use of org.apache.asterix.metadata.entities.Dataset in project asterixdb by apache.
the class MetadataBuiltinFunctions method addMetadataBuiltinFunctions.
public static void addMetadataBuiltinFunctions() {
BuiltinFunctions.addFunction(BuiltinFunctions.DATASET, new IResultTypeComputer() {
@Override
public IAType computeType(ILogicalExpression expression, IVariableTypeEnvironment env, IMetadataProvider<?, ?> mp) throws AlgebricksException {
AbstractFunctionCallExpression f = (AbstractFunctionCallExpression) expression;
if (f.getArguments().size() != 1) {
throw new AlgebricksException("dataset arity is 1, not " + f.getArguments().size());
}
ILogicalExpression a1 = f.getArguments().get(0).getValue();
IAType t1 = (IAType) env.getType(a1);
if (t1.getTypeTag() == ATypeTag.ANY) {
return BuiltinType.ANY;
}
if (t1.getTypeTag() != ATypeTag.STRING) {
throw new AlgebricksException("Illegal type " + t1 + " for dataset() argument.");
}
String datasetArg = ConstantExpressionUtil.getStringConstant(a1);
if (datasetArg == null) {
return BuiltinType.ANY;
}
MetadataProvider metadata = (MetadataProvider) mp;
Pair<String, String> datasetInfo = getDatasetInfo(metadata, datasetArg);
String dataverseName = datasetInfo.first;
String datasetName = datasetInfo.second;
if (dataverseName == null) {
throw new AlgebricksException("Unspecified dataverse!");
}
Dataset dataset = metadata.findDataset(dataverseName, datasetName);
if (dataset == null) {
throw new AlgebricksException("Could not find dataset " + datasetName + " in dataverse " + dataverseName);
}
String tn = dataset.getItemTypeName();
IAType t2 = metadata.findType(dataset.getItemTypeDataverseName(), tn);
if (t2 == null) {
throw new AlgebricksException("No type for dataset " + datasetName);
}
return t2;
}
}, true);
BuiltinFunctions.addPrivateFunction(BuiltinFunctions.FEED_COLLECT, new IResultTypeComputer() {
@Override
public IAType computeType(ILogicalExpression expression, IVariableTypeEnvironment env, IMetadataProvider<?, ?> mp) throws AlgebricksException {
AbstractFunctionCallExpression f = (AbstractFunctionCallExpression) expression;
if (f.getArguments().size() != BuiltinFunctions.FEED_COLLECT.getArity()) {
throw new AlgebricksException("Incorrect number of arguments -> arity is " + BuiltinFunctions.FEED_COLLECT.getArity() + ", not " + f.getArguments().size());
}
ILogicalExpression a1 = f.getArguments().get(5).getValue();
IAType t1 = (IAType) env.getType(a1);
if (t1.getTypeTag() == ATypeTag.ANY) {
return BuiltinType.ANY;
}
if (t1.getTypeTag() != ATypeTag.STRING) {
throw new AlgebricksException("Illegal type " + t1 + " for feed-ingest argument.");
}
String typeArg = ConstantExpressionUtil.getStringConstant(a1);
if (typeArg == null) {
return BuiltinType.ANY;
}
MetadataProvider metadata = (MetadataProvider) mp;
Pair<String, String> argInfo = getDatasetInfo(metadata, typeArg);
String dataverseName = argInfo.first;
String typeName = argInfo.second;
if (dataverseName == null) {
throw new AlgebricksException("Unspecified dataverse!");
}
IAType t2 = metadata.findType(dataverseName, typeName);
if (t2 == null) {
throw new AlgebricksException("Unknown type " + typeName);
}
return t2;
}
}, true);
BuiltinFunctions.addFunction(BuiltinFunctions.FEED_INTERCEPT, new IResultTypeComputer() {
@Override
public IAType computeType(ILogicalExpression expression, IVariableTypeEnvironment env, IMetadataProvider<?, ?> mp) throws AlgebricksException {
AbstractFunctionCallExpression f = (AbstractFunctionCallExpression) expression;
if (f.getArguments().size() != 1) {
throw new AlgebricksException("dataset arity is 1, not " + f.getArguments().size());
}
ILogicalExpression a1 = f.getArguments().get(0).getValue();
IAType t1 = (IAType) env.getType(a1);
if (t1.getTypeTag() == ATypeTag.ANY) {
return BuiltinType.ANY;
}
if (t1.getTypeTag() != ATypeTag.STRING) {
throw new AlgebricksException("Illegal type " + t1 + " for dataset() argument.");
}
String datasetArg = ConstantExpressionUtil.getStringConstant(a1);
if (datasetArg == null) {
return BuiltinType.ANY;
}
MetadataProvider metadata = (MetadataProvider) mp;
Pair<String, String> datasetInfo = getDatasetInfo(metadata, datasetArg);
String dataverseName = datasetInfo.first;
String datasetName = datasetInfo.second;
if (dataverseName == null) {
throw new AlgebricksException("Unspecified dataverse!");
}
Dataset dataset = metadata.findDataset(dataverseName, datasetName);
if (dataset == null) {
throw new AlgebricksException("Could not find dataset " + datasetName + " in dataverse " + dataverseName);
}
String tn = dataset.getItemTypeName();
IAType t2 = metadata.findType(dataset.getItemTypeDataverseName(), tn);
if (t2 == null) {
throw new AlgebricksException("No type for dataset " + datasetName);
}
return t2;
}
}, true);
}
use of org.apache.asterix.metadata.entities.Dataset in project asterixdb by apache.
