1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="mainworkflows_preprocessing_main">
|
2
|
<parameters>
|
3
|
<property>
|
4
|
<name>remove_sideproducts</name>
|
5
|
<value>true</value>
|
6
|
<description>flag indicating inference side products will be erased</description>
|
7
|
</property>
|
8
|
<!-- import concepts related -->
|
9
|
<property>
|
10
|
<name>import_islookup_service_location</name>
|
11
|
<description>IS Lookup service location</description>
|
12
|
</property>
|
13
|
<property>
|
14
|
<name>import_project_concepts_context_ids_csv</name>
|
15
|
<value>fet-fp7,fet-h2020</value>
|
16
|
<description>comma separated list of concepts context identifiers to be picked by ISLookup</description>
|
17
|
</property>
|
18
|
<!-- import project related -->
|
19
|
<property>
|
20
|
<name>import_database_service_location</name>
|
21
|
<description>Database service (not WSDL) location URL</description>
|
22
|
</property>
|
23
|
<property>
|
24
|
<name>import_database_dbname</name>
|
25
|
<value>dnet_openaireplus_node0_t</value>
|
26
|
<description>database name</description>
|
27
|
</property>
|
28
|
<!-- import datacite related, export datacite & wos related -->
|
29
|
<property>
|
30
|
<name>import_mdstore_service_location</name>
|
31
|
<description>MDStore service (not WSDL) location URL</description>
|
32
|
</property>
|
33
|
<property>
|
34
|
<name>import_dataset_mdstore_ids_csv</name>
|
35
|
<description>dataset MDStore identifier</description>
|
36
|
</property>
|
37
|
<property>
|
38
|
<name>import_wos_mdstore_id</name>
|
39
|
<description>WoS MDStore identifier</description>
|
40
|
</property>
|
41
|
<!-- import content related -->
|
42
|
<!-- currently disabled, input_document_content property is handled as input holding DocumentContent datastore -->
|
43
|
<property>
|
44
|
<name>import_content_object_store_location</name>
|
45
|
<value>$UNDEFINED$</value>
|
46
|
<description>object store service location required for content retrieval</description>
|
47
|
</property>
|
48
|
<property>
|
49
|
<name>import_content_wos_plaintext_objectstores_csv</name>
|
50
|
<value>$UNDEFINED$</value>
|
51
|
<description>CSV list of objectstore ids to be approved during WoS plaintext import.</description>
|
52
|
</property>
|
53
|
<property>
|
54
|
<name>import_content_datacite_objectstores_csv</name>
|
55
|
<value>$UNDEFINED$</value>
|
56
|
<description>CSV list of objectstore ids to be approved during datacite import.</description>
|
57
|
</property>
|
58
|
<!-- import content mime types -->
|
59
|
<property>
|
60
|
<name>import_content_mimetypes_pdf</name>
|
61
|
<value>pdf,application/pdf</value>
|
62
|
<description>pdf mime types</description>
|
63
|
</property>
|
64
|
<property>
|
65
|
<name>import_content_mimetypes_text</name>
|
66
|
<value>text,text/plain</value>
|
67
|
<description>text mime types</description>
|
68
|
</property>
|
69
|
<property>
|
70
|
<name>import_content_mimetypes_html</name>
|
71
|
<value>text/html</value>
|
72
|
<description>html mime types</description>
|
73
|
</property>
|
74
|
<property>
|
75
|
<name>import_content_mimetypes_xml_pmc</name>
|
76
|
<value>xml</value>
|
77
|
<description>xml pmc types</description>
|
78
|
</property>
|
79
|
<property>
|
80
|
<name>import_content_mimetypes_wos</name>
|
81
|
<value>file::WoS</value>
|
82
|
<description>WoS mime types</description>
|
83
|
</property>
|
84
|
<!-- import timeouts related -->
|
85
|
<property>
|
86
|
<name>import_resultset_client_read_timeout</name>
