
|
<prepare>
|
|
90
|
<delete path="${nameNode}${workingDir}/text-producer" />
|
|
91
|
<mkdir path="${nameNode}${workingDir}/text-producer" />
|
|
92
|
</prepare>
|
|
93
|
<configuration>
|
|
94
|
<property>
|
|
95
|
<name>mapreduce.inputformat.class</name>
|
|
96
|
<value>org.apache.avro.mapreduce.AvroKeyInputFormat</value>
|
|
97
|
</property>
|
|
98
|
<property>
|
|
99
|
<name>mapreduce.outputformat.class</name>
|
|
100
|
<value>org.apache.avro.mapreduce.AvroKeyOutputFormat</value>
|
|
101
|
</property>
|
|
102
|
<property>
|
|
103
|
<name>mapred.mapoutput.key.class</name>
|
|
104
|
<value>org.apache.avro.mapred.AvroKey</value>
|
|
105
|
</property>
|
|
106
|
<property>
|
|
107
|
<name>mapred.mapoutput.value.class</name>
|
|
108
|
<value>org.apache.hadoop.io.NullWritable</value>
|
|
109
|
</property>
|
|
110
|
<property>
|
|
111
|
<name>mapred.reduce.tasks</name>
|
|
112
|
<value>0</value>
|
|
113
|
</property>
|
|
114
|
<property>
|
|
115
|
<name>mapred.output.key.class</name>
|
|
116
|
<value>org.apache.avro.mapred.AvroKey</value>
|
|
117
|
</property>
|
|
118
|
<property>
|
|
119
|
<name>mapred.output.value.class</name>
|
|
120
|
<value>org.apache.avro.mapred.AvroValue</value>
|
|
121
|
</property>
|
|
122
|
<property>
|
|
123
|
<name>mapred.output.key.comparator.class</name>
|
|
124
|
<value>org.apache.avro.hadoop.io.AvroKeyComparator</value>
|
|
125
|
</property>
|
|
126
|
<property>
|
|
127
|
<name>io.serializations</name>
|
|
128
|
<value>org.apache.hadoop.io.serializer.WritableSerialization,org.apache.hadoop.io.serializer.avro.AvroSpecificSerialization,org.apache.hadoop.io.serializer.avro.AvroReflectSerialization,org.apache.avro.hadoop.io.AvroSerialization
|
|
129
|
</value>
|
|
130
|
</property>
|
|
131
|
<property>
|
|
132
|
<name>mapred.output.value.groupfn.class</name>
|
|
133
|
<value>org.apache.avro.hadoop.io.AvroKeyComparator</value>
|
|
134
|
</property>
|
|
135
|
<property>
|
|
136
|
<name>rpc.engine.org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolPB
|
|
137
|
</name>
|
|
138
|
<value>org.apache.hadoop.ipc.ProtobufRpcEngine</value>
|
|
139
|
</property>
|
|
140
|
<property>
|
|
141
|
<name>mapred.mapper.new-api</name>
|
|
142
|
<value>true</value>
|
|
143
|
</property>
|
|
144
|
<property>
|
|
145
|
<name>mapred.reducer.new-api</name>
|
|
146
|
<value>true</value>
|
|
147
|
</property>
|
|
148
|
<property>
|
|
149
|
<name>mapreduce.map.class</name>
|
|
150
|
<value>eu.dnetlib.iis.mainworkflows.converters.DocumentClasspathToTextConverter</value>
|
|
151
|
</property>
|
|
152
|
|
|
153
|
<property>
|
|
154
|
<name>avro.schema.input.key</name>
|
|
155
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.mainworkflows.schemas.DocumentContentClasspath']}</value>
|
|
156
|
</property>
|
|
157
|
<property>
|
|
158
|
<name>avro.serialization.key.reader.schema</name>
|
|
159
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.metadataextraction.schemas.DocumentText']}</value>
|
|
160
|
</property>
|
|
161
|
<property>
|
|
162
|
<name>avro.serialization.key.writer.schema</name>
|
|
163
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.metadataextraction.schemas.DocumentText']}</value>
|
|
164
|
</property>
|
|
165
|
<property>
|
|
166
|
<name>avro.serialization.value.reader.schema</name>
|
|
167
|
<value>${wf:actionData('generate-schema')['org.apache.avro.Schema.Type.NULL']}</value>
|
|
168
|
</property>
|
|
169
|
<property>
|
|
170
|
<name>avro.serialization.value.writer.schema</name>
|
|
171
|
<value>${wf:actionData('generate-schema')['org.apache.avro.Schema.Type.NULL']}</value>
|
|
172
|
</property>
|
|
173
|
<property>
|
|
174
|
<name>avro.schema.output.key</name>
|
|
175
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.metadataextraction.schemas.DocumentText']}</value>
|
|
176
|
</property>
|
|
177
|
|
|
178
|
<property>
|
|
179
|
<name>mapred.input.dir</name>
|
|
180
|
<value>${workingDir}/producer/document_text_classpath</value>
|
|
181
|
</property>
|
|
182
|
<property>
|
|
183
|
<name>mapred.output.dir</name>
|
|
184
|
<value>${workingDir}/text-producer/document_text</value>
|
|
185
|
</property>
|
|
186
|
</configuration>
|
|
187
|
