|
1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="export_actionmanager">
|
|
2
|
|
|
3
|
<parameters>
|
|
4
|
<property>
|
|
5
|
<name>input_document_metadata</name>
|
|
6
|
<value>$UNDEFINED$</value>
|
|
7
|
</property>
|
|
8
|
<property>
|
|
9
|
<name>input_document_to_project</name>
|
|
10
|
<value>$UNDEFINED$</value>
|
|
11
|
</property>
|
|
12
|
<property>
|
|
13
|
<name>input_document_to_project_concepts</name>
|
|
14
|
<value>$UNDEFINED$</value>
|
|
15
|
</property>
|
|
16
|
<property>
|
|
17
|
<name>input_document_to_dataset</name>
|
|
18
|
<value>$UNDEFINED$</value>
|
|
19
|
</property>
|
|
20
|
<property>
|
|
21
|
<name>input_document_to_research_initiatives</name>
|
|
22
|
<value>$UNDEFINED$</value>
|
|
23
|
</property>
|
|
24
|
<property>
|
|
25
|
<name>input_document_to_pdb</name>
|
|
26
|
<value>$UNDEFINED$</value>
|
|
27
|
</property>
|
|
28
|
<property>
|
|
29
|
<name>input_document_to_document_classes</name>
|
|
30
|
<value>$UNDEFINED$</value>
|
|
31
|
</property>
|
|
32
|
<property>
|
|
33
|
<name>input_citations</name>
|
|
34
|
<value>$UNDEFINED$</value>
|
|
35
|
</property>
|
|
36
|
<property>
|
|
37
|
<name>input_document_similarity</name>
|
|
38
|
<value>$UNDEFINED$</value>
|
|
39
|
</property>
|
|
40
|
<property>
|
|
41
|
<name>input_document_websiteusage_similarity</name>
|
|
42
|
<value>$UNDEFINED$</value>
|
|
43
|
</property>
|
|
44
|
<property>
|
|
45
|
<name>input_document_statistics</name>
|
|
46
|
<value>$UNDEFINED$</value>
|
|
47
|
</property>
|
|
48
|
<property>
|
|
49
|
<name>input_author_statistics</name>
|
|
50
|
<value>$UNDEFINED$</value>
|
|
51
|
</property>
|
|
52
|
<property>
|
|
53
|
<name>input_project_statistics</name>
|
|
54
|
<value>$UNDEFINED$</value>
|
|
55
|
</property>
|
|
56
|
|
|
57
|
<property>
|
|
58
|
<name>action_hbase_table_name</name>
|
|
59
|
<description>action manager hbase table name</description>
|
|
60
|
</property>
|
|
61
|
<property>
|
|
62
|
<name>action_hbase_table_initialize</name>
|
|
63
|
<value>false</value>
|
|
64
|
<description>flag indicating input table should be initialized, set to false by default</description>
|
|
65
|
</property>
|
|
66
|
<!-- actionset id section -->
|
|
67
|
<property>
|
|
68
|
<name>action_set_id</name>
|
|
69
|
<value>$UNDEFINED$</value>
|
|
70
|
<description>default action-set identifier of exported data</description>
|
|
71
|
</property>
|
|
72
|
<property>
|
|
73
|
<name>action_set_id_person_similarities_websiteusage</name>
|
|
74
|
<value>$UNDEFINED$</value>
|
|
75
|
<description>person_similarities_websiteusage action-set identifier of exported data</description>
|
|
76
|
</property>
|
|
77
|
<property>
|
|
78
|
<name>action_set_id_person_statistics</name>
|
|
79
|
<value>$UNDEFINED$</value>
|
|
80
|
<description>person_statistics action-set identifier of exported data</description>
|
|
81
|
</property>
|
|
82
|
<property>
|
|
83
|
<name>action_set_id_project_statistics</name>
|
|
84
|
<value>$UNDEFINED$</value>
|
|
85
|
<description>project_statistics action-set identifier of exported data</description>
|
|
86
|
</property>
|
|
87
|
<property>
|
|
88
|
<name>action_set_id_document_similarities_standard</name>
|
|
89
|
<value>$UNDEFINED$</value>
|
|
90
|
<description>document_similarities_standard action-set identifier of exported data</description>
|
|
91
|
</property>
|
|
92
|
<property>
|
|
93
|
<name>action_set_id_document_similarities_websiteusage</name>
|
|
94
|
<value>$UNDEFINED$</value>
|
|
95
|
<description>document_similarities_websiteusage action-set identifier of exported data</description>
|
|
96
|
</property>
|
|
97
|
<property>
|
|
98
|
<name>action_set_id_document_affiliations</name>
|
|
99
|
<value>$UNDEFINED$</value>
|
|
100
|
<description>document_affiliations action-set identifier of exported data</description>
|
|
101
|
</property>
|
|
102
|
<property>
|
|
103
|
<name>action_set_id_document_statistics</name>
|
|
104
|
<value>$UNDEFINED$</value>
|
|
105
|