|
87
|
<value>60000</value>
|
88
|
<description>resultset client read timeout</description>
|
89
|
</property>
|
90
|
<property>
|
91
|
<name>import_content_connection_timeout</name>
|
92
|
<value>60000</value>
|
93
|
<description>import content connection timeout</description>
|
94
|
</property>
|
95
|
<property>
|
96
|
<name>import_content_read_timeout</name>
|
97
|
<value>60000</value>
|
98
|
<description>import content read timeout</description>
|
99
|
</property>
|
100
|
<!-- metadata extraction related -->
|
101
|
<property>
|
102
|
<name>metadataextraction_excluded_checksums</name>
|
103
|
<value>$UNDEFINED$</value>
|
104
|
<description>list of content checksums excluded from metadataextraction processing</description>
|
105
|
</property>
|
106
|
<property>
|
107
|
<name>metadataextraction_max_file_size_mb</name>
|
108
|
<value>500</value>
|
109
|
<description>maximum allowed file size in Megabytes</description>
|
110
|
</property>
|
111
|
<property>
|
112
|
<name>metadataextraction_default_cache_location</name>
|
113
|
<value>/cache/metadataextraction</value>
|
114
|
<description>metadata extraction cache location, path pointing to root cache directory holding meta.json file</description>
|
115
|
</property>
|
116
|
<!-- export related -->
|
117
|
<property>
|
118
|
<name>active_export_to_hbase</name>
|
119
|
<value>true</value>
|
120
|
<description>flag indicating hbase export should be performed</description>
|
121
|
</property>
|
122
|
<property>
|
123
|
<name>active_export_to_json</name>
|
124
|
<value>false</value>
|
125
|
<description>flag indicating json export should be performed</description>
|
126
|
</property>
|
127
|
<property>
|
128
|
<name>export_action_hbase_table_name</name>
|
129
|
<description>action manager hbase table name</description>
|
130
|
</property>
|
131
|
<property>
|
132
|
<name>export_action_hbase_table_initialize</name>
|
133
|
<description>flag indicating input table should be initialized</description>
|
134
|
</property>
|
135
|
<!-- action set id properties -->
|
136
|
<property>
|
137
|
<name>export_action_set_id</name>
|
138
|
<value>$UNDEFINED$</value>
|
139
|
<description>action-set identifier of exported data</description>
|
140
|
</property>
|
141
|
<property>
|
142
|
<name>export_action_set_id_document_referencedProjects</name>
|
143
|
<value>$UNDEFINED$</value>
|
144
|
<description>document_referencedProjects action-set identifier of exported data</description>
|
145
|
</property>
|
146
|
<property>
|
147
|
<name>export_action_set_id_document_referencedDatasets</name>
|
148
|
<value>$UNDEFINED$</value>
|
149
|
<description>document_referencedDatasets action-set identifier of exported data</description>
|
150
|
</property>
|
151
|
<property>
|
152
|
<name>export_action_set_id_entity_wos</name>
|
153
|
<description>action-set identifier of exported data containing wos entities</description>
|
154
|
</property>
|
155
|
<property>
|
156
|
<name>export_action_set_id_entity_dataset</name>
|
157
|
<description>action-set identifier of exported data containing dataset entities</description>
|
158
|
</property>
|
159
|
<property>
|
160
|
<name>export_action_hbase_remote_zookeeper_quorum</name>
|
161
|
<value>$UNDEFINED$</value>
|
162
|
<description>external hbase zookeeper quorum, set to empty value by default which means data will be exported to local hbase instance</description>
|
163
|
</property>
|
164
|
<property>
|
165
|
<name>export_action_hbase_remote_zookeeper_clientport</name>
