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 testParseOpenTrials() throws Exception {
266

    
267
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("opentrials_datacite1.xml"));
268
	}
269

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

    
273
		final List<Row> rows = Lists.newArrayList();
274
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("pangaeODF1.xml")));
275
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("pangaeODF2.xml")));
276
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("pangaeOAF.xml")));
277
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCordaPangaea.xml")));
278

    
279
		printAll(mapAll(buildTable(rows)));
280
	}
281

    
282
	@Test
283
	public void testODF() throws Exception {
284

    
285
		final List<Row> rows = Lists.newArrayList();
286
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("odf.xml")));
287
		// printAll(mapAll(buildTable(rows)));
288
	}
289

    
290
	@Test
291
	public void testPangaea() throws Exception {
292

    
293
		final List<Row> rows = Lists.newArrayList();
294
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("pangaeOAF2.xml")));
295
		printAll(mapAll(buildTable(rows)));
296
	}
297

    
298
	@Test
299
	public void testZenodo() throws Exception {
300

    
301
		final List<Row> rows = Lists.newArrayList();
302
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("zenodoData.xml")));
303
		printAll(mapAll(buildTable(rows)));
304
	}
305

    
306
	@Test
307
	public void testZenodoSoftware() throws Exception {
308

    
309
		final List<Row> rows = Lists.newArrayList();
310
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("softwareZenodo_odf.xml")));
311
		printAll(mapAll(buildTable(rows)));
312
	}
313

    
314
	@Test
315
	public void testLinkCorda() throws Exception {
316

    
317
		final List<Row> rows = Lists.newArrayList();
318
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
319
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordCorda.xml")));
320

    
321
		printAll(mapAll(buildTable(rows)));
322
	}
323

    
324
	@Test
325
	public void testLinkFCT() throws Exception {
326

    
327
		final List<Row> rows = Lists.newArrayList();
328
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordFCT.xml")));
329
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordFCT.xml")));
330

    
331
		printAll(mapAll(buildTable(rows)));
332
	}
333

    
334
	@Test
335
	public void testLinkARC() throws Exception {
336

    
337
		final List<Row> rows = Lists.newArrayList();
338
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordARC.xml")));
339
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordARC.xml")));
340

    
341
		printAll(mapAll(buildTable(rows)));
342
	}
343

    
344
	@Test
345
	public void testLinkWT() throws Exception {
346

    
347
		final List<Row> rows = Lists.newArrayList();
348
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordWT.xml")));
349
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordWT.xml")));
350

    
351
		printAll(mapAll(buildTable(rows)));
352
	}
353

    
354
	@Test
355
	public void testLinkOrganization() throws Exception {
356

    
357
		final List<Row> rows = Lists.newArrayList();
358
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organization.xml")));
359
		rows.addAll(asRows(loadFromTransformationProfile("projectorganization_2_hbase.xsl"), load("project_organization.xml")));
360
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
361

    
362
		printAll(mapAll(buildTable(rows)));
363
	}
364

    
365
	@Test
366
	public void testLinkOrganizationAffiliation() throws Exception {
367

    
368
		final List<Row> rows = Lists.newArrayList();
369
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organization.xml")));
370
		rows.addAll(asRows(loadFromTransformationProfile("resultorganization_2_hbase.xsl"), load("result_organization.xml")));
371
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("record.xml")));
372

    
373
		printAll(mapAll(buildTable(rows)));
374
	}
375

    
376
	@Test
377
	public void testDuplicates() throws Exception {
378
		final String mergeId = "50|dedup_wf_001::08ed625d07e5738b794ff14d6773fd9f";
379
		final List<Row> rows = Lists.newArrayList();
380

    
381
		final Function<Row, Row> f = rowIn -> {
382

    
383
			final List<Column<String, byte[]>> cols = Lists.newArrayList();
384
			for (final Column<String, byte[]> col : rowIn.getColumns()) {
385
				if (col.getName().equals("body")) {
386
					cols.add(new Column(col.getName(), col.getValue()));
387

    
388
				}
389
			}
390
			return new Row("result", rowIn.getKey(), cols);
391
		};
392

    
393
		final List<Row> puma1 = asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordPuma1.xml"), f);
394
		puma1.add(new Row("resultResult_dedup_isMergedIn", mergeId));
395

