1
|
<workflow-app xmlns="uri:oozie:workflow:0.4" name="mainworkflows_primary_main">
|
2
|
|
3
|
<parameters>
|
4
|
<property>
|
5
|
<name>remove_sideproducts</name>
|
6
|
<value>true</value>
|
7
|
<description>flag indicating inference side products will be erased</description>
|
8
|
</property>
|
9
|
<property>
|
10
|
<name>active_existence_filter</name>
|
11
|
<value>true</value>
|
12
|
<description>flag indicating contents should be filtered against metadata entries retrieved from InformationSpace.
|
13
|
This way only contents having metadata representation will be processed.
|
14
|
To be disabled when processing new contents which metadata is not available in hbase.</description>
|
15
|
</property>
|
16
|
<!-- processing modes -->
|
17
|
<property>
|
18
|
<name>active_metadataextraction_export</name>
|
19
|
<value>false</value>
|
20
|
<description>flag indicating metadata extraction export should be enabled</description>
|
21
|
</property>
|
22
|
<property>
|
23
|
<name>active_referenceextraction_project</name>
|
24
|
<value>false</value>
|
25
|
<description>flag indicating project reference extraction should be enabled</description>
|
26
|
</property>
|
27
|
<property>
|
28
|
<name>active_referenceextraction_dataset</name>
|
29
|
<value>false</value>
|
30
|
<description>flag indicating dataset reference extraction should be enabled</description>
|
31
|
</property>
|
32
|
<property>
|
33
|
<!-- currently disabled by default -->
|
34
|
<name>active_referenceextraction_researchinitiative</name>
|
35
|
<value>false</value>
|
36
|
<description>flag indicating researchinitiative reference extraction should be enabled</description>
|
37
|
</property>
|
38
|
<property>
|
39
|
<name>active_documentsclassification</name>
|
40
|
<value>false</value>
|
41
|
<description>flag indicating documents classification should be enabled</description>
|
42
|
</property>
|
43
|
<property>
|
44
|
<name>active_documentssimilarity</name>
|
45
|
<value>false</value>
|
46
|
<description>flag indicating documents similarity should be enabled</description>
|
47
|
</property>
|
48
|
<property>
|
49
|
<name>active_citationmatching</name>
|
50
|
<value>false</value>
|
51
|
<description>flag indicating citation matching should be enabled</description>
|
52
|
</property>
|
53
|
<property>
|
54
|
<name>active_statistics</name>
|
55
|
<value>false</value>
|
56
|
<description>flag indicating statistics generation should be enabled</description>
|
57
|
</property>
|
58
|
<property>
|
59
|
<name>active_websiteusage_analysis</name>
|
60
|
<value>false</value>
|
61
|
<description>flag indicating logs should be imported from HDFS log file into avro datastore
|
62
|
and website usage analysis should be performed</description>
|
63
|
</property>
|
64
|
<property>
|
65
|
<name>active_export</name>
|
66
|
<value>true</value>
|
67
|
<description>flag indicating export should be performed</description>
|
68
|
</property>
|
69
|
<!-- import concepts related -->
|
70
|
<property>
|
71
|
<name>import_islookup_service_location</name>
|
72
|
<description>IS Lookup service location</description>
|
73
|
</property>
|
74
|
<property>
|
75
|
<name>import_project_concepts_context_ids_csv</name>
|
76
|
<value>fet-fp7,fet-h2020</value>
|
77
|
<description>comma separated list of concepts context identifiers to be picked by ISLookup</description>
|
78
|
</property>
|
79
|
<!-- import metadata related -->
|
80
|
<property>
|
81
|
<name>import_hbase_input_table</name>
|
82
|
<description>HBase input table holding InformationSpace, available on local cluster</description>
|
83
|
</property>
|
84
|
<property>
|
85
|
<name>import_hbase_approved_datasources_csv</name>
|
86
|
<value>$UNDEFINED$</value>
|
87
|
<description>CSV list of datasource ids to be approved during import. Applied on result and person entities.</description>
|
88
|
</property>
|
89
|
<!-- import datacite related -->
|
90
|
<property>
|
91
|
<name>import_mdstore_service_location</name>
|
92
|
<value>$UNDEFINED$</value>
|
93
|
<description>MDStore service (not WSDL) location URL</description>
|
94
|
</property>
|
95
|
<property>
|
96
|
<name>import_dataset_mdstore_ids_csv</name>
|
97
|
<value>$UNDEFINED$</value>
|
98
|
<description>MDStore identifier</description>
|
99
|
</property>
|
100
|
<!-- import content related -->
|
101
|
<property>
|
102
|
<name>import_content_object_store_location</name>
|
103
|
<value>$UNDEFINED$</value>
|
104
|
<description>object store service location required for content retrieval</description>
|
105
|
</property>
|
106
|
<property>
|
107
|
<name>import_content_objectstores_csv</name>
|
108
|
<value>$UNDEFINED$</value>
|
109
|
