deniskuzZ commented on code in PR #4131:
URL: https://github.com/apache/hive/pull/4131#discussion_r1149123408
##########
iceberg/iceberg-handler/src/main/java/org/apache/iceberg/mr/hive/HiveIcebergStorageHandler.java:
##########
@@ -349,6 +365,96 @@ public Map<String, String> getBasicStatistics(Partish
partish) {
return stats;
}
+
+ @Override
+ public boolean canSetColStatistics() {
+ String statsSource = HiveConf.getVar(conf,
HiveConf.ConfVars.HIVE_USE_STATS_FROM).toLowerCase();
+ return statsSource.equals(PUFFIN);
+ }
+
+ @Override
+ public boolean
canProvideColStatistics(org.apache.hadoop.hive.ql.metadata.Table tbl) {
+
+ org.apache.hadoop.hive.ql.metadata.Table hmsTable = tbl;
+ TableDesc tableDesc = Utilities.getTableDesc(hmsTable);
+ Table table = Catalogs.loadTable(conf, tableDesc.getProperties());
+ if (table.currentSnapshot() != null) {
+ String statsSource = HiveConf.getVar(conf,
HiveConf.ConfVars.HIVE_COL_STATS_SOURCE).toLowerCase();
+ String statsPath = table.location() + STATS + table.name() +
table.currentSnapshot().snapshotId();
+ if (statsSource.equals(PUFFIN)) {
+ try (FileSystem fs = new Path(table.location()).getFileSystem(conf)) {
+ if (fs.exists(new Path(statsPath))) {
+ return true;
+ }
+ } catch (IOException e) {
+ LOG.warn(e.getMessage());
+ }
+ }
+ }
+ return false;
+ }
+
+ @Override
+ public List<ColumnStatisticsObj>
getColStatistics(org.apache.hadoop.hive.ql.metadata.Table tbl) {
+
+ org.apache.hadoop.hive.ql.metadata.Table hmsTable = tbl;
+ TableDesc tableDesc = Utilities.getTableDesc(hmsTable);
+ Table table = Catalogs.loadTable(conf, tableDesc.getProperties());
+ String statsSource = HiveConf.getVar(conf,
HiveConf.ConfVars.HIVE_COL_STATS_SOURCE).toLowerCase();
+ switch (statsSource) {
+ case ICEBERG:
+ // Place holder for iceberg stats
+ break;
+ case PUFFIN:
+ String snapshotId = table.name() +
table.currentSnapshot().snapshotId();
+ String statsPath = table.location() + STATS + snapshotId;
+ LOG.info("Using stats from puffin file at:" + statsPath);
+ try (PuffinReader reader =
Puffin.read(table.io().newInputFile(statsPath)).build()) {
+ BlobMetadata blobMetadata = reader.fileMetadata().blobs().get(0);
+ Map<BlobMetadata, List<ColumnStatistics>> collect =
+
Streams.stream(reader.readAll(ImmutableList.of(blobMetadata))).collect(Collectors.toMap(Pair::first,
+ blobMetadataByteBufferPair -> SerializationUtils.deserialize(
+
ByteBuffers.toByteArray(blobMetadataByteBufferPair.second()))));
+
+ return
collect.entrySet().stream().iterator().next().getValue().get(0).getStatsObj();
+ } catch (IOException e) {
+ LOG.info(String.valueOf(e));
+ }
+ break;
+ default:
+ // fall back to metastore
+ }
+ return null;
+ }
+
+
+ @Override
+ public boolean setColStatistics(org.apache.hadoop.hive.ql.metadata.Table
table,
+ List<ColumnStatistics> colStats) {
+ TableDesc tableDesc = Utilities.getTableDesc(table);
+ Table tbl = Catalogs.loadTable(conf, tableDesc.getProperties());
+ String snapshotId = tbl.name() + tbl.currentSnapshot().snapshotId();
+ byte[] serializeColStats = SerializationUtils.serialize((Serializable)
colStats);
+
+ try (PuffinWriter writer =
Puffin.write(tbl.io().newOutputFile(tbl.location() + STATS + snapshotId))
Review Comment:
looks like not, per `getColStatsForPartCol` comment: currently, metastore
does not store column stats for the partition column. Also in the HMS column
stats table is called `TAB_COL_STATS` which doesn't track the per-partition
stats.
--
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]
