vinishjail97 commented on code in PR #806:
URL: https://github.com/apache/incubator-xtable/pull/806#discussion_r2839192552
##########
xtable-core/src/test/java/org/apache/xtable/parquet/ITParquetConversionSource.java:
##########
@@ -153,7 +152,7 @@ private static ConversionConfig getTableSyncConfig(
.name(tableName)
.formatName(formatName)
// set the metadata path to the data path as the
default (required by Hudi)
- .basePath(table.getDataPath())
+ .basePath(table.getBasePath())
Review Comment:
Nit: the comment on line 154 still says "set the metadata path to the
**data** path" but the code now uses `getBasePath()`. Update the comment to
match.
##########
xtable-core/src/test/java/org/apache/xtable/parquet/ITParquetConversionSource.java:
##########
@@ -319,43 +268,62 @@ public void
testFilePartitionedData(TableFormatPartitionDataHolder tableFormatPa
new Timestamp(System.currentTimeMillis() + 1500)));
Dataset<Row> dfAppend = sparkSession.createDataFrame(dataToAppend,
schema);
- dfAppend
- .withColumn(
- "year",
functions.year(functions.col("timestamp").cast(DataTypes.TimestampType)))
- .withColumn(
- "month",
-
functions.date_format(functions.col("timestamp").cast(DataTypes.TimestampType),
"MM"))
- .write()
- .mode(SaveMode.Append)
- .partitionBy("year", "month")
- .parquet(dataPathPart);
- GenericTable tableAppend;
- tableAppend =
- GenericTable.getInstance(
- tableName, Paths.get(dataPathPart), sparkSession, jsc,
sourceTableFormat, true);
- try (GenericTable tableToCloseAppended = tableAppend) {
- ConversionConfig conversionConfigAppended =
- getTableSyncConfig(
- sourceTableFormat,
- SyncMode.FULL,
- tableName,
- tableAppend,
- targetTableFormats,
- xTablePartitionConfig,
- null);
- ConversionController conversionControllerAppended =
- new ConversionController(jsc.hadoopConfiguration());
- conversionControllerAppended.sync(conversionConfigAppended,
conversionSourceProvider);
+ writeData(dfAppend, dataPath, xTablePartitionConfig);
+ ConversionConfig conversionConfigAppended =
+ getTableSyncConfig(
+ sourceTableFormat,
+ tableFormatPartitionDataHolder.getSyncMode(),
+ tableName,
+ table,
+ targetTableFormats,
+ xTablePartitionConfig,
+ null);
+ ConversionController conversionControllerAppended =
+ new ConversionController(jsc.hadoopConfiguration());
+ conversionControllerAppended.sync(conversionConfigAppended,
conversionSourceProvider);
+ checkDatasetEquivalenceWithFilter(
+ sourceTableFormat, table, targetTableFormats, isPartitioned);
+ }
+ }
+
+ private void writeData(Dataset<Row> df, String dataPath, String
partitionConfig) {
+ if (partitionConfig != null) {
+ // extract partition columns from config
+ String[] partitionCols =
+ Arrays.stream(partitionConfig.split(":")[2].split("/"))
+ .map(s -> s.split("=")[0])
+ .toArray(String[]::new);
+ // add partition columns to dataframe
+ for (String partitionCol : partitionCols) {
+ if (partitionCol.equals("year")) {
Review Comment:
The `if/else if` chain only handles "year", "month", and "day". If the
partition config ever contains a different column, no column gets added to the
DataFrame but `partitionBy` still references it — causing a confusing
`AnalysisException`. Add an `else` with `throw new
IllegalArgumentException("Unsupported partition column: " + partitionCol)` to
fail fast.
--
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]
