Skip to content

Spark: Fix RewriteManifestsSparkAction sortBy() for dotted partition field names #17428

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
agnes-xinyi-lu wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Spark: Fix RewriteManifestsSparkAction sortBy() for dotted partition field names #17428

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
74097f6
Spark 4.1:Spark: Fix RewriteManifestsSparkAction sortBy() for dotted …
agnes-xinyi-lu Jul 30, 2026
2879c3c
resolve comments and apply fix to all spark versions
agnes-xinyi-lu Jul 30, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 9 additions & 2 deletions ...3.5/spark/src/main/java/org/apache/iceberg/spark/actions/RewriteManifestsSparkAction.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -314,10 +314,17 @@ private Column sortColumn() {
partitionFieldClustering);

// Map the top level partition column names to the column name referenced within the manifest
// entry dataframe
// entry dataframe. Backtick-quote the partition field name (escaping any embedded backtick)
// because it may itself contain a literal '.' (e.g. when partitioning on a nested source
// column) - the partition struct is always flat, so unquoted, col() would misparse that dot
// as a further level of nesting instead of treating the whole name as one field.
Column[] partitionColumns =
partitionFieldClustering.stream()
.map(p -> col(DATA_FILE_PARTITION_COLUMN_NAME + "." + p))
.map(
p ->
col(
String.format(
"%s.`%s`", DATA_FILE_PARTITION_COLUMN_NAME, p.replace("`", "``"))))
.toArray(Column[]::new);

// Form a new temporary column to cluster manifests on, based on the custom clustering columns
Expand Down
41 changes: 41 additions & 0 deletions ...v3.5/spark/src/test/java/org/apache/iceberg/spark/actions/TestRewriteManifestsAction.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -756,6 +756,47 @@ public void testRewriteManifestsPartitionedTableWithCustomSorting() throws IOExc
assertThat(manifestsDf.count()).as("There should be at least 2 manifests").isGreaterThan(1L);
}

@TestTemplate
public void
testRewriteManifestsPartitionedTableWithCustomSortingOnFieldNameContainingDotAndBacktick()
throws IOException {
// Regression test: partitioning on a nested source column (struct "a" with child field
// "b`c") used to make sortBy() fail at execution time, since the resulting partition field is
// named "a.b`c" but the partition struct is flat - col() misparsed the dot as a further
// nesting level instead of treating "a.b`c" as a single field name, and even once quoted, the
// embedded backtick must itself be escaped or the quoting is malformed.
Schema schema =
new Schema(
optional(1, "a", Types.StructType.of(optional(2, "b`c", Types.StringType.get()))),
optional(3, "ds", Types.StringType.get()));
PartitionSpec spec = PartitionSpec.builderFor(schema).identity("a.b`c").build();
Map<String, String> options = Maps.newHashMap();
options.put(TableProperties.FORMAT_VERSION, String.valueOf(formatVersion));
options.put(TableProperties.SNAPSHOT_ID_INHERITANCE_ENABLED, snapshotIdInheritanceEnabled);
Table table = TABLES.create(schema, spec, options, tableLocation);

// write two manifests so the rewrite doesn't short-circuit via the "already optimal" early
// return, and sortColumn() actually gets exercised
ManifestFile manifest1 =
writeManifest(table, ImmutableList.of(newDataFile(table, TestHelpers.Row.of("val1"))));
table.newFastAppend().appendManifest(manifest1).commit();

ManifestFile manifest2 =
writeManifest(table, ImmutableList.of(newDataFile(table, TestHelpers.Row.of("val2"))));
table.newFastAppend().appendManifest(manifest2).commit();

RewriteManifests.Result result =
SparkActions.get()
.rewriteManifests(table)
.rewriteIf(manifest -> true)
.sortBy(ImmutableList.of("a.b`c"))
.option(RewriteManifestsSparkAction.USE_CACHING, useCaching)
.execute();

assertThat(result.rewrittenManifests()).hasSize(2);
assertThat(result.addedManifests()).hasSizeGreaterThanOrEqualTo(1);
}

