|
1 |
/**
|
|
2 |
* Copyright (C) 2009 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 |
package eu.etaxonomy.cdm.io.pesi.out;
|
|
10 |
|
|
11 |
import java.sql.SQLException;
|
|
12 |
import java.util.Arrays;
|
|
13 |
import java.util.List;
|
|
14 |
import java.util.Set;
|
|
15 |
import java.util.UUID;
|
|
16 |
|
|
17 |
import org.apache.log4j.Logger;
|
|
18 |
import org.springframework.stereotype.Component;
|
|
19 |
import org.springframework.transaction.TransactionStatus;
|
|
20 |
|
|
21 |
import eu.etaxonomy.cdm.common.CdmUtils;
|
|
22 |
import eu.etaxonomy.cdm.io.common.mapping.out.DbExportIgnoreMapper;
|
|
23 |
import eu.etaxonomy.cdm.io.common.mapping.out.DbFixedIntegerMapper;
|
|
24 |
import eu.etaxonomy.cdm.io.common.mapping.out.DbFixedStringMapper;
|
|
25 |
import eu.etaxonomy.cdm.io.common.mapping.out.IdIncMapper;
|
|
26 |
import eu.etaxonomy.cdm.io.common.mapping.out.MethodMapper;
|
|
27 |
import eu.etaxonomy.cdm.io.pesi.erms.ErmsLinkImport;
|
|
28 |
import eu.etaxonomy.cdm.io.pesi.erms.ErmsTransformer;
|
|
29 |
import eu.etaxonomy.cdm.model.common.CdmBase;
|
|
30 |
import eu.etaxonomy.cdm.model.common.Extension;
|
|
31 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase;
|
|
32 |
import eu.etaxonomy.cdm.profiler.ProfilerController;
|
|
33 |
/**
|
|
34 |
* The export class for PESI ecology notes (marine, brackish, fresh, terrestrial) coming from ERMS tu table.<p>
|
|
35 |
* @author a.mueller
|
|
36 |
* @since 28.09.2019
|
|
37 |
*/
|
|
38 |
@Component
|
|
39 |
public class PesiEcologyAndLinkExport extends PesiExportBase {
|
|
40 |
|
|
41 |
private static final long serialVersionUID = -2567615286288369111L;
|
|
42 |
private static final Logger logger = Logger.getLogger(PesiEcologyAndLinkExport.class);
|
|
43 |
|
|
44 |
private static final Class<? extends CdmBase> standardMethodParameter = TaxonBase.class;
|
|
45 |
|
|
46 |
private static int modCount = 1000;
|
|
47 |
private static final String dbTableName = "Note";
|
|
48 |
private static final String pluralString = "ecology or link notes";
|
|
49 |
private static final String parentPluralString = "Taxa";
|
|
50 |
|
|
51 |
public PesiEcologyAndLinkExport() {
|
|
52 |
super();
|
|
53 |
}
|
|
54 |
|
|
55 |
int countNotes = 0;
|
|
56 |
int countUrls = 0;
|
|
57 |
String currentValue;
|
|
58 |
|
|
59 |
@Override
|
|
60 |
public Class<? extends CdmBase> getStandardMethodParameter() {
|
|
61 |
return standardMethodParameter;
|
|
62 |
}
|
|
63 |
|
|
64 |
@Override
|
|
65 |
protected void doInvoke(PesiExportState state) {
|
|
66 |
try {
|
|
67 |
logger.info("*** Started making " + pluralString + " ...");
|
|
68 |
|
|
69 |
// Stores whether this invoke was successful or not.