</map-reduce>
|
|
188
|
<ok to="mainworkflows_primary_processing" />
|
|
189
|
<error to="fail" />
|
|
190
|
</action>
|
|
191
|
|
|
192
|
<action name="mainworkflows_primary_processing">
|
|
193
|
<sub-workflow>
|
|
194
|
<app-path>${wf:appPath()}/mainworkflows_primary_processing</app-path>
|
|
195
|
<propagate-configuration/>
|
|
196
|
<configuration>
|
|
197
|
<property>
|
|
198
|
<name>active_referenceextraction_project</name>
|
|
199
|
<value>true</value>
|
|
200
|
</property>
|
|
201
|
<property>
|
|
202
|
<name>active_referenceextraction_dataset</name>
|
|
203
|
<value>true</value>
|
|
204
|
</property>
|
|
205
|
<property>
|
|
206
|
<name>active_referenceextraction_researchinitiative</name>
|
|
207
|
<value>true</value>
|
|
208
|
</property>
|
|
209
|
<property>
|
|
210
|
<name>active_referenceextraction_pdb</name>
|
|
211
|
<value>false</value>
|
|
212
|
</property>
|
|
213
|
<property>
|
|
214
|
<name>active_documentsclassification</name>
|
|
215
|
<value>true</value>
|
|
216
|
</property>
|
|
217
|
<property>
|
|
218
|
<name>active_citationmatching</name>
|
|
219
|
<value>true</value>
|
|
220
|
</property>
|
|
221
|
<!-- disabling not tested KDMs -->
|
|
222
|
<property>
|
|
223
|
<name>active_documentssimilarity</name>
|
|
224
|
<value>false</value>
|
|
225
|
</property>
|
|
226
|
<property>
|
|
227
|
<name>active_statistics</name>
|
|
228
|
<value>false</value>
|
|
229
|
</property>
|
|
230
|
<property>
|
|
231
|
<name>input_document_metadata</name>
|
|
232
|
<value>${workingDir}/producer/document_metadata</value>
|
|
233
|
</property>
|
|
234
|
<property>
|
|
235
|
<name>input_document_to_project</name>
|
|
236
|
<value>${workingDir}/producer/document_to_project</value>
|
|
237
|
</property>
|
|
238
|
<property>
|
|
239
|
<name>input_document_text</name>
|
|
240
|
<value>${workingDir}/text-producer/document_text</value>
|
|
241
|
</property>
|
|
242
|
<property>
|
|
243
|
<name>input_document_text_wos</name>
|
|
244
|
<value>${workingDir}/producer/document_text_wos</value>
|
|
245
|
</property>
|
|
246
|
<property>
|
|
247
|
<name>input_project</name>
|
|
248
|
<value>${workingDir}/producer/project</value>
|
|
249
|
</property>
|
|
250
|
<property>
|
|
251
|
<name>input_project_concept</name>
|
|
252
|
<value>${workingDir}/producer/project_concept</value>
|
|
253
|
</property>
|
|
254
|
<property>
|
|
255
|
<name>input_person</name>
|
|
256
|
<value>${workingDir}/producer/person</value>
|
|
257
|
</property>
|
|
258
|
<property>
|
|
259
|
<name>input_dataset</name>
|
|
260
|
<value>${workingDir}/producer/dataset</value>
|
|
261
|
</property>
|
|
262
|
<property>
|
|
263
|
<name>input_extracted_document_metadata</name>
|
|
264
|
<value>${workingDir}/producer/extracted_document_metadata</value>
|
|
265
|
</property>
|
|
266
|
<property>
|
|
267
|
<name>input_citation_pmc</name>
|
|
268
|
<value>${workingDir}/producer/citation</value>
|
|
269
|
</property>
|
|
270
|
<property>
|
|
271
|
<name>output_document_to_project</name>
|
|
272
|
<value>${workingDir}/exported/document_to_project</value>
|
|
273
|
</property>
|
|
274
|
<property>
|
|
275
|
<name>output_document_to_project_concepts</name>
|
|
276
|
<value>${workingDir}/exported/document_to_project_concepts</value>
|
|
277
|
</property>
|
|
278
|
<property>
|
|
279
|
<name>output_document_to_dataset</name>
|
|
280
|
<value>${workingDir}/exported/document_to_dataset</value>
|
|
281
|
</property>
|
|
282
|
<property>
|
|
283
|
<name>output_document_to_research_initiatives</name>
|
|
284
|
<value>${workingDir}/exported/document_to_research_initiatives</value>
|
|
285
|
</property>
|
|
286
|
<property>
|
|
287
|
<name>output_document_to_pdb</name>
|
|
288
|
<value>${workingDir}/exported/document_to_pdb</value>
|
|
289
|
</property>
|
|
290
|
<property>
|
|
291
|
<name>output_document_to_document_classes</name>
|
|
292
|
<value>${workingDir}/exported/document_to_document_classes</value>
|
|
293
|
</property>
|
|
294
|
<property>
|
|
295
|
<name>output_citation</name>
|
|
296
|
<value>${workingDir}/exported/citation</value>
|
|
297
|
</property>
|
|
298
|
<property>
|
|
299
|
<name>output_document_similarity</name>
|
|
300
|