|
166
|
<value>$UNDEFINED$</value>
|
167
|
<description>external hbase zookeeper client port, required only whe zookeeper quorum property is set</description>
|
168
|
</property>
|
169
|
<!-- trust level threshold section -->
|
170
|
<property>
|
171
|
<name>export_trust_level_threshold</name>
|
172
|
<value>$UNDEFINED$</value>
|
173
|
<description>default trust level threshold of exported data</description>
|
174
|
</property>
|
175
|
<property>
|
176
|
<name>export_trust_level_threshold_document_referencedProjects</name>
|
177
|
<value>$UNDEFINED$</value>
|
178
|
<description>document_referencedProjects trust level threshold</description>
|
179
|
</property>
|
180
|
<property>
|
181
|
<name>export_trust_level_threshold_document_referencedDatasets</name>
|
182
|
<value>$UNDEFINED$</value>
|
183
|
<description>document_referencedDatasets trust level threshold</description>
|
184
|
</property>
|
185
|
<!-- -->
|
186
|
<!-- working directory related -->
|
187
|
<property>
|
188
|
<name>execution_environment</name>
|
189
|
<value>preprocessing</value>
|
190
|
<description>execution environment used for workingDir creation</description>
|
191
|
</property>
|
192
|
<property>
|
193
|
<name>workingDir</name>
|
194
|
<value>/user/${user.name}/iis/working_dirs/${execution_environment}</value>
|
195
|
<description>working directory</description>
|
196
|
</property>
|
197
|
</parameters>
|
198
|
|
199
|
<global>
|
200
|
<job-tracker>${jobTracker}</job-tracker>
|
201
|
<name-node>${nameNode}</name-node>
|
202
|
<configuration>
|
203
|
<property>
|
204
|
<name>mapred.job.queue.name</name>
|
205
|
<value>${queueName}</value>
|
206
|
</property>
|
207
|
</configuration>
|
208
|
</global>
|
209
|
|
210
|
<start to="init-workingDir"/>
|
211
|
|
212
|
<action name="init-workingDir">
|
213
|
<fs>
|
214
|
<delete path="${nameNode}${workingDir}" />
|
215
|
<mkdir path="${nameNode}${workingDir}" />
|
216
|
</fs>
|
217
|
<ok to="copy-version"/>
|
218
|
<error to="fail"/>
|
219
|
</action>
|
220
|
|
221
|
<action name="copy-version">
|
222
|
<distcp xmlns="uri:oozie:distcp-action:0.1">
|
223
|
<job-tracker>${jobTracker}</job-tracker>
|
224
|
<name-node>${nameNode}</name-node>
|
225
|
<arg>${wf:appPath()}/version.properties</arg>
|
226
|
<arg>${nameNode}${workingDir}</arg>
|
227
|
</distcp>
|
228
|
<ok to="import"/>
|
229
|
<error to="fail"/>
|
230
|
</action>
|
231
|
|
232
|
<action name="import">
|
233
|
<sub-workflow>
|
234
|
<app-path>${wf:appPath()}/mainworkflows_preprocessing_import</app-path>
|
235
|
<propagate-configuration/>
|
236
|
<configuration>
|
237
|
<property>
|
238
|
<name>workingDir</name>
|
239
|
<value>${workingDir}/mainworkflows_preprocessing_import/working_dir</value>
|
240
|
</property>
|
241
|
<!-- import project related -->
|
242
|
<property>
|
243
|
<name>database_service_location</name>
|
244
|
<value>${import_database_service_location}</value>
|
245
|
</property>
|
246
|
<property>
|
247
|
<name>database_dbname</name>
|
248
|
<value>${import_database_dbname}</value>
|
249
|
</property>
|
250
|
<!-- project concept related -->
|
251
|
<property>
|
252
|
<name>islookup_service_location</name>
|
253
|
<value>${import_islookup_service_location}</value>
|
254
|
</property>
|
255
|
<property>
|
256
|
<name>project_concepts_context_ids_csv</name>
|
257
|
<value>${import_project_concepts_context_ids_csv}</value>
|
258
|
</property>
|
259
|
<!-- import datacite related -->
|
260
|
<property>
|
261
|
