@Override
public <E> Dataset<E> load(String name) {
Preconditions.checkArgument(name != null, "Name can not be null");
logger.debug("Loading dataset:{}", name);
DatasetDescriptor descriptor = metadataProvider.load(name);
descriptor = addRepositoryUri(descriptor);
FileSystemDataset<E> ds =
new FileSystemDataset.Builder()
.name(name)
.configuration(conf)
.descriptor(descriptor)
.partitionKey(
descriptor.isPartitioned()
? org.kitesdk.data.impl.Accessor.getDefault().newPartitionKey()
: null)
.partitionListener(getPartitionListener())
.build();
logger.debug("Loaded dataset:{}", ds);
return ds;
}
@Override
public <E> Dataset<E> create(String name, DatasetDescriptor descriptor) {
Preconditions.checkArgument(name != null, "Name can not be null");
Preconditions.checkArgument(descriptor != null, "Descriptor can not be null");
Preconditions.checkArgument(
descriptor.getLocation() == null,
"Descriptor location cannot be set; " + "it is assigned by the MetadataProvider");
DatasetDescriptor newDescriptor = metadataProvider.create(name, descriptor);
newDescriptor = addRepositoryUri(newDescriptor);
final URI location = newDescriptor.getLocation();
if (location == null) {
throw new DatasetRepositoryException(
"[BUG] MetadataProvider did not assign a location to dataset:" + name);
}
ensureExists(newDescriptor, conf);
logger.debug(
"Created dataset:{} schema:{} datasetPath:{}",
new Object[] {name, newDescriptor.getSchema(), location.toString()});
return new FileSystemDataset.Builder()
.name(name)
.configuration(conf)
.descriptor(newDescriptor)
.partitionKey(
newDescriptor.isPartitioned()
? org.kitesdk.data.impl.Accessor.getDefault().newPartitionKey()
: null)
.partitionListener(getPartitionListener())
.build();
}
@Override
public boolean delete(String name) {
Preconditions.checkArgument(name != null, "Name can not be null");
logger.debug("Deleting dataset:{}", name);
DatasetDescriptor descriptor;
try {
descriptor = metadataProvider.load(name);
descriptor = addRepositoryUri(descriptor);
} catch (DatasetNotFoundException ex) {
return false;
}
boolean changed;
try {
// don't care about the return value here -- if it already doesn't exist
// we still need to delete the data directory
changed = metadataProvider.delete(name);
} catch (MetadataProviderException ex) {
throw new DatasetRepositoryException("Failed to delete descriptor for name:" + name, ex);
}
final Path dataLocation = new Path(descriptor.getLocation());
final FileSystem fs = fsForPath(dataLocation, conf);
try {
if (fs.exists(dataLocation)) {
if (fs.delete(dataLocation, true)) {
changed = true;
} else {
throw new DatasetRepositoryException(
"Failed to delete dataset name:" + name + " location:" + dataLocation);
}
}
} catch (IOException e) {
throw new DatasetRepositoryException(
"Internal failure when removing location:" + dataLocation);
}
return changed;
}
@Override
public Collection<String> list() {
return metadataProvider.list();
}
@Override
public boolean exists(String name) {
Preconditions.checkArgument(name != null, "Name can not be null");
return metadataProvider.exists(name);
}
@Override
public <E> Dataset<E> update(String name, DatasetDescriptor descriptor) {
Preconditions.checkArgument(name != null, "Dataset name cannot be null");
Preconditions.checkArgument(descriptor != null, "DatasetDescriptro cannot be null");
DatasetDescriptor oldDescriptor = metadataProvider.load(name);
// oldDescriptor is valid if load didn't throw NoSuchDatasetException
if (!oldDescriptor.getFormat().equals(descriptor.getFormat())) {
throw new DatasetRepositoryException(
"Cannot change dataset format from "
+ oldDescriptor.getFormat()
+ " to "
+ descriptor.getFormat());
}
final URI oldLocation = oldDescriptor.getLocation();
if ((oldLocation != null) && !(oldLocation.equals(descriptor.getLocation()))) {
throw new DatasetRepositoryException("Cannot change the dataset's location");
}
if (oldDescriptor.isPartitioned() != descriptor.isPartitioned()) {
throw new DatasetRepositoryException(
"Cannot change an unpartitioned dataset to " + " partitioned or vice versa.");
} else if (oldDescriptor.isPartitioned()
&& descriptor.isPartitioned()
&& !oldDescriptor.getPartitionStrategy().equals(descriptor.getPartitionStrategy())) {
throw new DatasetRepositoryException(
"Cannot change partition strategy from "
+ oldDescriptor.getPartitionStrategy()
+ " to "
+ descriptor.getPartitionStrategy());
}
// check can read records written with old schema using new schema
final Schema oldSchema = oldDescriptor.getSchema();
final Schema newSchema = descriptor.getSchema();
if (!SchemaValidationUtil.canRead(oldSchema, newSchema)) {
throw new IncompatibleSchemaException(
"New schema cannot read data "
+ "written using "
+ "old schema. New schema: "
+ newSchema.toString(true)
+ "\nOld schema: "
+ oldSchema.toString(true));
}
DatasetDescriptor updatedDescriptor = metadataProvider.update(name, descriptor);
updatedDescriptor = addRepositoryUri(updatedDescriptor);
logger.debug(
"Updated dataset:{} schema:{} datasetPath:{}",
new Object[] {
name, updatedDescriptor.getSchema(), updatedDescriptor.getLocation().toString()
});
return new FileSystemDataset.Builder()
.name(name)
.configuration(conf)
.descriptor(updatedDescriptor)
.partitionKey(
updatedDescriptor.isPartitioned()
? org.kitesdk.data.impl.Accessor.getDefault().newPartitionKey()
: null)
.partitionListener(getPartitionListener())
.build();
}