<description>document_statistics action-set identifier of exported data</description>
|
|
106
|
</property>
|
|
107
|
<property>
|
|
108
|
<name>action_set_id_document_classes</name>
|
|
109
|
<value>$UNDEFINED$</value>
|
|
110
|
<description>document_classes action-set identifier of exported data</description>
|
|
111
|
</property>
|
|
112
|
<property>
|
|
113
|
<name>action_set_id_document_referencedProjects</name>
|
|
114
|
<value>$UNDEFINED$</value>
|
|
115
|
<description>document_referencedProjects action-set identifier of exported data</description>
|
|
116
|
</property>
|
|
117
|
<property>
|
|
118
|
<name>action_set_id_document_referencedDatasets</name>
|
|
119
|
<value>$UNDEFINED$</value>
|
|
120
|
<description>document_referencedDatasets action-set identifier of exported data</description>
|
|
121
|
</property>
|
|
122
|
<property>
|
|
123
|
<name>action_set_id_document_referencedDocuments</name>
|
|
124
|
<value>$UNDEFINED$</value>
|
|
125
|
<description>document_referencedDocuments action-set identifier of exported data</description>
|
|
126
|
</property>
|
|
127
|
<property>
|
|
128
|
<name>action_set_id_document_research_initiative</name>
|
|
129
|
<value>$UNDEFINED$</value>
|
|
130
|
<description>document research initiative action-set identifier of exported data</description>
|
|
131
|
</property>
|
|
132
|
<property>
|
|
133
|
<name>action_set_id_document_pdb</name>
|
|
134
|
<value>$UNDEFINED$</value>
|
|
135
|
<description>document to protein databank action-set identifier of exported data</description>
|
|
136
|
</property>
|
|
137
|
<!-- -->
|
|
138
|
<!-- trust level threshold section -->
|
|
139
|
<property>
|
|
140
|
<name>trust_level_threshold</name>
|
|
141
|
<value>$UNDEFINED$</value>
|
|
142
|
<description>default trust level threshold of exported data</description>
|
|
143
|
</property>
|
|
144
|
<property>
|
|
145
|
<name>trust_level_threshold_document_classes</name>
|
|
146
|
<value>$UNDEFINED$</value>
|
|
147
|
<description>document_classes trust level threshold</description>
|
|
148
|
</property>
|
|
149
|
<property>
|
|
150
|
<name>trust_level_threshold_document_referencedProjects</name>
|
|
151
|
<value>$UNDEFINED$</value>
|
|
152
|
<description>document_referencedProjects trust level threshold</description>
|
|
153
|
</property>
|
|
154
|
<property>
|
|
155
|
<name>trust_level_threshold_document_referencedDatasets</name>
|
|
156
|
<value>$UNDEFINED$</value>
|
|
157
|
<description>document_referencedDatasets trust level threshold</description>
|
|
158
|
</property>
|
|
159
|
<property>
|
|
160
|
<name>trust_level_threshold_document_pdb</name>
|
|
161
|
<value>$UNDEFINED$</value>
|
|
162
|
<description>document to protein databank trust level threshold</description>
|
|
163
|
</property>
|
|
164
|
<!-- -->
|
|
165
|
<property>
|
|
166
|
<name>action_hbase_remote_zookeeper_quorum</name>
|
|
167
|
<value>$UNDEFINED$</value>
|
|
168
|
<description>external hbase zookeeper quorum, set to empty value by default which means data will be exported to local hbase instance</description>
|
|
169
|
</property>
|
|
170
|
<property>
|
|
171
|
<name>action_hbase_remote_zookeeper_clientport</name>
|
|
172
|
<value>$UNDEFINED$</value>
|
|
173
|
<description>external hbase zookeeper client port, required only whe zookeeper quorum property is set</description>
|
|
174
|
</property>
|
|
175
|
<property>
|
|
176
|
<name>documentssimilarity_threshold</name>
|
|
177
|
<value>$UNDEFINED$</value>
|
|
178
|
<description>documents similarity threshold value below which similarity export is omitted</description>
|
|
179
|
</property>
|
|
180
|
<property>
|
|
181
|
<name>referenceextraction_pdb_url_root</name>
|
|
182
|
<value>$UNDEFINED$</value>
|
|
183
|
<description>protein databank URL root part to be concatenated with pdb identifier when forming final URL</description>
|
|
184
|
</property>
|
|
185
|
</parameters>
|
|
186
|
|
|
187
|
<global>
|
|
188
|
<job-tracker>${jobTracker}</job-tracker>
|
|
189
|
