1
|
/**
|
2
|
* Copyright (C) 2007 EDIT
|
3
|
* European Distributed Institute of Taxonomy
|
4
|
* http://www.e-taxonomy.eu
|
5
|
*
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
8
|
*/
|
9
|
|
10
|
package eu.etaxonomy.cdm.io.excel.common;
|
11
|
|
12
|
import java.io.ByteArrayInputStream;
|
13
|
import java.io.FileNotFoundException;
|
14
|
import eu.etaxonomy.cdm.common.URI;
|
15
|
import java.util.List;
|
16
|
import java.util.Map;
|
17
|
import java.util.UUID;
|
18
|
|
19
|
import org.apache.commons.lang.StringUtils;
|
20
|
import org.apache.log4j.Logger;
|
21
|
import org.springframework.transaction.support.DefaultTransactionStatus;
|
22
|
|
23
|
import eu.etaxonomy.cdm.common.CdmUtils;
|
24
|
import eu.etaxonomy.cdm.common.ExcelUtils;
|
25
|
import eu.etaxonomy.cdm.io.common.CdmImportBase;
|
26
|
import eu.etaxonomy.cdm.io.distribution.excelupdate.ExcelDistributionUpdateConfigurator;
|
27
|
import eu.etaxonomy.cdm.io.excel.taxa.NormalExplicitImportConfigurator;
|
28
|
import eu.etaxonomy.cdm.io.excel.taxa.TaxonListImportConfigurator;
|
29
|
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase;
|
30
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
31
|
import eu.etaxonomy.cdm.model.common.TimePeriod;
|
32
|
import eu.etaxonomy.cdm.model.name.NomenclaturalCode;
|
33
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
34
|
import eu.etaxonomy.cdm.model.reference.ReferenceFactory;
|
35
|
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
36
|
import eu.etaxonomy.cdm.strategy.parser.TimePeriodParser;
|
37
|
|
38
|
/**
|
39
|
* @author a.babadshanjan
|
40
|
* @since 17.12.2008
|
41
|
*/
|
42
|
public abstract class ExcelImportBase<STATE extends ExcelImportState<CONFIG, ROW>, CONFIG extends ExcelImportConfiguratorBase, ROW extends ExcelRowBase>
|
43
|
extends CdmImportBase<CONFIG, STATE> {
|
44
|
|
45
|
private static final long serialVersionUID = 2759164811664484732L;
|
46
|
private static final Logger logger = Logger.getLogger(ExcelImportBase.class);
|
47
|
|
48
|
protected static final String SCIENTIFIC_NAME_COLUMN = "ScientificName";
|
49
|
|
50
|
private List<Map<String, String>> recordList = null;
|
51
|
|
52
|
private ExcelImportConfiguratorBase configurator = null;
|
53
|
|
54
|
|
55
|
/** Reads data from an Excel file and stores them into a CDM DB.
|
56
|
*
|
57
|
* @param config
|
58
|
* @param stores (not used)
|
59
|
*/
|
60
|
@Override
|
61
|
protected void doInvoke(STATE state){
|
62
|
|
63
|
logger.debug("Importing excel data");
|
64
|
|
65
|
//cleanup state from prior session
|
66
|
state.setSourceReference(null);
|
67
|
|
68
|
configurator = state.getConfig();
|
69
|
|
70
|
NomenclaturalCode nc = getConfigurator().getNomenclaturalCode();
|
71
|
if (nc == null && requiresNomenclaturalCode()) {
|
72
|
logger.error("Nomenclatural code could not be determined. Skip invoke.");
|
73
|
state.setUnsuccessfull();
|
74
|
return;
|
75
|
}
|
76
|
URI source = null;
|
77
|
|
78
|
byte[] data = null;
|
79
|
// read and save all rows of the excel worksheet
|
80
|
if ((state.getConfig() instanceof NormalExplicitImportConfigurator
|
81
|
|| state.getConfig() instanceof ExcelDistributionUpdateConfigurator
|
82
|
