Project

General

Profile

1
package eu.dnetlib.data.transform;
2

    
3
import static org.junit.Assert.assertFalse;
4
import static org.junit.Assert.assertNotNull;
5
import static org.junit.Assert.assertTrue;
6

    
7
import java.io.BufferedReader;
8
import java.io.IOException;
9
import java.io.InputStream;
10
import java.io.InputStreamReader;
11
import java.io.StringReader;
12
import java.io.UnsupportedEncodingException;
13
import java.util.ArrayList;
14
import java.util.HashMap;
15
import java.util.Iterator;
16
import java.util.List;
17
import java.util.Map;
18
import java.util.Map.Entry;
19
import java.util.Set;
20
import java.util.zip.GZIPInputStream;
21

    
22
import javax.xml.transform.TransformerConfigurationException;
23
import javax.xml.transform.TransformerFactoryConfigurationError;
24

    
25
import org.apache.commons.io.IOUtils;
26
import org.apache.commons.lang.StringUtils;
27
import org.apache.commons.logging.Log;
28
import org.apache.commons.logging.LogFactory;
29
import org.dom4j.Document;
30
import org.dom4j.DocumentException;
31
import org.dom4j.io.SAXReader;
32
import org.json.JSONObject;
33
import org.junit.Before;
34
import org.junit.Ignore;
35
import org.junit.Test;
36
import org.springframework.core.io.ByteArrayResource;
37
import org.springframework.core.io.Resource;
38

    
39
import com.google.common.base.Function;
40
import com.google.common.collect.Iterables;
41
import com.google.common.collect.Lists;
42
import com.google.common.collect.Maps;
43
import com.google.common.collect.Sets;
44
import com.google.protobuf.InvalidProtocolBufferException;
45
import com.googlecode.protobuf.format.JsonFormat;
46
import com.googlecode.protobuf.format.JsonFormat.ParseException;
47

    
48
import eu.dnetlib.actionmanager.actions.ActionFactory;
49
import eu.dnetlib.actionmanager.actions.XsltInfoPackageAction;
50
import eu.dnetlib.actionmanager.common.Agent;
51
import eu.dnetlib.actionmanager.common.Operation;
52
import eu.dnetlib.actionmanager.common.Provenance;
53
import eu.dnetlib.data.mapreduce.hbase.index.config.Context;
54
import eu.dnetlib.data.mapreduce.hbase.index.config.ContextMapper;
55
import eu.dnetlib.data.mapreduce.hbase.index.config.EntityConfigTable;
56
import eu.dnetlib.data.mapreduce.hbase.index.config.IndexConfig;
57
import eu.dnetlib.data.mapreduce.hbase.index.config.IndexConfigTest;
58
import eu.dnetlib.data.mapreduce.hbase.index.config.LinkDescriptor;
59
import eu.dnetlib.data.mapreduce.hbase.index.config.RelClasses;
60
import eu.dnetlib.data.mapreduce.hbase.index.config.RelClassesTest;
61
import eu.dnetlib.data.mapreduce.util.OafDecoder;
62
import eu.dnetlib.data.mapreduce.util.OafRowKeyDecoder;
63
import eu.dnetlib.data.mapreduce.util.RelDescriptor;
64
import eu.dnetlib.data.mapreduce.util.UpdateMerger;
65
import eu.dnetlib.data.mapreduce.util.XmlRecordFactory;
66
import eu.dnetlib.data.mapreduce.util.XmlRecordFactoryTest;
67
import eu.dnetlib.data.proto.KindProtos.Kind;
68
import eu.dnetlib.data.proto.OafProtos.Oaf;
69
import eu.dnetlib.data.proto.TypeProtos.Type;
70
import eu.dnetlib.miscutils.functional.xml.IndentXmlString;
71

    
72
public class XsltRowTransformerFactoryTest {
73

    
74
	private static final Log log = LogFactory.getLog(XsltRowTransformerFactoryTest.class);
75
	private static String basePathProfiles = "/eu/dnetlib/test/profiles/TransformationRuleDSResources/TransformationRuleDSResourceType/2hbase/";
76
	private XsltRowTransformerFactory factory;
77
	private EntityConfigTable entityConfigTable;
78

    
79
	@Before
80
	public void setUp() throws Exception {
81
		factory = new XsltRowTransformerFactory();
82
		entityConfigTable = IndexConfig.load(IndexConfigTest.config).getConfigMap();
83
	}
84

    
85
	@Test
86
	@Ignore // need to reimplement because claimUpdates_2_hbase.xsl was removed
87
	public void testParseOafClaimUpdate() throws Exception {
88
		doTest(loadFromTransformationProfile("claimUpdates_2_hbase.xsl"), load("recordClaimUpdate.xml"));
89
	}
90

    
91
	@Test
92
	@Ignore // need to reimplement because claimUpdates_2_hbase.xsl was removed
93
	public void testParseClaimUpdate() throws Exception {
94

