Revision 892efc69
Added by Andreas Kohlbecker almost 14 years ago
| app-import/src/main/java/eu/etaxonomy/cdm/io/CichorieaeImageImport.java | ||
|---|---|---|
| 10 | 10 |
package eu.etaxonomy.cdm.io; |
| 11 | 11 |
|
| 12 | 12 |
import java.io.File; |
| 13 |
import java.io.FileNotFoundException; |
|
| 14 | 13 |
import java.io.IOException; |
| 15 | 14 |
import java.net.MalformedURLException; |
| 16 | 15 |
import java.net.URI; |
| 17 | 16 |
import java.net.URL; |
| 18 |
import java.util.ArrayList; |
|
| 19 |
import java.util.HashMap; |
|
| 20 | 17 |
import java.util.List; |
| 21 |
import java.util.Set; |
|
| 22 | 18 |
import java.util.UUID; |
| 23 | 19 |
|
| 24 | 20 |
import org.apache.log4j.Logger; |
| 25 |
import org.apache.sanselan.ImageInfo; |
|
| 26 |
import org.apache.sanselan.ImageReadException; |
|
| 27 |
import org.apache.sanselan.Sanselan; |
|
| 21 |
import org.joda.time.DateTime; |
|
| 28 | 22 |
import org.springframework.stereotype.Component; |
| 29 | 23 |
|
| 30 |
import eu.etaxonomy.cdm.api.service.INameService; |
|
| 31 | 24 |
import eu.etaxonomy.cdm.app.images.AbstractImageImporter; |
| 32 | 25 |
import eu.etaxonomy.cdm.app.images.ImageImportConfigurator; |
| 33 | 26 |
import eu.etaxonomy.cdm.common.CdmUtils; |
| 34 |
import eu.etaxonomy.cdm.common.ExcelUtils; |
|
| 35 | 27 |
import eu.etaxonomy.cdm.common.mediaMetaData.ImageMetaData; |
| 36 |
import eu.etaxonomy.cdm.common.mediaMetaData.MetaDataFactory; |
|
| 37 |
import eu.etaxonomy.cdm.common.mediaMetaData.MimeType; |
|
| 28 |
import eu.etaxonomy.cdm.model.agent.AgentBase; |
|
| 38 | 29 |
import eu.etaxonomy.cdm.model.common.Language; |
| 39 |
import eu.etaxonomy.cdm.model.description.Feature;
|
|
| 30 |
import eu.etaxonomy.cdm.model.common.LanguageString;
|
|
| 40 | 31 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
| 41 | 32 |
import eu.etaxonomy.cdm.model.description.TextData; |
| 42 | 33 |
import eu.etaxonomy.cdm.model.media.ImageFile; |
| 43 | 34 |
import eu.etaxonomy.cdm.model.media.Media; |
| 44 |
import eu.etaxonomy.cdm.model.media.MediaRepresentation; |
|
| 45 |
import eu.etaxonomy.cdm.model.name.TaxonNameBase; |
|
| 46 | 35 |
import eu.etaxonomy.cdm.model.reference.ReferenceBase; |
| 47 | 36 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
| 48 | 37 |
import eu.etaxonomy.cdm.model.taxon.TaxonBase; |
| ... | ... | |
| 57 | 46 |
public class CichorieaeImageImport extends AbstractImageImporter {
|
| 58 | 47 |
private static final Logger logger = Logger.getLogger(CichorieaeImageImport.class); |
| 59 | 48 |
|
| 60 |
private static final String URL = "URL"; |
|
| 61 |
private static final String URL_APP = "URL_APP"; |
|
| 62 |
private static final String NAME = "NAME"; |
|
| 63 |
//private static final String CODE = "CODE"; |
|
| 64 |
|
|
| 65 |
// /** |
|
| 66 |
// * Imports images from an Excel file. |
|
| 67 |
// */ |
|
| 68 |
// protected boolean invokeImageImport_ (ImageImportConfigurator config){
|
|
| 69 |
// |
|
| 70 |
// ArrayList<HashMap<String, String>> contents; |
|
| 71 |
// try {
|
|
| 72 |
// contents = ExcelUtils.parseXLS(config.getSource().toString()); |
|
| 73 |
// } catch (FileNotFoundException e1) {
|
|
| 74 |