    
396
		final List<Row> puma2 = asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordPuma2.xml"), f);
397
		puma2.add(new Row("resultResult_dedup_isMergedIn", mergeId));
398

    
399
		rows.addAll(puma1);
400
		rows.addAll(puma2);
401

    
402
		final List<Oaf> duplicates = Lists.newArrayList();
403
		duplicates.add(getOafBody(puma1));
404
		duplicates.add(getOafBody(puma2));
405
		final Oaf.Builder oafMerge = OafEntityMerger.merge(mergeId, duplicates);
406

    
407
		final Row mergeRow = new Row("result", mergeId, Lists.newArrayList(new Column("body", oafMerge.build().toByteArray())));
408

    
409
		rows.add(mergeRow);
410

    
411
		printAll(mapAll(buildTable(rows)));
412
	}
413

    
414
	private Oaf getOafBody(final List<Row> rows) throws InvalidProtocolBufferException {
415
		for (final Row row : rows) {
416
			if (StringUtils.startsWith(row.getKey(), "50")) { return Oaf.parseFrom(row.getColumn("body").getValue());
417

    
418
			}
419
		}
420
		return null;
421
	}
422

    
423
	@Test
424
	public void testParseDoajOAF() throws Exception {
425

    
426
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("doajUniversityRecord.xml"));
427
	}
428

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

    
432
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("datasourceNative.xml"));
433
	}
434

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

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

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

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

    
447
	@Test
448
	public void testFromMongodbCompressedDump() throws Exception {
449
		doTestJsonGz(loadFromTransformationProfile("oaf2hbase.xml"), load("mdstore_cleaned.json.gz"));
450
	}
451

    
452
	@Test
453
	public void testLoadFromTransformationProfile() throws IOException {
454
		final InputStream in = loadFromTransformationProfile("oaf2hbase.xml");
455
		log.info(IOUtils.toString(in));
456
	}
457

    
458
	@Test
459
	public void test_template() throws Exception {
460
		final String xslt = IOUtils.toString(loadFromTransformationProfile("oaf2hbase.xml"));
461
		final XsltRowTransformer transformer = factory.getTransformer(xslt);
462
		assertNotNull(transformer);
463

    
464
		final String record = IOUtils.toString(load("record.xml"));
465
		final List<Row> rows = transformer.apply(record);
466

    
467
		System.out.println(rows);
468
	}
469

    
470
	@Test
471
	public void testWrongCharsOrganization() throws Exception {
472
		final List<Row> rows = Lists.newArrayList();
473
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("organizationWrongChars.xml")));
474
		printAll(mapAll(buildTable(rows)));
475
	}
476

    
477
	@Test
478
	public void testParseProjectWithFunderOriginalName() throws Exception {
479

    
480
		doTest(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectWithFunderOriginalName.xml"));
481
	}
482

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

    
486
		final List<Row> rows = Lists.newArrayList();
487
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectWithFunderOriginalName.xml")));
488
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordFunderOriginalName.xml")));
489

    
490
		printAll(mapAll(buildTable(rows)));
491
	}
492

    
493
	@Test
494
	public void testProjectExtraInfo() throws Exception {
495
		final List<Row> rows = Lists.newArrayList();
496
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordExtraInfo.xml")));
497
		printAll(mapAll(buildTable(rows)));
498
	}
499

    
500
	@Test
501
	public void testParseSoftwareFromODF() throws Exception {
502
		final List<Row> rows = Lists.newArrayList();
503
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("softwareODF.xml")));
504
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
505
		printAll(mapAll(buildTable(rows)));
506
	}
507

    
508
	@Test
509
	public void testParseSoftwareFromOAF() throws Exception {
510
		final List<Row> rows = Lists.newArrayList();
511
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordOAFsoftwareCLOSED.xml")));
512
		printAll(mapAll(buildTable(rows)));
513
	}
514

    
515
	@Test
516
	public void testParsePubFromODF() throws Exception {
517
		final List<Row> rows = Lists.newArrayList();
518
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("recordODFPub.xml")));
519
		printAll(mapAll(buildTable(rows)));
520
	}
521