<name>mdstore_service_location</name>
|
262
|
<value>${import_mdstore_service_location}</value>
|
263
|
</property>
|
264
|
<property>
|
265
|
<name>dataset_mdstore_ids_csv</name>
|
266
|
<value>${import_dataset_mdstore_ids_csv}</value>
|
267
|
</property>
|
268
|
<!-- import content related -->
|
269
|
<property>
|
270
|
<name>objectstore_service_location</name>
|
271
|
<value>${import_content_object_store_location}</value>
|
272
|
</property>
|
273
|
<property>
|
274
|
<name>approved_objectstores_csv</name>
|
275
|
<value>${import_content_datacite_objectstores_csv}</value>
|
276
|
</property>
|
277
|
<property>
|
278
|
<name>wos_plaintext_objectstores_csv</name>
|
279
|
<value>${import_content_wos_plaintext_objectstores_csv}</value>
|
280
|
</property>
|
281
|
<property>
|
282
|
<name>mimetypes_pdf</name>
|
283
|
<value>${import_content_mimetypes_pdf}</value>
|
284
|
</property>
|
285
|
<property>
|
286
|
<name>mimetypes_text</name>
|
287
|
<value>${import_content_mimetypes_text}</value>
|
288
|
</property>
|
289
|
<property>
|
290
|
<name>mimetypes_html</name>
|
291
|
<value>${import_content_mimetypes_html}</value>
|
292
|
</property>
|
293
|
<property>
|
294
|
<name>mimetypes_xml_pmc</name>
|
295
|
<value>${import_content_mimetypes_xml_pmc}</value>
|
296
|
</property>
|
297
|
<property>
|
298
|
<name>mimetypes_wos</name>
|
299
|
<value>${import_content_mimetypes_wos}</value>
|
300
|
</property>
|
301
|
<!-- import timeouts related -->
|
302
|
<property>
|
303
|
<name>resultset_client_read_timeout</name>
|
304
|
<value>${import_resultset_client_read_timeout}</value>
|
305
|
</property>
|
306
|
<property>
|
307
|
<name>content_connection_timeout</name>
|
308
|
<value>${import_content_connection_timeout}</value>
|
309
|
</property>
|
310
|
<property>
|
311
|
<name>content_read_timeout</name>
|
312
|
<value>${import_content_read_timeout}</value>
|
313
|
</property>
|
314
|
<!-- metadata extraction related are autmatically propagated-->
|
315
|
<!-- output parameters -->
|
316
|
<property>
|
317
|
<name>output_extracted_document_metadata</name>
|
318
|
<value>${workingDir}/mainworkflows_preprocessing_import/extracted_document_metadata</value>
|
319
|
</property>
|
320
|
<property>
|
321
|
<name>output_metadataimport_root</name>
|
322
|
<value>${workingDir}/mainworkflows_preprocessing_import/metadataimport</value>
|
323
|
</property>
|
324
|
<property>
|
325
|
<name>output_dataset</name>
|
326
|
<value>${workingDir}/mainworkflows_preprocessing_import/dataset</value>
|
327
|
</property>
|
328
|
<property>
|
329
|
<name>output_dataset_to_mdstore</name>
|
330
|
<value>${workingDir}/mainworkflows_preprocessing_import/dataset_to_mdstore</value>
|
331
|
</property>
|
332
|
<property>
|
333
|
<name>output_document_text</name>
|
334
|
<value>${workingDir}/mainworkflows_preprocessing_import/document-text</value>
|
335
|
</property>
|
336
|
<property>
|
337
|
<name>output_wos_text</name>
|
338
|
<value>${workingDir}/mainworkflows_preprocessing_import/wos-text</value>
|
339
|
</property>
|
340
|
<property>
|
341
|
<name>output_project_concept</name>
|
342
|
<value>${workingDir}/mainworkflows_preprocessing_import/project-concept</value>
|
343
|
</property>
|
344
|
<property>
|
345
|
<name>output_faults</name>
|
346
|
<value>${workingDir}/mainworkflows_preprocessing_import/faults</value>
|
347
|
</property>
|
348
|
<!-- TODO output_wos is part of common/import and should replace output_wos_text!!!
|
349
|
Currently this directory is not used!!!
