1
|
package eu.dnetlib.data.collective.transformation.engine;
|
2
|
|
3
|
import java.io.StringReader;
|
4
|
import java.util.HashMap;
|
5
|
import java.util.Iterator;
|
6
|
import java.util.LinkedHashMap;
|
7
|
import java.util.LinkedList;
|
8
|
import java.util.List;
|
9
|
import java.util.Map;
|
10
|
|
11
|
import javax.xml.xpath.XPath;
|
12
|
import javax.xml.xpath.XPathConstants;
|
13
|
import javax.xml.xpath.XPathExpressionException;
|
14
|
import javax.xml.xpath.XPathFactory;
|
15
|
|
16
|
import net.sf.saxon.instruct.TerminationException;
|
17
|
|
18
|
import org.apache.commons.logging.Log;
|
19
|
import org.apache.commons.logging.LogFactory;
|
20
|
import org.springframework.core.io.Resource;
|
21
|
import org.w3c.dom.Node;
|
22
|
import org.xml.sax.InputSource;
|
23
|
|
24
|
import eu.dnetlib.common.profile.ResourceDao;
|
25
|
import eu.dnetlib.data.collective.transformation.IDatabaseConnector;
|
26
|
import eu.dnetlib.data.collective.transformation.TransformationException;
|
27
|
import eu.dnetlib.data.collective.transformation.VocabularyRegistry;
|
28
|
import eu.dnetlib.data.collective.transformation.core.xsl.ext.TransformationFunctionProxy;
|
29
|
import eu.dnetlib.data.collective.transformation.engine.core.ITransformation;
|
30
|
import eu.dnetlib.data.collective.transformation.engine.functions.Convert;
|
31
|
// import eu.dnetlib.data.collective.transformation.engine.functions.Dblookup;
|
32
|
import eu.dnetlib.data.collective.transformation.engine.functions.Extract;
|
33
|
import eu.dnetlib.data.collective.transformation.engine.functions.IFeatureExtraction;
|
34
|
import eu.dnetlib.data.collective.transformation.engine.functions.ProcessingException;
|
35
|
import eu.dnetlib.data.collective.transformation.engine.functions.RegularExpression;
|
36
|
import eu.dnetlib.data.collective.transformation.engine.functions.RetrieveValue;
|
37
|
import eu.dnetlib.data.collective.transformation.engine.functions.RetrieveValue.FUNCTION;
|
38
|
import eu.dnetlib.data.collective.transformation.rulelanguage.Argument;
|
39
|
import eu.dnetlib.data.collective.transformation.rulelanguage.Argument.Type;
|
40
|
import eu.dnetlib.data.collective.transformation.rulelanguage.util.FunctionCall;
|
41
|
import eu.dnetlib.data.collective.transformation.utils.BlacklistConsumer;
|
42
|
|
43
|
/**
|
44
|
* @author jochen
|
45
|
*
|
46
|
*/
|
47
|
public class SimpleTransformationEngine {
|
48
|
|
49
|
private static Log log = LogFactory.getLog(SimpleTransformationEngine.class);
|
50
|
private ITransformation transformation;
|
51
|
private VocabularyRegistry vocabularyRegistry;
|
52
|
private IDatabaseConnector databaseConnector;
|
53
|
private ResourceDao resourceDao;
|
54
|
private IFeatureExtraction featureExtraction;
|
55
|
private final List<String> mdRecords = new LinkedList<String>();
|
56
|
private long totalTransformedRecords = 0;
|
57
|
private long totalIgnoredRecords = 0;
|
58
|
private String mappingFile;
|
59
|
private boolean stylesheetParamsCalculated = false;
|
60
|
private boolean preprocessingDone = false;
|
61
|
private Map<String, String> stylesheetParams = new LinkedHashMap<String, String>();
|
62
|
private Resource blacklistApi;
|
63
|
private List<String> blacklistedRecords = new LinkedList<String>();
|
64
|
|
65
|
|
66
|
/**
|
67
|
* execute any preprocessings declared in the transformation script prior starting the transformation of records
|
68
|
*/
|
69
|
public void preprocess(String dataSourceId) {