    
522
	@Test
523
	public void testParseSoftwareFromOAFOpen() throws Exception {
524
		final List<Row> rows = Lists.newArrayList();
525
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("recordOAFsoftwareOPEN.xml")));
526
		printAll(mapAll(buildTable(rows)));
527
	}
528

    
529
	@Test
530
	public void testParseSoftwareBiotool() throws Exception {
531
		final List<Row> rows = Lists.newArrayList();
532
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("biotoolSw.xml")));
533
		printAll(mapAll(buildTable(rows)));
534
	}
535

    
536
	@Test
537
	public void testParseOafWithExternalRef() throws Exception {
538
		final List<Row> rows = Lists.newArrayList();
539
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("oafWithExternalReference.xml")));
540
		printAll(mapAll(buildTable(rows)));
541
	}
542

    
543
	@Test
544
	public void testParseOafWithCommunity() throws Exception {
545
		final List<Row> rows = Lists.newArrayList();
546
		rows.addAll(asRows(loadFromTransformationProfile("oaf2hbase.xml"), load("oafWithCommunity.xml")));
547
		printAll(mapAll(buildTable(rows)));
548
	}
549

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

    
555
		final ActionFactory actionFactory = new ActionFactory();
556

    
557
		final Map<String, Resource> xslts = Maps.newHashMap();
558

    
559
		xslts.put("oaf2hbase", new ByteArrayResource(IOUtils.toString(loadFromTransformationProfile("oaf2hbase.xml")).getBytes()));
560
		actionFactory.setXslts(xslts);
561

    
562
		final XsltInfoPackageAction pa = actionFactory.generateInfoPackageAction(
563
				"oaf2hbase",
564
				"rawset-id",
565
				new Agent("agent-id", "agent-name", Agent.AGENT_TYPE.algo),
566
				Operation.UPDATE,
567
				IOUtils.toString(load("oafUpdateWithCommunity.xml")),
568
				Provenance.sysimport_mining_aggregator,
569
				"0.9");
570

    
571
		final String qualifier = "update_" + System.nanoTime();
572

    
573
		IOUtils.readLines(load("country_updates.json")).forEach(line -> {
574

    
575
			final Oaf.Builder oaf = Oaf.newBuilder();
576

    
577
			try {
578
				JsonFormat.merge(line, oaf);
579
			} catch (final JsonFormat.ParseException e) {
580
				throw new IllegalArgumentException(e);
581
			}
582

    
583
			final Column<String, byte[]> col = new Column<>("update_" + System.nanoTime(), oaf.build().toByteArray());
584
			rows.add(new Row("result", oaf.getEntity().getId(), Lists.newArrayList(col)));
585
		});
586

    
587
		pa.asAtomicActions().forEach(a -> {
588
			final Column<String, byte[]> col = new Column<>("update_" + System.nanoTime(), a.getTargetValue());
589
			rows.add(new Row(a.getTargetColumnFamily(), a.getTargetRowKey(), Lists.newArrayList(col)));
590
		});
591

    
592
		/*
593
		 * rows.forEach(r -> { log.info(r); });
594
		 */
595

    
596
		mapAll(buildTable(rows)).entrySet().forEach(b -> {
597
			log.info(b.getKey());
598
			log.info(b.getValue());
599
		});
600
	}
601

    
602
	@Test
603
	public void testParseCrisPub() throws Exception {
604
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("cris_pub1.xml"));
605
	}
606

    
607
	@Test
608
	public void testBioToolSwRowJson() throws Exception {
609
		doTestJsonRow(IOUtils.toString(load("biotoolSwRow.json")));
610
	}
611

    
612
	@Test
613
	public void testParseVirta() throws Exception {
614
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("virta.xml"));
615
	}
616

    
617
	@Test
618
	public void testParseJournal() throws Exception {
619
		doTest(loadFromTransformationProfile("datasources_2_hbase.xsl"), load("datasourceWithISSN.xml"));
620
	}
621

    
622
	@Test
623
	public void testGuidelines4Qeios() throws Exception {
624

    
625
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("guidelines4_qeios1.xml"));
626
	}
627

    
628
	@Test
629
	public void testGuidelines4Aria() throws Exception {
630

    
631
		doTest(loadFromTransformationProfile("odf2hbase.xml"), load("guidelines4_aria.xml"));
632
	}
633