|
350
|
-->
|
351
|
<property>
|
352
|
<name>output_wos</name>
|
353
|
<value>${workingDir}/mainworkflows_preprocessing_import/wos</value>
|
354
|
</property>
|
355
|
</configuration>
|
356
|
</sub-workflow>
|
357
|
<ok to="referenceextraction_forking"/>
|
358
|
<error to="fail" />
|
359
|
</action>
|
360
|
|
361
|
<fork name="referenceextraction_forking">
|
362
|
<path start="referenceextraction_dataset"/>
|
363
|
<path start="ingest_webcrawl_fundings"/>
|
364
|
</fork>
|
365
|
|
366
|
<action name="referenceextraction_dataset">
|
367
|
<sub-workflow>
|
368
|
<app-path>${wf:appPath()}/referenceextraction_dataset</app-path>
|
369
|
<propagate-configuration/>
|
370
|
<configuration>
|
371
|
<property>
|
372
|
<name>workingDir</name>
|
373
|
<value>${workingDir}/referenceextraction_dataset/working_dir</value>
|
374
|
</property>
|
375
|
<property>
|
376
|
<name>input_document_text</name>
|
377
|
<value>${workingDir}/mainworkflows_preprocessing_import/document-text</value>
|
378
|
</property>
|
379
|
<property>
|
380
|
<name>input_dataset</name>
|
381
|
<value>${workingDir}/mainworkflows_preprocessing_import/dataset</value>
|
382
|
</property>
|
383
|
<property>
|
384
|
<name>output_document_to_dataset</name>
|
385
|
<value>${workingDir}/referenceextraction_dataset/document_datasets</value>
|
386
|
</property>
|
387
|
</configuration>
|
388
|
</sub-workflow>
|
389
|
<ok to="referenceextraction_joining"/>
|
390
|
<error to="fail" />
|
391
|
</action>
|
392
|
|
393
|
<action name="ingest_webcrawl_fundings">
|
394
|
<sub-workflow>
|
395
|
<app-path>${wf:appPath()}/ingest_webcrawl_fundings</app-path>
|
396
|
<propagate-configuration/>
|
397
|
<configuration>
|
398
|
<property>
|
399
|
<name>workingDir</name>
|
400
|
<value>${workingDir}/ingest_webcrawl_fundings/working_dir</value>
|
401
|
</property>
|
402
|
<property>
|
403
|
<name>input</name>
|
404
|
<value>${workingDir}/mainworkflows_preprocessing_import/wos-text</value>
|
405
|
</property>
|
406
|
<property>
|
407
|
<name>output</name>
|
408
|
<value>${workingDir}/ingest_webcrawl_fundings/output</value>
|
409
|
</property>
|
410
|
</configuration>
|
411
|
</sub-workflow>
|
412
|
<ok to="referenceextraction_project"/>
|
413
|
<error to="fail" />
|
414
|
</action>
|
415
|
|
416
|
<action name="referenceextraction_project">
|
417
|
<sub-workflow>
|
418
|
<app-path>${wf:appPath()}/referenceextraction_project</app-path>
|
419
|
<propagate-configuration/>
|
420
|
<configuration>
|
421
|
<property>
|
422
|
<name>workingDir</name>
|
423
|
<value>${workingDir}/referenceextraction_project/working_dir</value>
|
424
|
</property>
|
425
|
<property>
|
426
|
<name>input_document_text</name>
|
427
|
<value>${workingDir}/ingest_webcrawl_fundings/output</value>
|
428
|
</property>
|
429
|
<property>
|
430
|
<name>input_project</name>
|
431
|
<value>${workingDir}/mainworkflows_preprocessing_import/metadataimport/project</value>
|
432
|
</property>
|
433
|
<property>
|
434
|
<name>output_document_to_project</name>
|
435
|
<value>${workingDir}/referenceextraction_project/document_projects</value>
|
436
|
</property>
|
437
|
</configuration>
|
438
|
</sub-workflow>
|
439
|
<ok to="transformers_project_toconcept"/>
|
440
|
<error to="fail" />
|
441
|
</action>
|
442
|
|
443
|
<action name="transformers_project_toconcept">
|
444
|
<sub-workflow>
|
445
|
<app-path>${wf:appPath()}/transformers_project_toconcept</app-path>
|
446
|
<propagate-configuration/>
|
447
|
<configuration>
|
448
|
<property>
|
449
|
<name>workingDir</name>
|
450
|
<value>${workingDir}/transformers_project_toconcept/working_dir</value>
|
451
|
</property>