|
|
70 |
boolean success = true;
|
|
71 |
|
|
72 |
// Get specific mappings: (CDM) TaxonBase.marker -> (PESI) Note (ecology)
|
|
73 |
PesiExportMapping mapping = getMapping();
|
|
74 |
mapping.initialize(state);
|
|
75 |
|
|
76 |
PesiExportMapping urlMapping = getUrlMapping();
|
|
77 |
mapping.initialize(state);
|
|
78 |
|
|
79 |
//All
|
|
80 |
success &= doPhase01(state, mapping, urlMapping);
|
|
81 |
|
|
82 |
logger.info("*** Finished Making " + pluralString + " ..." + getSuccessString(success));
|
|
83 |
|
|
84 |
if (!success){
|
|
85 |
state.getResult().addError("An unknown problem occurred");
|
|
86 |
}
|
|
87 |
return;
|
|
88 |
} catch (SQLException e) {
|
|
89 |
e.printStackTrace();
|
|
90 |
logger.error(e.getMessage());
|
|
91 |
state.getResult().addException(e, e.getMessage());
|
|
92 |
}
|
|
93 |
}
|
|
94 |
|
|
95 |
//PHASE 01: All
|
|
96 |
private boolean doPhase01(PesiExportState state, PesiExportMapping mapping, PesiExportMapping urlMapping) throws SQLException {
|
|
97 |
|
|
98 |
// logger.info("PHASE 1 (ecology...");
|
|
99 |
int count = 0;
|
|
100 |
int pastCount = 0;
|
|
101 |
boolean success = true;
|
|
102 |
int limit = state.getConfig().getLimitSave();
|
|
103 |
|
|
104 |
List<TaxonBase> taxonList = null;
|
|
105 |
|
|
106 |
TransactionStatus txStatus = startTransaction(true);
|
|
107 |
|
|
108 |
if (logger.isDebugEnabled()){
|
|
109 |
logger.info("Started new transaction. Fetching some " + parentPluralString + " (max: " + limit + ") ...");
|
|
110 |
logger.debug("Start snapshot, before starting loop");
|
|
111 |
ProfilerController.memorySnapshot();
|
|
112 |
}
|
|
113 |
|
|
114 |
List<String> propPath = Arrays.asList(new String[]{"markers.*"});
|
|
115 |
int partitionCount = 0;
|
|
116 |
while ((taxonList = getNextTaxonPartition(TaxonBase.class, limit, partitionCount++, propPath )) != null ) {
|
|
117 |
|
|
118 |
if (logger.isDebugEnabled()) {
|
|
119 |
logger.info("Fetched " + taxonList.size() + " " + parentPluralString + ". Exporting...");
|
|
120 |
}
|
|
121 |
|
|
122 |
for (TaxonBase<?> taxon : taxonList) {
|
|
123 |
doCount(count++, modCount, pluralString);
|
|
124 |
state.setCurrentTaxon(taxon);
|
|
125 |
if (!taxon.getMarkers().isEmpty()){
|
|
126 |
success &= handleSingleEcologyTaxon(taxon, mapping);
|
|
127 |
}
|
|
128 |
if (!taxon.getExtensions().isEmpty()){
|
|
129 |
success &= handleSingleLinkTaxon(taxon, urlMapping);
|
|
130 |
}
|
|
131 |
}
|
|
132 |
taxonList = null;
|
|
133 |
state.setCurrentTaxon(null);
|
|
134 |
|
|
135 |
// Commit transaction
|
|
136 |
commitTransaction(txStatus);
|
|
137 |
logger.info("Exported " + (count - pastCount) + " " + parentPluralString + ". Total taxa: " + count + ". Ecology notes: " + countNotes + ". Link notes: " + countUrls);
|
|
138 |
pastCount = count;
|
|
139 |
if (logger.isDebugEnabled()) {
|
|
140 |
ProfilerController.memorySnapshot();
|
|
141 |
}
|
|
142 |
// Start transaction
|
|
143 |
txStatus = startTransaction(true);
|
|
144 |
if(logger.isDebugEnabled()) {
|
|
145 |
logger.info("Started new transaction. Fetching some " + pluralString + " (max: " + limit + ") for description import ...");
|
|
146 |
}
|
|
147 |
}
|
|
148 |
|
|
149 |
|
|
150 |
// Commit transaction
|
|
151 |
commitTransaction(txStatus);
|
|
152 |
logger.debug("Committed transaction.");
|
|
153 |
return success;
|
|
154 |
}
|
|
155 |
|
|
156 |
private boolean handleSingleEcologyTaxon(TaxonBase<?> taxon, PesiExportMapping mapping) {
|
|
157 |
|
|
158 |
boolean success = true;
|
|
159 |
String ecologyStr = getEcologyString(taxon);
|
|
160 |
if (isNotBlank(ecologyStr)){
|
|
161 |
success &= mapping.invoke(taxon);
|
|
162 |
countNotes++;
|
|
163 |
}
|
|
164 |
return success;
|
|
165 |
}
|
|
166 |
|
|
167 |
private static String getEcologyString(TaxonBase<?> taxon) {
|
|
168 |
String ecologyStr = null;
|
|
169 |
ecologyStr = CdmUtils.concat(", ", ecologyStr, createEcologyStr(taxon, "marine", ErmsTransformer.uuidMarkerMarine));
|
|
170 |
ecologyStr = CdmUtils.concat(", ", ecologyStr, createEcologyStr(taxon, "brackish", ErmsTransformer.uuidMarkerBrackish));
|
|
171 |
ecologyStr = CdmUtils.concat(", ", ecologyStr, createEcologyStr(taxon, "fresh", ErmsTransformer.uuidMarkerFreshwater));
