manuzhang commented on code in PR #7653:
URL: https://github.com/apache/iceberg/pull/7653#discussion_r1266380690
##########
spark/v3.1/spark/src/main/java/org/apache/iceberg/spark/actions/BaseDeleteOrphanFilesSparkAction.java:
##########
@@ -296,4 +334,160 @@ private static FlatMapFunction<Iterator<String>, String>
listDirsRecursively(
return files.iterator();
};
}
+
+ @VisibleForTesting
+ static List<String> findOrphanFiles(
+ SparkSession spark,
+ Dataset<Row> actualFileDF,
+ Dataset<Row> validFileDF,
+ Map<String, String> equalSchemes,
+ Map<String, String> equalAuthorities,
+ PrefixMismatchMode prefixMismatchMode) {
+ Dataset<FileMetadata> actualFileMetadataDS =
+ actualFileDF.mapPartitions(
+ toFileMetadata(equalSchemes, equalAuthorities),
Encoders.bean(FileMetadata.class));
+ Dataset<FileMetadata> validFileMetadataDS =
+ validFileDF.mapPartitions(
+ toFileMetadata(equalSchemes, equalAuthorities),
Encoders.bean(FileMetadata.class));
+
+ SetAccumulator<Pair<String, String>> conflicts = new SetAccumulator<>();
+ spark.sparkContext().register(conflicts);
+
+ Column joinCond =
actualFileMetadataDS.col("path").equalTo(validFileMetadataDS.col("path"));
+
+ List<String> orphanFiles =
+ actualFileMetadataDS
+ .joinWith(validFileMetadataDS, joinCond, "leftouter")
+ .mapPartitions(findOrphanFiles(prefixMismatchMode, conflicts),
Encoders.STRING())
+ .collectAsList();
+
+ if (prefixMismatchMode == PrefixMismatchMode.ERROR &&
!conflicts.value().isEmpty()) {
+ throw new ValidationException(
+ "Unable to determine whether certain files are orphan. "
+ + "Metadata references files that match listed/provided files
except for authority/scheme. "
+ + "Please, inspect the conflicting authorities/schemes and
provide which of them are equal "
+ + "by further configuring the action via equalSchemes() and
equalAuthorities() methods. "
+ + "Set the prefix mismatch mode to 'NONE' to ignore remaining
locations with conflicting "
+ + "authorities/schemes or to 'DELETE' iff you are ABSOLUTELY
confident that remaining conflicting "
+ + "authorities/schemes are different. It will be impossible to
recover deleted files. "
+ + "Conflicting authorities/schemes: %s.",
+ conflicts.value());
+ }
+
+ return orphanFiles;
+ }
+
+ private static MapPartitionsFunction<Tuple2<FileMetadata, FileMetadata>,
String> findOrphanFiles(
+ PrefixMismatchMode mode, SetAccumulator<Pair<String, String>> conflicts)
{
+ return rows -> {
+ Iterator<String> transformed =
+ Iterators.transform(
+ rows,
Review Comment:
```
> Task :iceberg-spark:iceberg-spark-3.1_2.12:spotlessJavaCheck FAILED
FAILURE: Build failed with an exception.
* What went wrong:
Execution failed for task
':iceberg-spark:iceberg-spark-3.1_2.12:spotlessJavaCheck'.
> The following files had format violations:
src/main/java/org/apache/iceberg/spark/actions/BaseDeleteOrphanFilesSparkAction.java
@@ -394,7 +394,9 @@
······PrefixMismatchMode·mode,·SetAccumulator<Pair<String,·String>>·conflicts)·{
····return·rows·->·{
······Iterator<String>·transformed·=
-··········Iterators.transform(rows,·row·->·{
+··········Iterators.transform(
+··············rows,
+··············row·->·{
················FileMetadata·actual·=·row._1;
················FileMetadata·valid·=·row._2;
@@ -429,7 +431,9 @@
··private·static·MapPartitionsFunction<Row,·FileMetadata>·toFileMetadata(
······Map<String,·String>·equalSchemesMap,·Map<String,·String>·equalAuthoritiesMap)·{
····return·rows·->
-········Iterators.transform(rows,·row·->·{
+········Iterators.transform(
+············rows,
+············row·->·{
··············String·location·=·row.getString(0);
··············URI·uri·=·new·Path(location).toUri();
··············String·scheme·=·equalSchemesMap.getOrDefault(uri.getScheme(),·uri.getScheme());
Run './gradlew :iceberg-spark:iceberg-spark-3.1_2.12:spotlessApply' to fix
these violations.