|| state.getConfig() instanceof TaxonListImportConfigurator) && (state.getConfig().getStream() != null)
|
83
|
){
|
84
|
data = state.getConfig().getStream();
|
85
|
} else{
|
86
|
source = state.getConfig().getSource();
|
87
|
}
|
88
|
|
89
|
String sheetName = getWorksheetName(state.getConfig());
|
90
|
|
91
|
if (data != null){
|
92
|
try {
|
93
|
ByteArrayInputStream stream = new ByteArrayInputStream(data);
|
94
|
recordList = ExcelUtils.parseXLS(stream, sheetName);
|
95
|
} catch (Exception e) {
|
96
|
throw new RuntimeException(e);
|
97
|
}
|
98
|
}else{
|
99
|
try {
|
100
|
recordList = ExcelUtils.parseXLS(source, sheetName);
|
101
|
} catch (FileNotFoundException e) {
|
102
|
String message = "File not found: " + source;
|
103
|
warnProgress(state, message, e);
|
104
|
logger.error(message);
|
105
|
state.setUnsuccessfull();
|
106
|
return;
|
107
|
}
|
108
|
}
|
109
|
|
110
|
handleRecordList(state, source);
|
111
|
logger.debug("End excel data import");
|
112
|
return;
|
113
|
}
|
114
|
|
115
|
protected boolean requiresNomenclaturalCode() {
|
116
|
return true;
|
117
|
}
|
118
|
|
119
|
private void handleRecordList(STATE state, URI source) {
|
120
|
Integer startingLine = 2;
|
121
|
if (recordList != null) {
|
122
|
Map<String,String> record = null;
|
123
|
|
124
|
state.setTransactionStatus(startTransaction());
|
125
|
|
126
|
//first pass
|
127
|
state.setCurrentLine(startingLine);
|
128
|
for (int i = 0; i < recordList.size(); i++) {
|
129
|
record = recordList.get(i);
|
130
|
analyzeRecord(record, state);
|
131
|
state.setOriginalRecord(record);
|
132
|
try {
|
133
|
firstPass(state);
|
134
|
//for debugging only
|
135
|
// if (i % 1000 == 0){
|
136
|
// try {
|
137
|
// System.out.println(i);
|
138
|
// getSession().flush();
|
139
|
// } catch (Exception e) {
|
140
|
// e.printStackTrace();
|
141
|
// }
|
142
|
// }
|
143
|
DefaultTransactionStatus defStatus = (DefaultTransactionStatus) state.getTransactionStatus();
|
144
|
if (defStatus.isRollbackOnly()){
|
145
|
logger.warn("Rollback only in line: " + i);
|
146
|
}
|
147
|
} catch (Exception e) {
|
148
|
e.printStackTrace();
|
149
|
}finally{
|
150
|
state.incCurrentLine();
|
151
|
}
|
152
|
}
|
153
|
//second pass
|
154
|
state.setCurrentLine(startingLine);
|
155
|
for (int i = 0; i < recordList.size(); i++) {
|
156
|
record = recordList.get(i);
|
157
|
analyzeRecord(record, state);
|
158
|
state.setOriginalRecord(record);
|
159
|
secondPass(state);
|
160
|
state.incCurrentLine();
|
161
|
}
|
162
|
if (configurator.isDeduplicateReferences()){
|
163
|
getReferenceService().deduplicate(Reference.class, null, null);
|
164
|
}
|
165
|
if (configurator.isDeduplicateAuthors()){
|
166
|
getAgentService().deduplicate(TeamOrPersonBase.class, null, null);
|
167
|
}
|
168
|
commitTransaction(state.getTransactionStatus());
|
169
|
}else{
|
170
|
logger.warn("No records found in " + source);
|
171
|
}
|
172
|
return;
|
173
|
}
|
174
|
|
175
|
/**
|
176
|
* To define a worksheet name other then the one defined in the configurator
|
177
|
* override this method with a non <code>null</code> return value.
|
178
|
* If <code>null</code> is returned the first worksheet is taken.
|
179
|
|
180
|
* @return worksheet name. <code>null</null> if no worksheet is defined.