<description>CSV list of object stores identifiers to be processed</description>
|
110
|
</property>
|
111
|
<property>
|
112
|
<name>import_content_mimetypes_pdf</name>
|
113
|
<value>pdf,application/pdf</value>
|
114
|
<description>pdf mime types</description>
|
115
|
</property>
|
116
|
<property>
|
117
|
<name>import_content_mimetypes_text</name>
|
118
|
<value>text,text/plain</value>
|
119
|
<description>text mime types</description>
|
120
|
</property>
|
121
|
<property>
|
122
|
<name>import_content_mimetypes_html</name>
|
123
|
<value>text/html</value>
|
124
|
<description>html mime types</description>
|
125
|
</property>
|
126
|
<property>
|
127
|
<name>import_content_mimetypes_xml_pmc</name>
|
128
|
<value>xml</value>
|
129
|
<description>xml pmc types</description>
|
130
|
</property>
|
131
|
<property>
|
132
|
<name>import_content_mimetypes_wos</name>
|
133
|
<value>file::WoS</value>
|
134
|
<description>WoS types</description>
|
135
|
</property>
|
136
|
<!-- import timeouts related -->
|
137
|
<property>
|
138
|
<name>import_resultset_client_read_timeout</name>
|
139
|
<value>60000</value>
|
140
|
<description>resultset client read timeout</description>
|
141
|
</property>
|
142
|
<property>
|
143
|
<name>import_content_connection_timeout</name>
|
144
|
<value>60000</value>
|
145
|
<description>import content connection timeout</description>
|
146
|
</property>
|
147
|
<property>
|
148
|
<name>import_content_read_timeout</name>
|
149
|
<value>60000</value>
|
150
|
<description>import content read timeout</description>
|
151
|
</property>
|
152
|
<!-- import logs related -->
|
153
|
<property>
|
154
|
<name>portal_logs_location</name>
|
155
|
<value>/cache/portal-piwik-logs</value>
|
156
|
<description>portal log files HDFS location</description>
|
157
|
</property>
|
158
|
<!-- metadata extraction related -->
|
159
|
<property>
|
160
|
<name>metadataextraction_excluded_checksums</name>
|
161
|
<value>$UNDEFINED$</value>
|
162
|
<description>list of content checksums excluded from metadataextraction processing</description>
|
163
|
</property>
|
164
|
<property>
|
165
|
<name>metadataextraction_max_file_size_mb</name>
|
166
|
<value>500</value>
|
167
|
<description>maximum allowed file size in Megabytes</description>
|
168
|
</property>
|
169
|
<property>
|
170
|
<name>metadataextraction_default_cache_location</name>
|
171
|
<value>/cache/metadataextraction</value>
|
172
|
<description>metadata extraction HDFS cache location</description>
|
173
|
</property>
|
174
|
<property>
|
175
|
<name>metadataextraction_processing_mode</name>
|
176
|
<value>StreamingMetadataExtractorMapper</value>
|
177
|
<description>metadata extraction processing mode</description>
|
178
|
</property>
|
179
|
<property>
|
180
|
<name>metadataextraction_input_classname</name>
|
181
|
<value>eu.dnetlib.iis.importer.auxiliary.schemas.DocumentContentUrl</value>
|
182
|
<description>metadata extraction input classname</description>
|
183
|
</property>
|
184
|
<!-- export related -->
|
185
|
<property>
|
186
|
<name>export_action_hbase_table_name</name>
|
187
|
<description>action manager hbase table name</description>
|
188
|
</property>
|
189
|
<property>
|
190
|
<name>export_action_hbase_table_initialize</name>
|
191
|
<description>flag indicating input table should be initialized</description>
|
192
|
</property>
|
193
|
<!-- action set id properties -->
|
194
|
<property>
|
195
|
<name>export_action_set_id</name>
|
196
|
<value>$UNDEFINED$</value>
|
197
|
<description>action-set identifier of exported data</description>
|
198
|
</property>
|
199
|
<property>
|
200
|
<name>export_action_set_id_dataset_similarities_websiteusage</name>
|
201
|
<value>$UNDEFINED$</value>
|
202
|
<description>dataset_similarities_websiteusage action-set identifier of exported data</description>
|
203
|
</property>
|
204
|
<property>
|
205
|
<name>export_action_set_id_person_similarities_websiteusage</name>
|
206
|
<value>$UNDEFINED$</value>
|
207
|
<description>person_similarities_websiteusage action-set identifier of exported data</description>
|
208
|
</property>
|
209
|
<property>
|
210
|
<name>export_action_set_id_person_statistics</name>
|
211
|
<value>$UNDEFINED$</value>
|
212
|
<description>person_statistics action-set identifier of exported data</description>
|
213
|
</property>
|
214
|
<property>
|
215
|
<name>export_action_set_id_project_statistics</name>
|
216
|
<value>$UNDEFINED$</value>
|
217
|
<description>project_statistics action-set identifier of exported data</description>
|
218
|
</property>
|
219
|
<property>
|
220
|
<name>export_action_set_id_document_similarities_standard</name>
|
221
