private boolean nameDatasetRegExp(InvDatasetImpl dataset) {
boolean isMatch;
Matcher matcher;
// Test for a match on the urlPath
if (dataset.getUrlPath() != null) {
logger.debug("nameDatasetRegExp(): try naming on urlPath <{}>", dataset.getUrlPath());
matcher = this.regExpPattern.matcher(dataset.getUrlPath());
isMatch = matcher.find();
} else {
matcher = this.regExpPattern.matcher(dataset.getName());
isMatch = matcher.find();
}
if (isMatch) {
// Test for substitution.
StringBuffer resultingName = new StringBuffer();
matcher.appendReplacement(resultingName, this.substitutePattern);
resultingName.delete(0, matcher.start());
if (resultingName.length() != 0) {
logger.debug("nameDatasetRegExp(): Setting name to \"" + resultingName + "\".");
dataset.setName(resultingName.toString());
return true;
} else {
logger.debug("nameDatasetRegExp(): No name for regEx substitution.");
return false;
}
}
if (logger.isDebugEnabled())
logger.debug(
"nameDatasetRegExp(): Neither URL <"
+ dataset.getUrlPath()
+ "> or name <"
+ dataset.getName()
+ "> matched pattern <"
+ this.matchPattern
+ "> .");
return false;
}
public InvCatalog getDirCatalog(
CrawlableDataset catalogCrDs,
String filterPattern,
boolean sortInIncreasingOrder,
String addIdBase,
boolean addDatasetSize,
String dsNameMatchPattern,
String startTimeSubstitutionPattern,
String duration) {
// Setup the filter
CrawlableDatasetFilter filter = null;
if (filterPattern != null) {
// Include atomic datasets that match the given filter string.
MultiSelectorFilter.Selector selector =
new MultiSelectorFilter.Selector(
new RegExpMatchOnNameFilter(filterPattern), true, true, false);
filter = new MultiSelectorFilter(selector);
} else {
filter = new RegExpMatchOnNameFilter(".*");
}
List enhancerList = null;
if (dsNameMatchPattern != null && startTimeSubstitutionPattern != null && duration != null) {
enhancerList = new ArrayList();
enhancerList.add(
RegExpAndDurationTimeCoverageEnhancer.getInstanceToMatchOnDatasetName(
dsNameMatchPattern, startTimeSubstitutionPattern, duration));
}
CatalogBuilder catBuilder =
new StandardCatalogBuilder(
prefixPath,
null,
collectionCrDs,
filter,
service,
addIdBase,
null,
null,
addDatasetSize,
new LexigraphicByNameSorter(sortInIncreasingOrder),
null,
enhancerList,
null,
new BooleanCatalogRefExpander(!this.createCatalogRefs));
InvCatalog catalog;
try {
catalog = catBuilder.generateCatalog(catalogCrDs);
} catch (IOException e) {
throw new IllegalArgumentException("Could not generate catalog: " + e.getMessage());
}
// log.debug( "getDirCatalog(): directory=" + directory + "; filterPattern=" + filterPattern
// + "; sortInIncreasingOrder=" + sortInIncreasingOrder + "; addIdBase="+addIdBase+";
// dsNameMatchPattern=" + dsNameMatchPattern + "; startTimeSubstitutionPattern=" +
// startTimeSubstitutionPattern + "; duration=" + duration );
// if ( !directory.isDirectory() )
// {
// String tmpMsg = "Given directory is not a directory <" + directory.getAbsolutePath() +
// ">.";
// log.warn( tmpMsg);
// throw new IllegalArgumentException( tmpMsg );
// }
// DatasetSource dsSource = DatasetSource.newDatasetSource( directory.getName(),
// DatasetSourceType.LOCAL,
// DatasetSourceStructure.DIRECTORY_TREE,
// directory.getAbsolutePath(),
// this.service );
// dsSource.setPrefixUrlPath( this.prefixPath );
// dsSource.setCreateCatalogRefs( this.createCatalogRefs );
// dsSource.setAddDatasetSize( addDatasetSize );
// if ( filterPattern != null )
// {
// DatasetFilter datasetFilter = new DatasetFilter( dsSource, "Filter files on \"" +
// filterPattern + "\"",
// DatasetFilter.Type.REGULAR_EXPRESSION,
// filterPattern );
// datasetFilter.setMatchPatternTarget( "name");
// dsSource.addDatasetFilter( datasetFilter );
// datasetFilter = new DatasetFilter( dsSource, "Allow all dirs",
// DatasetFilter.Type.REGULAR_EXPRESSION,
// "", true, false, false );
// datasetFilter.setMatchPatternTarget( "name" );
// dsSource.addDatasetFilter( datasetFilter );
// }
// dsSource.setDatasetSorter( new DatasetSorter( sortInIncreasingOrder));
// if ( dsNameMatchPattern != null
// && startTimeSubstitutionPattern != null
// && duration != null)
// {
// dsSource.addDatasetEnhancer( DatasetEnhancer1.createAddTimeCoverageEnhancer(
// dsNameMatchPattern, startTimeSubstitutionPattern, duration));
// }
//
// if ( addIdBase != null)
// {
// dsSource.addDatasetEnhancer( DatasetEnhancer1.createAddIdEnhancer( addIdBase));
// }
//
// InvCatalog cat = null;
// try
// {
// cat = dsSource.fullExpand();
// }
// catch ( IOException e )
// {
// throw new IllegalArgumentException( "Given directory is not a collection dataset <" +
// directory.getAbsolutePath() + ">: " + e.getMessage() );
// }
InvDataset topDs = catalog.getDataset();
if (collectionCrDs.getPath().equals(catalogCrDs.getPath()) && this.serviceTitle != null)
// if ( topDs.getName().equals( "" ) && this.serviceTitle != null )
{
logger.warn(
"getDirCatalog(): top dataset name is null, setting to serviceTitle <"
+ this.serviceTitle
+ ">");
((InvDatasetImpl) topDs).setName(this.serviceTitle);
}
return (catalog);
}