Revision c95e3daf
Added by Patrick Plitzner almost 9 years ago
| cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/specimen/abcd206/in/Abcd206Import.java | ||
|---|---|---|
| 9 | 9 |
|
| 10 | 10 |
package eu.etaxonomy.cdm.io.specimen.abcd206.in; |
| 11 | 11 |
|
| 12 |
import java.io.InputStream; |
|
| 13 | 12 |
import java.net.MalformedURLException; |
| 14 | 13 |
import java.util.ArrayList; |
| 15 | 14 |
import java.util.HashMap; |
| ... | ... | |
| 19 | 18 |
import java.util.Set; |
| 20 | 19 |
import java.util.UUID; |
| 21 | 20 |
|
| 22 |
import javax.xml.parsers.DocumentBuilder; |
|
| 23 |
import javax.xml.parsers.DocumentBuilderFactory; |
|
| 24 |
|
|
| 25 | 21 |
import org.apache.commons.lang.StringUtils; |
| 26 | 22 |
import org.apache.log4j.Logger; |
| 27 | 23 |
import org.springframework.stereotype.Component; |
| 28 |
import org.w3c.dom.Document; |
|
| 29 | 24 |
import org.w3c.dom.Element; |
| 30 | 25 |
import org.w3c.dom.NodeList; |
| 31 | 26 |
|
| ... | ... | |
| 214 | 209 |
} |
| 215 | 210 |
} |
| 216 | 211 |
|
| 217 |
InputStream source = state.getConfig().getSource();
|
|
| 218 |
NodeList unitsList = getUnitsNodeList(source);
|
|
| 212 |
NodeList unitsList = AbcdParseUtility.getUnitsNodeList(state);
|
|
| 213 |
prefix = AbcdParseUtility.getPrefix(state);
|
|
| 219 | 214 |
|
| 220 | 215 |
if (unitsList != null) {
|
| 221 | 216 |
String message = "nb units to insert: " + unitsList.getLength(); |
| ... | ... | |
| 299 | 294 |
return; |
| 300 | 295 |
} |
| 301 | 296 |
|
| 302 |
/** |
|
| 303 |
* Return the list of root nodes for an ABCD 2.06 XML file |
|
| 304 |
* @param fileName: the file's location |
|
| 305 |
* @return the list of root nodes ("Unit")
|
|
| 306 |
*/ |
|
| 307 |
protected NodeList getUnitsNodeList(InputStream inputStream) {
|
|
| 308 |
NodeList unitList = null; |
|
| 309 |
try {
|
|
| 310 |
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance(); |
|
| 311 |
DocumentBuilder builder = factory.newDocumentBuilder(); |
|
| 312 |
|
|
| 313 |
Document document = builder.parse(inputStream); |
|
| 314 |
Element root = document.getDocumentElement(); |
|
| 315 |
unitList = root.getElementsByTagName("Unit");
|
|
| 316 |
if (unitList.getLength() == 0) {
|
|
| 317 |
unitList = root.getElementsByTagName("abcd:Unit");
|
|
| 318 |
prefix = "abcd:"; |
|
| 319 |
} |
|
| 320 |
if (unitList.getLength() == 0) {
|
|
| 321 |
unitList = root.getElementsByTagName("abcd21:Unit");
|
|
| 322 |
prefix = "abcd21:"; |
|
| 323 |
} |
|
| 324 |
} catch (Exception e) {
|
|
| 325 |
logger.warn(e); |
|
| 326 |
} |
|
| 327 |
return unitList; |
|
| 328 |
} |
|
| 329 |
|
|
| 330 | 297 |
/** |
| 331 | 298 |
* Handle a single unit |
| 332 | 299 |
* @param state |
| cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/specimen/abcd206/in/AbcdParseUtility.java | ||
|---|---|---|
| 9 | 9 |
*/ |
| 10 | 10 |
package eu.etaxonomy.cdm.io.specimen.abcd206.in; |
| 11 | 11 |
|
| 12 |
import java.io.InputStream; |
|
| 12 | 13 |
import java.net.URI; |
| 13 | 14 |
import java.util.Date; |
| 14 | 15 |
|
| 16 |
import javax.xml.parsers.DocumentBuilder; |
|
| 17 |
import javax.xml.parsers.DocumentBuilderFactory; |
