1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="mainworkflows_primary_main">
|
2
|
|
3
|
<parameters>
|
4
|
<!-- processing modes -->
|
5
|
<property>
|
6
|
<name>active_metadataextraction_export</name>
|
7
|
<value>false</value>
|
8
|
<description>flag indicating metadata extraction export should be enabled</description>
|
9
|
</property>
|
10
|
<property>
|
11
|
<name>active_referenceextraction_project</name>
|
12
|
<value>false</value>
|
13
|
<description>flag indicating project reference extraction should be enabled</description>
|
14
|
</property>
|
15
|
<property>
|
16
|
<name>active_referenceextraction_dataset</name>
|
17
|
<value>false</value>
|
18
|
<description>flag indicating dataset reference extraction should be enabled</description>
|
19
|
</property>
|
20
|
<property>
|
21
|
<!-- currently disabled by default -->
|
22
|
<name>active_referenceextraction_researchinitiative</name>
|
23
|
<value>false</value>
|
24
|
<description>flag indicating researchinitiative reference extraction should be enabled</description>
|
25
|
</property>
|
26
|
<property>
|
27
|
<name>active_documentsclassification</name>
|
28
|
<value>false</value>
|
29
|
<description>flag indicating documents classification should be enabled</description>
|
30
|
</property>
|
31
|
<property>
|
32
|
<name>active_documentssimilarity</name>
|
33
|
<value>false</value>
|
34
|
<description>flag indicating documents similarity should be enabled</description>
|
35
|
</property>
|
36
|
<property>
|
37
|
<name>active_citationmatching</name>
|
38
|
<value>false</value>
|
39
|
<description>flag indicating citation matching should be enabled</description>
|
40
|
</property>
|
41
|
<property>
|
42
|
<name>active_statistics</name>
|
43
|
<value>false</value>
|
44
|
<description>flag indicating statistics generation should be enabled</description>
|
45
|
</property>
|
46
|
<!-- import metadata related -->
|
47
|
<property>
|
48
|
<name>import_hbase_input_table</name>
|
49
|
<description>HBase input table holding InformationSpace, available on local cluster</description>
|
50
|
</property>
|
51
|
<property>
|
52
|
<name>import_hbase_approved_datasources_csv</name>
|
53
|
<value>$UNDEFINED$</value>
|
54
|
<description>CSV list of datasource ids to be approved during import. Applied on result and person entities.</description>
|
55
|
</property>
|
56
|
<!-- import datacite related -->
|
57
|
<property>
|
58
|
<name>import_mdstore_service_location</name>
|
59
|
<value>$UNDEFINED$</value>
|
60
|
<description>MDStore service (not WSDL) location URL</description>
|
61
|
</property>
|
62
|
<property>
|
63
|
<name>import_dataset_mdstore_ids_csv</name>
|
64
|
<value>$UNDEFINED$</value>
|
65
|
<description>MDStore identifier</description>
|
66
|
</property>
|
67
|
<!-- import content related -->
|
68
|
<property>
|
69
|
<name>import_content_object_store_location</name>
|
70
|
<value>$UNDEFINED$</value>
|
71
|
<description>object store service location required for content retrieval</description>
|
72
|
</property>
|
73
|
<property>
|
74
|
<name>import_content_objectstores_csv</name>
|
75
|
<value>$UNDEFINED$</value>
|
76
|
<description>CSV list of object stores identifiers to be processed</description>
|
77
|
</property>
|
78
|
<property>
|
79
|
<name>import_content_mimetypes_pdf</name>
|
80
|
<value>pdf,application/pdf</value>
|
81
|
<description>pdf mime types</description>
|
82
|
</property>
|
83
|
<property>
|
84
|
<name>import_content_mimetypes_text</name>
|
85
|
<value>text,text/plain</value>
|
86
|
<description>text mime types</description>
|
87
|
</property>
|
88
|
<property>
|
89
|
<name>import_content_mimetypes_xml_pmc</name>
|
90
|
<value>xml</value>
|
91
|
<description>xml pmc types</description>
|
92
|
</property>
|
93
|
<property>
|
94
|
<name>import_content_mimetypes_wos</name>
|
95
|
<value>file::WoS</value>
|
96
|
<description>WoS types</description>