<name-node>${nameNode}</name-node>
|
|
190
|
<configuration>
|
|
191
|
<property>
|
|
192
|
<name>mapred.job.queue.name</name>
|
|
193
|
<value>${queueName}</value>
|
|
194
|
</property>
|
|
195
|
<property>
|
|
196
|
<name>mapreduce.inputformat.class</name>
|
|
197
|
<value>org.apache.avro.mapreduce.AvroKeyInputFormat</value>
|
|
198
|
</property>
|
|
199
|
<property>
|
|
200
|
<name>mapreduce.outputformat.class</name>
|
|
201
|
<value>org.apache.hadoop.mapreduce.lib.output.NullOutputFormat</value>
|
|
202
|
</property>
|
|
203
|
<property>
|
|
204
|
<name>mapred.reduce.tasks</name>
|
|
205
|
<value>0</value>
|
|
206
|
</property>
|
|
207
|
<property>
|
|
208
|
<name>io.serializations</name>
|
|
209
|
<value>org.apache.hadoop.io.serializer.WritableSerialization,org.apache.hadoop.io.serializer.avro.AvroSpecificSerialization,org.apache.hadoop.io.serializer.avro.AvroReflectSerialization,org.apache.avro.hadoop.io.AvroSerialization
|
|
210
|
</value>
|
|
211
|
</property>
|
|
212
|
<property>
|
|
213
|
<name>rpc.engine.org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolPB
|
|
214
|
</name>
|
|
215
|
<value>org.apache.hadoop.ipc.ProtobufRpcEngine</value>
|
|
216
|
</property>
|
|
217
|
<property>
|
|
218
|
<name>mapred.mapper.new-api</name>
|
|
219
|
<value>true</value>
|
|
220
|
</property>
|
|
221
|
<property>
|
|
222
|
<name>mapred.reducer.new-api</name>
|
|
223
|
<value>true</value>
|
|
224
|
</property>
|
|
225
|
<!-- disabling speculative execution for HBase writes on map phase -->
|
|
226
|
<property>
|
|
227
|
<name>mapred.map.tasks.speculative.execution</name>
|
|
228
|
<value>false</value>
|
|
229
|
</property>
|
|
230
|
<property>
|
|
231
|
<!-- supported since hadoop 2.3.0 -->
|
|
232
|
<name>mapreduce.map.speculative</name>
|
|
233
|
<value>false</value>
|
|
234
|
</property>
|
|
235
|
<property>
|
|
236
|
<name>mapreduce.map.class</name>
|
|
237
|
<value>eu.dnetlib.iis.export.actionmanager.HBaseActionManagerBasedExporterMapper</value>
|
|
238
|
</property>
|
|
239
|
<property>
|
|
240
|
<name>export.action.hbase.table.name</name>
|
|
241
|
<value>${action_hbase_table_name}</value>
|
|
242
|
</property>
|
|
243
|
<property>
|
|
244
|
<name>export.action.hbase.remote.zookeeper.quorum</name>
|
|
245
|
<value>${action_hbase_remote_zookeeper_quorum}</value>
|
|
246
|
</property>
|
|
247
|
<property>
|
|
248
|
<name>export.action.hbase.remote.zookeeper.clientport</name>
|
|
249
|
<value>${action_hbase_remote_zookeeper_clientport}</value>
|
|
250
|
</property>
|
|
251
|
<!-- actionset related -->
|
|
252
|
<property>
|
|
253
|
<name>export.action.setid</name>
|
|
254
|
<value>${action_set_id}</value>
|
|
255
|
</property>
|
|
256
|
<property>
|
|
257
|
<name>export.action.setid.person_similarities_websiteusage</name>
|
|
258
|
<value>${action_set_id_person_similarities_websiteusage}</value>
|
|
259
|
</property>
|
|
260
|
<property>
|
|
261
|
<name>export.action.setid.person_statistics</name>
|
|
262
|
<value>${action_set_id_person_statistics}</value>
|
|
263
|
</property>
|
|
264
|
<property>
|
|
265
|
<name>export.action.setid.project_statistics</name>
|
|
266
|
<value>${action_set_id_project_statistics}</value>
|
|
267
|
</property>
|
|
268
|
<property>
|
|
269
|
<name>export.action.setid.document_similarities_standard</name>
|
|
270
|
<value>${action_set_id_document_similarities_standard}</value>
|
|
271
|
</property>
|
|
272
|
<property>
|
|
273
|
<name>export.action.setid.document_similarities_websiteusage</name>
|
|
274
|
<value>${action_set_id_document_similarities_websiteusage}</value>
|
|
275
|
</property>
|
|
276
|
<property>
|
|
277
|
<name>export.action.setid.document_affiliations</name>
|
|
278
|
<value>${action_set_id_document_affiliations}</value>
|
|
279
|
</property>
|
|
280
|
<property>
|
|
281
|
<name>export.action.setid.document_statistics</name>
|
|
282
|
<value>${action_set_id_document_statistics}</value>
|
|
283
|
</property>
|
|
284
|
<property>
|
|
285
|
<name>export.action.setid.document_classes</name>