|
70
|
for (Map<String, String> preprocMap : this.transformation.getRuleLanguageParser().getPreprocessings()) {
|
71
|
Iterator<String> it = preprocMap.keySet().iterator();
|
72
|
while (it.hasNext()) {
|
73
|
String function = it.next();
|
74
|
// if (function.equals("dblookup")) {
|
75
|
// Dblookup fun = new Dblookup();
|
76
|
// fun.setDbConnector(databaseConnector);
|
77
|
// try {
|
78
|
// log.debug("preprocessingMap value: " + preprocMap.get(function));
|
79
|
// TransformationFunctionProxy.getInstance().setLookupRecord(fun.getResults(preprocMap.get(function)));
|
80
|
// } catch (Exception e) {
|
81
|
// log.debug(e.getMessage());
|
82
|
// throw new IllegalStateException(e);
|
83
|
// }
|
84
|
// }
|
85
|
if (function.equals("blacklist")) {
|
86
|
BlacklistConsumer bc = new BlacklistConsumer();
|
87
|
try{
|
88
|
blacklistedRecords = bc.getBlackList(blacklistApi.getURL() + dataSourceId);
|
89
|
}catch(Exception e){
|
90
|
throw new IllegalStateException("error in preprocess: " + e.getMessage());
|
91
|
}
|
92
|
}
|
93
|
}
|
94
|
}
|
95
|
log.debug("preprocessing done.");
|
96
|
}
|
97
|
|
98
|
/**
|
99
|
* check if blacklistedRecords exist and if so check if the current record is blacklisted by its objIdentifier
|
100
|
* @param aRecord
|
101
|
* @return
|
102
|
* @throws XPathExpressionException
|
103
|
* @throws ProcessingException
|
104
|
*/
|
105
|
private boolean isBlacklistRecord(String aRecord){
|
106
|
if (blacklistedRecords.size() == 0) return false;
|
107
|
XPath xpath = XPathFactory.newInstance().newXPath();
|
108
|
try{
|
109
|
Node root = (Node) xpath.evaluate("/", new InputSource(new StringReader(aRecord)), XPathConstants.NODE);
|
110
|
String objId = xpath.evaluate("//*[local-name()='objIdentifier']", root);
|
111
|
if (blacklistedRecords.contains(objId)) return true;
|
112
|
}catch(Exception e){
|
113
|
throw new IllegalStateException("error in isBlacklistRecord: " + e.getMessage());
|
114
|
}
|
115
|
return false;
|
116
|
}
|
117
|
|
118
|
/**
|
119
|
* transforms a source record
|
120
|
*
|
121
|
* @param sourceRecord
|
122
|
* the record to transform
|
123
|
* @return transformed record
|
124
|
*/
|
125
|
public String transform(final String sourceRecord) {
|
126
|
List<String> objectRecords = new LinkedList<String>();
|
127
|
objectRecords.add(sourceRecord);
|
128
|
int index = 0;
|
129
|
mdRecords.clear();
|
130
|
initTransformationFunction();
|
131
|
|
132
|
if (!stylesheetParamsCalculated) {
|
133
|
try{
|
134
|
calculateStylesheetParams(sourceRecord);
|
135
|
}catch(Exception e){
|
136
|
throw new IllegalStateException("error in calculateStylesheetParams" + e.getMessage());
|
137
|
}
|
138
|
}
|
139
|
|
140
|
if (!preprocessingDone){
|
141
|
// xpath sourceRecord dataSourceid
|
142
|
preprocess(stylesheetParams.get("varBlacklistDataSourceId"));
|
143
|
preprocessingDone = true;
|
144
|
}
|
145
|
|
146
|
if (isBlacklistRecord(sourceRecord)){
|
147
|
try{
|
148
|
mdRecords.add(transformation.transformRecord(sourceRecord, ITransformation.XSLSyntaxcheckfailed));
|
149
|
}catch(Exception e){
|
150
|
log.fatal(sourceRecord);
|
151
|
throw new IllegalStateException(e);
|
152
|
}
|
153
|
}else if (!transformation.getRuleLanguageParser().isXslStylesheet()) {
|
154
|
// iterate over all rules which are functionCalls
|
155
|
log.debug("functionCalls size: " + transformation.getRuleLanguageParser().getFunctionCalls().size());
|
156
|