|
| 18 |
|
|
| 15 | 19 |
import org.apache.log4j.Logger; |
| 16 | 20 |
import org.joda.time.DateTime; |
| 21 |
import org.w3c.dom.Document; |
|
| 22 |
import org.w3c.dom.Element; |
|
| 17 | 23 |
import org.w3c.dom.Node; |
| 18 | 24 |
import org.w3c.dom.NodeList; |
| 19 | 25 |
|
| ... | ... | |
| 96 | 102 |
return date; |
| 97 | 103 |
} |
| 98 | 104 |
|
| 105 |
/** |
|
| 106 |
* Return the list of root nodes for an ABCD XML file |
|
| 107 |
* @param fileName: the file's location |
|
| 108 |
* @return the list of root nodes ("Unit")
|
|
| 109 |
*/ |
|
| 110 |
public static NodeList getUnitsNodeList(Abcd206ImportState state) {
|
|
| 111 |
InputStream inputStream = state.getConfig().getSource(); |
|
| 112 |
NodeList unitList = null; |
|
| 113 |
try {
|
|
| 114 |
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance(); |
|
| 115 |
DocumentBuilder builder = factory.newDocumentBuilder(); |
|
| 116 |
|
|
| 117 |
Document document = builder.parse(inputStream); |
|
| 118 |
Element root = document.getDocumentElement(); |
|
| 119 |
unitList = root.getElementsByTagName("Unit");
|
|
| 120 |
if (unitList.getLength() == 0) {
|
|
| 121 |
unitList = root.getElementsByTagName("abcd:Unit");
|
|
| 122 |
} |
|
| 123 |
if (unitList.getLength() == 0) {
|
|
| 124 |
unitList = root.getElementsByTagName("abcd21:Unit");
|
|
| 125 |
} |
|
| 126 |
} catch (Exception e) {
|
|
| 127 |
logger.warn(e); |
|
| 128 |
} |
|
| 129 |
return unitList; |
|
| 130 |
} |
|
| 131 |
|
|
| 132 |
/** |
|
| 133 |
* Return the prefix an ABCD XML file |
|
| 134 |
* @param fileName: the file's location |
|
| 135 |
* @return the prefix |
|
| 136 |
*/ |
|
| 137 |
public static String getPrefix(Abcd206ImportState state) {
|
|
| 138 |
InputStream inputStream = state.getConfig().getSource(); |
|
| 139 |
NodeList unitList = null; |
|
| 140 |
try {
|
|
| 141 |
DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance(); |
|
| 142 |
DocumentBuilder builder = factory.newDocumentBuilder(); |
|
| 143 |
|
|
| 144 |
Document document = builder.parse(inputStream); |
|
| 145 |
Element root = document.getDocumentElement(); |
|
| 146 |
unitList = root.getElementsByTagName("Unit");
|
|
| 147 |
if (unitList.getLength()>0) {
|
|
| 148 |
return null; |
|
| 149 |
} |
|
| 150 |
unitList = root.getElementsByTagName("abcd:Unit");
|
|
| 151 |
if (unitList.getLength()>0) {
|
|
| 152 |
return "abcd:"; |
|
| 153 |
} |
|
| 154 |
unitList = root.getElementsByTagName("abcd21:Unit");
|
|
| 155 |
if (unitList.getLength() == 0) {
|
|
| 156 |
return "abcd21:"; |
|
| 157 |
} |
|
| 158 |
} catch (Exception e) {
|
|
| 159 |
logger.warn(e); |
|
| 160 |
} |
|
| 161 |
return null; |
|
| 162 |
} |
|
| 163 |
|
|
| 99 | 164 |
} |
| cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/specimen/abcd206/in/UnitAssociationParser.java | ||
|---|---|---|
| 9 | 9 |
*/ |
| 10 | 10 |
package eu.etaxonomy.cdm.io.specimen.abcd206.in; |
| 11 | 11 |
|
| 12 |
import java.util.List; |
|
| 12 |
import java.io.IOException; |
|
| 13 |
import java.net.URI; |
|
| 13 | 14 |
|
| 15 |
import org.apache.http.client.ClientProtocolException; |
|
| 16 |