|
|
286
|
<value>${action_set_id_document_classes}</value>
|
|
287
|
</property>
|
|
288
|
<property>
|
|
289
|
<name>export.action.setid.document_referencedProjects</name>
|
|
290
|
<value>${action_set_id_document_referencedProjects}</value>
|
|
291
|
</property>
|
|
292
|
<property>
|
|
293
|
<name>export.action.setid.document_referencedDatasets</name>
|
|
294
|
<value>${action_set_id_document_referencedDatasets}</value>
|
|
295
|
</property>
|
|
296
|
<property>
|
|
297
|
<name>export.action.setid.document_referencedDocuments</name>
|
|
298
|
<value>${action_set_id_document_referencedDocuments}</value>
|
|
299
|
</property>
|
|
300
|
<property>
|
|
301
|
<name>export.action.setid.document_research_initiative</name>
|
|
302
|
<value>${action_set_id_document_research_initiative}</value>
|
|
303
|
</property>
|
|
304
|
<property>
|
|
305
|
<name>export.action.setid.document_pdb</name>
|
|
306
|
<value>${action_set_id_document_pdb}</value>
|
|
307
|
</property>
|
|
308
|
<!-- trust level threshold related -->
|
|
309
|
<property>
|
|
310
|
<name>export.trust.level.threshold</name>
|
|
311
|
<value>${trust_level_threshold}</value>
|
|
312
|
</property>
|
|
313
|
<property>
|
|
314
|
<name>export.trust.level.threshold.document_classes</name>
|
|
315
|
<value>${trust_level_threshold_document_classes}</value>
|
|
316
|
</property>
|
|
317
|
<property>
|
|
318
|
<name>export.trust.level.threshold.document_referencedProjects</name>
|
|
319
|
<value>${trust_level_threshold_document_referencedProjects}</value>
|
|
320
|
</property>
|
|
321
|
<property>
|
|
322
|
<name>export.trust.level.threshold.document_referencedDatasets</name>
|
|
323
|
<value>${trust_level_threshold_document_referencedDatasets}</value>
|
|
324
|
</property>
|
|
325
|
<property>
|
|
326
|
<name>export.trust.level.threshold.document_pdb</name>
|
|
327
|
<value>${trust_level_threshold_document_pdb}</value>
|
|
328
|
</property>
|
|
329
|
</configuration>
|
|
330
|
</global>
|
|
331
|
|
|
332
|
<start to="generate-schema" />
|
|
333
|
|
|
334
|
<!-- watch out for oozie.action.max.output.data limit set in oozie-site.xml
|
|
335
|
it cannot be exceeded when capturing output of schemas generated as properties.
|
|
336
|
When exceeded: either limit can be raised or multiple schema generating actions defined.
|
|
337
|
-->
|
|
338
|
<action name="generate-schema">
|
|
339
|
<java>
|
|
340
|
<main-class>eu.dnetlib.iis.core.javamapreduce.hack.AvroSchemaGenerator</main-class>
|
|
341
|
<arg>eu.dnetlib.iis.export.schemas.DocumentMetadata</arg>
|
|
342
|
<arg>eu.dnetlib.iis.referenceextraction.project.schemas.DocumentToProject</arg>
|
|
343
|
<arg>eu.dnetlib.iis.referenceextraction.dataset.schemas.DocumentToDataSet</arg>
|
|
344
|
<arg>eu.dnetlib.iis.export.schemas.DocumentToConceptIds</arg>
|
|
345
|
<arg>eu.dnetlib.iis.documentsclassification.schemas.DocumentToDocumentClasses</arg>
|
|
346
|
<arg>eu.dnetlib.iis.export.schemas.Citations</arg>
|
|
347
|
<arg>eu.dnetlib.iis.documentssimilarity.schemas.DocumentSimilarity</arg>
|
|
348
|
<arg>eu.dnetlib.iis.statistics.schemas.DocumentToDocumentStatistics</arg>
|
|
349
|
<arg>eu.dnetlib.iis.statistics.schemas.AuthorToAuthorStatistics</arg>
|
|
350
|
<arg>eu.dnetlib.iis.statistics.schemas.ProjectToProjectStatistics</arg>
|
|
351
|
<arg>eu.dnetlib.iis.websiteusage.schemas.DocumentsWithWebsiteUsageSimilarities</arg>
|
|
352
|
<capture-output />
|
|
353
|
</java>
|
|
354
|
<ok to="table-initializer" />
|
|
355
|
<error to="fail" />
|
|
356
|
</action>
|
|
357
|
|
|
358
|
<action name="table-initializer">
|
|
359
|
<java>
|
|
360
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
|
361
|
<arg>eu.dnetlib.iis.export.actionmanager.TableCreatorProcess</arg>
|
|
362
|
<arg>-Pexport.action.hbase.table.name=${action_hbase_table_name}</arg>
|
|
363
|
<arg>-Pexport.action.hbase.table.initialize=${action_hbase_table_initialize}</arg>
|
|
364
|
<arg>-Pexport.action.hbase.remote.zookeeper.quorum=${action_hbase_remote_zookeeper_quorum}</arg>