    
95
		final List<Row> rows = Lists.newArrayList();
96
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordClaim.xml")));
97
		rows.addAll(asRows(loadFromTransformationProfile("claimUpdates_2_hbase.xsl"), load("recordClaimUpdate.xml")));
98

    
99
		printAll(mapAll(buildTable(rows)));
100
	}
101

    
102
	@Test
103
	public void testParseClaimRel() throws Exception {
104

    
105
		doTest(loadFromTransformationProfile("claimRels_2_hbase.xml"), load("recordClaimRel.xml"));
106
	}
107

    
108
	@Test
109
	public void testParseFp7IctPUB() throws Exception {
110

    
111
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("ec_fp7_ict.xml"));
112
	}
113

    
114
	@Test
115
	public void testParseRecordCrossref() throws Exception {
116

    
117
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("recordCrossref.xml"));
118
	}
119

    
120
	@Test
121
	public void testParseDatasetPUB() throws Exception {
122

    
123
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordDatasetPUB.xml"));
124
	}
125

    
126
	@Test
127
	public void testParseSoftwareEgiApp() throws Exception {
128

    
129
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("softwareEgiApp.xml"));
130
	}
131

    
132
	@Test
133
	public void testParseSoftwareEgiApp2() throws Exception {
134

    
135
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("softwareEgiApp2.xml"));
136
	}
137

    
138
	@Test
139
	public void testParseOrpEgiApp() throws Exception {
140

    
141
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("orpEgiApp.xml"));
142
	}
143

    
144
	@Test
145
	public void testParseSoftwareDOECODE() throws Exception {
146

    
147
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("doecode.xml"));
148
	}
149

    
150
	@Test
151
	public void testParseDatasetLindat() throws Exception {
152

    
153
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("datasetLindat.xml"));
154
	}
155

    
156
	@Test
157
	public void testParseDatasetNeuroVault() throws Exception {
158

    
159
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordNeuroVault.xml"));
160
	}
161

    
162
	@Test
163
	public void testParseDatasetNeuroVault2() throws Exception {
164

    
165
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordNeuroVault2.xml"));
166
	}
167

    
168
	@Test
169
	public void testParseClaim() throws Exception {
170

    
171
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("recordClaim.xml"));
172
	}
173

    
174
	@Test
175
	public void testParseClaimDedup() throws Exception {
176

    
177
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("recordClaimedDedup.xml"));
178
	}
179

    
180
	@Test
181
	public void testParseClaimDataset() throws Exception {
182

    
183
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordClaimDataset.xml"));
184
	}
185

    
186
	@Test
187
	public void testParseACM() throws Exception {
188

    
189
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("recordACM.xml"));
190
	}
191

    
192
	@Test
193
	public void testParseASB() throws Exception {
194

    
195
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("recordASB.xml"));
196
	}
197

    
198
	@Test
199
	public void testParseProjectCorda() throws Exception {
200

    
201
		doTest(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml"));
202
	}
203

    
204
	@Test
205
	public void testParseProjectFCT() throws Exception {
206

    
207
		doTest(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordFCT.xml"));
208
	}
209

    
210
	@Test
211
	public void testParseProjectWithSummaryAndAmount() throws Exception {
212

    
213
		doTest(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectWithSummaryAndAmount.xml"));
214
	}
215

    
216
	@Test
217
	public void testLinkOrganizationWithBudget() throws Exception {
218

    
219
		final List<Row> rows = Lists.newArrayList();
220
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organization.xml")));
221
		rows.addAll(asRows(loadFromTransformationProfile("projectorganization_2_hbase.xsl"), load("project_org_budget.xml")));
222
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectWithSummaryAndAmount.xml")));
223

    
224
		printAll(mapAll(buildTable(rows)));
225
	}
226

    
227

    
228
	@Test
229
	public void testParseOaf() throws Exception {
230

    
231
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("record.xml"));
232
	}
233

    
234
	@Test
235
	public void testParseOafPublication() throws Exception {
236

    
237
		doTest(loadFromTransformationProfile("oaf_entity2hbase.xml"), load("record.xml"));
238
	}
239

    
240
	@Test
241
	public void testParseLindat() throws Exception {
242

    
243
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordLindat.xml"));
244
	}
245

    
246
	@Test
247
	public void testParseDatacite() throws Exception {
248

    
249
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordDatacite.xml"));
250
	}
251

    
252
	@Test
253
	public void testParseDatacite2() throws Exception {
254

    
255
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("recordDatacite2.xml"));
256
	}
257

    
258
	@Test
259
	public void testParseDataciteNewES() throws Exception {
260

    
261
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("dataciteNew.xml"));
262
	}
263

    
264
	@Test
265
	public void testParseDatacite2hostedby() throws Exception {
266

    
267
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("datacite2hostedby.xml"));
268
	}
269

    
270
	@Test
271
	public void testParseOpenTrials() throws Exception {
272

    
273
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("opentrials_datacite1.xml"));
274
	}
275