// logger.error("FileNotFound: " + config.getSource().toString());
|
|
| 75 |
// return false; |
|
| 76 |
// } |
|
| 77 |
// |
|
| 78 |
// for (HashMap<String, String> row : contents){
|
|
| 79 |
// |
|
| 80 |
// String taxonName = row.get(CichorieaeImageImport.NAME).trim(); |
|
| 81 |
// |
|
| 82 |
// INameService nameService = getNameService(); |
|
| 83 |
// List<TaxonBase> taxa = taxonService.searchTaxaByName(taxonName, config.getSourceReference()); |
|
| 84 |
// |
|
| 85 |
// if(taxa.size() == 0){
|
|
| 86 |
// logger.warn("no taxon with this name found: " + taxonName);
|
|
| 87 |
// }else if(taxa.size() > 1){
|
|
| 88 |
// logger.warn("multiple taxa with this name found: " + taxonName);
|
|
| 89 |
// }else{
|
|
| 90 |
// Taxon taxon = (Taxon) taxa.get(0); |
|
| 91 |
// |
|
| 92 |
// taxonService.saveTaxon(taxon); |
|
| 93 |
// |
|
| 94 |
// TextData feature = TextData.NewInstance(); |
|
| 95 |
// |
|
| 96 |
// logger.info("Importing image for taxon: " + taxa);
|
|
| 97 |
// |
|
| 98 |
// |
|
| 99 |
// ImageMetaData imageMetaData = new ImageMetaData(); |
|
| 100 |
// |
|
| 101 |
// |
|
| 102 |
// try {
|
|
| 103 |
// URL url = new URL(row.get(CichorieaeImageImport.URL).trim()); |
|
| 104 |
// |
|
| 105 |
// imageMetaData.readFrom(url); |
|
| 106 |
// |
|
| 107 |
// ImageFile image = ImageFile.NewInstance(url.toString(), null, imageMetaData); |
|
| 108 |
// |
|
| 109 |
// MediaRepresentation representation = MediaRepresentation.NewInstance(imageMetaData.getMimeType(), null); |
|
| 110 |
// representation.addRepresentationPart(image); |
|
| 111 |
// |
|
| 112 |
// Media media = Media.NewInstance(); |
|
| 113 |
// media.addRepresentation(representation); |
|
| 114 |
// |
|
| 115 |
// feature.addMedia(media); |
|
| 116 |
// |
|
| 117 |
// feature.setType(Feature.IMAGE()); |
|
| 118 |
// |
|
| 119 |
// TaxonDescription description = TaxonDescription.NewInstance(taxon); |
|
| 120 |
// |
|
| 121 |
// description.addElement(feature); |
|
| 122 |
// |
|
| 123 |
// } catch (MalformedURLException e) {
|
|
| 124 |
// logger.error("Malformed URL", e);
|
|
| 125 |
// } |
|
| 126 |
// |
|
| 127 |
// } |
|
| 128 |
// } |
|
| 129 |
// return true; |
|
| 130 |
// |
|
| 131 |
// } |
|
| 132 | 49 |
|
| 133 | 50 |
/** |
| 134 | 51 |
* Imports images from a directory. |
| ... | ... | |
| 142 | 59 |
if (source.isDirectory()){
|
| 143 | 60 |
for (File file : source.listFiles() ){
|
| 144 | 61 |
if (file.isFile()){
|
| 145 |
String name = file.getName(); |
|
| 146 |
String[] fileNameParts = name.split("\\.");
|
|
| 147 |
if (fileNameParts.length < 2){
|
|
| 148 |
logger.warn("No file extension found for: " + name);
|
|
| 149 |
continue; |
|
| 150 |
} |
|
| 151 |
String extension = fileNameParts[fileNameParts.length - 1]; |
|
| 152 |
if (! "jpg".equalsIgnoreCase(extension)) {
|
|
| 153 |
logger.warn("Extension not recognized: " + extension);
|
|
| 154 |
// Sometimes occurs here "Thumbs.db" |
|
| 62 |
String fileName = file.getName(); |
|
| 63 |
String taxonName = getTaxonName(fileName); |
|
| 64 |
if (taxonName == null){
|
|
| 155 | 65 |
continue; |
| 156 | 66 |
} |
| 157 |
String firstPart = name.substring(0, name.length() - extension.length() - 1); |
|
