[
https://issues.apache.org/jira/browse/DRILL-4264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16138704#comment-16138704
]
ASF GitHub Bot commented on DRILL-4264:
---------------------------------------
Github user vvysotskyi commented on a diff in the pull request:
https://github.com/apache/drill/pull/909#discussion_r134686740
--- Diff:
exec/java-exec/src/main/java/org/apache/drill/exec/physical/impl/ScanBatch.java
---
@@ -359,30 +361,109 @@ public Mutator(OperatorExecContext oContext,
BufferAllocator allocator, VectorCo
public <T extends ValueVector> T addField(MaterializedField field,
Class<T> clazz) throws
SchemaChangeException {
// Check if the field exists.
- ValueVector v = fieldVectorMap.get(field.getPath());
- if (v == null || v.getClass() != clazz) {
+ ValueVector vector = fieldVectorMap.get(field.getName());
+ ValueVector childVector = vector;
+ // if a vector does not exist yet, creates value vector, or if it
exists and has map type, omit this code
+ if (vector == null || (vector.getClass() != clazz
+ && (vector.getField().getType().getMinorType() != MinorType.MAP
+ || field.getType().getMinorType() != MinorType.MAP))) {
// Field does not exist--add it to the map and the output
container.
- v = TypeHelper.getNewVector(field, allocator, callBack);
- if (!clazz.isAssignableFrom(v.getClass())) {
+ vector = TypeHelper.getNewVector(field, allocator, callBack);
+ childVector = vector;
+ // gets inner field if the map was created the first time
+ if (field.getType().getMinorType() == MinorType.MAP) {
+ childVector = getChildVectorByField(vector, field);
+ } else if (!clazz.isAssignableFrom(vector.getClass())) {
throw new SchemaChangeException(
String.format(
"The class that was provided, %s, does not correspond to
the "
+ "expected vector type of %s.",
- clazz.getSimpleName(), v.getClass().getSimpleName()));
+ clazz.getSimpleName(),
vector.getClass().getSimpleName()));
}
- final ValueVector old = fieldVectorMap.put(field.getPath(), v);
+ final ValueVector old = fieldVectorMap.put(field.getName(),
vector);
if (old != null) {
old.clear();
container.remove(old);
}
- container.add(v);
+ container.add(vector);
// Added new vectors to the container--mark that the schema has
changed.
schemaChanged = true;
}
+ // otherwise, checks that field and existing vector have a map type
+ // and adds child fields from the field to the vector
+ else if (field.getType().getMinorType() == MinorType.MAP
+ && vector.getField().getType().getMinorType() ==
MinorType.MAP
+ && !field.getChildren().isEmpty()) {
+ // an incoming field contains only single child since it determines
+ // full name path of the field in the schema
+ childVector = addNestedChildToMap((MapVector) vector,
Iterables.getLast(field.getChildren()));
+ schemaChanged = true;
+ }
- return clazz.cast(v);
+ return clazz.cast(childVector);
+ }
+
+ /**
+ * Finds and returns value vector which path corresponds to the
specified field.
+ * If required vector is nested in the map, gets and returns this
vector from the map.
+ *
+ * @param valueVector vector that should be checked
+ * @param field field that corresponds to required vector
+ * @return value vector whose path corresponds to the specified field
+ *
+ * @throws SchemaChangeException if the field does not correspond to
the found vector
+ */
+ private ValueVector getChildVectorByField(ValueVector valueVector,
+ MaterializedField field)
throws SchemaChangeException {
+ if (field.getChildren().isEmpty()) {
+ if (valueVector.getField().equals(field)) {
+ return valueVector;
+ } else {
+ throw new SchemaChangeException(
+ String.format(
+ "The field that was provided, %s, does not correspond to the
"
+ + "expected vector type of %s.",
+ field, valueVector.getClass().getSimpleName()));
+ }
+ } else {
+ // an incoming field contains only single child since it determines
+ // full name path of the field in the schema
+ MaterializedField childField =
Iterables.getLast(field.getChildren());
+ return getChildVectorByField(((MapVector)
valueVector).getChild(childField.getName()), childField);
+ }
+ }
+
+ /**
+ * Adds new vector with the specified field to the specified map
vector.
+ * If a field has few levels of nesting, corresponding maps will be
created.
+ *
+ * @param valueVector a map where new vector will be added
+ * @param field a field that corresponds to the vector which
should be created
+ * @return vector that was created
+ *
+ * @throws SchemaChangeException if the field has a child and its type
is not map
+ */
+ private ValueVector addNestedChildToMap(MapVector valueVector,
--- End diff --
This method was designed to handle the case when we have a map inside other
map and we want to add a field into the nested map. But considering your second
comment, this method also is not needed.
> Allow field names to include dots
> ---------------------------------
>
> Key: DRILL-4264
> URL: https://issues.apache.org/jira/browse/DRILL-4264
> Project: Apache Drill
> Issue Type: Improvement
> Components: Execution - Codegen
> Reporter: Alex
> Assignee: Volodymyr Vysotskyi
> Labels: doc-impacting
> Fix For: 1.12.0
>
>
> If you have some json data like this...
> {code:javascript}
> {
> "0.0.1":{
> "version":"0.0.1",
> "date_created":"2014-03-15"
> },
> "0.1.2":{
> "version":"0.1.2",
> "date_created":"2014-05-21"
> }
> }
> {code}
> ... there is no way to select any of the rows since their identifiers contain
> dots and when trying to select them, Drill throws the following error:
> Error: SYSTEM ERROR: UnsupportedOperationException: Unhandled field reference
> "0.0.1"; a field reference identifier must not have the form of a qualified
> name
> This must be fixed since there are many json data files containing dots in
> some of the keys (e.g. when specifying version numbers etc)
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