|
97
|
</property>
|
98
|
<!-- import timeouts related -->
|
99
|
<property>
|
100
|
<name>import_resultset_client_read_timeout</name>
|
101
|
<value>60000</value>
|
102
|
<description>resultset client read timeout</description>
|
103
|
</property>
|
104
|
<property>
|
105
|
<name>import_content_connection_timeout</name>
|
106
|
<value>60000</value>
|
107
|
<description>import content connection timeout</description>
|
108
|
</property>
|
109
|
<property>
|
110
|
<name>import_content_read_timeout</name>
|
111
|
<value>60000</value>
|
112
|
<description>import content read timeout</description>
|
113
|
</property>
|
114
|
<!-- metadata extraction related -->
|
115
|
<property>
|
116
|
<name>metadataextraction_excluded_ids</name>
|
117
|
<value>$UNDEFINED$</value>
|
118
|
<description>list of content identifiers excluded from metadataextraction processing</description>
|
119
|
</property>
|
120
|
<property>
|
121
|
<name>metadataextraction_default_cache_location</name>
|
122
|
<value>/cache/metadataextraction</value>
|
123
|
<description>metadata extraction HDFS cache location</description>
|
124
|
</property>
|
125
|
<property>
|
126
|
<name>metadataextraction_processing_mode</name>
|
127
|
<value>StreamingMetadataExtractorMapper</value>
|
128
|
<description>metadata extraction processing mode</description>
|
129
|
</property>
|
130
|
<property>
|
131
|
<name>metadataextraction_input_classname</name>
|
132
|
<value>eu.dnetlib.iis.importer.auxiliary.schemas.DocumentContentUrl</value>
|
133
|
<description>metadata extraction input classname</description>
|
134
|
</property>
|
135
|
<!-- export related -->
|
136
|
<property>
|
137
|
<name>export_action_hbase_table_name</name>
|
138
|
<description>action manager hbase table name</description>
|
139
|
</property>
|
140
|
<property>
|
141
|
<name>export_action_hbase_table_initialize</name>
|
142
|
<description>flag indicating input table should be initialized</description>
|
143
|
</property>
|
144
|
<!-- action set id properties -->
|
145
|
<property>
|
146
|
<name>export_action_set_id</name>
|
147
|
<value>$UNDEFINED$</value>
|
148
|
<description>action-set identifier of exported data</description>
|
149
|
</property>
|
150
|
<property>
|
151
|
<name>export_action_set_id_dataset_similarities_websiteusage</name>
|
152
|
<value>$UNDEFINED$</value>
|
153
|
<description>dataset_similarities_websiteusage action-set identifier of exported data</description>
|
154
|
</property>
|
155
|
<property>
|
156
|
<name>export_action_set_id_person_similarities_websiteusage</name>
|
157
|
<value>$UNDEFINED$</value>
|
158
|
<description>person_similarities_websiteusage action-set identifier of exported data</description>
|
159
|
</property>
|
160
|
<property>
|
161
|
<name>export_action_set_id_person_statistics</name>
|
162
|
<value>$UNDEFINED$</value>
|
163
|
<description>person_statistics action-set identifier of exported data</description>
|
164
|
</property>
|
165
|
<property>
|
166
|
<name>export_action_set_id_project_statistics</name>
|
167
|
<value>$UNDEFINED$</value>
|
168
|
<description>project_statistics action-set identifier of exported data</description>
|
169
|
</property>
|
170
|
<property>
|
171
|
<name>export_action_set_id_document_similarities_standard</name>
|
172
|
<value>$UNDEFINED$</value>
|
173
|
<description>document_similarities_standard action-set identifier of exported data</description>
|
174
|
</property>
|
175
|
<property>
|
176
|
<name>export_action_set_id_document_similarities_websiteusage</name>
|
177
|
<value>$UNDEFINED$</value>
|
178
|
<description>document_similarities_websiteusage action-set identifier of exported data</description>
|
179
|
</property>
|
180
|
<property>
|
181
|
<name>export_action_set_id_document_extractedMetadata</name>
|
182
|
<value>$UNDEFINED$</value>
|
183
|
<description>document_extractedMetadata action-set identifier of exported data</description>