```
##########
spark/v3.1/spark/src/main/java/org/apache/iceberg/spark/actions/BaseDeleteOrphanFilesSparkAction.java:
##########
@@ -296,4 +334,160 @@ private static FlatMapFunction<Iterator<String>, String>
listDirsRecursively(
return files.iterator();
};
}
+
+ @VisibleForTesting
+ static List<String> findOrphanFiles(
+ SparkSession spark,
+ Dataset<Row> actualFileDF,
+ Dataset<Row> validFileDF,
+ Map<String, String> equalSchemes,
+ Map<String, String> equalAuthorities,
+ PrefixMismatchMode prefixMismatchMode) {
+ Dataset<FileMetadata> actualFileMetadataDS =
+ actualFileDF.mapPartitions(
+ toFileMetadata(equalSchemes, equalAuthorities),
Encoders.bean(FileMetadata.class));
+ Dataset<FileMetadata> validFileMetadataDS =
+ validFileDF.mapPartitions(
+ toFileMetadata(equalSchemes, equalAuthorities),
Encoders.bean(FileMetadata.class));
+
+ SetAccumulator<Pair<String, String>> conflicts = new SetAccumulator<>();
+ spark.sparkContext().register(conflicts);
+
+ Column joinCond =
actualFileMetadataDS.col("path").equalTo(validFileMetadataDS.col("path"));
+
+ List<String> orphanFiles =
+ actualFileMetadataDS
+ .joinWith(validFileMetadataDS, joinCond, "leftouter")
+ .mapPartitions(findOrphanFiles(prefixMismatchMode, conflicts),
Encoders.STRING())
+ .collectAsList();
+
+ if (prefixMismatchMode == PrefixMismatchMode.ERROR &&
!conflicts.value().isEmpty()) {
+ throw new ValidationException(
+ "Unable to determine whether certain files are orphan. "
+ + "Metadata references files that match listed/provided files
except for authority/scheme. "
+ + "Please, inspect the conflicting authorities/schemes and
provide which of them are equal "
+ + "by further configuring the action via equalSchemes() and
equalAuthorities() methods. "
+ + "Set the prefix mismatch mode to 'NONE' to ignore remaining
locations with conflicting "
+ + "authorities/schemes or to 'DELETE' iff you are ABSOLUTELY
confident that remaining conflicting "
+ + "authorities/schemes are different. It will be impossible to
recover deleted files. "
+ + "Conflicting authorities/schemes: %s.",
+ conflicts.value());
+ }
+
+ return orphanFiles;
+ }
+
+ private static MapPartitionsFunction<Tuple2<FileMetadata, FileMetadata>,
String> findOrphanFiles(
+ PrefixMismatchMode mode, SetAccumulator<Pair<String, String>> conflicts)
{
+ return rows -> {
+ Iterator<String> transformed =
+ Iterators.transform(
+ rows,
Review Comment:
```
> Task :iceberg-spark:iceberg-spark-3.1_2.12:spotlessJavaCheck FAILED
FAILURE: Build failed with an exception.
* What went wrong:
Execution failed for task
':iceberg-spark:iceberg-spark-3.1_2.12:spotlessJavaCheck'.
> The following files had format violations:
src/main/java/org/apache/iceberg/spark/actions/BaseDeleteOrphanFilesSparkAction.java
@@ -394,7 +394,9 @@
······PrefixMismatchMode·mode,·SetAccumulator<Pair<String,·String>>·conflicts)·{
····return·rows·->·{
······Iterator<String>·transformed·=
-··········Iterators.transform(rows,·row·->·{
+··········Iterators.transform(
+··············rows,
+··············row·->·{
················FileMetadata·actual·=·row._1;
················FileMetadata·valid·=·row._2;
@@ -429,7 +431,9 @@
··private·static·MapPartitionsFunction<Row,·FileMetadata>·toFileMetadata(
······Map<String,·String>·equalSchemesMap,·Map<String,·String>·equalAuthoritiesMap)·{
····return·rows·->
-········Iterators.transform(rows,·row·->·{
+········Iterators.transform(
+············rows,
+············row·->·{
··············String·location·=·row.getString(0);
··············URI·uri·=·new·Path(location).toUri();
··············String·scheme·=·equalSchemesMap.getOrDefault(uri.getScheme(),·uri.getScheme());
Run './gradlew :iceberg-spark:iceberg-spark-3.1_2.12:spotlessApply' to fix
these violations.
```
--
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]