import org.apache.log4j.Logger; |
|
| 14 | 17 |
import org.w3c.dom.Element; |
| 15 | 18 |
import org.w3c.dom.NodeList; |
| 16 | 19 |
|
| 17 | 20 |
import eu.etaxonomy.cdm.api.application.ICdmApplicationConfiguration; |
| 18 |
import eu.etaxonomy.cdm.model.agent.AgentBase; |
|
| 19 |
import eu.etaxonomy.cdm.model.agent.Institution; |
|
| 20 |
import eu.etaxonomy.cdm.model.occurrence.Collection; |
|
| 21 |
import eu.etaxonomy.cdm.persistence.query.MatchMode; |
|
| 21 |
import eu.etaxonomy.cdm.ext.occurrence.OccurenceQuery; |
|
| 22 |
import eu.etaxonomy.cdm.ext.occurrence.bioCase.BioCaseQueryServiceWrapper; |
|
| 22 | 23 |
|
| 23 | 24 |
/** |
| 24 | 25 |
* @author pplitzner |
| ... | ... | |
| 27 | 28 |
*/ |
| 28 | 29 |
public class UnitAssociationParser {
|
| 29 | 30 |
|
| 31 |
private static final Logger logger = Logger.getLogger(UnitAssociationParser.class); |
|
| 32 |
|
|
| 30 | 33 |
private final String prefix; |
| 31 | 34 |
|
| 32 | 35 |
private final Abcd206ImportReport report; |
| ... | ... | |
| 47 | 50 |
NodeList unitAssociationList = associations.getElementsByTagName(prefix+"UnitAssociation"); |
| 48 | 51 |
|
| 49 | 52 |
for(int i=0;i<unitAssociationList.getLength();i++){
|
| 50 |
// BioCaseQueryServiceWrapper serviceWrapper = new BioCaseQueryServiceWrapper(); |
|
| 51 |
} |
|
| 52 |
|
|
| 53 |
|
|
| 54 | 53 |
|
| 55 |
|
|
| 56 |
//FIXME: how to handle multiple unit assocations? |
|
| 57 |
// maybe check AssociationType but this needs to be stable |
|
| 58 |
// for only the first unitAssociation will be used |
|
| 59 |
if(unitAssociationList.getLength()>0 && unitAssociationList.item(0) instanceof Element){
|
|
| 60 | 54 |
Element unitAssociation = (Element)unitAssociationList.item(0); |
| 61 |
String collectionName = AbcdParseUtility.parseFirstTextContent(unitAssociation.getElementsByTagName(prefix+"SourceName")); |
|
| 62 |
List<Collection> matchingCollections = cdmAppController.getCollectionService().findByTitle(Collection.class, collectionName, MatchMode.EXACT, null, null, null, null, null).getRecords(); |
|
| 63 |
Collection collection; |
|
| 64 |
if(matchingCollections.size()==1){
|
|
| 65 |
collection = matchingCollections.iterator().next(); |
|
| 66 |
} |
|
| 67 |
else{
|
|
| 68 |
collection = Collection.NewInstance(); |
|
| 69 |
collection.setName(collectionName); |
|
| 55 |
|
|
| 56 |
//unit id |
|
| 57 |
String unitId = AbcdParseUtility.parseFirstTextContent(unitAssociation.getElementsByTagName(prefix+"UnitID")); |
|
| 58 |
//data access point |
|
| 59 |
URI datasetAccessPoint = AbcdParseUtility.parseFirstUri(unitAssociation.getElementsByTagName(prefix+"DatasetAccessPoint")); |
|
| 60 |
if(datasetAccessPoint==null){
|
|
| 61 |
datasetAccessPoint = AbcdParseUtility.parseFirstUri(unitAssociation.getElementsByTagName(prefix+"Comment")); |
|
| 70 | 62 |
} |
| 71 |
String institutionName = AbcdParseUtility.parseFirstTextContent(unitAssociation.getElementsByTagName(prefix+"SourceInstitutionCode")); |
|
| 72 |
List<AgentBase> matchingInstitutions = cdmAppController.getAgentService().findByTitle(Institution.class, institutionName, MatchMode.EXACT, null, null, null, null, null).getRecords(); |