|
|
172 |
ecologyStr = CdmUtils.concat(", ", ecologyStr, createEcologyStr(taxon, "terrestrial", ErmsTransformer.uuidMarkerTerrestrial));
|
|
173 |
return ecologyStr;
|
|
174 |
}
|
|
175 |
|
|
176 |
private static String createEcologyStr(TaxonBase<?> taxon, String strEcology, UUID markerUuid) {
|
|
177 |
Boolean value = taxon.markerValue(markerUuid);
|
|
178 |
if (value == null){
|
|
179 |
return null;
|
|
180 |
}else if (value == true){
|
|
181 |
return strEcology;
|
|
182 |
}else{
|
|
183 |
return "not " + strEcology;
|
|
184 |
}
|
|
185 |
}
|
|
186 |
|
|
187 |
private boolean handleSingleLinkTaxon(TaxonBase<?> taxon, PesiExportMapping mapping) {
|
|
188 |
|
|
189 |
boolean success = true;
|
|
190 |
Set<Extension> urlExtensions = taxon.getFilteredExtensions(ErmsTransformer.uuidExtErmsLink);
|
|
191 |
for (Extension extension : urlExtensions){
|
|
192 |
mapping.invoke(extension);
|
|
193 |
countUrls++;
|
|
194 |
}
|
|
195 |
return success;
|
|
196 |
}
|
|
197 |
|
|
198 |
protected boolean doDelete(PesiExportState state) {
|
|
199 |
//Note table is already filled by Description import
|
|
200 |
//=> we do not empty any table here
|
|
201 |
return true;
|
|
202 |
}
|
|
203 |
|
|
204 |
@SuppressWarnings("unused") //used by mapper
|
|
205 |
private static Integer getTaxonFk(TaxonBase<?> taxonBase, PesiExportState state) {
|
|
206 |
return state.getDbId(taxonBase);
|
|
207 |
}
|
|
208 |
|
|
209 |
@SuppressWarnings("unused") //used by mapper
|
|
210 |
private static Integer getCurrentTaxonFk(Extension extension, PesiExportState state) {
|
|
211 |
return state.getDbId(state.getCurrentTaxon());
|
|
212 |
}
|
|
213 |
|
|
214 |
@SuppressWarnings("unused") //used by mapper
|
|
215 |
private static String getNote_1(TaxonBase<?> taxon) {
|
|
216 |
return getEcologyString(taxon);
|
|
217 |
}
|
|
218 |
|
|
219 |
@SuppressWarnings("unused") //used by mapper
|
|
220 |
private static String getUrlNote_1(Extension extension, PesiExportState state) {
|
|
221 |
String value = extension.getValue();
|
|
222 |
if (value == null){
|
|
223 |
return null;
|
|
224 |
}else{
|
|
225 |
//TODO use regex grouping
|
|
226 |
String result = value.split(ErmsLinkImport.TOKEN_LINKTEXT)[0];
|
|
227 |
result = result.trim();
|
|
228 |
return result;
|
|
229 |
}
|
|
230 |
}
|
|
231 |
|
|
232 |
@SuppressWarnings("unused") //used by mapper
|
|
233 |
private static String getUrlNote_2(Extension extension, PesiExportState state) {
|
|
234 |
String value = extension.getValue();
|
|
235 |
if (value == null){
|
|
236 |
return null;
|
|
237 |
}else{
|
|
238 |
//TODO use regex grouping
|
|
239 |
String result = value.split(ErmsLinkImport.TOKEN_LINKTEXT)[1];
|
|
240 |
result = result.trim();
|
|
241 |
return result;
|
|
242 |
}
|
|
243 |
}
|
|
244 |
|
|
245 |
|
|
246 |
@SuppressWarnings("unused") //used by mapper
|
|
247 |
private static Integer getNoteCategoryFk(Extension extension, PesiExportState state) {
|
|
248 |
String linktext = getUrlNote_2(extension, state);
|
|
249 |
int catFk = categoryByLinkText(linktext);
|
|
250 |
return catFk;
|
|
251 |
}
|
|
252 |
|
|
253 |
private static int categoryByLinkText(String linktext) {
|
|
254 |
if (linktext == null){
|
|
255 |
return PesiTransformer.NoteCategory_undefined_link;
|
|
256 |
}else if (linktext.matches("(to fishbase|marine life inf).*")){
|
|
257 |
return PesiTransformer.NoteCategory_Link_to_general_information;
|
|
258 |
}else if (linktext.matches(".*(clemam|nemys|algaebase|fishbase).*")){
|
|
259 |
return PesiTransformer.NoteCategory_Link_to_taxonomy;
|
|
260 |
}else{
|
|
261 |
return PesiTransformer.NoteCategory_undefined_link;
|
|
262 |
}
|
|
263 |
}
|
|
264 |
|
|
265 |
@SuppressWarnings("unused") //used by mapper
|
|
266 |
private static String getNoteCategoryCache(Extension extension, PesiExportState state) {
|
|
267 |
int catFk = getNoteCategoryFk(extension, state);
|
|
268 |
String result = categoryByLinkText(catFk);
|
|
269 |
return result;
|
|
270 |
}
|
|
271 |
|
|
272 |
//******************************* MAPPINGS ********************************************
|
|
273 |
|
|
274 |
/**
|
|
275 |
* @param catFk
|
|