    
276
	@Test
277
	public void testLinkPangaea() throws Exception {
278

    
279
		final List<Row> rows = Lists.newArrayList();
280
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("pangaeODF1.xml")));
281
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("pangaeODF2.xml")));
282
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("pangaeOAF.xml")));
283
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCordaPangaea.xml")));
284

    
285
		printAll(mapAll(buildTable(rows)));
286
	}
287

    
288
	@Test
289
	public void testODF() throws Exception {
290

    
291
		final List<Row> rows = Lists.newArrayList();
292
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("odf.xml")));
293
		// printAll(mapAll(buildTable(rows)));
294
	}
295

    
296
	@Test
297
	public void testPangaea() throws Exception {
298

    
299
		final List<Row> rows = Lists.newArrayList();
300
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("pangaeOAF2.xml")));
301
		printAll(mapAll(buildTable(rows)));
302
	}
303

    
304
	@Test
305
	public void testZenodo() throws Exception {
306

    
307
		final List<Row> rows = Lists.newArrayList();
308
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("zenodoData.xml")));
309
		printAll(mapAll(buildTable(rows)));
310
	}
311

    
312
	@Test
313
	public void testZenodoSoftware() throws Exception {
314

    
315
		final List<Row> rows = Lists.newArrayList();
316
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("softwareZenodo_odf.xml")));
317
		printAll(mapAll(buildTable(rows)));
318
	}
319

    
320
	@Test
321
	public void testLinkCorda() throws Exception {
322

    
323
		final List<Row> rows = Lists.newArrayList();
324
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
325
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordCorda.xml")));
326

    
327
		printAll(mapAll(buildTable(rows)));
328
	}
329

    
330
	@Test
331
	public void testLinkFCT() throws Exception {
332

    
333
		final List<Row> rows = Lists.newArrayList();
334
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordFCT.xml")));
335
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordFCT.xml")));
336

    
337
		printAll(mapAll(buildTable(rows)));
338
	}
339

    
340
	@Test
341
	public void testLinkARC() throws Exception {
342

    
343
		final List<Row> rows = Lists.newArrayList();
344
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordARC.xml")));
345
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordARC.xml")));
346

    
347
		printAll(mapAll(buildTable(rows)));
348
	}
349

    
350
	@Test
351
	public void testLinkWT() throws Exception {
352

    
353
		final List<Row> rows = Lists.newArrayList();
354
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordWT.xml")));
355
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordWT.xml")));
356

    
357
		printAll(mapAll(buildTable(rows)));
358
	}
359

    
360
	@Test
361
	public void testLinkOrganization() throws Exception {
362

    
363
		final List<Row> rows = Lists.newArrayList();
364
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organization.xml")));
365
		rows.addAll(asRows(loadFromTransformationProfile("projectorganization_2_hbase.xsl"), load("project_organization.xml")));
366
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
367

    
368
		printAll(mapAll(buildTable(rows)));
369
	}
370

    
371
	@Test
372
	public void testLinkOrganizationAffiliation() throws Exception {
373

    
374
		final List<Row> rows = Lists.newArrayList();
375
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organization.xml")));
376
		rows.addAll(asRows(loadFromTransformationProfile("resultorganization_2_hbase.xsl"), load("result_organization.xml")));
377
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("record.xml")));
378

    
379
		printAll(mapAll(buildTable(rows)));
380
	}
381

    
382
	@Test
383
	public void testDuplicates() throws Exception {
384
		final String mergeId = "50|dedup_wf_001::08ed625d07e5738b794ff14d6773fd9f";
385
		final List<Row> rows = Lists.newArrayList();
386

    
387
		final Function<Row, Row> f = rowIn -> {
388

    
389
			final List<Column<String, byte[]>> cols = Lists.newArrayList();
390
			for (final Column<String, byte[]> col : rowIn.getColumns()) {
391
				if (col.getName().equals("body")) {
392
					cols.add(new Column(col.getName(), col.getValue()));
393

    
394
				}
395
			}
396
			return new Row("result", rowIn.getKey(), cols);
397
		};
398

    
399
		final List<Row> puma1 = asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordPuma1.xml"), f);
400
		puma1.add(new Row("resultResult_dedup_isMergedIn", mergeId));
401

    
402
		final List<Row> puma2 = asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordPuma2.xml"), f);
403
		puma2.add(new Row("resultResult_dedup_isMergedIn", mergeId));
404

    
405
		rows.addAll(puma1);
406
		rows.addAll(puma2);
407

    
408
		final List<Oaf> duplicates = Lists.newArrayList();
409
		duplicates.add(getOafBody(puma1));
410
		duplicates.add(getOafBody(puma2));
411
		final Oaf.Builder oafMerge = OafEntityMerger.merge(mergeId, duplicates);
412

    
413
		final Row mergeRow = new Row("result", mergeId, Lists.newArrayList(new Column("body", oafMerge.build().toByteArray())));
414