the class DatasetUtil method compactDatasetJobSpec.
public static JobSpecification compactDatasetJobSpec(Dataverse dataverse, String datasetName, MetadataProvider metadataProvider) throws AlgebricksException {
String dataverseName = dataverse.getDataverseName();
Dataset dataset = metadataProvider.findDataset(dataverseName, datasetName);
if (dataset == null) {
throw new AsterixException("Could not find dataset " + datasetName + " in dataverse " + dataverseName);
}
JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
Pair<IFileSplitProvider, AlgebricksPartitionConstraint> splitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset);
IIndexDataflowHelperFactory indexHelperFactory = new IndexDataflowHelperFactory(metadataProvider.getStorageComponentProvider().getStorageManager(), splitsAndConstraint.first);
LSMTreeIndexCompactOperatorDescriptor compactOp = new LSMTreeIndexCompactOperatorDescriptor(spec, indexHelperFactory);
AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, compactOp, splitsAndConstraint.second);
AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, compactOp, splitsAndConstraint.second);
spec.addRoot(compactOp);
return spec;
}
use of org.apache.asterix.metadata.entities.Dataset in project asterixdb by apache.
the class InvertedIndexPOperator method contributeRuntimeOperator.
@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
AbstractUnnestMapOperator unnestMapOp = (AbstractUnnestMapOperator) op;
ILogicalExpression unnestExpr = unnestMapOp.getExpressionRef().getValue();
if (unnestExpr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
throw new IllegalStateException();
}
AbstractFunctionCallExpression unnestFuncExpr = (AbstractFunctionCallExpression) unnestExpr;
if (unnestFuncExpr.getFunctionIdentifier() != BuiltinFunctions.INDEX_SEARCH) {
return;
}
InvertedIndexJobGenParams jobGenParams = new InvertedIndexJobGenParams();
jobGenParams.readFromFuncArgs(unnestFuncExpr.getArguments());
MetadataProvider metadataProvider = (MetadataProvider) context.getMetadataProvider();
Dataset dataset;
try {
dataset = metadataProvider.findDataset(jobGenParams.getDataverseName(), jobGenParams.getDatasetName());
} catch (MetadataException e) {
throw new AlgebricksException(e);
}
int[] keyIndexes = getKeyIndexes(jobGenParams.getKeyVarList(), inputSchemas);
int[] minFilterFieldIndexes = getKeyIndexes(unnestMapOp.getMinFilterVars(), inputSchemas);
int[] maxFilterFieldIndexes = getKeyIndexes(unnestMapOp.getMaxFilterVars(), inputSchemas);
boolean retainNull = false;
if (op.getOperatorTag() == LogicalOperatorTag.LEFT_OUTER_UNNEST_MAP) {
// By nature, LEFT_OUTER_UNNEST_MAP should generate null values for non-matching tuples.
retainNull = true;
}
// Build runtime.
Pair<IOperatorDescriptor, AlgebricksPartitionConstraint> invIndexSearch = buildInvertedIndexRuntime(metadataProvider, context, builder.getJobSpec(), unnestMapOp, opSchema, jobGenParams.getRetainInput(), retainNull, jobGenParams.getDatasetName(), dataset, jobGenParams.getIndexName(), jobGenParams.getSearchKeyType(), keyIndexes, jobGenParams.getSearchModifierType(), jobGenParams.getSimilarityThreshold(), minFilterFieldIndexes, maxFilterFieldIndexes, jobGenParams.getIsFullTextSearch());
// Contribute operator in hyracks job.
builder.contributeHyracksOperator(unnestMapOp, invIndexSearch.first);
builder.contributeAlgebricksPartitionConstraint(invIndexSearch.first, invIndexSearch.second);
ILogicalOperator srcExchange = unnestMapOp.getInputs().get(0).getValue();
builder.contributeGraphEdge(srcExchange, 0, unnestMapOp, 0);
}
use of org.apache.asterix.metadata.entities.Dataset in project asterixdb by apache.
the class RTreeSearchPOperator method contributeRuntimeOperator.
