dramaticlly commented on code in PR #7539:
URL: https://github.com/apache/iceberg/pull/7539#discussion_r1192504063
##########
core/src/main/java/org/apache/iceberg/BaseEntriesTable.java:
##########
@@ -125,31 +130,120 @@ ManifestFile manifest() {
@Override
public CloseableIterable<StructLike> rows() {
- // Project data-file fields
- CloseableIterable<StructLike> prunedRows;
- if (manifest.content() == ManifestContent.DATA) {
- prunedRows =
+ Types.NestedField readableMetricsField =
projection.findField(MetricsUtil.READABLE_METRICS);
+
+ if (readableMetricsField == null) {
+ CloseableIterable<StructLike> entryAsStruct =
CloseableIterable.transform(
- ManifestFiles.read(manifest, io).project(fileSchema).entries(),
- file -> (GenericManifestEntry<DataFile>) file);
+ entries(fileProjection),
+ entry -> (GenericManifestEntry<? extends ContentFile<?>>)
entry);
+
+ StructProjection structProjection = projectNonReadable(projection);
+ return CloseableIterable.transform(entryAsStruct,
structProjection::wrap);
} else {
- prunedRows =
- CloseableIterable.transform(
- ManifestFiles.readDeleteManifest(manifest, io, specsById)
- .project(fileSchema)
- .entries(),
- file -> (GenericManifestEntry<DeleteFile>) file);
+ Schema requiredFileProjection = requiredFileProjection();
+ Schema actualProjection = removeReadableMetrics(readableMetricsField);
+ StructProjection structProjection =
projectNonReadable(actualProjection);
+
+ return CloseableIterable.transform(
+ entries(requiredFileProjection),
+ entry -> withReadableMetrics(structProjection, entry,
readableMetricsField));
}
+ }
+
+ /**
+ * Remove virtual columns from the file projection and ensure that the
underlying metrics used
+ * to create those columns are part of the file projection
+ *
+ * @return file projection with required columns to read readable metrics
+ */
+ private Schema requiredFileProjection() {
+ Schema projectionForReadableMetrics =
+ new Schema(
+ MetricsUtil.READABLE_METRIC_COLS.stream()
+ .map(MetricsUtil.ReadableMetricColDefinition::originalCol)
+ .collect(Collectors.toList()));
+ return TypeUtil.join(fileProjection, projectionForReadableMetrics);
+ }
- // Project non-readable fields
- Schema readSchema = ManifestEntry.wrapFileSchema(fileSchema.asStruct());
- StructProjection projection = StructProjection.create(readSchema,
schema);
- return CloseableIterable.transform(prunedRows, projection::wrap);
+ private Schema removeReadableMetrics(Types.NestedField
readableMetricsField) {
+ Set<Integer> readableMetricsIds =
TypeUtil.getProjectedIds(readableMetricsField.type());
+ return TypeUtil.selectNot(projection, readableMetricsIds);
+ }
+
+ private StructProjection projectNonReadable(Schema projectedSchema) {
Review Comment:
agreed, I think before this change the comment seem to suggest
- entry only field (like snapshot_id, status, sequence_number) are readable
directly from ManifestEntry
- and wrap of files schema and data file as struct considered as
non-readable
But after our refactoring, this method is only about wrap file schema into
entry schema, and project entry schema to necessary columns. So I think
`structProjection` sounds good to me
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