|
|
365
|
<arg>-Pexport.action.hbase.remote.zookeeper.clientport=${action_hbase_remote_zookeeper_clientport}</arg>
|
|
366
|
</java>
|
|
367
|
<ok to="forking" />
|
|
368
|
<error to="fail" />
|
|
369
|
</action>
|
|
370
|
|
|
371
|
<fork name="forking">
|
|
372
|
<path start="decision-exporter-document-metadata"/>
|
|
373
|
<path start="decision-exporter-document-to-dataset"/>
|
|
374
|
<path start="decision-exporter-document-to-project"/>
|
|
375
|
<path start="decision-exporter-document-to-project-concepts"/>
|
|
376
|
<path start="decision-exporter-document-to-researchinitiatives"/>
|
|
377
|
<path start="decision-exporter-document-to-pdb"/>
|
|
378
|
<path start="decision-exporter-document-to-documentclasses"/>
|
|
379
|
<path start="decision-exporter-citation"/>
|
|
380
|
<path start="decision-exporter-document-similiarity"/>
|
|
381
|
<path start="decision-exporter-document-statistics"/>
|
|
382
|
<path start="decision-exporter-person-statistics"/>
|
|
383
|
<path start="decision-exporter-project-statistics"/>
|
|
384
|
<path start="decision-exporter-websiteusage"/>
|
|
385
|
</fork>
|
|
386
|
|
|
387
|
<decision name="decision-exporter-document-metadata">
|
|
388
|
<switch>
|
|
389
|
<case to="joining">${input_document_metadata eq "$UNDEFINED$"}</case>
|
|
390
|
<default to="exporter-document-metadata"/>
|
|
391
|
</switch>
|
|
392
|
</decision>
|
|
393
|
|
|
394
|
<decision name="decision-exporter-document-to-project">
|
|
395
|
<switch>
|
|
396
|
<case to="joining">${input_document_to_project eq "$UNDEFINED$"}</case>
|
|
397
|
<default to="exporter-document-to-project"/>
|
|
398
|
</switch>
|
|
399
|
</decision>
|
|
400
|
|
|
401
|
<decision name="decision-exporter-document-to-dataset">
|
|
402
|
<switch>
|
|
403
|
<case to="joining">${input_document_to_dataset eq "$UNDEFINED$"}</case>
|
|
404
|
<default to="exporter-document-to-dataset"/>
|
|
405
|
</switch>
|
|
406
|
</decision>
|
|
407
|
|
|
408
|
<decision name="decision-exporter-document-to-project-concepts">
|
|
409
|
<switch>
|
|
410
|
<case to="joining">${input_document_to_project_concepts eq "$UNDEFINED$"}</case>
|
|
411
|
<default to="exporter-document-to-project-concepts"/>
|
|
412
|
</switch>
|
|
413
|
</decision>
|
|
414
|
|
|
415
|
<decision name="decision-exporter-document-to-researchinitiatives">
|
|
416
|
<switch>
|
|
417
|
<case to="joining">${input_document_to_research_initiatives eq "$UNDEFINED$"}</case>
|
|
418
|
<default to="exporter-document-to-researchinitiatives"/>
|
|
419
|
</switch>
|
|
420
|
</decision>
|
|
421
|
|
|
422
|
<decision name="decision-exporter-document-to-pdb">
|
|
423
|
<switch>
|
|
424
|
<case to="joining">${input_document_to_pdb eq "$UNDEFINED$"}</case>
|
|
425
|
<default to="exporter-document-to-pdb"/>
|
|
426
|
</switch>
|
|
427
|
</decision>
|
|
428
|
|
|
429
|
<decision name="decision-exporter-document-to-documentclasses">
|
|
430
|
<switch>
|
|
431
|
<case to="joining">${input_document_to_document_classes eq "$UNDEFINED$"}</case>
|
|
432
|
<default to="exporter-document-to-documentclasses"/>
|
|
433
|
</switch>
|
|
434
|
</decision>
|
|
435
|
|
|
436
|
<decision name="decision-exporter-citation">
|
|
437
|
<switch>
|
|
438
|
<case to="joining">${input_citations eq "$UNDEFINED$"}</case>
|
|
439
|
<default to="exporter-citation"/>
|
|
440
|
</switch>
|
|
441
|
</decision>
|
|
442
|
|
|
443
|
<decision name="decision-exporter-document-similiarity">
|
|
444
|
<switch>
|
|
445
|
<case to="joining">${input_document_similarity eq "$UNDEFINED$"}</case>
|
|
446
|
<default to="exporter-document-similiarity"/>
|
|
447
|
</switch>
|
|
448
|
</decision>
|
|
449
|
|
|
450
|
<decision name="decision-exporter-websiteusage">
|
|
451
|
<switch>
|
|
452
|
<case to="joining">${input_document_websiteusage_similarity eq "$UNDEFINED$"}</case>
|
|
453
|
<default to="exporter-document-websiteusage-similiarity"/>
|
|
454
|
</switch>
|
|
455
|
</decision>
|
|
456
|
|
|
457
|