    
634
	@Test
635
	public void testJournalRecord() throws Exception {
636
		doTest(loadFromTransformationProfile("oaf2hbase.xml"), load("journalRecord.xml"));
637
	}
638

    
639
	@Test
640
	public void testSygmaDatacite() throws Exception {
641

    
642
		final List<Row> rows = Lists.newArrayList();
643
		rows.addAll(asRows(loadFromTransformationProfile("odf2hbase.xml"), load("sygma_datacite.xml")));
644
		rows.addAll(asRows(loadFromTransformationProfile("projects_2_hbase.xsl"), load("projectRecordCorda.xml")));
645

    
646
		printAll(mapAll(buildTable(rows)));
647
	}
648

    
649
	@Test
650
	public void testOpenOrganizations() throws Exception {
651

    
652
		final List<Row> rows = Lists.newArrayList();
653
		rows.addAll(asRows(loadFromTransformationProfile("organizations_2_hbase.xsl"), load("openorgs_sample.xml")));
654

    
655
		printAll(mapAll(buildTable(rows)));
656
	}
657

    
658
	private void doTestJsonRow(final String json) throws Exception {
659
		final Row row = asRowFromJson(json);
660
		log.info(row);
661
		final List<Row> rows = new ArrayList<>();
662
		rows.add(row);
663
		final Map<String, Map<String, Map<String, byte[]>>> table = buildTable(rows);
664
		final Map<String, XmlRecordFactory> builders = mapAll(table);
665
		printAll(builders);
666

    
667
	}
668

    
669
	private void doTest(final InputStream xsltStream, final InputStream recordStream) throws Exception {
670
		try {
671
			final List<Row> rows = asRows(xsltStream, recordStream);
672

    
673
			log.info(rows);
674

    
675
			final Map<String, Map<String, Map<String, byte[]>>> table = buildTable(rows);
676

    
677
			// System.out.println("\n" + table.toString());
678

    
679
			final Map<String, XmlRecordFactory> builders = mapAll(table);
680

    
681
			printAll(builders);
682
		} catch (final InvalidProtocolBufferException e) {
683
			throw new Exception(e);
684
		} catch (final TransformerConfigurationException e) {
685
			throw new Exception(e);
686
		} catch (final TransformerFactoryConfigurationError e) {
687
			throw new Exception(e);
688
		} catch (final DocumentException e) {
689
			throw new Exception(e);
690
		}
691
	}
692

    
693
	private void doTestJsonGz(final InputStream xsltStream, final InputStream recordStream) throws Exception {
694

    
695
		final Iterator<List<Row>> rowsIterator = asRowsJsonGzip(xsltStream, recordStream);
696

    
697
		int i = 0;
698
		while (rowsIterator.hasNext()) {
699
			final List<Row> rows = rowsIterator.next();
700
			i++;
701

    
702
			if (i % 10000 == 0) {
703
				System.out.println(i);
704
			}
705

    
706
			final Map<String, Map<String, Map<String, byte[]>>> table = buildTableDoaj(rows);
707

    
708
			for (final Map<String, Map<String, byte[]>> m : table.values()) {
709
				for (final Map<String, byte[]> mv : m.values()) {
710
					for (final byte[] v : mv.values()) {
711
						final OafDecoder d = OafDecoder.decode(v);
712
						assertNotNull(d);
713
						assertNotNull(d.getOaf());
714

    
715
						switch (d.getKind()) {
716
						case entity:
717
							assertNotNull(d.getMetadata());
718
							if (d.getOaf().getEntity().getType().equals(Type.result)) {
719
								System.out.println(d.getOaf());
720
							}
721
							break;
722
						case relation:
723
							assertNotNull(d.getRel());
724
							break;
725
						default:
726
							break;
727
						}
728
					}
729
				}
730
			}
731
		}
732
	}
733

    
734
	private List<Row> asRows(final InputStream xsltStream, final InputStream recordStream, final Function<Row, Row> p) throws Exception {
735
		return asRows(xsltStream, new HashMap<>(), recordStream, p);
736
	}
737

    
738
	private List<Row> asRows(final InputStream xsltStream, final InputStream recordStream) throws Exception {
739
		return asRows(xsltStream, new HashMap<>(), recordStream);
740
	}
741