@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
AbstractUnnestMapOperator unnestMap = (AbstractUnnestMapOperator) op;
ILogicalExpression unnestExpr = unnestMap.getExpressionRef().getValue();
if (unnestExpr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
throw new IllegalStateException();
}
AbstractFunctionCallExpression unnestFuncExpr = (AbstractFunctionCallExpression) unnestExpr;
FunctionIdentifier funcIdent = unnestFuncExpr.getFunctionIdentifier();
if (!funcIdent.equals(BuiltinFunctions.INDEX_SEARCH)) {
return;
}
RTreeJobGenParams jobGenParams = new RTreeJobGenParams();
jobGenParams.readFromFuncArgs(unnestFuncExpr.getArguments());
int[] keyIndexes = getKeyIndexes(jobGenParams.getKeyVarList(), inputSchemas);
int[] minFilterFieldIndexes = getKeyIndexes(unnestMap.getMinFilterVars(), inputSchemas);
int[] maxFilterFieldIndexes = getKeyIndexes(unnestMap.getMaxFilterVars(), inputSchemas);
MetadataProvider mp = (MetadataProvider) context.getMetadataProvider();
Dataset dataset = mp.findDataset(jobGenParams.getDataverseName(), jobGenParams.getDatasetName());
IVariableTypeEnvironment typeEnv = context.getTypeEnvironment(unnestMap);
List<LogicalVariable> outputVars = unnestMap.getVariables();
if (jobGenParams.getRetainInput()) {
outputVars = new ArrayList<LogicalVariable>();
VariableUtilities.getLiveVariables(unnestMap, outputVars);
}
boolean retainNull = false;
if (op.getOperatorTag() == LogicalOperatorTag.LEFT_OUTER_UNNEST_MAP) {
// By nature, LEFT_OUTER_UNNEST_MAP should generate null values for non-matching tuples.
retainNull = true;
}
Pair<IOperatorDescriptor, AlgebricksPartitionConstraint> rtreeSearch = mp.buildRtreeRuntime(builder.getJobSpec(), outputVars, opSchema, typeEnv, context, jobGenParams.getRetainInput(), retainNull, dataset, jobGenParams.getIndexName(), keyIndexes, minFilterFieldIndexes, maxFilterFieldIndexes);
builder.contributeHyracksOperator(unnestMap, rtreeSearch.first);
builder.contributeAlgebricksPartitionConstraint(rtreeSearch.first, rtreeSearch.second);
ILogicalOperator srcExchange = unnestMap.getInputs().get(0).getValue();
builder.contributeGraphEdge(srcExchange, 0, unnestMap, 0);
}
use of org.apache.asterix.metadata.entities.Dataset in project asterixdb by apache.
the class BTreeSearchPOperator method contributeRuntimeOperator.
@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
AbstractUnnestMapOperator unnestMap = (AbstractUnnestMapOperator) op;
ILogicalExpression unnestExpr = unnestMap.getExpressionRef().getValue();
if (unnestExpr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
throw new IllegalStateException();
}
AbstractFunctionCallExpression unnestFuncExpr = (AbstractFunctionCallExpression) unnestExpr;
FunctionIdentifier funcIdent = unnestFuncExpr.getFunctionIdentifier();
if (!funcIdent.equals(BuiltinFunctions.INDEX_SEARCH)) {
return;
}
BTreeJobGenParams jobGenParams = new BTreeJobGenParams();
jobGenParams.readFromFuncArgs(unnestFuncExpr.getArguments());
int[] lowKeyIndexes = getKeyIndexes(jobGenParams.getLowKeyVarList(), inputSchemas);
int[] highKeyIndexes = getKeyIndexes(jobGenParams.getHighKeyVarList(), inputSchemas);
int[] minFilterFieldIndexes = getKeyIndexes(unnestMap.getMinFilterVars(), inputSchemas);
int[] maxFilterFieldIndexes = getKeyIndexes(unnestMap.getMaxFilterVars(), inputSchemas);
MetadataProvider metadataProvider = (MetadataProvider) context.getMetadataProvider();
Dataset dataset = metadataProvider.findDataset(jobGenParams.getDataverseName(), jobGenParams.getDatasetName());
IVariableTypeEnvironment typeEnv = context.getTypeEnvironment(op);
// By nature, LEFT_OUTER_UNNEST_MAP should generate null values for non-matching tuples.
boolean retainMissing = op.getOperatorTag() == LogicalOperatorTag.LEFT_OUTER_UNNEST_MAP;
Pair<IOperatorDescriptor, AlgebricksPartitionConstraint> btreeSearch = metadataProvider.buildBtreeRuntime(builder.getJobSpec(), opSchema, typeEnv, context, jobGenParams.getRetainInput(), retainMissing, dataset, jobGenParams.getIndexName(), lowKeyIndexes, highKeyIndexes, jobGenParams.isLowKeyInclusive(), jobGenParams.isHighKeyInclusive(), minFilterFieldIndexes, maxFilterFieldIndexes);
builder.contributeHyracksOperator(unnestMap, btreeSearch.first);
builder.contributeAlgebricksPartitionConstraint(btreeSearch.first, btreeSearch.second);
ILogicalOperator srcExchange = unnestMap.getInputs().get(0).getValue();
builder.contributeGraphEdge(srcExchange, 0, unnestMap, 0);
}
Aggregations