| 158 |
logger.info(firstPart); |
|
| 159 |
String[] nameParts = firstPart.split("_");
|
|
| 160 |
if (nameParts.length < 3){
|
|
| 161 |
logger.warn("name string has less than 2 '_'");
|
|
| 162 |
continue; |
|
| 163 |
} |
|
| 164 |
String featureString = nameParts[nameParts.length-2]; |
|
| 165 |
logger.debug("FeatureString: " + featureString);
|
|
| 166 |
String detailString = nameParts[nameParts.length-1]; |
|
| 167 |
logger.debug("detailString: " + detailString);
|
|
| 168 |
|
|
| 169 |
String taxonName = ""; |
|
| 170 |
for (int i= 0; i < nameParts.length-2; i++){
|
|
| 171 |
taxonName += nameParts[i] + " "; |
|
| 172 |
} |
|
| 173 |
taxonName = taxonName.trim(); |
|
| 174 |
logger.info("Taxon name: " + taxonName);
|
|
| 175 |
|
|
| 176 |
String _s_ = " s "; |
|
| 177 |
String subsp = " subsp. "; |
|
| 178 |
if (taxonName.contains(_s_)) {
|
|
| 179 |
taxonName = taxonName.replace(_s_, subsp); |
|
| 180 |
logger.info("Taxon name: " + taxonName);
|
|
| 181 |
} |
|
| 182 |
|
|
| 183 | 67 |
List<TaxonBase> taxa = taxonService.searchTaxaByName(taxonName, config.getSourceReference()); |
| 184 |
|
|
| 185 |
Taxon taxon = null; |
|
| 186 |
|
|
| 187 | 68 |
if(taxa.size() == 0){
|
| 188 | 69 |
logger.warn("no taxon with this name found: " + taxonName);
|
| 189 |
} |
|
| 190 |
|
|
| 191 |
else {
|
|
| 192 |
if(taxa.size() > 1) {
|
|
| 193 |
if (logger.isDebugEnabled()) {
|
|
| 194 |
logger.debug("multiple taxa with this name found: " + taxonName);
|
|
| 195 |
} |
|
| 196 |
for (TaxonBase taxonBase : taxa) {
|
|
| 197 |
Taxon tax = (Taxon)taxonBase; |
|
| 198 |
if (tree.isTaxonInTree(tax)) {
|
|
| 199 |
taxon = tax; |
|
| 200 |
break; |
|
| 201 |
} |
|
| 202 |
} |
|
| 203 |
|
|
| 204 |
} else {
|
|
| 205 |
taxon = (Taxon) taxa.get(0); |
|
| 206 |
} |
|
| 207 |
taxonService.saveOrUpdate(taxon); |
|
| 208 |
|
|
| 209 |
TextData textData = TextData.NewInstance(); |
|
| 210 |
logger.info("Importing image for taxon: " + taxa);
|
|
| 211 |
//ImageMetaData imageMetaData =ImageMetaData.newInstance(); |
|
| 212 |
MetaDataFactory metaDataFactory = MetaDataFactory.getInstance(); |
|
| 213 |
|
|
| 214 |
|
|
| 215 |
|
|
| 216 |
try {
|
|
| 217 |
String urlPrefix = "http://media.bgbm.org/erez/erez?src=EditWP6/photos/"; |
|
| 218 |
String urlString = urlPrefix + name; |
|
| 219 |
logger.info(urlString); |
|
| 220 |
URL url = new URL(urlString); |
|
| 221 |
|
|
| 222 |
URI uri = CdmUtils.string2Uri(urlString); |
|
| 223 |
|
|
| 224 |
//imageinfo = Sanselan.getImageInfo(file); |
|
| 225 |
ImageMetaData imageMetaData =ImageMetaData.newInstance(); |
|
| 226 |
imageMetaData.readImageInfo(uri, 0); |
|
| 227 |
//ImageMetaData imageMetaData = (ImageMetaData) metaDataFactory.readMediaData(uri, MimeType.IMAGE); |
|
| 228 |
ImageFile image = ImageFile.NewInstance(url.toString(), null, imageMetaData.getHeight(), imageMetaData.getWidth()); |
|
| 229 |
|
|
| 230 |
MediaRepresentation representation = MediaRepresentation.NewInstance(imageMetaData.getMimeType(), null); |
|
| 231 |
representation.addRepresentationPart(image); |
|
| 232 |
Media media = Media.NewInstance(); |
|
| 233 |
media.addRepresentation(representation); |
|
| 234 |
|
|
| 235 |