|
452
|
<property>
|
453
|
<name>input_document_to_project</name>
|
454
|
<value>${workingDir}/referenceextraction_project/document_projects</value>
|
455
|
</property>
|
456
|
<property>
|
457
|
<name>input_project</name>
|
458
|
<value>${workingDir}/mainworkflows_preprocessing_import/metadataimport/project</value>
|
459
|
</property>
|
460
|
<property>
|
461
|
<name>input_concept</name>
|
462
|
<value>${workingDir}/mainworkflows_preprocessing_import/project-concept</value>
|
463
|
</property>
|
464
|
<property>
|
465
|
<name>output</name>
|
466
|
<value>${workingDir}/transformers_project_toconcept/out</value>
|
467
|
</property>
|
468
|
</configuration>
|
469
|
</sub-workflow>
|
470
|
<ok to="transformers_concept_to_researchinitiatives"/>
|
471
|
<error to="fail" />
|
472
|
</action>
|
473
|
|
474
|
<action name="transformers_concept_to_researchinitiatives">
|
475
|
<sub-workflow>
|
476
|
<app-path>${wf:appPath()}/transformers_export_researchinitiatives</app-path>
|
477
|
<propagate-configuration/>
|
478
|
<configuration>
|
479
|
<property>
|
480
|
<name>workingDir</name>
|
481
|
<value>${workingDir}/transformers_concept_to_researchinitiatives/working_dir</value>
|
482
|
</property>
|
483
|
<property>
|
484
|
<name>input_document_to_research_initiative</name>
|
485
|
<value>${workingDir}/transformers_project_toconcept/out</value>
|
486
|
</property>
|
487
|
<property>
|
488
|
<name>output_document_to_research_initiatives</name>
|
489
|
<value>${workingDir}/transformers_concept_to_researchinitiatives/out</value>
|
490
|
</property>
|
491
|
</configuration>
|
492
|
</sub-workflow>
|
493
|
<ok to="referenceextraction_joining"/>
|
494
|
<error to="fail" />
|
495
|
</action>
|
496
|
|
497
|
<join name="referenceextraction_joining" to="decision-export-to-hbase"/>
|
498
|
|
499
|
<decision name="decision-export-to-hbase">
|
500
|
<switch>
|
501
|
<case to="export-to-hbase">${active_export_to_hbase eq "true"}</case>
|
502
|
<default to="decision-export-to-json"/>
|
503
|
</switch>
|
504
|
</decision>
|
505
|
|
506
|
<action name="export-to-hbase">
|
507
|
<sub-workflow>
|
508
|
<app-path>${wf:appPath()}/mainworkflows_common_export</app-path>
|
509
|
<propagate-configuration/>
|
510
|
<configuration>
|
511
|
<property>
|
512
|
<name>workingDir</name>
|
513
|
<value>${workingDir}/mainworkflows_common_export/working_dir</value>
|
514
|
</property>
|
515
|
<!-- input ports -->
|
516
|
<property>
|
517
|
<name>input_document_to_project</name>
|
518
|
<value>${workingDir}/referenceextraction_project/document_projects</value>
|
519
|
</property>
|
520
|
<property>
|
521
|
<name>input_document_to_project_concepts</name>
|
522
|
<value>${workingDir}/transformers_concept_to_researchinitiatives/out</value>
|
523
|
</property>
|
524
|
<property>
|
525
|
<name>input_document_to_dataset</name>
|
526
|
<value>${workingDir}/referenceextraction_dataset/document_datasets</value>
|
527
|
</property>
|
528
|
<property>
|
529
|
<name>input_document_to_mdstore</name>
|
530
|
<value>${workingDir}/mainworkflows_preprocessing_import/dataset_to_mdstore</value>
|
531
|
</property>
|
532
|
<!-- entities exporting modes -->
|
533
|
<property>
|
534
|
<name>active_export_referenceddataset_datasets</name>
|
535
|
<value>true</value>
|
536
|
</property>
|
537
|
<property>
|
538
|
<name>active_export_referencedproject_entities</name>
|
539
|
<value>true</value>
|
540
|
</property>
|
541
|
<property>
|
542
|
<name>mdstore_service_location</name>
|
543
|
<value>${import_mdstore_service_location}</value>
|
544
|
</property>
|
545
|
<property>
|
546
|
<name>wos_mdstore_id</name>
|
547
|