@TestTemplate
public void testRewriteManifestsWithPredicate() throws IOException {
PartitionSpec spec = PartitionSpec.builderFor(SCHEMA).identity("c1").truncate("c2", 2).build();
Expand Down
11 changes: 9 additions & 2 deletions ...4.0/spark/src/main/java/org/apache/iceberg/spark/actions/RewriteManifestsSparkAction.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -314,10 +314,17 @@ private Column sortColumn() {
partitionFieldClustering);

// Map the top level partition column names to the column name referenced within the manifest
// entry dataframe
// entry dataframe. Backtick-quote the partition field name (escaping any embedded backtick)
// because it may itself contain a literal '.' (e.g. when partitioning on a nested source
// column) - the partition struct is always flat, so unquoted, col() would misparse that dot
// as a further level of nesting instead of treating the whole name as one field.
Column[] partitionColumns =
partitionFieldClustering.stream()
.map(p -> col(DATA_FILE_PARTITION_COLUMN_NAME + "." + p))
.map(
p ->
col(
String.format(
"%s.`%s`", DATA_FILE_PARTITION_COLUMN_NAME, p.replace("`", "``"))))
.toArray(Column[]::new);

// Form a new temporary column to cluster manifests on, based on the custom clustering columns
Expand Down
41 changes: 41 additions & 0 deletions ...v4.0/spark/src/test/java/org/apache/iceberg/spark/actions/TestRewriteManifestsAction.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -756,6 +756,47 @@ public void testRewriteManifestsPartitionedTableWithCustomSorting() throws IOExc
assertThat(manifestsDf.count()).as("There should be at least 2 manifests").isGreaterThan(1L);
}

@TestTemplate
public void
testRewriteManifestsPartitionedTableWithCustomSortingOnFieldNameContainingDotAndBacktick()
throws IOException {
// Regression test: partitioning on a nested source column (struct "a" with child field
// "b`c") used to make sortBy() fail at execution time, since the resulting partition field is
// named "a.b`c" but the partition struct is flat - col() misparsed the dot as a further
// nesting level instead of treating "a.b`c" as a single field name, and even once quoted, the
// embedded backtick must itself be escaped or the quoting is malformed.
Schema schema =
new Schema(
optional(1, "a", Types.StructType.of(optional(2, "b`c", Types.StringType.get()))),
optional(3, "ds", Types.StringType.get()));
PartitionSpec spec = PartitionSpec.builderFor(schema).identity("a.b`c").build();
Map<String, String> options = Maps.newHashMap();
options.put(TableProperties.FORMAT_VERSION, String.valueOf(formatVersion));
options.put(TableProperties.SNAPSHOT_ID_INHERITANCE_ENABLED, snapshotIdInheritanceEnabled);
Table table = TABLES.create(schema, spec, options, tableLocation);

// write two manifests so the rewrite doesn't short-circuit via the "already optimal" early
// return, and sortColumn() actually gets exercised
ManifestFile manifest1 =
writeManifest(table, ImmutableList.of(newDataFile(table, TestHelpers.Row.of("val1"))));
table.newFastAppend().appendManifest(manifest1).commit();

ManifestFile manifest2 =
writeManifest(table, ImmutableList.of(newDataFile(table, TestHelpers.Row.of("val2"))));
table.newFastAppend().appendManifest(manifest2).commit();

RewriteManifests.Result result =
SparkActions.get()
.rewriteManifests(table)
.rewriteIf(manifest -> true)
.sortBy(ImmutableList.of("a.b`c"))
.option(RewriteManifestsSparkAction.USE_CACHING, useCaching)
.execute();

assertThat(result.rewrittenManifests()).hasSize(2);
assertThat(result.addedManifests()).hasSizeGreaterThanOrEqualTo(1);
}

@TestTemplate
public void testRewriteManifestsWithPredicate() throws IOException {
PartitionSpec spec = PartitionSpec.builderFor(SCHEMA).identity("c1").truncate("c2", 2).build();
Expand Down
11 changes: 9 additions & 2 deletions ...4.1/spark/src/main/java/org/apache/iceberg/spark/actions/RewriteManifestsSparkAction.java
View file
Open in desktop

@uros-b uros-b Jul 30, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How about spark/v4.0 and spark/v3.5? Shall we fix there too?