for (FunctionCall functionCall : transformation.getRuleLanguageParser().getFunctionCalls()) {
|
157
|
preprocess(objectRecords, functionCall);
|
158
|
}
|
159
|
for (String record : objectRecords) {
|
160
|
// log.debug(record);
|
161
|
try {
|
162
|
log.debug("now run transformation for record with index: " + index);
|
163
|
try{
|
164
|
String transformedRecord = transformation.transformRecord(record, index);
|
165
|
mdRecords.add(transformedRecord);
|
166
|
} catch (TerminationException e){
|
167
|
log.debug("record transformation terminated.");
|
168
|
String failedRecord = transformation.transformRecord(record, ITransformation.XSLSyntaxcheckfailed);
|
169
|
log.debug(failedRecord);
|
170
|
totalIgnoredRecords++;
|
171
|
mdRecords.add(failedRecord);
|
172
|
}
|
173
|
} catch (TransformationException e) {
|
174
|
log.error(sourceRecord);
|
175
|
throw new IllegalStateException(e);
|
176
|
}
|
177
|
index++;
|
178
|
}
|
179
|
} else {
|
180
|
for (String record : objectRecords) {
|
181
|
// test for init params and assign values
|
182
|
try {
|
183
|
log.debug("now run transformation for record with index: " + index);
|
184
|
try{
|
185
|
String transformedRecord = transformation.transformRecord(record, stylesheetParams);
|
186
|
mdRecords.add(transformedRecord);
|
187
|
}catch(TerminationException e){
|
188
|
String failedRecord = transformation.transformRecord(record, ITransformation.XSLSyntaxcheckfailed);
|
189
|
totalIgnoredRecords++;
|
190
|
log.debug(failedRecord);
|
191
|
mdRecords.add(failedRecord);
|
192
|
}
|
193
|
} catch (TransformationException e) {
|
194
|
log.error(sourceRecord);
|
195
|
throw new IllegalStateException(e);
|
196
|
}
|
197
|
index++;
|
198
|
}
|
199
|
}
|
200
|
|
201
|
totalTransformedRecords = totalTransformedRecords + mdRecords.size();
|
202
|
log.debug("objRecordSize: " + objectRecords.size() + ", mdRecordSize: " + mdRecords.size() + ", ignoredRecordSize: " + totalIgnoredRecords);
|
203
|
return mdRecords.get(0);
|
204
|
}
|
205
|
|
206
|
private void calculateStylesheetParams(final String aRecord) throws XPathExpressionException, ProcessingException {
|
207
|
stylesheetParamsCalculated = true;
|
208
|
XPath xpath = XPathFactory.newInstance().newXPath();
|
209
|
Node root = (Node) xpath.evaluate("/", new InputSource(new StringReader(aRecord)), XPathConstants.NODE);
|
210
|
String datasourcePrefix = xpath.evaluate("//*[local-name()='datasourceprefix']", root);
|
211
|
String profileXquery = "collection('/db/DRIVER/RepositoryServiceResources')//RESOURCE_PROFILE[.//EXTRA_FIELDS/FIELD[key=\"NamespacePrefix\"][value=\"" + datasourcePrefix + "\"]]";
|
212
|
//String repositoryId = xpath.evaluate("//*[local-name()='repositoryId']", root);
|
213
|
log.debug("profileXquery: " + profileXquery);
|
214
|
// static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId",
|
215
|
// xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
|
216
|
RetrieveValue retrieveValue = new RetrieveValue();
|
217
|
retrieveValue.setResourceDao(resourceDao);
|
218
|
List<Argument> argList = new LinkedList<Argument>();
|
219
|
argList.add(new Argument(Type.VALUE, profileXquery));
|
220
|
Argument argXpath = new Argument(Type.INPUTFIELD, "//OFFICIAL_NAME");
|
221
|
argList.add(argXpath);
|
222
|
String varOfficialName = retrieveValue.executeSingleValue(FUNCTION.PROFILEFIELD.toString(), argList, null, new HashMap<String, String>());
|
223
|
stylesheetParams.put("varOfficialName", varOfficialName);
|