276 |
* @return
|
|
277 |
*/
|
|
278 |
private static String categoryByLinkText(int catFk) {
|
|
279 |
if(catFk == 22){
|
|
280 |
return "";
|
|
281 |
}else if (catFk == 23){
|
|
282 |
return "";
|
|
283 |
}else if (catFk == 24){
|
|
284 |
return "";
|
|
285 |
}else{
|
|
286 |
logger.warn("Link category fk not yet supported: " + catFk);
|
|
287 |
return null;
|
|
288 |
}
|
|
289 |
}
|
|
290 |
|
|
291 |
private PesiExportMapping getMapping() {
|
|
292 |
PesiExportMapping mapping = new PesiExportMapping(dbTableName);
|
|
293 |
|
|
294 |
mapping.addMapper(idIncMapper);
|
|
295 |
mapping.addMapper(MethodMapper.NewInstance("Note_1", this, standardMethodParameter));
|
|
296 |
|
|
297 |
mapping.addMapper(DbFixedIntegerMapper.NewInstance(PesiTransformer.NoteCategory_ecology, "NoteCategoryFk"));
|
|
298 |
mapping.addMapper(DbFixedStringMapper.NewInstance("ecology", "NoteCategoryCache"));
|
|
299 |
mapping.addMapper(DbFixedIntegerMapper.NewInstance(12, "LanguageFk"));
|
|
300 |
mapping.addMapper(DbFixedStringMapper.NewInstance("English", "LanguageCache"));
|
|
301 |
|
|
302 |
mapping.addMapper(MethodMapper.NewInstance("TaxonFk", this, standardMethodParameter, PesiExportState.class));
|
|
303 |
|
|
304 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("Note_2", "Note_2 not used for ecology fact"));
|
|
305 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("Region", "Region not used for ecology fact"));
|
|
306 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("SpeciesExpertGUID", "SpeciesExpertGUID not used for ecology fact"));
|
|
307 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("SpeciesExpertName", "SpeciesExpertName not used for ecology fact"));
|
|
308 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("LastAction", "LastAction not used for ecology fact"));
|
|
309 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("LastActionDate", "LastActionDate not used for ecology fact"));
|
|
310 |
|
|
311 |
return mapping;
|
|
312 |
}
|
|
313 |
private IdIncMapper idIncMapper = IdIncMapper.NewComputedInstance("NoteId");
|
|
314 |
private PesiExportMapping getUrlMapping() {
|
|
315 |
PesiExportMapping mapping = new PesiExportMapping(dbTableName);
|
|
316 |
|
|
317 |
mapping.addMapper(idIncMapper);
|
|
318 |
mapping.addMapper(MethodMapper.NewInstance("Note_1", this.getClass(), "getUrlNote_1", Extension.class));
|
|
319 |
mapping.addMapper(MethodMapper.NewInstance("Note_2", this.getClass(), "getUrlNote_2", Extension.class));
|
|
320 |
mapping.addMapper(MethodMapper.NewInstance("NoteCategoryFk", this.getClass(), Extension.class));
|
|
321 |
mapping.addMapper(MethodMapper.NewInstance("NoteCategoryCache", this.getClass(), Extension.class));
|
|
322 |
mapping.addMapper(MethodMapper.NewInstance("TaxonFk", this.getClass(), "getCurrentTaxonFk", Extension.class, PesiExportState.class));
|
|
323 |
|
|
324 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("LanguageFk", "LanguageFk not used for link fact"));
|
|
325 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("LanguageCache", "LanguageCache not used for link fact"));
|
|
326 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("Region", "Region not used for link fact"));
|
|
327 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("SpeciesExpertGUID", "SpeciesExpertGUID not used for link fact"));
|
|
328 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("SpeciesExpertName", "SpeciesExpertName not used for link fact"));
|
|
329 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("LastAction", "LastAction not used for link fact"));
|
|
330 |
mapping.addMapper(DbExportIgnoreMapper.NewInstance("LastActionDate", "LastActionDate not used for link fact"));
|
|
331 |
|
|
332 |
return mapping;
|
|
333 |
}
|
|
334 |
|
|
335 |
@Override
|
|
336 |
protected boolean doCheck(PesiExportState state) {
|
|
337 |
boolean result = true;
|
|
338 |
return result;
|
|
339 |
}
|
|
340 |
|
|
341 |
@Override
|
|
342 |
protected boolean isIgnore(PesiExportState state) {
|
|
343 |
return ! state.getConfig().isDoEcology();
|
|
344 |
}
|
|
345 |
}
|
ref #1444, ref #8508, ref #8509 fix link handling and add ecology export to ERMS pipeline