<decision name="decision-exporter-document-statistics">
|
|
458
|
<switch>
|
|
459
|
<case to="joining">${input_document_statistics eq "$UNDEFINED$"}</case>
|
|
460
|
<default to="exporter-document-statistics"/>
|
|
461
|
</switch>
|
|
462
|
</decision>
|
|
463
|
|
|
464
|
<decision name="decision-exporter-person-statistics">
|
|
465
|
<switch>
|
|
466
|
<case to="joining">${input_author_statistics eq "$UNDEFINED$"}</case>
|
|
467
|
<default to="exporter-person-statistics"/>
|
|
468
|
</switch>
|
|
469
|
</decision>
|
|
470
|
|
|
471
|
<decision name="decision-exporter-project-statistics">
|
|
472
|
<switch>
|
|
473
|
<case to="joining">${input_project_statistics eq "$UNDEFINED$"}</case>
|
|
474
|
<default to="exporter-project-statistics"/>
|
|
475
|
</switch>
|
|
476
|
</decision>
|
|
477
|
|
|
478
|
<action name="exporter-document-metadata">
|
|
479
|
<map-reduce>
|
|
480
|
<configuration>
|
|
481
|
<property>
|
|
482
|
<name>export.action.builder.factory.classname</name>
|
|
483
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentAffiliationsActionBuilderModuleFactory</value>
|
|
484
|
</property>
|
|
485
|
<property>
|
|
486
|
<name>avro.schema.input.key</name>
|
|
487
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.export.schemas.DocumentMetadata']}</value>
|
|
488
|
</property>
|
|
489
|
<property>
|
|
490
|
<name>mapred.input.dir</name>
|
|
491
|
<value>${input_document_metadata}</value>
|
|
492
|
</property>
|
|
493
|
</configuration>
|
|
494
|
</map-reduce>
|
|
495
|
<ok to="joining" />
|
|
496
|
<error to="fail" />
|
|
497
|
</action>
|
|
498
|
|
|
499
|
<action name="exporter-document-to-project">
|
|
500
|
<map-reduce>
|
|
501
|
<configuration>
|
|
502
|
<property>
|
|
503
|
<name>export.action.builder.factory.classname</name>
|
|
504
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToProjectActionBuilderModuleFactory</value>
|
|
505
|
</property>
|
|
506
|
<property>
|
|
507
|
<name>avro.schema.input.key</name>
|
|
508
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.referenceextraction.project.schemas.DocumentToProject']}</value>
|
|
509
|
</property>
|
|
510
|
<property>
|
|
511
|
<name>mapred.input.dir</name>
|
|
512
|
<value>${input_document_to_project}</value>
|
|
513
|
</property>
|
|
514
|
</configuration>
|
|
515
|
</map-reduce>
|
|
516
|
<ok to="joining" />
|
|
517
|
<error to="fail" />
|
|
518
|
</action>
|
|
519
|
|
|
520
|
<action name="exporter-document-to-dataset">
|
|
521
|
<map-reduce>
|
|
522
|
<configuration>
|
|
523
|
<property>
|
|
524
|
<name>export.action.builder.factory.classname</name>
|
|
525
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToDataSetActionBuilderModuleFactory</value>
|
|
526
|
</property>
|
|
527
|
<property>
|
|
528
|
<name>avro.schema.input.key</name>
|
|
529
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.referenceextraction.dataset.schemas.DocumentToDataSet']}</value>
|
|
530
|
</property>
|
|
531
|
<property>
|
|
532
|
<name>mapred.input.dir</name>
|
|
533
|
<value>${input_document_to_dataset}</value>
|
|
534
|
</property>
|
|
535
|
</configuration>
|
|
536
|
</map-reduce>
|
|
537
|
<ok to="joining" />
|
|
538
|
<error to="fail" />
|
|
539
|
</action>
|
|
540
|
|
|
541
|
<action name="exporter-document-to-project-concepts">
|
|
542
|
<map-reduce>
|
|
543
|
<configuration>
|
|
544
|
<property>
|
|
545
|
<name>export.action.builder.factory.classname</name>
|
|
546
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToProjectConceptsActionBuilderModuleFactory</value>
|
|
547
|
</property>
|
|
548
|
<property>
|
|
549
|
<name>avro.schema.input.key</name>
|
|
550
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.export.schemas.DocumentToConceptIds']}</value>
|
|
551
|
</property>
|
|
552
|
<property>
|
|
553
|
<name>mapred.input.dir</name>
|
|
554
|
<value>${input_document_to_project_concepts}</value>
|
|
555
|
</property>
|
|
556
|
</configuration>
|
|
557
|
</map-reduce>
|
|
558
|
<ok to="joining" />
|
|
559
|
<error to="fail" />
|
|
560
|