|
<value>$UNDEFINED$</value>
|
222
|
<description>document_similarities_standard action-set identifier of exported data</description>
|
223
|
</property>
|
224
|
<property>
|
225
|
<name>export_action_set_id_document_similarities_websiteusage</name>
|
226
|
<value>$UNDEFINED$</value>
|
227
|
<description>document_similarities_websiteusage action-set identifier of exported data</description>
|
228
|
</property>
|
229
|
<property>
|
230
|
<name>export_action_set_id_document_statistics</name>
|
231
|
<value>$UNDEFINED$</value>
|
232
|
<description>document_statistics action-set identifier of exported data</description>
|
233
|
</property>
|
234
|
<property>
|
235
|
<name>export_action_set_id_document_classes</name>
|
236
|
<value>$UNDEFINED$</value>
|
237
|
<description>document_classes action-set identifier of exported data</description>
|
238
|
</property>
|
239
|
<property>
|
240
|
<name>export_action_set_id_document_referencedProjects</name>
|
241
|
<value>$UNDEFINED$</value>
|
242
|
<description>document_referencedProjects action-set identifier of exported data</description>
|
243
|
</property>
|
244
|
<property>
|
245
|
<name>export_action_set_id_document_referencedDatasets</name>
|
246
|
<value>$UNDEFINED$</value>
|
247
|
<description>document_referencedDatasets action-set identifier of exported data</description>
|
248
|
</property>
|
249
|
<property>
|
250
|
<name>export_action_set_id_document_referencedDocuments</name>
|
251
|
<value>$UNDEFINED$</value>
|
252
|
<description>document_referencedDocuments action-set identifier of exported data</description>
|
253
|
</property>
|
254
|
<property>
|
255
|
<name>export_action_set_id_document_research_initiative</name>
|
256
|
<value>$UNDEFINED$</value>
|
257
|
<description>document research initiative action-set identifier of exported data</description>
|
258
|
</property>
|
259
|
<!-- -->
|
260
|
<property>
|
261
|
<name>export_action_set_id_entity_dataset</name>
|
262
|
<value>$UNDEFINED$</value>
|
263
|
<description>action-set identifier of exported data containing dataset entities</description>
|
264
|
</property>
|
265
|
<property>
|
266
|
<name>export_action_hbase_remote_zookeeper_quorum</name>
|
267
|
<value>$UNDEFINED$</value>
|
268
|
<description>external hbase zookeeper quorum, set to empty value by default which means data will be exported to local hbase instance</description>
|
269
|
</property>
|
270
|
<property>
|
271
|
<name>export_action_hbase_remote_zookeeper_clientport</name>
|
272
|
<value>$UNDEFINED$</value>
|
273
|
<description>external hbase zookeeper client port, required only whe zookeeper quorum property is set</description>
|
274
|
</property>
|
275
|
<property>
|
276
|
<name>export_documentssimilarity_threshold</name>
|
277
|
<value>$UNDEFINED$</value>
|
278
|
<description>documents similarity threshold value below which similarity export is omitted</description>
|
279
|
</property>
|
280
|
<!-- working directory related -->
|
281
|
<property>
|
282
|
<name>execution_environment</name>
|
283
|
<value>primary</value>
|
284
|
<description>execution environment used for workingDir creation</description>
|
285
|
</property>
|
286
|
<property>
|
287
|
<name>workingDir</name>
|
288
|
<value>/user/${user.name}/iis/working_dirs/${execution_environment}</value>
|
289
|
<description>working directory</description>
|
290
|
</property>
|
291
|
</parameters>
|
292
|
|
293
|
<global>
|
294
|
<job-tracker>${jobTracker}</job-tracker>
|
295
|
<name-node>${nameNode}</name-node>
|
296
|
<configuration>
|
297
|
<property>
|
298
|
<name>mapred.job.queue.name</name>
|
299
|
<value>${queueName}</value>
|
300
|
</property>
|
301
|
</configuration>
|
302
|
</global>
|
303
|
|
304
|
<start to="init-workingDir"/>
|
305
|
|
306
|
<action name="init-workingDir">
|
307
|
<fs>
|
308
|
<delete path="${nameNode}${workingDir}" />
|
309
|
<mkdir path="${nameNode}${workingDir}" />
|
310
|
</fs>
|
311
|
<ok to="copy-version"/>
|
312
|
<error to="fail"/>
|
313
|
</action>
|
314
|
|
315
|
<action name="copy-version">
|
316
|
<distcp xmlns="uri:oozie:distcp-action:0.1">
|
317
|
<!-- distcp doesn't support global parameters so we need to provide them explicitly -->
|
318
|
<job-tracker>${jobTracker}</job-tracker>
|
319
|
<name-node>${nameNode}</name-node>
|
320
|
<configuration>
|
321
|
<property>
|
322
|
<name>mapred.job.queue.name</name>
|
323
|
<value>${queueName}</value>
|
324
|
</property>
|
325
|
</configuration>
|
326
|
<arg>${wf:appPath()}/version.properties</arg>
|
327
|
<arg>${nameNode}${workingDir}</arg>
|
328
|
</distcp>
|
329
|
<ok to="import"/>
|
330
|
<error to="fail"/>
|
331
|
</action>
|
332
|
|
333
|