|
184
|
</property>
|
185
|
<property>
|
186
|
<name>export_action_set_id_document_statistics</name>
|
187
|
<value>$UNDEFINED$</value>
|
188
|
<description>document_statistics action-set identifier of exported data</description>
|
189
|
</property>
|
190
|
<property>
|
191
|
<name>export_action_set_id_document_classes</name>
|
192
|
<value>$UNDEFINED$</value>
|
193
|
<description>document_classes action-set identifier of exported data</description>
|
194
|
</property>
|
195
|
<property>
|
196
|
<name>export_action_set_id_document_referencedProjects</name>
|
197
|
<value>$UNDEFINED$</value>
|
198
|
<description>document_referencedProjects action-set identifier of exported data</description>
|
199
|
</property>
|
200
|
<property>
|
201
|
<name>export_action_set_id_document_referencedDatasets</name>
|
202
|
<value>$UNDEFINED$</value>
|
203
|
<description>document_referencedDatasets action-set identifier of exported data</description>
|
204
|
</property>
|
205
|
<property>
|
206
|
<name>export_action_set_id_document_referencedDocuments</name>
|
207
|
<value>$UNDEFINED$</value>
|
208
|
<description>document_referencedDocuments action-set identifier of exported data</description>
|
209
|
</property>
|
210
|
<property>
|
211
|
<name>export_action_set_id_document_research_initiative</name>
|
212
|
<value>$UNDEFINED$</value>
|
213
|
<description>document research initiative action-set identifier of exported data</description>
|
214
|
</property>
|
215
|
<!-- -->
|
216
|
<property>
|
217
|
<name>export_action_set_id_entity_dataset</name>
|
218
|
<value>$UNDEFINED$</value>
|
219
|
<description>action-set identifier of exported data containing dataset entities</description>
|
220
|
</property>
|
221
|
<property>
|
222
|
<name>export_action_hbase_remote_zookeeper_quorum</name>
|
223
|
<value>$UNDEFINED$</value>
|
224
|
<description>external hbase zookeeper quorum, set to empty value by default which means data will be exported to local hbase instance</description>
|
225
|
</property>
|
226
|
<property>
|
227
|
<name>export_action_hbase_remote_zookeeper_clientport</name>
|
228
|
<value>$UNDEFINED$</value>
|
229
|
<description>external hbase zookeeper client port, required only whe zookeeper quorum property is set</description>
|
230
|
</property>
|
231
|
<property>
|
232
|
<name>export_documentssimilarity_threshold</name>
|
233
|
<value>$UNDEFINED$</value>
|
234
|
<description>documents similarity threshold value below which similarity export is omitted</description>
|
235
|
</property>
|
236
|
</parameters>
|
237
|
|
238
|
<global>
|
239
|
<job-tracker>${jobTracker}</job-tracker>
|
240
|
<name-node>${nameNode}</name-node>
|
241
|
<configuration>
|
242
|
<property>
|
243
|
<name>mapred.job.queue.name</name>
|
244
|
<value>${queueName}</value>
|
245
|
</property>
|
246
|
</configuration>
|
247
|
</global>
|
248
|
|
249
|
<start to="import" />
|
250
|
|
251
|
<action name="import">
|
252
|
<sub-workflow>
|
253
|
<app-path>${wf:appPath()}/mainworkflows_common_import</app-path>
|
254
|
<propagate-configuration/>
|
255
|
<configuration>
|
256
|
<property>
|
257
|
<name>workingDir</name>
|
258
|
<value>${workingDir}/mainworkflows_common_import/working_dir</value>
|
259
|
</property>
|
260
|
<!-- importing modes -->
|
261
|
<property>
|
262
|
<name>active_import_metadata</name>
|
263
|
<value>true</value>
|
264
|
</property>
|
265
|
<property>
|
266
|
<name>active_import_dataset</name>
|
267
|
<value>${active_referenceextraction_dataset}</value>
|
268
|
</property>
|
269
|
<property>
|
270
|
<name>active_ingest_pmc_citations</name>
|
271
|
<value>false</value>
|
272
|
<!-- FIXME: enable after fixing PMC ingestion -->
|
273
|
<!--
|
274
|
<value>${active_citationmatching}</value>
|
275
|
-->
|
276
|
</property>
|
277
|
<!-- import metadata related -->
|
278
|
<property>
|
279
|