textData.addMedia(media); |
|
| 236 |
textData.putText(taxonName, Language.ENGLISH()); |
|
| 237 |
|
|
| 238 |
textData.setType(Feature.IMAGE()); |
|
| 239 |
TaxonDescription taxonDescription = taxon.getOrCreateImageGallery(sourceRef == null ? null :sourceRef.getTitleCache()); |
|
| 240 |
|
|
| 241 |
taxonDescription.addElement(textData); |
|
| 242 |
|
|
| 243 |
} catch (MalformedURLException e) {
|
|
| 244 |
logger.error("Malformed URL", e);
|
|
| 245 |
} catch (IOException e) {
|
|
| 246 |
// TODO Auto-generated catch block |
|
| 247 |
e.printStackTrace(); |
|
| 248 |
} |
|
| 70 |
} else {
|
|
| 71 |
handleTaxa(tree, sourceRef, fileName, taxonName, taxa); |
|
| 249 | 72 |
} |
| 250 | 73 |
}else{
|
| 251 | 74 |
logger.warn("File is not a file (but a directory?): " + file.getName());
|
| ... | ... | |
| 258 | 81 |
return true; |
| 259 | 82 |
|
| 260 | 83 |
} |
| 84 |
|
|
| 85 |
private String getTaxonName(String fileName){
|
|
| 86 |
String[] fileNameParts = fileName.split("\\.");
|
|
| 87 |
if (fileNameParts.length < 2){
|
|
| 88 |
logger.warn("No file extension found for: " + fileName);
|
|
| 89 |
return null; |
|
| 90 |
} |
|
| 91 |
String extension = fileNameParts[fileNameParts.length - 1]; |
|
| 92 |
if (! "jpg".equalsIgnoreCase(extension)) {
|
|
| 93 |
logger.warn("Extension not recognized: " + extension);
|
|
| 94 |
// Sometimes occurs here "Thumbs.db" |
|
| 95 |
return null; |
|
| 96 |
} |
|
| 97 |
String firstPart = fileName.substring(0, fileName.length() - extension.length() - 1); |
|
| 98 |
logger.info(firstPart); |
|
| 99 |
String[] nameParts = firstPart.split("_");
|
|
| 100 |
if (nameParts.length < 3){
|
|
| 101 |
logger.warn("name string has less than 2 '_'");
|
|
| 102 |
return null; |
|
| 103 |
} |
|
| 104 |
|
|
| 105 |
String featureString = nameParts[nameParts.length-2]; |
|
| 106 |
logger.debug("FeatureString: " + featureString);
|
|
| 107 |
String detailString = nameParts[nameParts.length-1]; |
|
| 108 |
logger.debug("detailString: " + detailString);
|
|
| 109 |
|
|
| 110 |
String taxonName = ""; |
|
| 111 |
for (int i= 0; i < nameParts.length-2; i++){
|
|
| 112 |
taxonName += nameParts[i] + " "; |
|
| 113 |
} |
|
| 114 |
taxonName = taxonName.trim(); |
|
| 115 |
logger.info("Taxon name: " + taxonName);
|
|
| 116 |
|
|
| 117 |
String _s_ = " s "; |
|
| 118 |
String subsp = " subsp. "; |
|
| 119 |
if (taxonName.contains(_s_)) {
|
|
| 120 |
taxonName = taxonName.replace(_s_, subsp); |
|
| 121 |
logger.info("Taxon name: " + taxonName);
|
|
| 122 |
} |
|
| 123 |
return taxonName; |
|
| 124 |
} |
|
| 125 |
|
|
| 126 |
|
|
| 127 |
/** |
|
| 128 |
* @param tree |
|
| 129 |
* @param sourceRef |
|
| 130 |
* @param name |
|
| 131 |
* @param taxonName |
|
| 132 |
* @param taxa |
|
| 133 |
* @param taxon |
|
| 134 |
*/ |
|
| 135 |
private void handleTaxa(TaxonomicTree tree, ReferenceBase sourceRef, String fileName, String taxonName, List<TaxonBase> taxa) {
|
|
| 136 |
|
|
| 137 |
Taxon taxon = getTaxon(tree, taxonName, taxa); |
|
| 138 |
TaxonDescription imageGallery = taxon.getOrCreateImageGallery(sourceRef == null ? null :sourceRef.getTitleCache()); |
|
| 139 |
TextData textData = imageGallery.getOrCreateImageTextData(); |