<action name="import">
|
334
|
<sub-workflow>
|
335
|
<app-path>${wf:appPath()}/mainworkflows_common_import</app-path>
|
336
|
<propagate-configuration/>
|
337
|
<configuration>
|
338
|
<property>
|
339
|
<name>workingDir</name>
|
340
|
<value>${workingDir}/mainworkflows_common_import/working_dir</value>
|
341
|
</property>
|
342
|
<!-- importing modes -->
|
343
|
<property>
|
344
|
<name>active_existence_filter</name>
|
345
|
<value>${active_existence_filter}</value>
|
346
|
</property>
|
347
|
<property>
|
348
|
<name>active_import_metadata</name>
|
349
|
<value>true</value>
|
350
|
</property>
|
351
|
<property>
|
352
|
<name>active_import_dataset</name>
|
353
|
<value>${active_referenceextraction_dataset}</value>
|
354
|
</property>
|
355
|
<property>
|
356
|
<name>active_import_concept</name>
|
357
|
<value>${active_referenceextraction_project}</value>
|
358
|
</property>
|
359
|
<property>
|
360
|
<name>active_ingest_pmc_citations</name>
|
361
|
<value>${active_citationmatching}</value>
|
362
|
</property>
|
363
|
<!-- import metadata related -->
|
364
|
<property>
|
365
|
<name>hbase_input_table</name>
|
366
|
<value>${import_hbase_input_table}</value>
|
367
|
</property>
|
368
|
<property>
|
369
|
<name>hbase_approved_datasources_csv</name>
|
370
|
<value>${import_hbase_approved_datasources_csv}</value>
|
371
|
</property>
|
372
|
<!-- import datacite related -->
|
373
|
<property>
|
374
|
<name>mdstore_service_location</name>
|
375
|
<value>${import_mdstore_service_location}</value>
|
376
|
</property>
|
377
|
<property>
|
378
|
<name>dataset_mdstore_ids_csv</name>
|
379
|
<value>${import_dataset_mdstore_ids_csv}</value>
|
380
|
</property>
|
381
|
<!-- project concept related -->
|
382
|
<property>
|
383
|
<name>islookup_service_location</name>
|
384
|
<value>${import_islookup_service_location}</value>
|
385
|
</property>
|
386
|
<property>
|
387
|
<name>project_concepts_context_ids_csv</name>
|
388
|
<value>${import_project_concepts_context_ids_csv}</value>
|
389
|
</property>
|
390
|
<!-- import content related -->
|
391
|
<property>
|
392
|
<name>objectstore_service_location</name>
|
393
|
<value>${import_content_object_store_location}</value>
|
394
|
</property>
|
395
|
<property>
|
396
|
<name>approved_objectstores_csv</name>
|
397
|
<value>${import_content_objectstores_csv}</value>
|
398
|
</property>
|
399
|
<property>
|
400
|
<name>mimetypes_pdf</name>
|
401
|
<value>${import_content_mimetypes_pdf}</value>
|
402
|
</property>
|
403
|
<property>
|
404
|
<name>mimetypes_text</name>
|
405
|
<value>${import_content_mimetypes_text}</value>
|
406
|
</property>
|
407
|
<property>
|
408
|
<name>mimetypes_html</name>
|
409
|
<value>${import_content_mimetypes_html}</value>
|
410
|
</property>
|
411
|
<property>
|
412
|
<name>mimetypes_xml_pmc</name>
|
413
|
<value>${import_content_mimetypes_xml_pmc}</value>
|
414
|
</property>
|
415
|
<property>
|
416
|
<name>mimetypes_wos</name>
|
417
|
<value>${import_content_mimetypes_wos}</value>
|
418
|
</property>
|
419
|
<!-- import timeouts related -->
|
420
|
<property>
|
421
|
<name>resultset_client_read_timeout</name>
|
422
|
<value>${import_resultset_client_read_timeout}</value>
|
423
|
</property>
|
424
|
<property>
|
425
|
<name>content_connection_timeout</name>
|
426
|
<value>${import_content_connection_timeout}</value>
|
427
|
</property>
|
428
|
<property>
|
429
|
<name>content_read_timeout</name>
|
430
|
<value>${import_content_read_timeout}</value>
|
431
|
</property>
|
432
|
<!-- metadata extraction related -->
|
433
|
<property>
|
434
|
<name>metadataextraction_excluded_checksums</name>
|
435
|
<value>${metadataextraction_excluded_checksums}</value>
|
436
|
</property>
|
437
|
<property>
|
438
|
<name>metadataextraction_max_file_size_mb</name>
|
439
|
<value>${metadataextraction_max_file_size_mb}</value>
|
440
|
</property>
|
441
|
<property>
|
442
|
<name>metadataextraction_default_cache_location</name>
|
443
|
<value>${metadataextraction_default_cache_location}</value>
|
444
|
</property>
|
445
|
<!-- metadatainput and metadataextraction output subdirectory names -->
|
446
|
<property>
|
447
|
<name>metadataimport_output_name_document_meta</name>
|
448
|
<value>docmeta</value>
|
449
|
</property>
|
450
|
<property>
|
451
|
<name>metadataimport_output_name_document_project</name>
|
452
|
<value>docproject</value>
|
453
|
</property>
|
454
|
<property>
|
455
|
<name>metadataimport_output_name_project</name>
|
456
|
<value>project</value>
|
457
|
</property>
|
458
|
<property>
|
459
|
<name>metadataimport_output_name_person</name>
|
460
|
<value>person</value>
|
461
|
</property>
|
462
|
<property>
|
463
|