224
|
argList.remove(argXpath);
|
225
|
argXpath = new Argument(Type.INPUTFIELD, "//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value");
|
226
|
argList.add(argXpath);
|
227
|
String varDataSourceId = retrieveValue.executeSingleValue(FUNCTION.PROFILEFIELD.toString(), argList, null, new HashMap<String, String>());
|
228
|
stylesheetParams.put("varDataSourceId", varDataSourceId);
|
229
|
argList.remove(argXpath);
|
230
|
argXpath = new Argument(Type.INPUTFIELD, "//CONFIGURATION/DATASOURCE_TYPE");
|
231
|
argList.add(argXpath);
|
232
|
String varDsType = retrieveValue.executeSingleValue(FUNCTION.PROFILEFIELD.toString(), argList, null, new HashMap<String, String>());
|
233
|
stylesheetParams.put("varDsType", varDsType);
|
234
|
argList.remove(argXpath);
|
235
|
|
236
|
// if blacklist
|
237
|
for (Map<String, String> preprocMap : this.transformation.getRuleLanguageParser().getPreprocessings()) {
|
238
|
Iterator<String> it = preprocMap.keySet().iterator();
|
239
|
while (it.hasNext()) {
|
240
|
String function = it.next();
|
241
|
if (function.equals("blacklist")) {
|
242
|
argXpath = new Argument(Type.INPUTFIELD, preprocMap.get(function)); // blacklistDataSourceIdXpath
|
243
|
argList.add(argXpath);
|
244
|
String varBlacklistDataSourceId = retrieveValue.executeSingleValue(FUNCTION.PROFILEFIELD.toString(), argList, null, new HashMap<String, String>());
|
245
|
stylesheetParams.put("varBlacklistDataSourceId", varBlacklistDataSourceId);
|
246
|
argList.remove(argXpath);
|
247
|
}
|
248
|
}
|
249
|
}
|
250
|
}
|
251
|
|
252
|
private void initTransformationFunction() {
|
253
|
if (this.vocabularyRegistry == null) { throw new IllegalStateException("vocabularyReg is null"); }
|
254
|
Convert convertFunction = new Convert();
|
255
|
convertFunction.setVocabularyRegistry(this.vocabularyRegistry);
|
256
|
TransformationFunctionProxy.getInstance().setConvertFunction(convertFunction);
|
257
|
|
258
|
}
|
259
|
|
260
|
/**
|
261
|
* preprocesses function if function is configured resp.
|
262
|
*
|
263
|
* @param records
|
264
|
* list of object records
|
265
|
* @param aFunctionCall
|
266
|
*/
|
267
|
private void preprocess(final List<String> records, final FunctionCall aFunctionCall) {
|
268
|
try {
|
269
|
log.debug("preprocess");
|
270
|
if (transformation.getRuleLanguageParser() == null) { throw new IllegalStateException("rulelanguageparser not initialised"); }
|
271
|
if (transformation.getRuleLanguageParser().getNamespaceDeclarations() == null) { throw new IllegalStateException("nsDecl is null"); }
|
272
|
PreProcessor preProc = new PreProcessor();
|
273
|
preProc.setConvertFunction(TransformationFunctionProxy.getInstance().getConvertFunction());
|
274
|
RetrieveValue retrieveValue = new RetrieveValue();
|
275
|
retrieveValue.setResourceDao(resourceDao);
|
276
|
preProc.setRetrieveFunction(retrieveValue);
|
277
|
RegularExpression regExpr = new RegularExpression();
|
278
|
preProc.setRegExprFunction(regExpr);
|
279
|
TransformationFunctionProxy functionProxy = TransformationFunctionProxy.getInstance();
|
280
|
preProc.setFunctionProxy(functionProxy);
|
281
|
Extract extractFunction = new Extract();
|
282
|
extractFunction.setFeatureExtraction(featureExtraction);
|
283
|
preProc.setExtractFunction(extractFunction);
|
284
|
if (aFunctionCall.doPreprocess() || aFunctionCall.isStatic()) {
|
285
|
// log.debug("now call preprocess with: " + aFunctionCall.getExternalFunctionName() + " " + aFunctionCall.getUuid());