</action>
|
|
561
|
|
|
562
|
<action name="exporter-document-to-researchinitiatives">
|
|
563
|
<map-reduce>
|
|
564
|
<configuration>
|
|
565
|
<property>
|
|
566
|
<name>export.action.builder.factory.classname</name>
|
|
567
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToConceptIdsActionBuilderModuleFactory</value>
|
|
568
|
</property>
|
|
569
|
<property>
|
|
570
|
<name>avro.schema.input.key</name>
|
|
571
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.export.schemas.DocumentToConceptIds']}</value>
|
|
572
|
</property>
|
|
573
|
<property>
|
|
574
|
<name>mapred.input.dir</name>
|
|
575
|
<value>${input_document_to_research_initiatives}</value>
|
|
576
|
</property>
|
|
577
|
</configuration>
|
|
578
|
</map-reduce>
|
|
579
|
<ok to="joining" />
|
|
580
|
<error to="fail" />
|
|
581
|
</action>
|
|
582
|
|
|
583
|
<action name="exporter-document-to-pdb">
|
|
584
|
<map-reduce>
|
|
585
|
<configuration>
|
|
586
|
<property>
|
|
587
|
<name>export.action.builder.factory.classname</name>
|
|
588
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToPdbActionBuilderModuleFactory</value>
|
|
589
|
</property>
|
|
590
|
<property>
|
|
591
|
<name>avro.schema.input.key</name>
|
|
592
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.export.schemas.DocumentToConceptIds']}</value>
|
|
593
|
</property>
|
|
594
|
<property>
|
|
595
|
<name>mapred.input.dir</name>
|
|
596
|
<value>${input_document_to_pdb}</value>
|
|
597
|
</property>
|
|
598
|
<property>
|
|
599
|
<name>export.referenceextraction.pdb.url.root</name>
|
|
600
|
<value>${referenceextraction_pdb_url_root}</value>
|
|
601
|
</property>
|
|
602
|
</configuration>
|
|
603
|
</map-reduce>
|
|
604
|
<ok to="joining" />
|
|
605
|
<error to="fail" />
|
|
606
|
</action>
|
|
607
|
|
|
608
|
<action name="exporter-document-to-documentclasses">
|
|
609
|
<map-reduce>
|
|
610
|
<configuration>
|
|
611
|
<property>
|
|
612
|
<name>export.action.builder.factory.classname</name>
|
|
613
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToDocumentClassesActionBuilderModuleFactory</value>
|
|
614
|
</property>
|
|
615
|
<property>
|
|
616
|
<name>avro.schema.input.key</name>
|
|
617
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.documentsclassification.schemas.DocumentToDocumentClasses']}</value>
|
|
618
|
</property>
|
|
619
|
<property>
|
|
620
|
<name>mapred.input.dir</name>
|
|
621
|
<value>${input_document_to_document_classes}</value>
|
|
622
|
</property>
|
|
623
|
</configuration>
|
|
624
|
</map-reduce>
|
|
625
|
<ok to="joining" />
|
|
626
|
<error to="fail" />
|
|
627
|
</action>
|
|
628
|
|
|
629
|
<action name="exporter-citation">
|
|
630
|
<map-reduce>
|
|
631
|
<configuration>
|
|
632
|
<property>
|
|
633
|
<name>export.action.builder.factory.classname</name>
|
|
634
|
<value>eu.dnetlib.iis.export.actionmanager.module.CitationsActionBuilderModuleFactory</value>
|
|
635
|
</property>
|
|
636
|
<property>
|
|
637
|
<name>avro.schema.input.key</name>
|
|
638
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.export.schemas.Citations']}</value>
|
|
639
|
</property>
|
|
640
|
<property>
|
|
641
|
<name>mapred.input.dir</name>
|
|
642
|
<value>${input_citations}</value>
|
|
643
|
</property>
|
|
644
|
</configuration>
|
|
645
|
</map-reduce>
|
|
646
|
<ok to="joining" />
|
|
647
|
<error to="fail" />
|
|
648
|
</action>
|
|
649
|
|
|
650
|
<action name="exporter-document-similiarity">
|
|
651
|
<map-reduce>
|
|
652
|
<configuration>
|
|
653
|
<property>
|
|
654
|
<name>export.action.builder.factory.classname</name>
|
|
655
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentSimilarityActionBuilderModuleFactory</value>
|
|
656
|
</property>
|
|
657
|
<property>
|
|
658
|
<name>avro.schema.input.key</name>
|
|
659
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.documentssimilarity.schemas.DocumentSimilarity']}</value>
|
|
660
|
</property>
|
|
661
|
<property>
|
|
662
|
<name>mapred.input.dir</name>
|
|
663
|