|
181
|
*/
|
182
|
protected String getWorksheetName(CONFIG config) {
|
183
|
return config.getWorksheetName();
|
184
|
}
|
185
|
|
186
|
@Override
|
187
|
protected boolean doCheck(STATE state) {
|
188
|
boolean result = true;
|
189
|
logger.warn("No check implemented for Excel import");
|
190
|
return result;
|
191
|
}
|
192
|
|
193
|
/**
|
194
|
*
|
195
|
*
|
196
|
* @param record
|
197
|
* @return
|
198
|
*/
|
199
|
protected abstract void analyzeRecord(Map<String,String> record, STATE state);
|
200
|
|
201
|
protected abstract void firstPass(STATE state);
|
202
|
protected abstract void secondPass(STATE state);
|
203
|
|
204
|
|
205
|
public ExcelImportConfiguratorBase getConfigurator() {
|
206
|
return configurator;
|
207
|
}
|
208
|
|
209
|
|
210
|
protected int floatString2IntValue(String value) {
|
211
|
int intValue = 0;
|
212
|
try {
|
213
|
Float fobj = new Float(Float.parseFloat(value));
|
214
|
intValue = fobj.intValue();
|
215
|
if (logger.isDebugEnabled()) { logger.debug("Value formatted: " + intValue); }
|
216
|
} catch (NumberFormatException ex) {
|
217
|
logger.error(value + " is not an integer");
|
218
|
}
|
219
|
return intValue;
|
220
|
}
|
221
|
|
222
|
protected String floatString2IntStringValue(String value) {
|
223
|
int i = floatString2IntValue(value);
|
224
|
return String.valueOf(i);
|
225
|
}
|
226
|
|
227
|
|
228
|
/**
|
229
|
* @param start
|
230
|
* @param end
|
231
|
* @return
|
232
|
*/
|
233
|
protected TimePeriod getTimePeriod(String start, String end) {
|
234
|
String strPeriod = CdmUtils.concat(" - ", start, end);
|
235
|
TimePeriod result = TimePeriodParser.parseString(strPeriod);
|
236
|
return result;
|
237
|
}
|
238
|
|
239
|
/**
|
240
|
* Returns the value of the record map for the given key.
|
241
|
* The value is trimmed and empty values are set to <code>null</code>.
|
242
|
* @param record
|
243
|
* @param originalKey
|
244
|
* @return the value
|
245
|
*/
|
246
|
protected static String getValue(Map<String, String> record, String originalKey) {
|
247
|
String value = record.get(originalKey);
|
248
|
if (! StringUtils.isBlank(value)) {
|
249
|
if (logger.isDebugEnabled()) { logger.debug(originalKey + ": " + value); }
|
250
|
value = CdmUtils.removeDuplicateWhitespace(value.trim()).toString();
|
251
|
return value;
|
252
|
}else{
|
253
|
return null;
|
254
|
}
|
255
|
}
|
256
|
|
257
|
protected String getValue(STATE state, String key){
|
258
|
key = state.getConfig().replaceColumnLabel(key);
|
259
|
return getValue(state.getOriginalRecord(), key);
|
260
|
}
|
261
|
|
262
|
/**
|
263
|
* Returns the taxon for the given CDM uuid. If no taxon exists for the given id
|
264
|
* no record is returned. If a name cache, name title cache (full name) or
|
265
|
* taxon title cache column is given the name is checked against the given columns.
|
266
|
* If they don't manage it is logged as a warning in import result.
|
267
|
* <BR>If clazz is given, only objects of the given class are loaded.
|
268
|
*
|
269
|
*
|
270
|
* @param state
|
271
|
* @param colTaxonUuid taxon uuid column
|
272
|
* @param colNameCache name cache column (if exists)
|
273
|
* @param colNameTitleCache name title cache column (if exists)
|
274
|
* @param colTaxonTitleCache taxon title cache column (if exists)
|
275
|
* @param clazz the clazz null
|
276
|
* @param line the row, for debug information
|
277
|
* @return the taxon to load
|
278
|
*/
|
279
|
protected <T extends TaxonBase<?>> T getTaxonByCdmId(STATE state, String colTaxonUuid,
|
280
|
String colNameCache, String colNameTitleCache, String colTaxonTitleCache,
|
281
|
Class<T> clazz, String line) {
|
282
|
|
283
|
Map<String, String> record = getRecord(state);
|
284
|
String strUuidTaxon = record.get(colTaxonUuid);
|
285
|
if (strUuidTaxon != null){
|
286
|
UUID uuidTaxon;
|
287
|
try {
|
288
|
uuidTaxon = UUID.fromString(strUuidTaxon);
|
289
|
} catch (Exception e) {
|
290
|
state.getResult().addError("Taxon uuid has incorrect format. Taxon could not be loaded. Data not imported.", null, line);
|
291
|
return null;
|
292
|
}
|
293
|
TaxonBase<?> result = getTaxonService().find(uuidTaxon);
|
294
|
//TODO load only objects of correct class
|
295
|
