public InvCatalog getDirCatalog(
File directory,
String filterPattern,
boolean sortInIncreasingOrder,
String addIdBase,
boolean addDatasetSize,
String dsNameMatchPattern,
String startTimeSubstitutionPattern,
String duration) {
CrawlableDataset catalogCrDs;
try {
catalogCrDs =
CrawlableDatasetFactory.createCrawlableDataset(directory.getAbsolutePath(), null, null);
} catch (IOException e) {
// @todo Should throw an IOException!
throw new IllegalArgumentException("IOException while creating dataset: " + e.getMessage());
} catch (ClassNotFoundException e) {
throw new IllegalArgumentException("Did not find class: " + e.getMessage());
} catch (NoSuchMethodException e) {
throw new IllegalArgumentException(
"Required constructor not found in class: " + e.getMessage());
} catch (IllegalAccessException e) {
throw new IllegalArgumentException(
"Did not have necessary access to class: " + e.getMessage());
} catch (InvocationTargetException e) {
throw new IllegalArgumentException(
"Could not invoke required method in class: " + e.getMessage());
} catch (InstantiationException e) {
throw new IllegalArgumentException("Could not instatiate class: " + e.getMessage());
}
if (!catalogCrDs.isCollection())
throw new IllegalArgumentException(
"catalog directory is not a directory <" + serviceBaseUrlDir.getAbsolutePath() + ">.");
return getDirCatalog(
catalogCrDs,
filterPattern,
sortInIncreasingOrder,
addIdBase,
addDatasetSize,
dsNameMatchPattern,
startTimeSubstitutionPattern,
duration);
}
public CrawlableDataset getDescendant(String relativePath) {
if (!this.isCollection) throw new IllegalStateException("Dataset not a collection.");
String[] pathSegments = CrawlableDatasetUtils.getPathSegments(relativePath);
if (!CrawlableDatasetUtils.isValidRelativePath(pathSegments))
throw new IllegalArgumentException(String.format("Ill-formed relative path [%s]", path));
boolean singleLevelPath = pathSegments.length == 1;
CrawlableDataset curCrDs = this.childrenMap.get(pathSegments[0]);
if (curCrDs != null) {
if (singleLevelPath) return curCrDs;
return curCrDs.getDescendant(CrawlableDatasetUtils.stepDownRelativePath(pathSegments));
} else {
curCrDs = new MockCrawlableDataset(this.getPath() + "/" + pathSegments[0], singleLevelPath);
((MockCrawlableDataset) curCrDs).setExists(false);
if (singleLevelPath) return curCrDs;
return curCrDs.getDescendant(CrawlableDatasetUtils.stepDownRelativePath(pathSegments));
}
}
/**
* Construct a DirectoryScanner given information about the data server to be cataloged.
*
* @param service - the service for the datasets found in the directory scanned.
* @param serviceTitle - a title for the service (used as the title of the top-level dataset).
* @param serviceBaseUrlDir - the local directory to which the service's base URL references.
* @param prefixPath - path name to append to the urlPath of resulting datasets, can be null.
* @param createCatalogRefs - if true, generate a catalogRef for each directory, otherwise,
* recurse into directories.
* @throws IllegalArgumentException if invalid service type is given or root directory is not a
* directory.
*/
public DirectoryScanner(
InvService service,
String serviceTitle,
File serviceBaseUrlDir,
String prefixPath,
boolean createCatalogRefs) {
this.service = service;
this.serviceTitle = serviceTitle;
this.serviceBaseUrlDir = serviceBaseUrlDir;
try {
collectionCrDs =
CrawlableDatasetFactory.createCrawlableDataset(
serviceBaseUrlDir.getAbsolutePath(), null, null);
} catch (IOException e) {
// @todo Should throw an IOException!