@agnes-xinyi-lu agnes-xinyi-lu Jul 30, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

done.

Original file line number Diff line number Diff line change
Expand Up @@ -314,10 +314,17 @@ private Column sortColumn() {
partitionFieldClustering);

// Map the top level partition column names to the column name referenced within the manifest
// entry dataframe
// entry dataframe. Backtick-quote the partition field name (escaping any embedded backtick)
// because it may itself contain a literal '.' (e.g. when partitioning on a nested source
// column) - the partition struct is always flat, so unquoted, col() would misparse that dot
// as a further level of nesting instead of treating the whole name as one field.
Column[] partitionColumns =
partitionFieldClustering.stream()
.map(p -> col(DATA_FILE_PARTITION_COLUMN_NAME + "." + p))
.map(
p ->
col(
String.format(
"%s.`%s`", DATA_FILE_PARTITION_COLUMN_NAME, p.replace("`", "``"))))
.toArray(Column[]::new);

// Form a new temporary column to cluster manifests on, based on the custom clustering columns
Expand Down
41 changes: 41 additions & 0 deletions ...v4.1/spark/src/test/java/org/apache/iceberg/spark/actions/TestRewriteManifestsAction.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -756,6 +756,47 @@ public void testRewriteManifestsPartitionedTableWithCustomSorting() throws IOExc
assertThat(manifestsDf.count()).as("There should be at least 2 manifests").isGreaterThan(1L);
}

@TestTemplate
public void
testRewriteManifestsPartitionedTableWithCustomSortingOnFieldNameContainingDotAndBacktick()
throws IOException {
// Regression test: partitioning on a nested source column (struct "a" with child field
// "b`c") used to make sortBy() fail at execution time, since the resulting partition field is
// named "a.b`c" but the partition struct is flat - col() misparsed the dot as a further
// nesting level instead of treating "a.b`c" as a single field name, and even once quoted, the
// embedded backtick must itself be escaped or the quoting is malformed.
Schema schema =
new Schema(
optional(1, "a", Types.StructType.of(optional(2, "b`c", Types.StringType.get()))),
optional(3, "ds", Types.StringType.get()));
PartitionSpec spec = PartitionSpec.builderFor(schema).identity("a.b`c").build();
Map<String, String> options = Maps.newHashMap();
options.put(TableProperties.FORMAT_VERSION, String.valueOf(formatVersion));
options.put(TableProperties.SNAPSHOT_ID_INHERITANCE_ENABLED, snapshotIdInheritanceEnabled);
Table table = TABLES.create(schema, spec, options, tableLocation);

// write two manifests so the rewrite doesn't short-circuit via the "already optimal" early
// return, and sortColumn() actually gets exercised
ManifestFile manifest1 =
writeManifest(table, ImmutableList.of(newDataFile(table, TestHelpers.Row.of("val1"))));
table.newFastAppend().appendManifest(manifest1).commit();

ManifestFile manifest2 =
writeManifest(table, ImmutableList.of(newDataFile(table, TestHelpers.Row.of("val2"))));
table.newFastAppend().appendManifest(manifest2).commit();

RewriteManifests.Result result =
SparkActions.get()
.rewriteManifests(table)
.rewriteIf(manifest -> true)
.sortBy(ImmutableList.of("a.b`c"))
.option(RewriteManifestsSparkAction.USE_CACHING, useCaching)
.execute();

assertThat(result.rewrittenManifests()).hasSize(2);
assertThat(result.addedManifests()).hasSizeGreaterThanOrEqualTo(1);
}

@TestTemplate
public void testRewriteManifestsWithPredicate() throws IOException {
PartitionSpec spec = PartitionSpec.builderFor(SCHEMA).identity("c1").truncate("c2", 2).build();
Expand Down
Loading