    
415
		rows.add(mergeRow);
416

    
417
		printAll(mapAll(buildTable(rows)));
418
	}
419

    
420
	private Oaf getOafBody(final List<Row> rows) throws InvalidProtocolBufferException {
421
		for (final Row row : rows) {
422
			if (StringUtils.startsWith(row.getKey(), "50")) { return Oaf.parseFrom(row.getColumn("body").getValue());
423

    
424
			}
425
		}
426
		return null;
427
	}
428

    
429
	@Test
430
	public void testParseDoajOAF() throws Exception {
431

    
432
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("doajUniversityRecord.xml"));
433
	}
434

    
435
	@Test
436
	public void testParseDatasource() throws Exception {
437

    
438
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("datasourceNative.xml"));
439
	}
440

    
441
	@Test
442
	public void testParseDatasourcePiwik() throws Exception {
443

    
444
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("datasourcePiwik.xml"));
445
	}
446

    
447
	@Test
448
	public void testParseDataDatasource() throws Exception {
449

    
450
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("dataDatasource.xml"));
451
	}
452

    
453
	@Test
454
	public void testFromMongodbCompressedDump() throws Exception {
455
		doTestJsonGz(loadFromTransformationProfile("oaf2hbase.xml"), load("mdstore_cleaned.json.gz"));
456
	}
457

    
458
	@Test
459
	public void testLoadFromTransformationProfile() throws IOException {
460
		final InputStream in = loadFromTransformationProfile("oaf2hbase.xml");
461
		log.info(IOUtils.toString(in));
462
	}
463

    
464
	@Test
465
	public void test_template() throws Exception {
466
		final String xslt = IOUtils.toString(loadFromTransformationProfile("oaf2hbase.xml"));
467
		final XsltRowTransformer transformer = factory.getTransformer(xslt);
468
		assertNotNull(transformer);
469

    
470
		final String record = IOUtils.toString(load("record.xml"));
471
		final List<Row> rows = transformer.apply(record);
472

    
473
		System.out.println(rows);
474
	}
475

    
476
	@Test
477
	public void testWrongCharsOrganization() throws Exception {
478
		final List<Row> rows = Lists.newArrayList();
479
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organizationWrongChars.xml")));
480
		printAll(mapAll(buildTable(rows)));
481
	}
482

    
483
	@Test
484
	public void testParseProjectWithFunderOriginalName() throws Exception {
485

    
486
		doTest(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectWithFunderOriginalName.xml"));
487
	}
488

    
489
	@Test
490
	public void testLinkFunderOriginalName() throws Exception {
491

    
492
		final List<Row> rows = Lists.newArrayList();
493
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectWithFunderOriginalName.xml")));
494
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordFunderOriginalName.xml")));
495

    
496
		printAll(mapAll(buildTable(rows)));
497
	}
498

    
499
	@Test
500
	public void testProjectExtraInfo() throws Exception {
501
		final List<Row> rows = Lists.newArrayList();
502
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordExtraInfo.xml")));
503
		printAll(mapAll(buildTable(rows)));
504
	}
505

    
506
	@Test
507
	public void testParseSoftwareFromODF() throws Exception {
508
		final List<Row> rows = Lists.newArrayList();
509
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("softwareODF.xml")));
510
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
511
		printAll(mapAll(buildTable(rows)));
512
	}
513

    
514
	@Test
515
	public void testParseSoftwareFromOAF() throws Exception {
516
		final List<Row> rows = Lists.newArrayList();
517
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordOAFsoftwareCLOSED.xml")));
518
		printAll(mapAll(buildTable(rows)));
519
	}
520

    
521
	@Test
522
	public void testParsePubFromODF() throws Exception {
523
		final List<Row> rows = Lists.newArrayList();
524
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("recordODFPub.xml")));
525
		printAll(mapAll(buildTable(rows)));
526
	}
527

    
528
	@Test
529
	public void testParseSoftwareFromOAFOpen() throws Exception {
530
		final List<Row> rows = Lists.newArrayList();
531
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordOAFsoftwareOPEN.xml")));
532
		printAll(mapAll(buildTable(rows)));
533
	}
534

    
535
	@Test
536
	public void testParseSoftwareBiotool() throws Exception {
537
		final List<Row> rows = Lists.newArrayList();
538
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("biotoolSw.xml")));
539
		printAll(mapAll(buildTable(rows)));
540
	}
541

    
542
	@Test
543
	public void testParseOafWithExternalRef() throws Exception {
544
		final List<Row> rows = Lists.newArrayList();
545
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("oafWithExternalReference.xml")));
546
		printAll(mapAll(buildTable(rows)));
547
	}
548

    
549
	@Test
550
	public void testParseOafWithCommunity() throws Exception {
551
		final List<Row> rows = Lists.newArrayList();
552
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("oafWithCommunity.xml")));
553
		printAll(mapAll(buildTable(rows)));
554
	}
555

