File Lock Issue in ParquetRewriter class in Apache Parquet Hadoop 1.14.1 on Windows 11 22H2

[cetindogu]

Describe the bug, including details regarding any error messages, version, and platform.

java 21, Apache Parquet Hadoop 1.14.1 windows 11

after successfully pass the try block, my inputs(.parquet files) are still used by another process. so i can't delete the inputs (the parquet files)

java.nio.file.FileSystemException: test\sample\dayhour-150-0-6a57c328-8610-46f8-bf7e-311f8605def0.parquet: This process cannot access the file because the file is being used by another process

    RewriteOptions rewriteOptions = new RewriteOptions.Builder(confForLocal, paths, outputPath).build();

    try (ParquetRewriter rewriter = new ParquetRewriter(rewriteOptions)) {
        rewriter.processBlocks();
    } catch (IOException e) {
        logger.error("Cannot merge parquet files", e);
        return Optional.empty();
    }

    if (config.isDeleteOriginalParquetsAfterMerge()) {
        deleteMergedParquets(inputs);
    }

 private void deleteMergedParquets(List<String> inputs) {
    for (String input : inputs) {
        try {
            Files.delete(Paths.get(input));             
            logger.info("Deleting {} file since this file is merged with another one", input);
        } catch (IOException e) {
            logger.error("Cannot remove files for {}", input, e);
        }
    }
}

Component(s)

No response

[cetindogu]

in ParquetReWriter at line 256 there is reader = inputFiles.poll();

this line must be reader.close(); reader = inputFiles.poll();

[cetindogu]

is there anyone to check my pull request ? https://github.com/apache/parquet-java/pull/3002

[cetindogu]

also null checked

[cetindogu]

i moved "reader.close();" command to one line up (in same line with "if" clouse)

https://github.com/apache/parquet-java/pull/3002

[cetindogu]

https://github.com/apache/parquet-java/pull/3002

mvc spotless:apply is run.

[cetindogu]

bug is fixed with pull request https://github.com/apache/parquet-java/pull/3002 so next release version should be waited.