amogh-jahagirdar commented on code in PR #16699:
URL: https://github.com/apache/iceberg/pull/16699#discussion_r3377080595
##########
spark/v3.5/spark/src/test/java/org/apache/iceberg/spark/actions/TestRewriteManifestsAction.java:
##########
@@ -196,6 +196,144 @@ public void testRewriteManifestsPreservesOptionalFields()
throws IOException {
}
}
+ @TestTemplate
+ public void testRewriteV3ManifestsPreservesFirstRowId() {
+ assumeThat(formatVersion).isGreaterThanOrEqualTo(3);
+
+ PartitionSpec spec = PartitionSpec.unpartitioned();
+ Map<String, String> options = Maps.newHashMap();
+ options.put(TableProperties.FORMAT_VERSION, String.valueOf(formatVersion));
+ Table table = TABLES.create(SCHEMA, spec, options, tableLocation);
+
+ writeRecords(Lists.newArrayList(new ThreeColumnRecord(1, null, "AAAA")));
+ writeRecords(Lists.newArrayList(new ThreeColumnRecord(2, "CCCC", "CCCC")));
+ table.refresh();
+
+ assertThat(table.currentSnapshot().dataManifests(table.io())).hasSize(2);
+
+ List<Row> rowsBefore =
+ spark
+ .read()
+ .format("iceberg")
+ .load(tableLocation)
+ .selectExpr("_row_id", "_last_updated_sequence_number", "*")
+ .orderBy("_row_id")
+ .collectAsList();
+ assertThat(rowsBefore).extracting(r ->
r.<Long>getAs("_row_id")).doesNotContainNull();
+
+ SparkActions.get()
+ .rewriteManifests(table)
+ .rewriteIf(manifest -> true)
+ .option(RewriteManifestsSparkAction.USE_CACHING, useCaching)
+ .execute();
+
+ assertThat(table.currentSnapshot().dataManifests(table.io())).hasSize(1);
+
+ List<Row> rowsAfter =
+ spark
+ .read()
+ .format("iceberg")
+ .load(tableLocation)
+ .selectExpr("_row_id", "_last_updated_sequence_number", "*")
+ .orderBy("_row_id")
+ .collectAsList();
+
+ assertThat(rowsAfter).containsExactlyElementsOf(rowsBefore);
+ }
+
+ @TestTemplate
+ public void testRewriteV3PartitionedManifestsPreservesFirstRowId() {
+ assumeThat(formatVersion).isGreaterThanOrEqualTo(3);
+
+ PartitionSpec spec =
PartitionSpec.builderFor(SCHEMA).identity("c1").build();
+ Map<String, String> options = Maps.newHashMap();
+ options.put(TableProperties.FORMAT_VERSION, String.valueOf(formatVersion));
+ Table table = TABLES.create(SCHEMA, spec, options, tableLocation);
+
+ writeRecords(Lists.newArrayList(new ThreeColumnRecord(1, "AAAA", "AAAA")));
+ writeRecords(Lists.newArrayList(new ThreeColumnRecord(2, "BBBB", "BBBB")));
+ table.refresh();
+
+ assertThat(table.currentSnapshot().dataManifests(table.io())).hasSize(2);
+
+ List<Row> rowsBefore =
+ spark
+ .read()
+ .format("iceberg")
+ .load(tableLocation)
+ .selectExpr("_row_id", "_last_updated_sequence_number", "*")
+ .orderBy("_row_id")
+ .collectAsList();
+ assertThat(rowsBefore).extracting(r ->
r.<Long>getAs("_row_id")).doesNotContainNull();
+
+ SparkActions.get()
+ .rewriteManifests(table)
+ .rewriteIf(manifest -> true)
+ .option(RewriteManifestsSparkAction.USE_CACHING, useCaching)
+ .execute();
+
+ assertThat(table.currentSnapshot().dataManifests(table.io())).hasSize(1);
+
+ List<Row> rowsAfter =
+ spark
+ .read()
+ .format("iceberg")
+ .load(tableLocation)
+ .selectExpr("_row_id", "_last_updated_sequence_number", "*")
+ .orderBy("_row_id")
+ .collectAsList();
+
+ assertThat(rowsAfter).containsExactlyElementsOf(rowsBefore);
+ }
+
+ @TestTemplate
+ public void testRewriteManifestsAfterV2ToV3Upgrade() {
+ assumeThat(formatVersion).isGreaterThanOrEqualTo(3);
+
+ PartitionSpec spec = PartitionSpec.unpartitioned();
+ Map<String, String> options = Maps.newHashMap();
+ options.put(TableProperties.FORMAT_VERSION, "2");
+ Table table = TABLES.create(SCHEMA, spec, options, tableLocation);
+
+ ThreeColumnRecord record1 = new ThreeColumnRecord(1, null, "AAAA");
+ ThreeColumnRecord record2 = new ThreeColumnRecord(2, "CCCC", "CCCC");
+ writeRecords(Lists.newArrayList(record1));
+ writeRecords(Lists.newArrayList(record2));
+ table.refresh();
+
+ assertThat(table.currentSnapshot().dataManifests(table.io())).hasSize(2);
+
+ table
+ .updateProperties()
+ .set(TableProperties.FORMAT_VERSION, String.valueOf(formatVersion))
+ .commit();
+
+ SparkActions.get()
+ .rewriteManifests(table)
+ .rewriteIf(manifest -> true)
+ .option(RewriteManifestsSparkAction.USE_CACHING, useCaching)
+ .execute();
+
+ assertThat(table.currentSnapshot().dataManifests(table.io())).hasSize(1);
+
+ List<Row> rowsAfter =
+ spark
+ .read()
+ .format("iceberg")
+ .load(tableLocation)
+ .selectExpr("_row_id", "_last_updated_sequence_number", "*")
+ .orderBy("_row_id")
+ .collectAsList();
+
+ assertThat(rowsAfter)
+ .extracting(r -> r.<Long>getAs("_row_id"))
+ .doesNotContainNull()
+ .doesNotHaveDuplicates();
+ assertThat(rowsAfter)
+ .extracting(r -> new ThreeColumnRecord(r.getAs("c1"), r.getAs("c2"),
r.getAs("c3")))
+ .containsExactlyInAnyOrder(record1, record2);
Review Comment:
I think the assertion after the rewrite manifests that there is exactly 2
manifests before and an assertion that 1 manifest afterwards addresses any
issue coming from the caching catalog; if we were reading stale metadata the
latter assertion would fail.
--
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]