    
742
	private List<Row> asRows(final InputStream xsltStream, final Map<String, Object> params, final InputStream recordStream) throws Exception {
743
		return asRows(xsltStream, params, recordStream, null);
744
	}
745

    
746
	private List<Row> asRows(final InputStream xsltStream, final Map<String, Object> params, final InputStream recordStream, final Function<Row, Row> p)
747
			throws Exception {
748
		final String xslt = IOUtils.toString(xsltStream);
749
		final XsltRowTransformer transformer = factory.getTransformer(xslt, params);
750
		assertNotNull(transformer);
751

    
752
		final String record = IOUtils.toString(recordStream);
753
		final List<Row> rows = transformer.apply(record);
754

    
755
		assertNotNull(rows);
756
		assertFalse(rows.isEmpty());
757
		return p == null ? rows : Lists.newArrayList(Iterables.transform(rows, p));
758
	}
759

    
760
	private Iterator<List<Row>> asRowsJsonGzip(final InputStream xsltStream, final InputStream recordStreamJsonGzip) throws Exception {
761
		final String xslt = IOUtils.toString(xsltStream);
762
		final XsltRowTransformer transformer = factory.getTransformer(xslt);
763
		assertNotNull(transformer);
764
		assertNotNull(recordStreamJsonGzip);
765

    
766
		final GZIPInputStream stream = new GZIPInputStream(recordStreamJsonGzip);
767
		assertNotNull(stream);
768
		final BufferedReader inStream = new BufferedReader(new InputStreamReader(stream));
769
		assertNotNull(inStream);
770
		return new Iterator<List<Row>>() {
771

    
772
			String jsonRecord = null;
773

    
774
			@Override
775
			public boolean hasNext() {
776
				try {
777
					return (jsonRecord = inStream.readLine()) != null;
778
				} catch (final IOException e) {
779
					throw new RuntimeException(e);
780
				}
781
			}
782

    
783
			@Override
784
			public List<Row> next() {
785

    
786
				final JSONObject jsonObj = new JSONObject(jsonRecord);
787
				final String body = jsonObj.getString("body");
788
				try {
789
					assertTrue(StringUtils.isNotBlank(body));
790
					// System.out.println(body);
791
					final List<Row> rows = transformer.apply(body);
792
					assertNotNull(rows);
793
					assertFalse(rows.isEmpty());
794
					return rows;
795
				} catch (final Throwable e) {
796
					System.err.println("error transforming document: " + body);
797
					throw new RuntimeException(e);
798
				}
799
			}
800

    
801
			@Override
802
			public void remove() {
803
				throw new UnsupportedOperationException();
804
			}
805

    
806
		};
807

    
808
	}
809

    
810
	private Map<String, Map<String, Map<String, byte[]>>> buildTableDoaj(final List<Row> rows) throws UnsupportedEncodingException {
811
		final Map<String, Map<String, Map<String, byte[]>>> table = Maps.newHashMap();
812

    
813
		for (final Row row : rows) {
814
			final String rowKey = row.getKey();
815
			final String cf = row.getColumnFamily();
816
			if (!table.containsKey(rowKey)) {
817
				table.put(rowKey, new HashMap<>());
818
			}
819
			if (!table.get(rowKey).containsKey(cf)) {
820
				table.get(rowKey).put(row.getColumnFamily(), new HashMap<>());
821
			}
822
			for (final Column<String, byte[]> c : row.getColumns()) {
823
				// System.out.println(String.format("ADDING K:%s CF:%s Q:%s", rowKey, cf, c.getName()));
824
				table.get(rowKey).get(cf).put(c.getName(), c.getValue());
825
				if (cf.equals("result") && c.getName().equals("body")) {
826
					// System.out.println(String.format("ADDING K:%s CF:%s Q:%s", rowKey, cf, c.getName()));
827
					assertTrue(StringUtils.isNotBlank(new String(c.getValue(), "UTF-8")));
828
				}
829
			}
830
		}
831
		return table;
832

    
833
	}
834

    
835
	protected Map<String, Map<String, Map<String, byte[]>>> buildTable(final List<Row> rows) throws UnsupportedEncodingException {
836
		final Map<String, Map<String, Map<String, byte[]>>> table = Maps.newHashMap();
837