<value>${input_document_similarity}</value>
|
|
664
|
</property>
|
|
665
|
<property>
|
|
666
|
<name>export.documentssimilarity.threshold</name>
|
|
667
|
<value>${documentssimilarity_threshold}</value>
|
|
668
|
</property>
|
|
669
|
|
|
670
|
</configuration>
|
|
671
|
</map-reduce>
|
|
672
|
<ok to="joining" />
|
|
673
|
<error to="fail" />
|
|
674
|
</action>
|
|
675
|
|
|
676
|
<action name="exporter-document-statistics">
|
|
677
|
<map-reduce>
|
|
678
|
<configuration>
|
|
679
|
<property>
|
|
680
|
<name>export.action.builder.factory.classname</name>
|
|
681
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentToDocumentStatisticsActionBuilderModuleFactory</value>
|
|
682
|
</property>
|
|
683
|
<property>
|
|
684
|
<name>avro.schema.input.key</name>
|
|
685
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.statistics.schemas.DocumentToDocumentStatistics']}</value>
|
|
686
|
</property>
|
|
687
|
<property>
|
|
688
|
<name>mapred.input.dir</name>
|
|
689
|
<value>${input_document_statistics}</value>
|
|
690
|
</property>
|
|
691
|
</configuration>
|
|
692
|
</map-reduce>
|
|
693
|
<ok to="joining" />
|
|
694
|
<error to="fail" />
|
|
695
|
</action>
|
|
696
|
|
|
697
|
<action name="exporter-person-statistics">
|
|
698
|
<map-reduce>
|
|
699
|
<configuration>
|
|
700
|
<property>
|
|
701
|
<name>export.action.builder.factory.classname</name>
|
|
702
|
<value>eu.dnetlib.iis.export.actionmanager.module.AuthorToAuthorStatisticsActionBuilderModuleFactory</value>
|
|
703
|
</property>
|
|
704
|
<property>
|
|
705
|
<name>avro.schema.input.key</name>
|
|
706
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.statistics.schemas.AuthorToAuthorStatistics']}</value>
|
|
707
|
</property>
|
|
708
|
<property>
|
|
709
|
<name>mapred.input.dir</name>
|
|
710
|
<value>${input_author_statistics}</value>
|
|
711
|
</property>
|
|
712
|
</configuration>
|
|
713
|
</map-reduce>
|
|
714
|
<ok to="joining" />
|
|
715
|
<error to="fail" />
|
|
716
|
</action>
|
|
717
|
|
|
718
|
<action name="exporter-project-statistics">
|
|
719
|
<map-reduce>
|
|
720
|
<configuration>
|
|
721
|
<property>
|
|
722
|
<name>export.action.builder.factory.classname</name>
|
|
723
|
<value>eu.dnetlib.iis.export.actionmanager.module.ProjectToProjectStatisticsActionBuilderModuleFactory</value>
|
|
724
|
</property>
|
|
725
|
<property>
|
|
726
|
<name>avro.schema.input.key</name>
|
|
727
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.statistics.schemas.ProjectToProjectStatistics']}</value>
|
|
728
|
</property>
|
|
729
|
<property>
|
|
730
|
<name>mapred.input.dir</name>
|
|
731
|
<value>${input_project_statistics}</value>
|
|
732
|
</property>
|
|
733
|
</configuration>
|
|
734
|
</map-reduce>
|
|
735
|
<ok to="joining" />
|
|
736
|
<error to="fail" />
|
|
737
|
</action>
|
|
738
|
|
|
739
|
<action name="exporter-document-websiteusage-similiarity">
|
|
740
|
<map-reduce>
|
|
741
|
<configuration>
|
|
742
|
<property>
|
|
743
|
<name>export.action.builder.factory.classname</name>
|
|
744
|
<value>eu.dnetlib.iis.export.actionmanager.module.DocumentsWithWebsiteUsageSimilaritiesActionBuilderModuleFactory</value>
|
|
745
|
</property>
|
|
746
|
<property>
|
|
747
|
<name>avro.schema.input.key</name>
|
|
748
|
<value>${wf:actionData('generate-schema')['eu.dnetlib.iis.websiteusage.schemas.DocumentsWithWebsiteUsageSimilarities']}</value>
|
|
749
|
</property>
|
|
750
|
<property>
|
|
751
|
<name>mapred.input.dir</name>
|
|
752
|
<value>${input_document_websiteusage_similarity}</value>
|
|
753
|
</property>
|
|
754
|
</configuration>
|
|
755
|
</map-reduce>
|
|
756
|
<ok to="joining" />
|
|
757
|
<error to="fail" />
|
|
758
|
</action>
|
|
759
|
|
|
760
|
<join name="joining" to="end"/>
|
|
761
|
|
|
762
|
<kill name="fail">
|
|
763
|
<message>Unfortunately, the process failed -- error message:
|
|
764
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
|
765
|
</kill>
|
|
766
|
<end name="end" />
|
|
767
|
|
|
768
|
</workflow-app>
|