|
| 140 |
logger.info("Importing image for taxon: " + taxa);
|
|
| 141 |
try {
|
|
| 142 |
Media media = getMedia(fileName, taxonName); |
|
| 143 |
textData.addMedia(media); |
|
| 144 |
} catch (MalformedURLException e) {
|
|
| 145 |
logger.error("Malformed URL", e);
|
|
| 146 |
} catch (IOException e) {
|
|
| 147 |
logger.error("IOException when handling image url");
|
|
| 148 |
} |
|
| 149 |
} |
|
| 150 |
|
|
| 151 |
|
|
| 152 |
/** |
|
| 153 |
* @param fileName |
|
| 154 |
* @param taxonName |
|
| 155 |
* @return |
|
| 156 |
* @throws MalformedURLException |
|
| 157 |
* @throws IOException |
|
| 158 |
*/ |
|
| 159 |
private Media getMedia(String fileName, String taxonName) throws MalformedURLException, IOException {
|
|
| 160 |
String urlPrefix = "http://media.bgbm.org/erez/erez?src=EditWP6/photos/"; |
|
| 161 |
String urlString = urlPrefix + fileName; |
|
| 162 |
logger.info(urlString); |
|
| 163 |
URL url = new URL(urlString); |
|
| 164 |
URI uri = CdmUtils.string2Uri(urlString); |
|
| 165 |
ImageMetaData imageMetaData =ImageMetaData.newInstance(); |
|
| 166 |
imageMetaData.readImageInfo(uri, 0); |
|
| 167 |
|
|
| 168 |
//String uri = url.toString(); |
|
| 169 |
|
|
| 170 |
String uriString = url.toString(); |
|
| 171 |
String mimeType = imageMetaData.getMimeType(); |
|
| 172 |
String suffix = null; |
|
| 173 |
int height = imageMetaData.getHeight(); |
|
| 174 |
int width = imageMetaData.getWidth(); |
|
| 175 |
Integer size = null; |
|
| 176 |
DateTime mediaCreated = null; |
|
| 177 |
AgentBase artist = null; |
|
| 178 |
|
|
| 179 |
|
|
| 180 |
ImageFile image = ImageFile.NewInstance(uriString, size, height, width); |
|
| 181 |
Media media = ImageFile.NewMediaInstance(mediaCreated, artist, uriString, mimeType, suffix, size, height, width); |
|
| 182 |
media.addTitle(LanguageString.NewInstance(taxonName, Language.LATIN())); |
|
| 183 |
|
|
| 184 |
return media; |
|
| 185 |
} |
|
| 186 |
|
|
| 187 |
/** |
|
| 188 |
* @param tree |
|
| 189 |
* @param taxonName |
|
| 190 |
* @param taxa |
|
| 191 |
* @return |
|
| 192 |
*/ |
|
| 193 |
private Taxon getTaxon(TaxonomicTree tree, String taxonName, |
|
| 194 |
List<TaxonBase> taxa) {
|
|
| 195 |
Taxon taxon = null; |
|
| 196 |
if(taxa.size() > 1) {
|
|
| 197 |
if (logger.isDebugEnabled()) {
|
|
| 198 |
logger.debug("multiple taxa with this name found: " + taxonName);
|
|
| 199 |
} |
|
| 200 |
for (TaxonBase taxonBase : taxa) {
|
|
| 201 |
Taxon tax = (Taxon)taxonBase; |
|
| 202 |
if (tree.isTaxonInTree(tax)) {
|
|
| 203 |
taxon = tax; |
|
| 204 |
break; |
|
| 205 |
} |
|
| 206 |
} |
|
| 207 |
if (taxon == null){
|
|
| 208 |
taxon = (Taxon)taxa.get(0); |
|
| 209 |
logger.warn("Taxon not found in preferred tree. Use " + taxon.getTitleCache() + " instead.");
|
|
| 210 |
} |
|
| 211 |
|
|
| 212 |
} else {
|
|
| 213 |
taxon = (Taxon) taxa.get(0); |
|
| 214 |
} |
|
| 215 |
if (taxon != null){
|
|
| 216 |
taxonService.saveOrUpdate(taxon); |
|
| 217 |
}else{
|
|
| 218 |
logger.warn("Taxon was null. Did not save taxon");
|
|
| 219 |
} |
|
| 220 |
return taxon; |
|
| 221 |
} |
|
| 261 | 222 |
} |
Also available in: Unified diff
merging /branches/cdmlib/SPRINT-Chichorieae1/ to trunk