<name>hbase_input_table</name>
|
280
|
<value>${import_hbase_input_table}</value>
|
281
|
</property>
|
282
|
<property>
|
283
|
<name>hbase_approved_datasources_csv</name>
|
284
|
<value>${import_hbase_approved_datasources_csv}</value>
|
285
|
</property>
|
286
|
<!-- import datacite related -->
|
287
|
<property>
|
288
|
<name>mdstore_service_location</name>
|
289
|
<value>${import_mdstore_service_location}</value>
|
290
|
</property>
|
291
|
<property>
|
292
|
<name>dataset_mdstore_ids_csv</name>
|
293
|
<value>${import_dataset_mdstore_ids_csv}</value>
|
294
|
</property>
|
295
|
<!-- import content related -->
|
296
|
<property>
|
297
|
<name>objectstore_service_location</name>
|
298
|
<value>${import_content_object_store_location}</value>
|
299
|
</property>
|
300
|
<property>
|
301
|
<name>approved_objectstores_csv</name>
|
302
|
<value>${import_content_objectstores_csv}</value>
|
303
|
</property>
|
304
|
<property>
|
305
|
<name>mimetypes_pdf</name>
|
306
|
<value>${import_content_mimetypes_pdf}</value>
|
307
|
</property>
|
308
|
<property>
|
309
|
<name>mimetypes_text</name>
|
310
|
<value>${import_content_mimetypes_text}</value>
|
311
|
</property>
|
312
|
<property>
|
313
|
<name>mimetypes_xml_pmc</name>
|
314
|
<value>${import_content_mimetypes_xml_pmc}</value>
|
315
|
</property>
|
316
|
<property>
|
317
|
<name>mimetypes_wos</name>
|
318
|
<value>${import_content_mimetypes_wos}</value>
|
319
|
</property>
|
320
|
<!-- import timeouts related -->
|
321
|
<property>
|
322
|
<name>resultset_client_read_timeout</name>
|
323
|
<value>${import_resultset_client_read_timeout}</value>
|
324
|
</property>
|
325
|
<property>
|
326
|
<name>content_connection_timeout</name>
|
327
|
<value>${import_content_connection_timeout}</value>
|
328
|
</property>
|
329
|
<property>
|
330
|
<name>content_read_timeout</name>
|
331
|
<value>${import_content_read_timeout}</value>
|
332
|
</property>
|
333
|
<!-- metadata extraction related -->
|
334
|
<property>
|
335
|
<name>metadataextraction_excluded_ids</name>
|
336
|
<value>${metadataextraction_excluded_ids}</value>
|
337
|
</property>
|
338
|
<property>
|
339
|
<name>metadataextraction_default_cache_location</name>
|
340
|
<value>${metadataextraction_default_cache_location}</value>
|
341
|
</property>
|
342
|
<!-- metadatainput and metadataextraction output subdirectory names -->
|
343
|
<property>
|
344
|
<name>metadataextraction_output_name_meta</name>
|
345
|
<value>meta</value>
|
346
|
</property>
|
347
|
<property>
|
348
|
<name>metadataimport_output_name_document_meta</name>
|
349
|
<value>docmeta</value>
|
350
|
</property>
|
351
|
<property>
|
352
|
<name>metadataimport_output_name_document_project</name>
|
353
|
<value>docproject</value>
|
354
|
</property>
|
355
|
<property>
|
356
|
<name>metadataimport_output_name_document_relation</name>
|
357
|
<value>docrelation</value>
|
358
|
</property>
|
359
|
<property>
|
360
|
<name>metadataimport_output_name_project</name>
|
361
|
<value>project</value>
|
362
|
</property>
|
363
|
<property>
|
364
|
<name>metadataimport_output_name_person</name>
|
365
|
<value>person</value>
|
366
|
</property>
|
367
|
<property>
|
368
|
<name>metadataimport_output_name_dataset_id</name>
|
369
|
<value>datasetid</value>
|
370
|
</property>
|
371
|
<property>
|
372
|
<name>metadataimport_output_name_dedup_mapping</name>
|
373
|
<value>dedupmapping</value>
|
374
|
</property>
|
375
|
<!-- output parameters -->
|
376
|
<property>
|
377
|
<name>output_metadataextraction_root</name>
|
378
|
<value>${workingDir}/mainworkflows_common_import/metadataextraction</value>
|
379
|
</property>
|
380
|
<property>
|
381
|
<name>output_metadataimport_root</name>
|
382
|
<value>${workingDir}/mainworkflows_common_import/metadataimport</value>
|
383
|
</property>
|
384
|
<property>
|
385
|
<name>output_dataset</name>
|
386
|
<value>${workingDir}/mainworkflows_common_import/dataset</value>