|
| 73 |
Institution institution; |
|
| 74 |
if(matchingInstitutions.size()==1){
|
|
| 75 |
institution = (Institution) matchingInstitutions.iterator().next(); |
|
| 63 |
String message = "Unable to load unit "+unitId+" from "+datasetAccessPoint; |
|
| 64 |
if(datasetAccessPoint!=null){
|
|
| 65 |
//association type |
|
| 66 |
NodeList associationTypeList = unitAssociation.getElementsByTagName(prefix+"AssociationType"); |
|
| 67 |
|
|
| 68 |
BioCaseQueryServiceWrapper serviceWrapper = new BioCaseQueryServiceWrapper(); |
|
| 69 |
OccurenceQuery query = new OccurenceQuery(unitId); |
|
| 70 |
try {
|
|
| 71 |
serviceWrapper.query(query, datasetAccessPoint); |
|
| 72 |
} catch (ClientProtocolException e) {
|
|
| 73 |
logger.error(message, e); |
|
| 74 |
} catch (IOException e) {
|
|
| 75 |
logger.error(message, e); |
|
| 76 |
} |
|
| 76 | 77 |
} |
| 77 | 78 |
else{
|
| 78 |
institution = Institution.NewInstance(); |
|
| 79 |
institution.setName(institutionName); |
|
| 79 |
report.addInfoMessage(message); |
|
| 80 | 80 |
} |
| 81 |
} |
|
| 81 | 82 |
|
| 82 |
String unitId = AbcdParseUtility.parseFirstTextContent(unitAssociation.getElementsByTagName(prefix+"UnitID")); |
|
| 83 |
NodeList associationTypeList = unitAssociation.getElementsByTagName(prefix+"AssociationType"); |
|
| 84 | 83 |
|
| 85 |
} |
|
| 84 |
|
|
| 85 |
|
|
| 86 |
// //FIXME: how to handle multiple unit assocations? |
|
| 87 |
// // maybe check AssociationType but this needs to be stable |
|
| 88 |
// // for only the first unitAssociation will be used |
|
| 89 |
// if(unitAssociationList.getLength()>0 && unitAssociationList.item(0) instanceof Element){
|
|
| 90 |
// String collectionName = AbcdParseUtility.parseFirstTextContent(unitAssociation.getElementsByTagName(prefix+"SourceName")); |
|
| 91 |
// List<Collection> matchingCollections = cdmAppController.getCollectionService().findByTitle(Collection.class, collectionName, MatchMode.EXACT, null, null, null, null, null).getRecords(); |
|
| 92 |
// Collection collection; |
|
| 93 |
// if(matchingCollections.size()==1){
|
|
| 94 |
// collection = matchingCollections.iterator().next(); |
|
| 95 |
// } |
|
| 96 |
// else{
|
|
| 97 |
// collection = Collection.NewInstance(); |
|
| 98 |
// collection.setName(collectionName); |
|
| 99 |
// } |
|
| 100 |
// String institutionName = AbcdParseUtility.parseFirstTextContent(unitAssociation.getElementsByTagName(prefix+"SourceInstitutionCode")); |
|
| 101 |
// List<AgentBase> matchingInstitutions = cdmAppController.getAgentService().findByTitle(Institution.class, institutionName, MatchMode.EXACT, null, null, null, null, null).getRecords(); |
|
| 102 |
// Institution institution; |
|
| 103 |
// if(matchingInstitutions.size()==1){
|
|
| 104 |
// institution = (Institution) matchingInstitutions.iterator().next(); |
|
| 105 |
// } |
|
| 106 |
// else{
|
|
| 107 |
// institution = Institution.NewInstance(); |
|
| 108 |
// institution.setName(institutionName); |
|
| 109 |
// } |
|
| 110 |
// } |
|
| 86 | 111 |
} |
| 87 | 112 |
} |
| 88 | 113 |
|
Also available in: Unified diff
Move unit parsing to utility class