<name>metadataimport_output_name_dedup_mapping</name>
|
464
|
<value>dedupmapping</value>
|
465
|
</property>
|
466
|
<!-- output parameters -->
|
467
|
<property>
|
468
|
<name>output_extracted_document_metadata</name>
|
469
|
<value>${workingDir}/mainworkflows_common_import/extracted_document_metadata</value>
|
470
|
</property>
|
471
|
<property>
|
472
|
<name>output_metadataimport_root</name>
|
473
|
<value>${workingDir}/mainworkflows_common_import/metadataimport</value>
|
474
|
</property>
|
475
|
<property>
|
476
|
<name>output_dataset</name>
|
477
|
<value>${workingDir}/mainworkflows_common_import/dataset</value>
|
478
|
</property>
|
479
|
<property>
|
480
|
<name>output_dataset_to_mdstore</name>
|
481
|
<value>${workingDir}/mainworkflows_common_import/dataset_to_mdstore</value>
|
482
|
</property>
|
483
|
<property>
|
484
|
<name>output_citation_pmc</name>
|
485
|
<value>${workingDir}/mainworkflows_common_import/citation_pmc</value>
|
486
|
</property>
|
487
|
<property>
|
488
|
<name>output_document_text</name>
|
489
|
<value>${workingDir}/mainworkflows_common_import/document-text</value>
|
490
|
</property>
|
491
|
<property>
|
492
|
<name>output_project_concept</name>
|
493
|
<value>${workingDir}/mainworkflows_common_import/project-concept</value>
|
494
|
</property>
|
495
|
<property>
|
496
|
<name>output_wos</name>
|
497
|
<value>${workingDir}/mainworkflows_common_import/wos</value>
|
498
|
</property>
|
499
|
</configuration>
|
500
|
</sub-workflow>
|
501
|
<ok to="decision-import_logs"/>
|
502
|
<error to="fail" />
|
503
|
</action>
|
504
|
|
505
|
<decision name="decision-import_logs">
|
506
|
<switch>
|
507
|
<case to="import_logs">${active_websiteusage_analysis eq "true"}</case>
|
508
|
<default to="mainworkflows_primary_processing"/>
|
509
|
</switch>
|
510
|
</decision>
|
511
|
|
512
|
<action name="import_logs">
|
513
|
<sub-workflow>
|
514
|
<app-path>${wf:appPath()}/import_logs</app-path>
|
515
|
<propagate-configuration/>
|
516
|
<configuration>
|
517
|
<property>
|
518
|
<name>workingDir</name>
|
519
|
<value>${workingDir}/websiteusage_analysis/import_logs/working_dir</value>
|
520
|
</property>
|
521
|
<property>
|
522
|
<name>input</name>
|
523
|
<value>${portal_logs_location}</value>
|
524
|
</property>
|
525
|
<property>
|
526
|
<name>output</name>
|
527
|
<value>${workingDir}/websiteusage_analysis/import_logs/output</value>
|
528
|
</property>
|
529
|
</configuration>
|
530
|
</sub-workflow>
|
531
|
<ok to="mainworkflows_primary_processing" />
|
532
|
<error to="fail" />
|
533
|
</action>
|
534
|
|
535
|
<action name="mainworkflows_primary_processing">
|
536
|
<sub-workflow>
|
537
|
<app-path>${wf:appPath()}/mainworkflows_primary_processing</app-path>
|
538
|
<propagate-configuration/>
|
539
|
<configuration>
|
540
|
<property>
|
541
|
<name>input_document_metadata</name>
|
542
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docmeta</value>
|
543
|
</property>
|
544
|
<property>
|
545
|
<name>input_document_to_project</name>
|
546
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docproject</value>
|
547
|
</property>
|
548
|
<property>
|
549
|
<name>input_document_text</name>
|
550
|
<value>${workingDir}/mainworkflows_common_import/document-text</value>
|
551
|
</property>
|
552
|
<property>
|
553
|
<name>input_document_text_wos</name>
|
554
|
<value>${workingDir}/mainworkflows_common_import/wos</value>
|
555
|
</property>
|
556
|
<property>
|
557
|
<name>input_project</name>
|
558
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/project</value>
|
559
|
</property>
|
560
|
<property>
|
561
|
<name>input_person</name>
|
562
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/person</value>
|
563
|
</property>
|
564
|
<property>
|
565
|
<name>input_dataset</name>
|
566
|
<value>${workingDir}/mainworkflows_common_import/dataset</value>
|
567
|
</property>
|
568
|
<property>
|
569
|
<name>input_extracted_document_metadata</name>
|
570
|
<value>${workingDir}/mainworkflows_common_import/extracted_document_metadata</value>
|
571
|
</property>
|
572
|
<property>
|
573
|
<name>input_citation_pmc</name>
|
574
|
<value>${workingDir}/mainworkflows_common_import/citation_pmc</value>
|
575
|
</property>
|
576
|
<property>
|
577
|
<name>input_project_concept</name>
|
578
|
<value>${workingDir}/mainworkflows_common_import/project-concept</value>
|
579
|
</property>
|
580
|
<property>
|
581
|
<name>output_document_to_project</name>
|
582
|
<value>${workingDir}/exported/document_to_project</value>
|
583
|
</property>
|
584
|
<property>
|
585
|
<name>output_document_to_project_concepts</name>
|
586
|
<value>${workingDir}/exported/document_to_project_concepts</value>