|
387
|
</property>
|
388
|
<property>
|
389
|
<name>output_citation_pmc</name>
|
390
|
<value>${workingDir}/mainworkflows_common_import/citation_pmc</value>
|
391
|
</property>
|
392
|
<property>
|
393
|
<name>output_document_text</name>
|
394
|
<value>${workingDir}/mainworkflows_common_import/document-text</value>
|
395
|
</property>
|
396
|
<property>
|
397
|
<name>output_wos</name>
|
398
|
<value>${workingDir}/mainworkflows_common_import/wos</value>
|
399
|
</property>
|
400
|
</configuration>
|
401
|
</sub-workflow>
|
402
|
<ok to="mainworkflows_primary_processing"/>
|
403
|
<error to="fail" />
|
404
|
</action>
|
405
|
|
406
|
<action name="mainworkflows_primary_processing">
|
407
|
<sub-workflow>
|
408
|
<app-path>${wf:appPath()}/mainworkflows_primary_processing</app-path>
|
409
|
<propagate-configuration/>
|
410
|
<configuration>
|
411
|
<property>
|
412
|
<name>input_document_metadata</name>
|
413
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docmeta</value>
|
414
|
</property>
|
415
|
<property>
|
416
|
<name>input_document_to_project</name>
|
417
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docproject</value>
|
418
|
</property>
|
419
|
<property>
|
420
|
<name>input_document_text</name>
|
421
|
<value>${workingDir}/mainworkflows_common_import/document-text</value>
|
422
|
</property>
|
423
|
<property>
|
424
|
<name>input_document_text_wos</name>
|
425
|
<value>${workingDir}/mainworkflows_common_import/wos</value>
|
426
|
</property>
|
427
|
<property>
|
428
|
<name>input_project</name>
|
429
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/project</value>
|
430
|
</property>
|
431
|
<property>
|
432
|
<name>input_person</name>
|
433
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/person</value>
|
434
|
</property>
|
435
|
<property>
|
436
|
<name>input_dataset</name>
|
437
|
<value>${workingDir}/mainworkflows_common_import/dataset</value>
|
438
|
</property>
|
439
|
<property>
|
440
|
<name>input_extracted_document_metadata</name>
|
441
|
<value>${workingDir}/mainworkflows_common_import/metadataextraction/meta</value>
|
442
|
</property>
|
443
|
<property>
|
444
|
<name>input_citation_pmc</name>
|
445
|
<value>${workingDir}/mainworkflows_common_import/citation_pmc</value>
|
446
|
</property>
|
447
|
<property>
|
448
|
<name>output_document_to_project</name>
|
449
|
<value>${workingDir}/exported_to_be_cleaned/document_to_project</value>
|
450
|
</property>
|
451
|
<property>
|
452
|
<name>output_document_to_dataset</name>
|
453
|
<value>${workingDir}/exported_to_be_cleaned/document_to_dataset</value>
|
454
|
</property>
|
455
|
<property>
|
456
|
<name>output_document_to_research_initiatives</name>
|
457
|
<value>${workingDir}/exported/document_to_research_initiatives</value>
|
458
|
</property>
|
459
|
<property>
|
460
|
<name>output_document_to_document_classes</name>
|
461
|
<value>${workingDir}/exported/document_to_document_classes</value>
|
462
|
</property>
|
463
|
<property>
|
464
|
<name>output_citation</name>
|
465
|
<value>${workingDir}/exported/citation</value>
|
466
|
</property>
|
467
|
<property>
|
468
|
<name>output_document_similarity</name>
|
469
|
<value>${workingDir}/exported/document_similarity</value>
|
470
|
</property>
|
471
|
<property>
|
472
|
<name>output_document_statistics</name>
|
473
|
<value>${workingDir}/exported/document_statistics</value>
|
474
|
</property>
|
475
|
<property>
|
476
|
<name>output_author_statistics</name>
|
477
|
<value>${workingDir}/exported/author_statistics</value>
|
478
|
</property>
|
479
|
<property>
|
480
|
<name>output_project_statistics</name>
|
481
|
<value>${workingDir}/exported/project_statistics</value>
|
482
|
</property>
|
483
|
</configuration>
|
484
|
</sub-workflow>
|
485
|
<!--
|
486
|
<ok to="transformers_export_document_skip_imported_data"/>
|
487
|
-->
|
488
|