throw new IllegalArgumentException("IOException while creating dataset: " + e.getMessage());
} catch (ClassNotFoundException e) {
throw new IllegalArgumentException("Did not find class: " + e.getMessage());
} catch (NoSuchMethodException e) {
throw new IllegalArgumentException(
"Required constructor not found in class: " + e.getMessage());
} catch (IllegalAccessException e) {
throw new IllegalArgumentException(
"Did not have necessary access to class: " + e.getMessage());
} catch (InvocationTargetException e) {
throw new IllegalArgumentException(
"Could not invoke required method in class: " + e.getMessage());
} catch (InstantiationException e) {
throw new IllegalArgumentException("Could not instatiate class: " + e.getMessage());
}
if (!collectionCrDs.isCollection())
throw new IllegalArgumentException(
"Base URL directory is not a directory <" + serviceBaseUrlDir.getAbsolutePath() + ">.");
this.prefixPath = prefixPath;
this.createCatalogRefs = createCatalogRefs;
// log.debug( "DirectoryScanner(): service="+service.getName()+";
// serviceTitle="+serviceTitle+"; serviceBaseUrlDir="+serviceBaseUrlDir+";
// createCatalogRefs="+createCatalogRefs);
// this.serviceTitle = serviceTitle;
// this.serviceBaseUrlDir = serviceBaseUrlDir;
// if ( this.serviceBaseUrlDir.getPath().indexOf( "*") == -1)
// {
// if ( ! this.serviceBaseUrlDir.isDirectory()) throw new IllegalArgumentException( "Base
// URL directory is not a directory <" + serviceBaseUrlDir.getAbsolutePath() + ">.");
// }
// this.service = new ResultService( service, serviceBaseUrlDir.getAbsolutePath());
//
// this.prefixPath = prefixPath;
// this.createCatalogRefs = createCatalogRefs;
}
/** Test ... */
public void testOne() {
createFiles();
// ******** DO TEST STUFF **********
/*
<filter logicalComp="OR">
<!-- Only grib1 files that are older than 60 seconds -->
<filter logicalComp="AND">
<filter>
<include wildcard="*.grib1"/>
</filter>
<filter lastModifiedLimit="60000" />
</filter>
<!-- Only nc files that are less then 60 seconds old -->
<filter logicalComp="AND">
<filter>
<include wildcard="*.nc"/>
</filter>
<filter logicalComp="NOT">
<filter lastModifiedLimit="60000"/>
</filter>
</filter>
</filter>
*/
CrawlableDatasetFilter includeGribFilter =
new MultiSelectorFilter(
new MultiSelectorFilter.Selector(
new WildcardMatchOnNameFilter("*.grib1"), true, true, false));
CrawlableDatasetFilter lastModAtLeast4MinPastFilter = new LastModifiedLimitFilter(240000);
CrawlableDatasetFilter oldGribFilter =
LogicalFilterComposer.getAndFilter(includeGribFilter, lastModAtLeast4MinPastFilter);
CrawlableDatasetFilter includeNcFilter =
new MultiSelectorFilter(
new MultiSelectorFilter.Selector(
new WildcardMatchOnNameFilter("*.nc"), true, true, false));
CrawlableDatasetFilter newNcFilter =
LogicalFilterComposer.getAndFilter(
includeNcFilter, LogicalFilterComposer.getNotFilter(lastModAtLeast4MinPastFilter));
CrawlableDatasetFilter oldGribOrNewNcFilter =
LogicalFilterComposer.getOrFilter(oldGribFilter, newNcFilter);
CrawlableDataset tmpDirCrDs = new CrawlableDatasetFile(tmpDir);
List crDsList = null;
try {
crDsList = tmpDirCrDs.listDatasets();
} catch (IOException e) {
assertTrue("I/O problem getting contained dataset list.", false);
deleteFiles();
return;
}
for (Iterator it = crDsList.iterator(); it.hasNext(); ) {
CrawlableDataset curCrDs = (CrawlableDataset) it.next();
if (oldGribOrNewNcFilter.accept(curCrDs)) {
if (!curCrDs.getName().equals("old.grib1") && !curCrDs.getName().equals("new.nc")) {
assertTrue("Matched wrong file <" + curCrDs.getPath() + ">.", false);
deleteFiles();
return;
}
}
}
// ******** DO TEST STUFF - END **********
deleteFiles();
}
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);
}
