/** Create String-based (nested) field expression keys on a composite type. */
public ExpressionKeys(String[] keyExpressions, TypeInformation<T> type) {
Preconditions.checkNotNull(keyExpressions, "Field expression cannot be null.");
this.keyFields = new ArrayList<>(keyExpressions.length);
if (type instanceof CompositeType) {
CompositeType<T> cType = (CompositeType<T>) type;
// extract the keys on their flat position
for (String keyExpr : keyExpressions) {
if (keyExpr == null) {
throw new InvalidProgramException("Expression key may not be null.");
}
// strip off whitespace
keyExpr = keyExpr.trim();
List<FlatFieldDescriptor> flatFields = cType.getFlatFields(keyExpr);
if (flatFields.size() == 0) {
throw new InvalidProgramException(
"Unable to extract key from expression '" + keyExpr + "' on key " + cType);
}
// check if all nested fields can be used as keys
for (FlatFieldDescriptor field : flatFields) {
if (!field.getType().isKeyType()) {
throw new InvalidProgramException(
"This type (" + field.getType() + ") cannot be used as key.");
}
}
// add flat fields to key fields
keyFields.addAll(flatFields);
}
} else {
if (!type.isKeyType()) {
throw new InvalidProgramException("This type (" + type + ") cannot be used as key.");
}
// check that all key expressions are valid
for (String keyExpr : keyExpressions) {
if (keyExpr == null) {
throw new InvalidProgramException("Expression key may not be null.");
}
// strip off whitespace
keyExpr = keyExpr.trim();
// check that full type is addressed
if (!(SELECT_ALL_CHAR.equals(keyExpr) || SELECT_ALL_CHAR_SCALA.equals(keyExpr))) {
throw new InvalidProgramException(
"Field expression must be equal to '"
+ SELECT_ALL_CHAR
+ "' or '"
+ SELECT_ALL_CHAR_SCALA
+ "' for non-composite types.");
}
// add full type as key
keyFields.add(new FlatFieldDescriptor(0, type));
}
}
}
@Override
protected TypeComparator<PojoContainingTuple> createComparator(boolean ascending) {
Assert.assertTrue(type instanceof CompositeType);
CompositeType<PojoContainingTuple> cType = (CompositeType<PojoContainingTuple>) type;
ExpressionKeys<PojoContainingTuple> keys =
new ExpressionKeys<PojoContainingTuple>(new String[] {"theTuple.*"}, cType);
boolean[] orders = new boolean[keys.getNumberOfKeyFields()];
Arrays.fill(orders, ascending);
return cType.createComparator(
keys.computeLogicalKeyPositions(), orders, 0, new ExecutionConfig());
}
/** Create int-based (non-nested) field position keys on a tuple type. */
public ExpressionKeys(int[] keyPositions, TypeInformation<T> type, boolean allowEmpty) {
if (!type.isTupleType() || !(type instanceof CompositeType)) {
throw new InvalidProgramException(
"Specifying keys via field positions is only valid "
+ "for tuple data types. Type: "
+ type);
}
if (type.getArity() == 0) {
throw new InvalidProgramException(
"Tuple size must be greater than 0. Size: " + type.getArity());
}
if (!allowEmpty && (keyPositions == null || keyPositions.length == 0)) {
throw new IllegalArgumentException("The grouping fields must not be empty.");
}
this.keyFields = new ArrayList<>();
if (keyPositions == null || keyPositions.length == 0) {
// use all tuple fields as key fields
keyPositions = createIncrIntArray(type.getArity());
} else {
rangeCheckFields(keyPositions, type.getArity() - 1);
}
Preconditions.checkArgument(
keyPositions.length > 0, "Grouping fields can not be empty at this point");
// extract key field types
CompositeType<T> cType = (CompositeType<T>) type;
this.keyFields = new ArrayList<>(type.getTotalFields());
// for each key position, find all (nested) field types
String[] fieldNames = cType.getFieldNames();
ArrayList<FlatFieldDescriptor> tmpList = new ArrayList<>();
for (int keyPos : keyPositions) {
tmpList.clear();
// get all flat fields
cType.getFlatFields(fieldNames[keyPos], 0, tmpList);
// check if fields are of key type
for (FlatFieldDescriptor ffd : tmpList) {
if (!ffd.getType().isKeyType()) {
throw new InvalidProgramException(
"This type (" + ffd.getType() + ") cannot be used as key.");
}
}
this.keyFields.addAll(tmpList);
}
}