<ok to="decision-transform-metadataextraction-for-export"/>
|
489
|
<error to="fail" />
|
490
|
</action>
|
491
|
|
492
|
<decision name="decision-transform-metadataextraction-for-export">
|
493
|
<switch>
|
494
|
<case to="transformer_export_documentmetadata">${active_metadataextraction_export eq "true"}</case>
|
495
|
<default to="skip-transformer_export_documentmetadata"/>
|
496
|
</switch>
|
497
|
</decision>
|
498
|
|
499
|
<action name="transformer_export_documentmetadata">
|
500
|
<sub-workflow>
|
501
|
<app-path>${wf:appPath()}/transformer_export_documentmetadata</app-path>
|
502
|
<configuration>
|
503
|
<property>
|
504
|
<name>jobTracker</name>
|
505
|
<value>${jobTracker}</value>
|
506
|
</property>
|
507
|
<property>
|
508
|
<name>nameNode</name>
|
509
|
<value>${nameNode}</value>
|
510
|
</property>
|
511
|
<property>
|
512
|
<name>queueName</name>
|
513
|
<value>${queueName}</value>
|
514
|
</property>
|
515
|
<!-- Working directory of the subworkflow -->
|
516
|
<property>
|
517
|
<name>workingDir</name>
|
518
|
<value>${workingDir}/transformer_export_documentmetadata/working_dir</value>
|
519
|
</property>
|
520
|
<property>
|
521
|
<name>input_extracted_metadata</name>
|
522
|
<value>${workingDir}/mainworkflows_common_import/metadataextraction/meta</value>
|
523
|
</property>
|
524
|
<property>
|
525
|
<name>output_metadata</name>
|
526
|
<value>${workingDir}/transformer_export_documentmetadata/output_metadata</value>
|
527
|
</property>
|
528
|
</configuration>
|
529
|
</sub-workflow>
|
530
|
<ok to="forking_skip_imported_data"/>
|
531
|
<error to="fail"/>
|
532
|
</action>
|
533
|
|
534
|
<action name="skip-transformer_export_documentmetadata">
|
535
|
<java>
|
536
|
<prepare>
|
537
|
<!-- notice: directory have to aligned with skipped action output -->
|
538
|
<delete path="${nameNode}${workingDir}/transformer_export_documentmetadata" />
|
539
|
<mkdir path="${nameNode}${workingDir}/transformer_export_documentmetadata/working_dir" />
|
540
|
</prepare>
|
541
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
542
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
543
|
<arg>-C{document_metadata,
|
544
|
eu.dnetlib.iis.export.schemas.DocumentMetadata,
|
545
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
546
|
<arg>-SworkingDir=${workingDir}/transformer_export_documentmetadata/working_dir</arg>
|
547
|
<!-- notice: directory have to aligned with skipped action output -->
|
548
|
<arg>-Odocument_metadata=${workingDir}/transformer_export_documentmetadata/output_metadata</arg>
|
549
|
</java>
|
550
|
<ok to="forking_skip_imported_data"/>
|
551
|
<error to="fail"/>
|
552
|
</action>
|
553
|
|
554
|
<fork name="forking_skip_imported_data">
|
555
|
<path start="transformer_export_documenttodataset"/>
|
556
|
<path start="transformer_export_documenttoproject"/>
|
557
|
</fork>
|
558
|
|
559
|
<action name="transformer_export_documenttodataset">
|
560
|
<sub-workflow>
|
561
|
<app-path>${wf:appPath()}/transformer_export_documenttodataset_without_imported_data</app-path>
|
562
|
<propagate-configuration/>
|
563
|
<configuration>
|
564
|
<property>
|
565
|
<name>workingDir</name>
|
566
|
<value>${workingDir}/transformer_export_documenttodataset_without_imported_data/working_dir</value>
|
567
|
</property>
|
568
|
<property>
|
569
|
<name>input_document_to_dataset</name>
|
570
|
<value>${workingDir}/exported_to_be_cleaned/document_to_dataset</value>
|
571
|
</property>
|
572
|
<property>
|
573
|
<name>input_document_relation</name>
|
574
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docrelation</value>
|
575
|
</property>
|
576
|
<property>
|
577
|
<name>output_document_to_dataset</name>
|
578
|
<value>${workingDir}/transformer_export_documenttodataset_without_imported_data/output</value>
|
579
|
</property>
|
580
|
</configuration>
|
581
|
</sub-workflow>
|
582
|
<ok to="joining_skip_imported_data"/>