    
556
	@Test
557
	public void testParseOafWithUpdates() throws Exception {
558
		final List<Row> rows = Lists.newArrayList();
559
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("oafWithCommunity.xml")));
560

    
561
		final ActionFactory actionFactory = new ActionFactory();
562

    
563
		final Map<String, Resource> xslts = Maps.newHashMap();
564

    
565
		xslts.put("oaf2hbase", new ByteArrayResource(IOUtils.toString(loadFromTransformationProfile("oaf2hbase.xml")).getBytes()));
566
		actionFactory.setXslts(xslts);
567

    
568
		final XsltInfoPackageAction pa = actionFactory.generateInfoPackageAction(
569
				"oaf2hbase",
570
				"rawset-id",
571
				new Agent("agent-id", "agent-name", Agent.AGENT_TYPE.algo),
572
				Operation.UPDATE,
573
				IOUtils.toString(load("oafUpdateWithCommunity.xml")),
574
				Provenance.sysimport_mining_aggregator,
575
				"0.9");
576

    
577
		final String qualifier = "update_" + System.nanoTime();
578

    
579
		IOUtils.readLines(load("country_updates.json")).forEach(line -> {
580

    
581
			final Oaf.Builder oaf = Oaf.newBuilder();
582

    
583
			try {
584
				JsonFormat.merge(line, oaf);
585
			} catch (final JsonFormat.ParseException e) {
586
				throw new IllegalArgumentException(e);
587
			}
588

    
589
			final Column<String, byte[]> col = new Column<>("update_" + System.nanoTime(), oaf.build().toByteArray());
590
			rows.add(new Row("result", oaf.getEntity().getId(), Lists.newArrayList(col)));
591
		});
592

    
593
		pa.asAtomicActions().forEach(a -> {
594
			final Column<String, byte[]> col = new Column<>("update_" + System.nanoTime(), a.getTargetValue());
595
			rows.add(new Row(a.getTargetColumnFamily(), a.getTargetRowKey(), Lists.newArrayList(col)));
596
		});
597

    
598
		/*
599
		 * rows.forEach(r -> { log.info(r); });
600
		 */
601

    
602
		mapAll(buildTable(rows)).entrySet().forEach(b -> {
603
			log.info(b.getKey());
604
			log.info(b.getValue());
605
		});
606
	}
607

    
608
	@Test
609
	public void testParseCrisPub() throws Exception {
610
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("cris_pub1.xml"));
611
	}
612

    
613
	@Test
614
	public void testBioToolSwRowJson() throws Exception {
615
		doTestJsonRow(IOUtils.toString(load("biotoolSwRow.json")));
616
	}
617

    
618
	@Test
619
	public void testParseVirta() throws Exception {
620
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("virta.xml"));
621
	}
622

    
623
	@Test
624
	public void testParseJournal() throws Exception {
625
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("datasourceWithISSN.xml"));
626
	}
627

    
628
	@Test
629
	public void testGuidelines4Qeios() throws Exception {
630

    
631
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("guidelines4_qeios1.xml"));
632
	}
633

    
634
	@Test
635
	public void testGuidelines4Aria() throws Exception {
636

    
637
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("guidelines4_aria.xml"));
638
	}
639

    
640
	@Test
641
	public void testJournalRecord() throws Exception {
642
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("journalRecord.xml"));
643
	}
644

    
645
	@Test
646
	public void testSygmaDatacite() throws Exception {
647

    
648
		final List<Row> rows = Lists.newArrayList();
649
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("sygma_datacite.xml")));
650
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
651

    
652
		printAll(mapAll(buildTable(rows)));
653
	}
654

    
655
	@Test
656
	public void testOpenOrganizations() throws Exception {
657

    
658
		final List<Row> rows = Lists.newArrayList();
659
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("openorgs_sample.xml")));
660

    
661
		printAll(mapAll(buildTable(rows)));
662
	}
663

    
664
	private void doTestJsonRow(final String json) throws Exception {
665
		final Row row = asRowFromJson(json);
666
		log.info(row);
667
		final List<Row> rows = new ArrayList<>();
668
		rows.add(row);
669
		final Map<String, Map<String, Map<String, byte[]>>> table = buildTable(rows);
670
		final Map<String, XmlRecordFactory> builders = mapAll(table);
671
		printAll(builders);
672

    
673
	}
674

    
675
	private void doTest(final InputStream xsltStream, final InputStream recordStream) throws Exception {
676
		try {
677
			final List<Row> rows = asRows(xsltStream, recordStream);
678

    
679
			log.info(rows);
680

    
681
			final Map<String, Map<String, Map<String, byte[]>>> table = buildTable(rows);
682