/** Create ExpressionKeys from String-expressions */
public ExpressionKeys(String[] expressionsIn, TypeInformation<T> type) {
Preconditions.checkNotNull(expressionsIn, "Field expression cannot be null.");
if (type instanceof AtomicType) {
if (!type.isKeyType()) {
throw new InvalidProgramException("This type (" + type + ") cannot be used as key.");
} else if (expressionsIn.length != 1
|| !(Keys.ExpressionKeys.SELECT_ALL_CHAR.equals(expressionsIn[0])
|| Keys.ExpressionKeys.SELECT_ALL_CHAR_SCALA.equals(expressionsIn[0]))) {
throw new InvalidProgramException(
"Field expression for atomic type must be equal to '*' or '_'.");
}
keyFields = new ArrayList<>(1);
keyFields.add(new FlatFieldDescriptor(0, type));
} else {
CompositeType<T> cType = (CompositeType<T>) type;
String[] expressions = removeDuplicates(expressionsIn);
if (expressionsIn.length != expressions.length) {
LOG.warn("The key expressions contained duplicates. They are now unique");
}
// extract the keys on their flat position
keyFields = new ArrayList<>(expressions.length);
for (String expression : expressions) {
List<FlatFieldDescriptor> keys =
cType.getFlatFields(expression); // use separate list to do a size check
for (FlatFieldDescriptor key : keys) {
TypeInformation<?> keyType = key.getType();
if (!keyType.isKeyType()) {
throw new InvalidProgramException(
"This type (" + key.getType() + ") cannot be used as key.");
}
if (!(keyType instanceof AtomicType || keyType instanceof CompositeType)) {
throw new InvalidProgramException(
"Field type is neither CompositeType nor AtomicType: " + keyType);
}
}
if (keys.size() == 0) {
throw new InvalidProgramException(
"Unable to extract key from expression '" + expression + "' on key " + cType);
}
keyFields.addAll(keys);
}
}
}
@Override
public void getFlatFields(String fieldExpression, int offset, List<FlatFieldDescriptor> result) {
Matcher matcher = PATTERN_NESTED_FIELDS_WILDCARD.matcher(fieldExpression);
if (!matcher.matches()) {
throw new InvalidFieldReferenceException(
"Invalid POJO field reference \"" + fieldExpression + "\".");
}
String field = matcher.group(0);
if (field.equals(ExpressionKeys.SELECT_ALL_CHAR)
|| field.equals(ExpressionKeys.SELECT_ALL_CHAR_SCALA)) {
// handle select all
int keyPosition = 0;
for (PojoField pField : fields) {
if (pField.type instanceof CompositeType) {
CompositeType<?> cType = (CompositeType<?>) pField.type;
cType.getFlatFields(
String.valueOf(ExpressionKeys.SELECT_ALL_CHAR), offset + keyPosition, result);
keyPosition += cType.getTotalFields() - 1;
} else {
result.add(
new NamedFlatFieldDescriptor(
pField.field.getName(), offset + keyPosition, pField.type));
}
keyPosition++;
}
return;
} else {
field = matcher.group(1);
}
// get field
int fieldPos = -1;
TypeInformation<?> fieldType = null;
for (int i = 0; i < fields.length; i++) {
if (fields[i].field.getName().equals(field)) {
fieldPos = i;
fieldType = fields[i].type;
break;
}
}
if (fieldPos == -1) {
throw new InvalidFieldReferenceException(
"Unable to find field \"" + field + "\" in type " + this + ".");
}
String tail = matcher.group(3);
if (tail == null) {
if (fieldType instanceof CompositeType) {
// forward offset
for (int i = 0; i < fieldPos; i++) {
offset += this.getTypeAt(i).getTotalFields();
}
// add all fields of composite type
((CompositeType<?>) fieldType).getFlatFields("*", offset, result);
return;
} else {
// we found the field to add
// compute flat field position by adding skipped fields
int flatFieldPos = offset;
for (int i = 0; i < fieldPos; i++) {
flatFieldPos += this.getTypeAt(i).getTotalFields();
}
result.add(new FlatFieldDescriptor(flatFieldPos, fieldType));
// nothing left to do
return;
}
} else {
if (fieldType instanceof CompositeType<?>) {
// forward offset
for (int i = 0; i < fieldPos; i++) {
offset += this.getTypeAt(i).getTotalFields();
}
((CompositeType<?>) fieldType).getFlatFields(tail, offset, result);
// nothing left to do
return;
} else {
throw new InvalidFieldReferenceException(
"Nested field expression \""
+ tail
+ "\" not possible on atomic type "
+ fieldType
+ ".");
}
}
}