    
838
		for (final Row row : rows) {
839
			final String rowKey = row.getKey();
840
			final String cf = row.getColumnFamily();
841
			if (!table.containsKey(rowKey)) {
842
				table.put(rowKey, new HashMap<>());
843
			}
844
			if (!table.get(rowKey).containsKey(cf)) {
845
				table.get(rowKey).put(row.getColumnFamily(), new HashMap<>());
846
			}
847
			for (final Column<String, byte[]> c : row.getColumns()) {
848
				System.out.println(String.format("ADDING K:%s CF:%s Q:%s", rowKey, cf, c.getName()));
849
				table.get(rowKey).get(cf).put(c.getName(), c.getValue());
850
				if (c.getName().equals("body")) {
851
					final String theBody = new String(c.getValue(), "UTF-8");
852
					assertTrue(StringUtils.isNotBlank(theBody));
853
					// System.out.println(theBody);
854
				}
855
			}
856
		}
857
		return table;
858

    
859
	}
860

    
861
	protected Map<String, XmlRecordFactory> mapAll(final Map<String, Map<String, Map<String, byte[]>>> table) throws Exception {
862

    
863
		final Map<String, XmlRecordFactory> builders = Maps.newHashMap();
864
		for (final Entry<String, Map<String, Map<String, byte[]>>> e : table.entrySet()) {
865
			map(builders, e.getKey(), e.getValue());
866
		}
867
		return builders;
868
	}
869

    
870
	// private Map<String, XmlRecordFactory> mapResultsOnly(final Map<String, Map<String, Map<String, byte[]>>> table) throws Exception {
871
	//
872
	// final Map<String, XmlRecordFactory> builders = Maps.newHashMap();
873
	// for (final Entry<String, Map<String, Map<String, byte[]>>> e : table.entrySet()) {
874
	// final Type type = OafRowKeyDecoder.decode(e.getKey()).getType();
875
	// if (type == Type.result) {
876
	// map(builders, e.getKey(), e.getValue());
877
	// }
878
	// }
879
	// return builders;
880
	// }
881

    
882
	private void map(final Map<String, XmlRecordFactory> builders, final String rowKey, final Map<String, Map<String, byte[]>> row) throws Exception {
883

    
884
		final Type type = OafRowKeyDecoder.decode(rowKey).getType();
885

    
886
		final Map<String, byte[]> familyMap = row.get(type.toString());
887

    
888
		if (familyMap == null) { return; }
889

    
890
		final byte[] bodyB = familyMap.get("body");
891

    
892
		if (bodyB != null) {
893
			ensureBuilder(builders, rowKey);
894

    
895
			final Oaf oaf = UpdateMerger.mergeBodyUpdates(familyMap);
896

    
897
			final OafDecoder mainEntity = OafDecoder.decode(oaf);
898

    
899
			builders.get(rowKey).setMainEntity(mainEntity);
900

    
901
			for (final LinkDescriptor ld : entityConfigTable.getDescriptors(type)) {
902

    
903
				final String it = ld.getRelDescriptor().getIt();
904
				final Map<String, byte[]> cols = row.get(it);
905

    
906
				if (cols != null && !cols.isEmpty()) {
907

    
908
					for (final byte[] oafB : cols.values()) {
909

    
910
						final Oaf.Builder relBuilder = Oaf.newBuilder(Oaf.parseFrom(oafB));
911

    
912
						if (ld.isSymmetric()) {
913
							final RelDescriptor rd = ld.getRelDescriptor();
914

    
915
							relBuilder.getRelBuilder().setCachedTarget(mainEntity.getEntity()).setRelType(rd.getRelType()).setSubRelType(rd.getSubRelType());
916
						}
917

    
918
						relBuilder.getRelBuilder().setChild(ld.isChild());
919

    
920
						final Oaf.Builder oafBuilder = Oaf.newBuilder().setKind(Kind.relation).setLastupdatetimestamp(System.currentTimeMillis());
921
						oafBuilder.mergeFrom(relBuilder.build());
922

    
923
						final String targetId = ld.isSymmetric() ? oafBuilder.getRel().getTarget() : oafBuilder.getRel().getSource();
924
						ensureBuilder(builders, targetId);
925
						final OafDecoder decoder = OafDecoder.decode(oafBuilder.build());
926