|
587
|
</property>
|
588
|
<property>
|
589
|
<name>output_document_to_dataset</name>
|
590
|
<value>${workingDir}/exported/document_to_dataset</value>
|
591
|
</property>
|
592
|
<property>
|
593
|
<name>output_document_to_research_initiatives</name>
|
594
|
<value>${workingDir}/exported/document_to_research_initiatives</value>
|
595
|
</property>
|
596
|
<property>
|
597
|
<name>output_document_to_document_classes</name>
|
598
|
<value>${workingDir}/exported/document_to_document_classes</value>
|
599
|
</property>
|
600
|
<property>
|
601
|
<name>output_citation</name>
|
602
|
<value>${workingDir}/exported/citation</value>
|
603
|
</property>
|
604
|
<property>
|
605
|
<name>output_document_similarity</name>
|
606
|
<value>${workingDir}/exported/document_similarity</value>
|
607
|
</property>
|
608
|
<property>
|
609
|
<name>output_document_statistics</name>
|
610
|
<value>${workingDir}/exported/document_statistics</value>
|
611
|
</property>
|
612
|
<property>
|
613
|
<name>output_author_statistics</name>
|
614
|
<value>${workingDir}/exported/author_statistics</value>
|
615
|
</property>
|
616
|
<property>
|
617
|
<name>output_project_statistics</name>
|
618
|
<value>${workingDir}/exported/project_statistics</value>
|
619
|
</property>
|
620
|
</configuration>
|
621
|
</sub-workflow>
|
622
|
<ok to="decision-websiteusage_analysis"/>
|
623
|
<error to="fail" />
|
624
|
</action>
|
625
|
|
626
|
<!-- website usage analysis block -->
|
627
|
<decision name="decision-websiteusage_analysis">
|
628
|
<switch>
|
629
|
<case to="websiteusage-idextractor">${active_websiteusage_analysis eq "true"}</case>
|
630
|
<default to="skip-websiteusage"/>
|
631
|
</switch>
|
632
|
</decision>
|
633
|
|
634
|
<action name="websiteusage-idextractor">
|
635
|
<sub-workflow>
|
636
|
<app-path>${wf:appPath()}/transformers_idextractor</app-path>
|
637
|
<propagate-configuration/>
|
638
|
<configuration>
|
639
|
<property>
|
640
|
<name>workingDir</name>
|
641
|
<value>${workingDir}/websiteusage_analysis/transformers_idextractor/working_dir</value>
|
642
|
</property>
|
643
|
<property>
|
644
|
<name>input_document_metadata</name>
|
645
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/docmeta</value>
|
646
|
</property>
|
647
|
<property>
|
648
|
<name>output_identifier</name>
|
649
|
<value>${workingDir}/websiteusage_analysis/transformers_idextractor/output</value>
|
650
|
</property>
|
651
|
</configuration>
|
652
|
</sub-workflow>
|
653
|
<ok to="mainworkflows_websiteusage_document"/>
|
654
|
<error to="fail"/>
|
655
|
</action>
|
656
|
|
657
|
<action name="mainworkflows_websiteusage_document">
|
658
|
<sub-workflow>
|
659
|
<app-path>${wf:appPath()}/mainworkflows_websiteusage_document</app-path>
|
660
|
<propagate-configuration/>
|
661
|
<configuration>
|
662
|
<property>
|
663
|
<name>workingDir</name>
|
664
|
<value>${workingDir}/websiteusage_analysis/working_dir</value>
|
665
|
</property>
|
666
|
<property>
|
667
|
<name>input_logs</name>
|
668
|
<value>${workingDir}/websiteusage_analysis/import_logs/output</value>
|
669
|
</property>
|
670
|
<property>
|
671
|
<name>input_id_mapping</name>
|
672
|
<value>${workingDir}/mainworkflows_common_import/metadataimport/dedupmapping</value>
|
673
|
</property>
|
674
|
<property>
|
675
|
<name>input_document_id</name>
|
676
|
<value>${workingDir}/websiteusage_analysis/transformers_idextractor/output</value>
|
677
|
</property>
|
678
|
<property>
|
679
|
<name>output</name>
|
680
|
<value>${workingDir}/websiteusage_analysis/output</value>
|
681
|
</property>
|
682
|
</configuration>
|
683
|
</sub-workflow>
|
684
|
<ok to="decision-transform-metadataextraction-for-export"/>
|
685
|
<error to="fail"/>
|
686
|
</action>
|
687
|
|
688
|
<action name="skip-websiteusage">
|
689
|
<java>
|
690
|
<prepare>
|
691
|
<!-- notice: directory have to aligned with skipped action output -->
|
692
|
<delete path="${nameNode}${workingDir}/websiteusage_analysis" />
|
693
|
<mkdir path="${nameNode}${workingDir}/websiteusage_analysis" />
|
694
|
</prepare>
|
695
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
696
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
697
|
<arg>-C{websiteusage_analysis_document,
|
698
|
eu.dnetlib.iis.websiteusage.schemas.DocumentsWithWebsiteUsageSimilarities,
|
699
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
700
|
<arg>-SworkingDir=${workingDir}/websiteusage_analysis/working_dir</arg>
|
701
|
<arg>-Owebsiteusage_analysis_document=${workingDir}/websiteusage_analysis/output</arg>
|
702
|
</java>
|
703
|
<ok to="decision-transform-metadataextraction-for-export"/>
|
704
|
<error to="fail"/>
|
705
|