|
583
|
<error to="fail" />
|
584
|
</action>
|
585
|
|
586
|
<action name="transformer_export_documenttoproject">
|
587
|
<sub-workflow>
|
588
|
<app-path>${wf:appPath()}/transformer_export_documenttoproject_without_imported_data</app-path>
|
589
|
<propagate-configuration/>
|
590
|
<configuration>
|
591
|
<property>
|
592
|
<name>workingDir</name>
|
593
|
<value>${workingDir}/transformer_export_documenttoproject_without_imported_data/working_dir</value>
|
594
|
</property>
|
595
|
<property>
|
596
|
<name>input_document_to_project</name>
|
597
|
<value>${workingDir}/exported_to_be_cleaned/document_to_project</value>
|
598
|
</property>
|
599
|
<property>
|
600
|
<name>input_imported_document_to_project</name>
|
601
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docproject</value>
|
602
|
</property>
|
603
|
<property>
|
604
|
<name>output_document_to_project</name>
|
605
|
<value>${workingDir}/transformer_export_documenttoproject_without_imported_data/output</value>
|
606
|
</property>
|
607
|
</configuration>
|
608
|
</sub-workflow>
|
609
|
<ok to="joining_skip_imported_data"/>
|
610
|
<error to="fail" />
|
611
|
</action>
|
612
|
|
613
|
<join name="joining_skip_imported_data" to="export"/>
|
614
|
|
615
|
<action name="export">
|
616
|
<sub-workflow>
|
617
|
<app-path>${wf:appPath()}/mainworkflows_common_export</app-path>
|
618
|
<propagate-configuration/>
|
619
|
<configuration>
|
620
|
<property>
|
621
|
<name>workingDir</name>
|
622
|
<value>${workingDir}/mainworkflows_common_export/working_dir</value>
|
623
|
</property>
|
624
|
<!-- input ports -->
|
625
|
<property>
|
626
|
<name>input_document_metadata</name>
|
627
|
<value>${workingDir}/transformer_export_documentmetadata/output_metadata</value>
|
628
|
</property>
|
629
|
<property>
|
630
|
<name>input_document_to_project</name>
|
631
|
<value>${workingDir}/transformer_export_documenttoproject_without_imported_data/output</value>
|
632
|
</property>
|
633
|
<property>
|
634
|
<name>input_document_to_dataset</name>
|
635
|
<value>${workingDir}/transformer_export_documenttodataset_without_imported_data/output</value>
|
636
|
</property>
|
637
|
<property>
|
638
|
<name>input_document_to_research_initiatives</name>
|
639
|
<value>${workingDir}/exported/document_to_research_initiatives</value>
|
640
|
</property>
|
641
|
<property>
|
642
|
<name>input_document_to_document_classes</name>
|
643
|
<value>${workingDir}/exported/document_to_document_classes</value>
|
644
|
</property>
|
645
|
<property>
|
646
|
<name>input_citation</name>
|
647
|
<value>${workingDir}/exported/citation</value>
|
648
|
</property>
|
649
|
<property>
|
650
|
<name>input_document_similarity</name>
|
651
|
<value>${workingDir}/exported/document_similarity</value>
|
652
|
</property>
|
653
|
<property>
|
654
|
<name>input_document_statistics</name>
|
655
|
<value>${workingDir}/exported/document_statistics</value>
|
656
|
</property>
|
657
|
<property>
|
658
|
<name>input_author_statistics</name>
|
659
|
<value>${workingDir}/exported/author_statistics</value>
|
660
|
</property>
|
661
|
<property>
|
662
|
<name>input_project_statistics</name>
|
663
|
<value>${workingDir}/exported/project_statistics</value>
|
664
|
</property>
|
665
|
<property>
|
666
|
<name>input_aux_dataset_existing_id</name>
|
667
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/datasetid</value>
|
668
|
</property>
|
669
|
<!-- entities exporting modes -->
|
670
|
<property>
|
671
|
<name>active_export_referenceddataset_datasets</name>
|
672
|
<value>${active_referenceextraction_dataset}</value>
|
673
|
</property>
|
674
|
<property>
|
675
|
<name>active_export_referencedproject_entities</name>
|
676
|
<value>false</value>
|
677
|
</property>
|
678
|
<property>
|
679
|
<name>mdstore_service_location</name>
|
680
|
<value>${import_mdstore_service_location}</value>
|
681
|
</property>
|
682
|
