[
https://issues.apache.org/jira/browse/PARQUET-2385?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17789879#comment-17789879
]
ASF GitHub Bot commented on PARQUET-2385:
-----------------------------------------
wgtmac commented on code in PR #1203:
URL: https://github.com/apache/parquet-mr/pull/1203#discussion_r1405536183
##########
parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetWriter.java:
##########
@@ -303,6 +303,32 @@ public ParquetWriter(Path file, Configuration conf,
WriteSupport<T> writeSupport
int maxPaddingSize,
ParquetProperties encodingProps,
FileEncryptionProperties encryptionProperties) throws IOException {
+ this(
+ file,
+ mode,
+ writeSupport,
+ compressionCodecName,
+ new CodecFactory(conf, encodingProps.getPageSizeThreshold()),
+ rowGroupSize,
+ validating,
+ conf,
+ maxPaddingSize,
+ encodingProps,
+ encryptionProperties);
+ }
+
+ ParquetWriter(
+ OutputFile file,
Review Comment:
It seems the indentation of constructor is 4 spaces elsewhere.
##########
parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetWriter.java:
##########
@@ -321,17 +347,17 @@ public ParquetWriter(Path file, Configuration conf,
WriteSupport<T> writeSupport
encodingProps.getPageWriteChecksumEnabled(), encryptionProperties);
fileWriter.start();
- this.codecFactory = new CodecFactory(conf,
encodingProps.getPageSizeThreshold());
+ this.codecFactory = codecFactory;
CompressionCodecFactory.BytesInputCompressor compressor =
codecFactory.getCompressor(compressionCodecName);
this.writer = new InternalParquetRecordWriter<T>(
- fileWriter,
- writeSupport,
- schema,
- writeContext.getExtraMetaData(),
- rowGroupSize,
- compressor,
- validating,
- encodingProps);
+ fileWriter,
Review Comment:
Could you please revert the irrelevant style change?
> Don't initialize CodecFactory in ParquetWriter
> ----------------------------------------------
>
> Key: PARQUET-2385
> URL: https://issues.apache.org/jira/browse/PARQUET-2385
> Project: Parquet
> Issue Type: Improvement
> Components: parquet-mr
> Reporter: Atour Mousavi Gourabi
> Priority: Major
>
> In ParquetWriter we initialize a CodecFactory, instead we should allow users
> to set their own via the builder as to provide a little more flexibility
> (analogous to PARQUET-2282).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