</action>
|
706
|
<!-- end of website usage analysis block -->
|
707
|
|
708
|
<decision name="decision-transform-metadataextraction-for-export">
|
709
|
<switch>
|
710
|
<case to="transformers_export_documentmetadata">${active_metadataextraction_export eq "true"}</case>
|
711
|
<default to="skip-transformers_export_documentmetadata"/>
|
712
|
</switch>
|
713
|
</decision>
|
714
|
|
715
|
<action name="transformers_export_documentmetadata">
|
716
|
<sub-workflow>
|
717
|
<app-path>${wf:appPath()}/transformers_export_documentmetadata</app-path>
|
718
|
<configuration>
|
719
|
<property>
|
720
|
<name>jobTracker</name>
|
721
|
<value>${jobTracker}</value>
|
722
|
</property>
|
723
|
<property>
|
724
|
<name>nameNode</name>
|
725
|
<value>${nameNode}</value>
|
726
|
</property>
|
727
|
<property>
|
728
|
<name>queueName</name>
|
729
|
<value>${queueName}</value>
|
730
|
</property>
|
731
|
<!-- Working directory of the subworkflow -->
|
732
|
<property>
|
733
|
<name>workingDir</name>
|
734
|
<value>${workingDir}/transformers_export_documentmetadata/working_dir</value>
|
735
|
</property>
|
736
|
<property>
|
737
|
<name>input_extracted_metadata</name>
|
738
|
<value>${workingDir}/mainworkflows_common_import/extracted_document_metadata</value>
|
739
|
</property>
|
740
|
<property>
|
741
|
<name>output_metadata</name>
|
742
|
<value>${workingDir}/transformers_export_documentmetadata/output_metadata</value>
|
743
|
</property>
|
744
|
</configuration>
|
745
|
</sub-workflow>
|
746
|
<ok to="decision-export"/>
|
747
|
<error to="fail"/>
|
748
|
</action>
|
749
|
|
750
|
<action name="skip-transformers_export_documentmetadata">
|
751
|
<java>
|
752
|
<prepare>
|
753
|
<!-- notice: directory have to aligned with skipped action output -->
|
754
|
<delete path="${nameNode}${workingDir}/transformers_export_documentmetadata" />
|
755
|
<mkdir path="${nameNode}${workingDir}/transformers_export_documentmetadata" />
|
756
|
</prepare>
|
757
|
<main-class>eu.dnetlib.iis.core.java.ProcessWrapper</main-class>
|
758
|
<arg>eu.dnetlib.iis.core.java.jsonworkflownodes.Producer</arg>
|
759
|
<arg>-C{document_metadata,
|
760
|
eu.dnetlib.iis.export.schemas.DocumentMetadata,
|
761
|
eu/dnetlib/iis/mainworkflows/data/empty.json}</arg>
|
762
|
<arg>-SworkingDir=${workingDir}/transformers_export_documentmetadata/working_dir</arg>
|
763
|
<!-- notice: directory have to aligned with skipped action output -->
|
764
|
<arg>-Odocument_metadata=${workingDir}/transformers_export_documentmetadata/output_metadata</arg>
|
765
|
</java>
|
766
|
<ok to="decision-export"/>
|
767
|
<error to="fail"/>
|
768
|
</action>
|
769
|
|
770
|
<decision name="decision-export">
|
771
|
<switch>
|
772
|
<case to="export">${active_export eq "true"}</case>
|
773
|
<default to="end"/>
|
774
|
</switch>
|
775
|
</decision>
|
776
|
|
777
|
<action name="export">
|
778
|
<sub-workflow>
|
779
|
<app-path>${wf:appPath()}/mainworkflows_common_export</app-path>
|
780
|
<propagate-configuration/>
|
781
|
<configuration>
|
782
|
<property>
|
783
|
<name>workingDir</name>
|
784
|
<value>${workingDir}/mainworkflows_common_export/working_dir</value>
|
785
|
</property>
|
786
|
<!-- input ports -->
|
787
|
<property>
|
788
|
<name>input_document_metadata</name>
|
789
|
<value>${workingDir}/transformers_export_documentmetadata/output_metadata</value>
|
790
|
</property>
|
791
|
<property>
|
792
|
<name>input_document_to_project</name>
|
793
|
<value>${workingDir}/exported/document_to_project</value>
|
794
|
</property>
|
795
|
<property>
|
796
|
<name>input_document_to_project_concepts</name>
|
797
|
<value>${workingDir}/exported/document_to_project_concepts</value>
|
798
|
</property>
|
799
|
<property>
|
800
|
<name>input_document_to_dataset</name>
|
801
|
<value>${workingDir}/exported/document_to_dataset</value>
|
802
|
</property>
|
803
|
<property>
|
804
|
<name>input_document_to_mdstore</name>
|
805
|
<value>${workingDir}/mainworkflows_common_import/dataset_to_mdstore</value>
|
806
|
</property>
|
807
|
<property>
|
808
|
<name>input_document_to_research_initiatives</name>
|
809
|
<value>${workingDir}/exported/document_to_research_initiatives</value>
|
810
|
</property>
|
811
|
<property>
|
812
|
<name>input_document_to_document_classes</name>
|
813
|
<value>${workingDir}/exported/document_to_document_classes</value>
|
814
|
</property>
|
815
|
<property>
|
816
|
<name>input_citations</name>
|
817
|
<value>${workingDir}/exported/citation</value>
|
818
|
</property>
|
819
|
<property>
|
820
|
<name>input_document_similarity</name>
|
821
|
<value>${workingDir}/exported/document_similarity</value>
|
822
|
</property>
|
823
|
<property>