|
286
|
preProc.preprocess(
|
287
|
aFunctionCall,
|
288
|
records,
|
289
|
transformation.getRuleLanguageParser().getNamespaceDeclarations(),
|
290
|
transformation.getStaticTransformationResults(),
|
291
|
transformation.getJobProperties(),
|
292
|
transformation.getRuleLanguageParser().getVariableMappingRules());
|
293
|
// log.debug("preprocess end");
|
294
|
} else {
|
295
|
log.debug("skip preprocessing for function: " + aFunctionCall.getExternalFunctionName());
|
296
|
}
|
297
|
|
298
|
} catch (Exception e) {
|
299
|
throw new IllegalStateException(e);
|
300
|
}
|
301
|
|
302
|
}
|
303
|
|
304
|
/**
|
305
|
* @param transformation
|
306
|
* the transformation to set
|
307
|
*/
|
308
|
public void setTransformation(final ITransformation transformation) {
|
309
|
this.transformation = transformation;
|
310
|
}
|
311
|
|
312
|
/**
|
313
|
* @return the transformation
|
314
|
*/
|
315
|
public ITransformation getTransformation() {
|
316
|
return transformation;
|
317
|
}
|
318
|
|
319
|
/**
|
320
|
* @param vocabularyRegistry
|
321
|
* the vocabularyRegistry to set
|
322
|
*/
|
323
|
public void setVocabularyRegistry(final VocabularyRegistry vocabularyRegistry) {
|
324
|
this.vocabularyRegistry = vocabularyRegistry;
|
325
|
}
|
326
|
|
327
|
/**
|
328
|
* @return the vocabularyRegistry
|
329
|
*/
|
330
|
public VocabularyRegistry getVocabularyRegistry() {
|
331
|
return vocabularyRegistry;
|
332
|
}
|
333
|
|
334
|
/**
|
335
|
* @return the resourceDao
|
336
|
*/
|
337
|
public ResourceDao getResourceDao() {
|
338
|
return resourceDao;
|
339
|
}
|
340
|
|
341
|
/**
|
342
|
* @param resourceDao
|
343
|
* the resourceDao to set
|
344
|
*/
|
345
|
public void setResourceDao(final ResourceDao resourceDao) {
|
346
|
this.resourceDao = resourceDao;
|
347
|
}
|
348
|
|
349
|
/**
|
350
|
* @param featureExtraction
|
351
|
* the featureExtraction to set
|
352
|
*/
|
353
|
public void setFeatureExtraction(final IFeatureExtraction featureExtraction) {
|
354
|
this.featureExtraction = featureExtraction;
|
355
|
}
|
356
|
|
357
|
/**
|
358
|
* @return the featureExtraction
|
359
|
*/
|
360
|
public IFeatureExtraction getFeatureExtraction() {
|
361
|
return featureExtraction;
|
362
|
}
|
363
|
|
364
|
/**
|
365
|
* @return the databaseConnector
|
366
|
*/
|
367
|
public IDatabaseConnector getDatabaseConnector() {
|
368
|
return databaseConnector;
|
369
|
}
|
370
|
|
371
|
/**
|
372
|
* @param databaseConnector
|
373
|
* the databaseConnector to set
|
374
|
*/
|
375
|
public void setDatabaseConnector(final IDatabaseConnector databaseConnector) {
|
376
|
this.databaseConnector = databaseConnector;
|
377
|
}
|
378
|
|
379
|
public long getTotalTransformedRecords() {
|
380
|
return this.totalTransformedRecords;
|
381
|
}
|
382
|
|
383
|
public long getTotalIgnoredRecords() {
|
384
|
return this.totalIgnoredRecords;
|
385
|
}
|
386
|
|
387
|
/**
|
388
|
* @return the mappingFile
|
389
|
*/
|
390
|
public String getMappingFile() {
|
391
|
return mappingFile;
|
392
|
}
|
393
|
|
394
|
/**
|
395
|
* @param mappingFile
|
396
|
* the mappingFile to set
|
397
|
*/
|
398
|
public void setMappingFile(final String mappingFile) {
|
399
|
this.mappingFile = mappingFile;
|
400
|
}
|
401
|
|
402
|
public Resource getBlacklistApi() {
|
403
|
return blacklistApi;
|
404
|
}
|
405
|
|
406
|
public void setBlacklistApi(Resource blacklistApi) {
|
407
|
this.blacklistApi = blacklistApi;
|
408
|
}
|
409
|
}
|