<value>${import_wos_mdstore_id}</value>
|
548
|
</property>
|
549
|
<!-- export related -->
|
550
|
<property>
|
551
|
<name>action_hbase_table_name</name>
|
552
|
<value>${export_action_hbase_table_name}</value>
|
553
|
</property>
|
554
|
<property>
|
555
|
<name>action_hbase_table_initialize</name>
|
556
|
<value>${export_action_hbase_table_initialize}</value>
|
557
|
</property>
|
558
|
<!-- action set id properties -->
|
559
|
<property>
|
560
|
<name>action_set_id</name>
|
561
|
<value>${export_action_set_id}</value>
|
562
|
</property>
|
563
|
<property>
|
564
|
<name>action_set_id_document_referencedProjects</name>
|
565
|
<value>${export_action_set_id_document_referencedProjects}</value>
|
566
|
</property>
|
567
|
<property>
|
568
|
<name>action_set_id_document_referencedDatasets</name>
|
569
|
<value>${export_action_set_id_document_referencedDatasets}</value>
|
570
|
</property>
|
571
|
<property>
|
572
|
<name>action_set_id_entity_wos</name>
|
573
|
<value>${export_action_set_id_entity_wos}</value>
|
574
|
</property>
|
575
|
<property>
|
576
|
<name>action_set_id_entity_dataset</name>
|
577
|
<value>${export_action_set_id_entity_dataset}</value>
|
578
|
</property>
|
579
|
<property>
|
580
|
<name>trust_level_threshold</name>
|
581
|
<value>${export_trust_level_threshold}</value>
|
582
|
</property>
|
583
|
<property>
|
584
|
<name>trust_level_threshold_document_referencedProjects</name>
|
585
|
<value>${export_trust_level_threshold_document_referencedProjects}</value>
|
586
|
</property>
|
587
|
<property>
|
588
|
<name>trust_level_threshold_document_referencedDatasets</name>
|
589
|
<value>${export_trust_level_threshold_document_referencedDatasets}</value>
|
590
|
</property>
|
591
|
<property>
|
592
|
<name>action_hbase_remote_zookeeper_quorum</name>
|
593
|
<value>${export_action_hbase_remote_zookeeper_quorum}</value>
|
594
|
</property>
|
595
|
<property>
|
596
|
<name>action_hbase_remote_zookeeper_clientport</name>
|
597
|
<value>${export_action_hbase_remote_zookeeper_clientport}</value>
|
598
|
</property>
|
599
|
</configuration>
|
600
|
</sub-workflow>
|
601
|
<ok to="decision-export-to-json"/>
|
602
|
<error to="fail" />
|
603
|
</action>
|
604
|
|
605
|
<decision name="decision-export-to-json">
|
606
|
<switch>
|
607
|
<case to="export-to-json">${active_export_to_json eq "true"}</case>
|
608
|
<default to="end"/>
|
609
|
</switch>
|
610
|
</decision>
|
611
|
|
612
|
<action name="export-to-json">
|
613
|
<sub-workflow>
|
614
|
<app-path>${wf:appPath()}/mainworkflows_common_export_to_json</app-path>
|
615
|
<propagate-configuration/>
|
616
|
<configuration>
|
617
|
<property>
|
618
|
<name>input_document_to_project</name>
|
619
|
<value>${workingDir}/referenceextraction_project/document_projects</value>
|
620
|
</property>
|
621
|
<property>
|
622
|
<name>input_document_to_project_concepts</name>
|
623
|
<value>${workingDir}/transformers_concept_to_researchinitiatives/out</value>
|
624
|
</property>
|
625
|
<property>
|
626
|
<name>input_document_to_dataset</name>
|
627
|
<value>${workingDir}/referenceextraction_dataset/document_datasets</value>
|
628
|
</property>
|
629
|
<property>
|
630
|
<name>output_root</name>
|
631
|
<value>${workingDir}/exported_as_json</value>
|
632
|
</property>
|
633
|
</configuration>
|
634
|
</sub-workflow>
|
635
|
<ok to="end"/>
|
636
|
<error to="fail" />
|
637
|
</action>
|
638
|
|
639
|
<kill name="fail">
|
640
|
<message>Unfortunately, the process failed -- error message:
|
641
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
642
|
</kill>
|
643
|
<end name="end" />
|
644
|
</workflow-app>
|