    
683
			// System.out.println("\n" + table.toString());
684

    
685
			final Map<String, XmlRecordFactory> builders = mapAll(table);
686

    
687
			printAll(builders);
688
		} catch (final InvalidProtocolBufferException e) {
689
			throw new Exception(e);
690
		} catch (final TransformerConfigurationException e) {
691
			throw new Exception(e);
692
		} catch (final TransformerFactoryConfigurationError e) {
693
			throw new Exception(e);
694
		} catch (final DocumentException e) {
695
			throw new Exception(e);
696
		}
697
	}
698

    
699
	private void doTestJsonGz(final InputStream xsltStream, final InputStream recordStream) throws Exception {
700

    
701
		final Iterator<List<Row>> rowsIterator = asRowsJsonGzip(xsltStream, recordStream);
702

    
703
		int i = 0;
704
		while (rowsIterator.hasNext()) {
705
			final List<Row> rows = rowsIterator.next();
706
			i++;
707

    
708
			if (i % 10000 == 0) {
709
				System.out.println(i);
710
			}
711

    
712
			final Map<String, Map<String, Map<String, byte[]>>> table = buildTableDoaj(rows);
713

    
714
			for (final Map<String, Map<String, byte[]>> m : table.values()) {
715
				for (final Map<String, byte[]> mv : m.values()) {
716
					for (final byte[] v : mv.values()) {
717
						final OafDecoder d = OafDecoder.decode(v);
718
						assertNotNull(d);
719
						assertNotNull(d.getOaf());
720

    
721
						switch (d.getKind()) {
722
						case entity:
723
							assertNotNull(d.getMetadata());
724
							if (d.getOaf().getEntity().getType().equals(Type.result)) {
725
								System.out.println(d.getOaf());
726
							}
727
							break;
728
						case relation:
729
							assertNotNull(d.getRel());
730
							break;
731
						default:
732
							break;
733
						}
734
					}
735
				}
736
			}
737
		}
738
	}
739

    
740
	private List<Row> asRows(final InputStream xsltStream, final InputStream recordStream, final Function<Row, Row> p) throws Exception {
741
		return asRows(xsltStream, new HashMap<>(), recordStream, p);
742
	}
743

    
744
	private List<Row> asRows(final InputStream xsltStream, final InputStream recordStream) throws Exception {
745
		return asRows(xsltStream, new HashMap<>(), recordStream);
746
	}
747

    
748
	private List<Row> asRows(final InputStream xsltStream, final Map<String, Object> params, final InputStream recordStream) throws Exception {
749
		return asRows(xsltStream, params, recordStream, null);
750
	}
751

    
752
	private List<Row> asRows(final InputStream xsltStream, final Map<String, Object> params, final InputStream recordStream, final Function<Row, Row> p)
753
			throws Exception {
754
		final String xslt = IOUtils.toString(xsltStream);
755
		final XsltRowTransformer transformer = factory.getTransformer(xslt, params);
756
		assertNotNull(transformer);
757

    
758
		final String record = IOUtils.toString(recordStream);
759
		final List<Row> rows = transformer.apply(record);
760

    
761
		assertNotNull(rows);
762
		assertFalse(rows.isEmpty());
763
		return p == null ? rows : Lists.newArrayList(Iterables.transform(rows, p));
764
	}
765

    
766
	private Iterator<List<Row>> asRowsJsonGzip(final InputStream xsltStream, final InputStream recordStreamJsonGzip) throws Exception {
767
		final String xslt = IOUtils.toString(xsltStream);
768
		final XsltRowTransformer transformer = factory.getTransformer(xslt);
769
		assertNotNull(transformer);
770
		assertNotNull(recordStreamJsonGzip);
771

    
772
		final GZIPInputStream stream = new GZIPInputStream(recordStreamJsonGzip);
773
		assertNotNull(stream);
774
		final BufferedReader inStream = new BufferedReader(new InputStreamReader(stream));
775
		assertNotNull(inStream);
776
		return new Iterator<List<Row>>() {
777

    
778
			String jsonRecord = null;
779

    
780
			@Override
781
			public boolean hasNext() {
782
				try {
783
					return (jsonRecord = inStream.readLine()) != null;
784
				} catch (final IOException e) {
785
					throw new RuntimeException(e);
786
				}
787
			}
788

    
789
			@Override
790
			public List<Row> next() {
791

    
792
				final JSONObject jsonObj = new JSONObject(jsonRecord);
793
				final String body = jsonObj.getString("body");
794
				try {
795
					assertTrue(StringUtils.isNotBlank(body));
796
					// System.out.println(body);
797
					final List<Row> rows = transformer.apply(body);
798
					assertNotNull(rows);
799
					assertFalse(rows.isEmpty());
800
					return rows;
801
				} catch (final Throwable e) {
802
					System.err.println("error transforming document: " + body);
803
					throw new RuntimeException(e);
804
				}
805
			}
806

    
807
			@Override
808
			public void remove() {
809
				throw new UnsupportedOperationException();
810
			}
811

    
812
		};
813

    
814
	}
815

    
816
	private Map<String, Map<String, Map<String, byte[]>>> buildTableDoaj(final List<Row> rows) throws UnsupportedEncodingException {
817
		final Map<String, Map<String, Map<String, byte[]>>> table = Maps.newHashMap();
818