<value>${workingDir}/exported/document_similarity</value>
|
|
301
|
</property>
|
|
302
|
<property>
|
|
303
|
<name>output_document_statistics</name>
|
|
304
|
<value>${workingDir}/exported/document_statistics</value>
|
|
305
|
</property>
|
|
306
|
<property>
|
|
307
|
<name>output_author_statistics</name>
|
|
308
|
<value>${workingDir}/exported/author_statistics</value>
|
|
309
|
</property>
|
|
310
|
<property>
|
|
311
|
<name>output_project_statistics</name>
|
|
312
|
<value>${workingDir}/exported/project_statistics</value>
|
|
313
|
</property>
|
|
314
|
</configuration>
|
|
315
|
</sub-workflow>
|
|
316
|
<ok to="consumer"/>
|
|
317
|
<error to="fail" />
|
|
318
|
</action>
|
|
319
|
|
|
320
|
<action name="consumer">
|
|
321
|
<java>
|
|
322
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
323
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.TestingConsumer</arg>
|
|
324
|
<!-- TODO prepare proper json generated datastores -->
|
|
325
|
<arg>-C{document_to_document_classes,
|
|
326
|
eu.dnetlib.iis.documentsclassification.schemas.DocumentToDocumentClasses,
|
|
327
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/output/document_classes.json}</arg>
|
|
328
|
<arg>-Idocument_to_document_classes=${workingDir}/exported/document_to_document_classes</arg>
|
|
329
|
<arg>-C{document_to_dataset,
|
|
330
|
eu.dnetlib.iis.referenceextraction.dataset.schemas.DocumentToDataSet,
|
|
331
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/output/document_to_dataset.json}</arg>
|
|
332
|
<arg>-Idocument_to_dataset=${workingDir}/exported/document_to_dataset</arg>
|
|
333
|
<arg>-C{document_to_project,
|
|
334
|
eu.dnetlib.iis.referenceextraction.project.schemas.DocumentToProject,
|
|
335
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/output/document_to_project.json}</arg>
|
|
336
|
<arg>-Idocument_to_project=${workingDir}/exported/document_to_project</arg>
|
|
337
|
<arg>-C{document_to_project_concepts,
|
|
338
|
eu.dnetlib.iis.export.schemas.DocumentToConceptIds,
|
|
339
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/output/document_to_project_concepts.json}</arg>
|
|
340
|
<arg>-Idocument_to_project_concepts=${workingDir}/exported/document_to_project_concepts</arg>
|
|
341
|
<arg>-C{document_to_research_initiatives,
|
|
342
|
eu.dnetlib.iis.export.schemas.DocumentToConceptIds,
|
|
343
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/output/document_to_research_initiatives.json}</arg>
|
|
344
|
<arg>-Idocument_to_research_initiatives=${workingDir}/exported/document_to_research_initiatives</arg>
|
|
345
|
<arg>-C{citation,
|
|
346
|
eu.dnetlib.iis.export.schemas.Citations,
|
|
347
|
eu/dnetlib/iis/mainworkflows/integration/primary/processing/output/citations.json}</arg>
|
|
348
|
<arg>-Icitation=${workingDir}/exported/citation</arg>
|
|
349
|
<arg>-C{document_similarity,
|
|
350
|
eu.dnetlib.iis.documentssimilarity.schemas.DocumentSimilarity,
|
|
351
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
352
|
<arg>-Idocument_similarity=${workingDir}/exported/document_similarity</arg>
|
|
353
|
<arg>-C{document_statistics,
|
|
354
|
eu.dnetlib.iis.statistics.schemas.DocumentToDocumentStatistics,
|
|
355
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
356
|
<arg>-Idocument_statistics=${workingDir}/exported/document_statistics</arg>
|
|
357
|
<arg>-C{author_statistics,
|
|
358
|
eu.dnetlib.iis.statistics.schemas.AuthorToAuthorStatistics,
|
|
359
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
360
|
<arg>-Iauthor_statistics=${workingDir}/exported/author_statistics</arg>
|
|
361
|
<arg>-C{project_statistics,
|
|
362
|
eu.dnetlib.iis.statistics.schemas.ProjectToProjectStatistics,
|
|
363
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
|
364
|
<arg>-Iproject_statistics=${workingDir}/exported/project_statistics</arg>
|
|
365
|
</java>
|
|
366
|
<ok to="end" />
|
|
367
|
<error to="fail" />
|
|
368
|
</action>
|
|
369
|
|
|
370
|
<kill name="fail">
|
|
371
|
<message>Unfortunately, the process failed -- error message:
|
|
372
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
373
|
</kill>
|
|
374
|
<end name="end" />
|
|
375
|
</workflow-app>
|