    
927
						if (ld.isChild()) {
928
							builders.get(targetId).addChild(type, decoder);
929
						} else {
930
							builders.get(targetId).addRelation(type, decoder);
931
						}
932
					}
933

    
934
				}
935
			}
936
		}
937

    
938
	}
939

    
940
	private void printAll(final Map<String, XmlRecordFactory> builders) throws DocumentException {
941
		print(Sets.newHashSet(Type.values()), builders, null);
942
	}
943

    
944
	private void print(final Set<Type> types, final Map<String, XmlRecordFactory> builders, final Map<Type, Set<String>> xpaths) throws DocumentException {
945
		final SAXReader r = new SAXReader();
946

    
947
		for (final Entry<String, XmlRecordFactory> e : builders.entrySet()) {
948
			final OafRowKeyDecoder kd = OafRowKeyDecoder.decode(e.getKey());
949

    
950
			if (!e.getValue().isValid()) { throw new IllegalArgumentException("invalid builder: " + e.getKey()); }
951
			if (types.contains(kd.getType())) {
952
				final String val = IndentXmlString.apply(e.getValue().build());
953

    
954
				if (xpaths != null && !xpaths.isEmpty() && xpaths.get(kd.getType()) != null) {
955
					final Document doc = r.read(new StringReader(val));
956

    
957
					log.debug("\n" + e.getKey());
958
					for (final String xpath : xpaths.get(kd.getType())) {
959
						log.debug(doc.valueOf(xpath));
960
					}
961
				} else {
962
					log.info(val);
963
				}
964
			}
965
		}
966
	}
967

    
968
	private void printNoIndent(final Map<String, XmlRecordFactory> builders) {
969
		for (final Entry<String, XmlRecordFactory> e : builders.entrySet()) {
970
			if (e.getValue().isValid()) {
971
				log.debug(e.getValue().build());
972
			} else {
973
				log.debug("invalid builder: " + e.getKey());
974
			}
975
		}
976
	}
977

    
978
	private void ensureBuilder(final Map<String, XmlRecordFactory> builders, final String rowKey) throws Exception {
979
		if (!builders.containsKey(rowKey)) {
980
			builders.put(rowKey, newBuilder());
981
		}
982
	}
983

    
984
	private XmlRecordFactory newBuilder() throws TransformerConfigurationException, TransformerFactoryConfigurationError, DocumentException {
985
		return new XmlRecordFactory(entityConfigTable, ContextMapper.fromXml(Context.xml),
986
				RelClasses.fromJSon(RelClassesTest.relClassesJson), XmlRecordFactoryTest.SCHEMA_LOCATION, true, false, false,
987
				XmlRecordFactoryTest.specialDatasourceTypes);
988
	}
989

    
990
	private InputStream load(final String fileName) {
991
		return getClass().getResourceAsStream(fileName);
992
	}
993

    
994
	private InputStream loadFromTransformationProfile(final String profilePath) {
995
		log.info("Loading xslt from: " + basePathProfiles + profilePath);
996
		final InputStream profile = getClass().getResourceAsStream(basePathProfiles + profilePath);
997
		final SAXReader saxReader = new SAXReader();
998
		Document doc = null;
999
		try {
1000
			doc = saxReader.read(profile);
1001
		} catch (final DocumentException e) {
1002
			e.printStackTrace();
1003
			throw new RuntimeException(e);
1004
		}
1005
		final String xslt = doc.selectSingleNode("//SCRIPT/CODE/*[local-name()='stylesheet']").asXML();
1006
		// log.info(xslt);
1007
		return IOUtils.toInputStream(xslt);
1008
	}
1009

    
1010
	private Row asRowFromJson(final String json) throws ParseException {
1011
		final Oaf.Builder oafBuilder = Oaf.newBuilder();
1012
		JsonFormat.merge(json, oafBuilder);
1013
		final Oaf oaf = oafBuilder.build();
1014
		final Row row = new Row("result", oaf.getEntity().getId());
1015
		final Column<String, byte[]> c = new Column<>("body", oaf.toByteArray());
1016
		row.setColumn("body", c);
1017
		return row;
1018

    
1019
	}
1020

    
1021
}
    (1-1/1)