    
819
		for (final Row row : rows) {
820
			final String rowKey = row.getKey();
821
			final String cf = row.getColumnFamily();
822
			if (!table.containsKey(rowKey)) {
823
				table.put(rowKey, new HashMap<>());
824
			}
825
			if (!table.get(rowKey).containsKey(cf)) {
826
				table.get(rowKey).put(row.getColumnFamily(), new HashMap<>());
827
			}
828
			for (final Column<String, byte[]> c : row.getColumns()) {
829
				// System.out.println(String.format("ADDING K:%s CF:%s Q:%s", rowKey, cf, c.getName()));
830
				table.get(rowKey).get(cf).put(c.getName(), c.getValue());
831
				if (cf.equals("result") && c.getName().equals("body")) {
832
					// System.out.println(String.format("ADDING K:%s CF:%s Q:%s", rowKey, cf, c.getName()));
833
					assertTrue(StringUtils.isNotBlank(new String(c.getValue(), "UTF-8")));
834
				}
835
			}
836
		}
837
		return table;
838

    
839
	}
840

    
841
	protected Map<String, Map<String, Map<String, byte[]>>> buildTable(final List<Row> rows) throws UnsupportedEncodingException {
842
		final Map<String, Map<String, Map<String, byte[]>>> table = Maps.newHashMap();
843

    
844
		for (final Row row : rows) {
845
			final String rowKey = row.getKey();
846
			final String cf = row.getColumnFamily();
847
			if (!table.containsKey(rowKey)) {
848
				table.put(rowKey, new HashMap<>());
849
			}
850
			if (!table.get(rowKey).containsKey(cf)) {
851
				table.get(rowKey).put(row.getColumnFamily(), new HashMap<>());
852
			}
853
			for (final Column<String, byte[]> c : row.getColumns()) {
854
				System.out.println(String.format("ADDING K:%s CF:%s Q:%s", rowKey, cf, c.getName()));
855
				table.get(rowKey).get(cf).put(c.getName(), c.getValue());
856
				if (c.getName().equals("body")) {
857
					final String theBody = new String(c.getValue(), "UTF-8");
858
					assertTrue(StringUtils.isNotBlank(theBody));
859
					// System.out.println(theBody);
860
				}
861
			}
862
		}
863
		return table;
864

    
865
	}
866

    
867
	protected Map<String, XmlRecordFactory> mapAll(final Map<String, Map<String, Map<String, byte[]>>> table) throws Exception {
868

    
869
		final Map<String, XmlRecordFactory> builders = Maps.newHashMap();
870
		for (final Entry<String, Map<String, Map<String, byte[]>>> e : table.entrySet()) {
871
			map(builders, e.getKey(), e.getValue());
872
		}
873
		return builders;
874
	}
875

    
876
	// private Map<String, XmlRecordFactory> mapResultsOnly(final Map<String, Map<String, Map<String, byte[]>>> table) throws Exception {
877
	//
878
	// final Map<String, XmlRecordFactory> builders = Maps.newHashMap();
879
	// for (final Entry<String, Map<String, Map<String, byte[]>>> e : table.entrySet()) {
880
	// final Type type = OafRowKeyDecoder.decode(e.getKey()).getType();
881
	// if (type == Type.result) {
882
	// map(builders, e.getKey(), e.getValue());
883
	// }
884
	// }
885
	// return builders;
886
	// }
887

    
888
	private void map(final Map<String, XmlRecordFactory> builders, final String rowKey, final Map<String, Map<String, byte[]>> row) throws Exception {
889

    
890
		final Type type = OafRowKeyDecoder.decode(rowKey).getType();
891

    
892
		final Map<String, byte[]> familyMap = row.get(type.toString());
893

    
894
		if (familyMap == null) { return; }
895

    
896
		final byte[] bodyB = familyMap.get("body");
897

    
898
		if (bodyB != null) {
899
			ensureBuilder(builders, rowKey);
900

    
901
			final Oaf oaf = UpdateMerger.mergeBodyUpdates(familyMap);
902

    
903
			final OafDecoder mainEntity = OafDecoder.decode(oaf);
904

    
905
			builders.get(rowKey).setMainEntity(mainEntity);
906

    
907
			for (final LinkDescriptor ld : entityConfigTable.getDescriptors(type)) {
908

    
909
				final String it = ld.getRelDescriptor().getIt();
910
				final Map<String, byte[]> cols = row.get(it);
911

    
912
				if (cols != null && !cols.isEmpty()) {
913

    
914
					for (final byte[] oafB : cols.values()) {
915

    
916
						final Oaf.Builder relBuilder = Oaf.newBuilder(Oaf.parseFrom(oafB));
917

    
918
						if (ld.isSymmetric()) {
919
							final RelDescriptor rd = ld.getRelDescriptor();
920