|
824
|
<name>input_document_statistics</name>
|
825
|
<value>${workingDir}/exported/document_statistics</value>
|
826
|
</property>
|
827
|
<property>
|
828
|
<name>input_document_websiteusage_similarity</name>
|
829
|
<value>${workingDir}/websiteusage_analysis/output</value>
|
830
|
</property>
|
831
|
<property>
|
832
|
<name>input_author_statistics</name>
|
833
|
<value>${workingDir}/exported/author_statistics</value>
|
834
|
</property>
|
835
|
<property>
|
836
|
<name>input_project_statistics</name>
|
837
|
<value>${workingDir}/exported/project_statistics</value>
|
838
|
</property>
|
839
|
<!-- entities exporting modes -->
|
840
|
<property>
|
841
|
<name>active_export_referenceddataset_datasets</name>
|
842
|
<value>${active_referenceextraction_dataset}</value>
|
843
|
</property>
|
844
|
<property>
|
845
|
<name>active_export_referencedproject_entities</name>
|
846
|
<value>false</value>
|
847
|
</property>
|
848
|
<property>
|
849
|
<name>mdstore_service_location</name>
|
850
|
<value>${import_mdstore_service_location}</value>
|
851
|
</property>
|
852
|
<!-- export related -->
|
853
|
<property>
|
854
|
<name>action_hbase_table_name</name>
|
855
|
<value>${export_action_hbase_table_name}</value>
|
856
|
</property>
|
857
|
<property>
|
858
|
<name>action_hbase_table_initialize</name>
|
859
|
<value>${export_action_hbase_table_initialize}</value>
|
860
|
</property>
|
861
|
<!-- action set id properties -->
|
862
|
<property>
|
863
|
<name>action_set_id</name>
|
864
|
<value>${export_action_set_id}</value>
|
865
|
</property>
|
866
|
<property>
|
867
|
<name>action_set_id_dataset_similarities_websiteusage</name>
|
868
|
<value>${export_action_set_id_dataset_similarities_websiteusage}</value>
|
869
|
</property>
|
870
|
<property>
|
871
|
<name>action_set_id_person_similarities_websiteusage</name>
|
872
|
<value>${export_action_set_id_person_similarities_websiteusage}</value>
|
873
|
</property>
|
874
|
<property>
|
875
|
<name>action_set_id_person_statistics</name>
|
876
|
<value>${export_action_set_id_person_statistics}</value>
|
877
|
</property>
|
878
|
<property>
|
879
|
<name>action_set_id_project_statistics</name>
|
880
|
<value>${export_action_set_id_project_statistics}</value>
|
881
|
</property>
|
882
|
<property>
|
883
|
<name>action_set_id_document_similarities_standard</name>
|
884
|
<value>${export_action_set_id_document_similarities_standard}</value>
|
885
|
</property>
|
886
|
<property>
|
887
|
<name>action_set_id_document_similarities_websiteusage</name>
|
888
|
<value>${export_action_set_id_document_similarities_websiteusage}</value>
|
889
|
</property>
|
890
|
<property>
|
891
|
<name>action_set_id_document_statistics</name>
|
892
|
<value>${export_action_set_id_document_statistics}</value>
|
893
|
</property>
|
894
|
<property>
|
895
|
<name>action_set_id_document_classes</name>
|
896
|
<value>${export_action_set_id_document_classes}</value>
|
897
|
</property>
|
898
|
<property>
|
899
|
<name>action_set_id_document_referencedProjects</name>
|
900
|
<value>${export_action_set_id_document_referencedProjects}</value>
|
901
|
</property>
|
902
|
<property>
|
903
|
<name>action_set_id_document_referencedDatasets</name>
|
904
|
<value>${export_action_set_id_document_referencedDatasets}</value>
|
905
|
</property>
|
906
|
<property>
|
907
|
<name>action_set_id_document_referencedDocuments</name>
|
908
|
<value>${export_action_set_id_document_referencedDocuments}</value>
|
909
|
</property>
|
910
|
<property>
|
911
|
<name>action_set_id_document_research_initiative</name>
|
912
|
<value>${export_action_set_id_document_research_initiative}</value>
|
913
|
</property>
|
914
|
<property>
|
915
|
<name>action_set_id_entity_dataset</name>
|
916
|
<value>${export_action_set_id_entity_dataset}</value>
|
917
|
</property>
|
918
|
<property>
|
919
|
<name>action_hbase_remote_zookeeper_quorum</name>
|
920
|
<value>${export_action_hbase_remote_zookeeper_quorum}</value>
|
921
|
</property>
|
922
|
<property>
|
923
|
<name>action_hbase_remote_zookeeper_clientport</name>
|
924
|
<value>${export_action_hbase_remote_zookeeper_clientport}</value>
|
925
|
</property>
|
926
|
<property>
|
927
|
<name>documentssimilarity_threshold</name>
|
928
|
<value>${export_documentssimilarity_threshold}</value>
|
929
|
</property>
|
930
|
</configuration>
|
931
|
</sub-workflow>
|
932
|
<ok to="end"/>
|
933
|
<error to="fail" />
|
934
|
</action>
|
935
|
|
936
|
<kill name="fail">
|
937
|
<message>Unfortunately, the process failed -- error message:
|
938
|
[${wf:errorMessage(wf:lastErrorNode())}]</message>
|
939
|
</kill>
|
940
|
<end name="end" />
|
941
|
</workflow-app>
|