<property>
|
683
|
<name>dataset_mdstore_ids_csv</name>
|
684
|
<value>${import_dataset_mdstore_ids_csv}</value>
|
685
|
</property>
|
686
|
<!-- export related -->
|
687
|
<property>
|
688
|
<name>action_hbase_table_name</name>
|
689
|
<value>${export_action_hbase_table_name}</value>
|
690
|
</property>
|
691
|
<property>
|
692
|
<name>action_hbase_table_initialize</name>
|
693
|
<value>${export_action_hbase_table_initialize}</value>
|
694
|
</property>
|
695
|
<!-- action set id properties -->
|
696
|
<property>
|
697
|
<name>action_set_id</name>
|
698
|
<value>${export_action_set_id}</value>
|
699
|
</property>
|
700
|
<property>
|
701
|
<name>action_set_id_dataset_similarities_websiteusage</name>
|
702
|
<value>${export_action_set_id_dataset_similarities_websiteusage}</value>
|
703
|
</property>
|
704
|
<property>
|
705
|
<name>action_set_id_person_similarities_websiteusage</name>
|
706
|
<value>${export_action_set_id_person_similarities_websiteusage}</value>
|
707
|
</property>
|
708
|
<property>
|
709
|
<name>action_set_id_person_statistics</name>
|
710
|
<value>${export_action_set_id_person_statistics}</value>
|
711
|
</property>
|
712
|
<property>
|
713
|
<name>action_set_id_project_statistics</name>
|
714
|
<value>${export_action_set_id_project_statistics}</value>
|
715
|
</property>
|
716
|
<property>
|
717
|
<name>action_set_id_document_similarities_standard</name>
|
718
|
<value>${export_action_set_id_document_similarities_standard}</value>
|
719
|
</property>
|
720
|
<property>
|
721
|
<name>action_set_id_document_similarities_websiteusage</name>
|
722
|
<value>${export_action_set_id_document_similarities_websiteusage}</value>
|
723
|
</property>
|
724
|
<property>
|
725
|
<name>action_set_id_document_extractedMetadata</name>
|
726
|
<value>${export_action_set_id_document_extractedMetadata}</value>
|
727
|
</property>
|
728
|
<property>
|
729
|
<name>action_set_id_document_statistics</name>
|
730
|
<value>${export_action_set_id_document_statistics}</value>
|
731
|
</property>
|
732
|
<property>
|
733
|
<name>action_set_id_document_classes</name>
|
734
|
<value>${export_action_set_id_document_classes}</value>
|
735
|
</property>
|
736
|
<property>
|
737
|
<name>action_set_id_document_referencedProjects</name>
|
738
|
<value>${export_action_set_id_document_referencedProjects}</value>
|
739
|
</property>
|
740
|
<property>
|
741
|
<name>action_set_id_document_referencedDatasets</name>
|
742
|
<value>${export_action_set_id_document_referencedDatasets}</value>
|
743
|
</property>
|
744
|
<property>
|
745
|
<name>action_set_id_document_referencedDocuments</name>
|
746
|
<value>${export_action_set_id_document_referencedDocuments}</value>
|
747
|
</property>
|
748
|
<property>
|
749
|
<name>action_set_id_document_research_initiative</name>
|
750
|
<value>${export_action_set_id_document_research_initiative}</value>
|
751
|
</property>
|
752
|
<property>
|
753
|
<name>action_set_id_entity_dataset</name>
|
754
|
<value>${export_action_set_id_entity_dataset}</value>
|
755
|
</property>
|
756
|
<property>
|
757
|
<name>action_hbase_remote_zookeeper_quorum</name>
|
758
|
<value>${export_action_hbase_remote_zookeeper_quorum}</value>
|
759
|
</property>
|
760
|
<property>
|
761
|
<name>action_hbase_remote_zookeeper_clientport</name>
|
762
|
<value>${export_action_hbase_remote_zookeeper_clientport}</value>
|
763
|
</property>
|
764
|
<property>
|
765
|
<name>documentssimilarity_threshold</name>
|
766
|
<value>${export_documentssimilarity_threshold}</value>
|
767
|
</property>
|
768
|
</configuration>
|
769
|
</sub-workflow>
|
770
|
<ok to="end"/>
|
771
|
<error to="fail" />
|
772
|
</action>
|
773
|
|
774
|
<kill name="fail">
|
775
|
<message>Unfortunately, the process failed -- error message:
|
776
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
777
|
</kill>
|
778
|
<end name="end" />
|
779
|
</workflow-app>
|