    
921
							relBuilder.getRelBuilder().setCachedTarget(mainEntity.getEntity()).setRelType(rd.getRelType()).setSubRelType(rd.getSubRelType());
922
						}
923

    
924
						relBuilder.getRelBuilder().setChild(ld.isChild());
925

    
926
						final Oaf.Builder oafBuilder = Oaf.newBuilder().setKind(Kind.relation).setLastupdatetimestamp(System.currentTimeMillis());
927
						oafBuilder.mergeFrom(relBuilder.build());
928

    
929
						final String targetId = ld.isSymmetric() ? oafBuilder.getRel().getTarget() : oafBuilder.getRel().getSource();
930
						ensureBuilder(builders, targetId);
931
						final OafDecoder decoder = OafDecoder.decode(oafBuilder.build());
932

    
933
						if (ld.isChild()) {
934
							builders.get(targetId).addChild(type, decoder);
935
						} else {
936
							builders.get(targetId).addRelation(type, decoder);
937
						}
938
					}
939

    
940
				}
941
			}
942
		}
943

    
944
	}
945

    
946
	private void printAll(final Map<String, XmlRecordFactory> builders) throws DocumentException {
947
		print(Sets.newHashSet(Type.values()), builders, null);
948
	}
949

    
950
	private void print(final Set<Type> types, final Map<String, XmlRecordFactory> builders, final Map<Type, Set<String>> xpaths) throws DocumentException {
951
		final SAXReader r = new SAXReader();
952

    
953
		for (final Entry<String, XmlRecordFactory> e : builders.entrySet()) {
954
			final OafRowKeyDecoder kd = OafRowKeyDecoder.decode(e.getKey());
955

    
956
			if (!e.getValue().isValid()) { throw new IllegalArgumentException("invalid builder: " + e.getKey()); }
957
			if (types.contains(kd.getType())) {
958
				final String val = IndentXmlString.apply(e.getValue().build());
959

    
960
				if (xpaths != null && !xpaths.isEmpty() && xpaths.get(kd.getType()) != null) {
961
					final Document doc = r.read(new StringReader(val));
962

    
963
					log.debug("\n" + e.getKey());
964
					for (final String xpath : xpaths.get(kd.getType())) {
965
						log.debug(doc.valueOf(xpath));
966
					}
967
				} else {
968
					log.info(val);
969
				}
970
			}
971
		}
972
	}
973

    
974
	private void printNoIndent(final Map<String, XmlRecordFactory> builders) {
975
		for (final Entry<String, XmlRecordFactory> e : builders.entrySet()) {
976
			if (e.getValue().isValid()) {
977
				log.debug(e.getValue().build());
978
			} else {
979
				log.debug("invalid builder: " + e.getKey());
980
			}
981
		}
982
	}
983

    
984
	private void ensureBuilder(final Map<String, XmlRecordFactory> builders, final String rowKey) throws Exception {
985
		if (!builders.containsKey(rowKey)) {
986
			builders.put(rowKey, newBuilder());
987
		}
988
	}
989

    
990
	private XmlRecordFactory newBuilder() throws TransformerConfigurationException, TransformerFactoryConfigurationError, DocumentException {
991
		return new XmlRecordFactory(entityConfigTable, ContextMapper.fromXml(Context.xml),
992
				RelClasses.fromJSon(RelClassesTest.relClassesJson), XmlRecordFactoryTest.SCHEMA_LOCATION, true, false, false,
993
				XmlRecordFactoryTest.specialDatasourceTypes);
994
	}
995

    
996
	private InputStream load(final String fileName) {
997
		return getClass().getResourceAsStream(fileName);
998
	}
999

    
1000
	private InputStream loadFromTransformationProfile(final String profilePath) {
1001
		log.info("Loading xslt from: " + basePathProfiles + profilePath);
1002
		final InputStream profile = getClass().getResourceAsStream(basePathProfiles + profilePath);
1003
		final SAXReader saxReader = new SAXReader();
1004
		Document doc = null;
1005
		try {
1006
			doc = saxReader.read(profile);
1007
		} catch (final DocumentException e) {
1008
			e.printStackTrace();
1009
			throw new RuntimeException(e);
1010
		}
1011
		final String xslt = doc.selectSingleNode("//SCRIPT/CODE/*[local-name()='stylesheet']").asXML();
1012
		// log.info(xslt);
1013
		return IOUtils.toInputStream(xslt);
1014
	}
1015

    
1016
	private Row asRowFromJson(final String json) throws ParseException {
1017
		final Oaf.Builder oafBuilder = Oaf.newBuilder();
1018
		JsonFormat.merge(json, oafBuilder);
1019
		final Oaf oaf = oafBuilder.build();
1020
		final Row row = new Row("result", oaf.getEntity().getId());
1021
		final Column<String, byte[]> c = new Column<>("body", oaf.toByteArray());
1022
		row.setColumn("body", c);
1023
		return row;
1024

    
1025
	}
1026

    
1027
}
    (1-1/1)