if (result != null && clazz != null && !result.isInstanceOf(clazz)){
|
296
|
result = null;
|
297
|
}
|
298
|
|
299
|
|
300
|
if (result == null){
|
301
|
state.getResult().addError("Taxon for uuid "+strUuidTaxon+" could not be found in database. "
|
302
|
+ "Taxon could not be loaded. Data not imported.", null, line);
|
303
|
}else{
|
304
|
verifyName(state, colNameCache, colNameTitleCache, colTaxonTitleCache, line, record, result);
|
305
|
}
|
306
|
result = CdmBase.deproxy(result, clazz);
|
307
|
|
308
|
|
309
|
return CdmBase.deproxy(result, clazz);
|
310
|
}else{
|
311
|
String message = "No taxon identifier column found";
|
312
|
state.getResult().addWarning(message, null, line);
|
313
|
return null;
|
314
|
}
|
315
|
}
|
316
|
|
317
|
protected Map<String, String> getRecord(STATE state) {
|
318
|
Map<String, String> record = state.getOriginalRecord();
|
319
|
return record;
|
320
|
}
|
321
|
|
322
|
|
323
|
/**
|
324
|
* @see #getTaxonByCdmId(ExcelImportState, String, String, String, String, Class, String)
|
325
|
*/
|
326
|
protected void verifyName(STATE state, String colNameCache, String colNameTitleCache, String colTaxonTitleCache,
|
327
|
String line, Map<String, String> record, TaxonBase<?> result) {
|
328
|
//nameCache
|
329
|
String strExpectedNameCache = record.get(colNameCache);
|
330
|
String nameCache = result.getName() == null ? null : result.getName().getNameCache();
|
331
|
if (isNotBlank(strExpectedNameCache) && (!strExpectedNameCache.trim().equals(nameCache))){
|
332
|
String message = "Name cache (%s) does not match expected name (%s)";
|
333
|
message = String.format(message, nameCache==null? "null":nameCache, strExpectedNameCache);
|
334
|
state.getResult().addWarning(message, null, line);
|
335
|
}
|
336
|
//name title
|
337
|
String strExpectedNameTitleCache = record.get(colNameTitleCache);
|
338
|
String nameTitleCache = result.getName() == null ? null : result.getName().getTitleCache();
|
339
|
if (isNotBlank(strExpectedNameTitleCache) && (!strExpectedNameTitleCache.trim().equals(nameTitleCache))){
|
340
|
String message = "Name title cache (%s) does not match expected name (%s)";
|
341
|
message = String.format(message, nameTitleCache==null? "null":nameTitleCache, strExpectedNameTitleCache);
|
342
|
state.getResult().addWarning(message, null, line);
|
343
|
}
|
344
|
//taxon title cache
|
345
|
String strExpectedTaxonTitleCache = record.get(colTaxonTitleCache);
|
346
|
String taxonTitleCache = result.getTitleCache();
|
347
|
if (isNotBlank(strExpectedTaxonTitleCache) && (!strExpectedTaxonTitleCache.trim().equals(taxonTitleCache))){
|
348
|
String message = "Name cache (%s) does not match expected name (%s)";
|
349
|
message = String.format(message, taxonTitleCache==null? "null":taxonTitleCache, strExpectedTaxonTitleCache);
|
350
|
state.getResult().addWarning(message, null, line);
|
351
|
}
|
352
|
}
|
353
|
|
354
|
|
355
|
/**
|
356
|
* Non transaction save method to retrieve the source reference
|
357
|
* if either existent or not in the database (uses check for uuid).
|
358
|
*
|
359
|
* @param state
|
360
|
* @return the source reference
|
361
|
*/
|
362
|
protected Reference getSourceReference(STATE state) {
|
363
|
|
364
|
Reference sourceRef = state.getSourceReference();
|
365
|
if (sourceRef != null){
|
366
|
return sourceRef;
|
367
|
}
|
368
|
UUID uuid = state.getConfig().getSourceRefUuid();
|
369
|
if (uuid == null){
|
370
|
sourceRef = state.getConfig().getSourceReference();
|
371
|
if (sourceRef != null){
|
372
|
uuid = sourceRef.getUuid();
|
373
|
}
|
374
|
}
|
375
|
if (uuid != null){
|
376
|
Reference existingRef = getReferenceService().find(uuid);
|
377
|
if (existingRef != null){
|
378
|
sourceRef = existingRef;
|
379
|
}
|
380
|
// else if (sourceRef != null){
|
381
|
// getReferenceService().save(sourceRef);
|
382
|
// }
|
383
|
}
|
384
|
if (sourceRef == null){
|
385
|
sourceRef = ReferenceFactory.newGeneric();
|
386
|
String title = state.getConfig().getSourceNameString();
|
387
|
sourceRef.setTitle(title);
|
388
|
state.getConfig().setSourceReference(sourceRef);
|
389
|
}
|
390
|
state.setSourceReference(sourceRef);
|
391
|
|
392
|
return sourceRef;
|
393
|
}
|
394
|
}
|