Skip to content

append row groups to already exist parquet file #557

@jaredzhou

Description

@jaredzhou

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
The SerializedFileWriter can only write row groups to new file , I want to append data to already existing parquet file

Describe the solution you'd like
Since we dont want to break SerializedFileWriter api, we can create a new struct SerializedFileAppender, when new this struct, just fetch and truncate the existing parquet file meta and footer, the rest is just the same as SerializedFileWriter.

Describe alternatives you've considered
Maybe change the SerializedFileWriter to supprot append mode.

Additional context
the code maybe like this

impl<F: ParquetWriter + ChunkReader> SerializedFileAppender<F> {
    pub fn new(
        mut read_writer: F,
        schema: TypePtr,
        properties: WriterPropertiesPtr,
    ) -> Result<Self> {
        let mut total_num_rows = 0;
        let mut row_groups = vec![];
        if read_writer.len() != 0 {
            let (metadata, meta_footer_size) = footer::parse_metadata(&read_writer)?;
            total_num_rows = metadata.file_metadata().num_rows();

            read_writer.seek(SeekFrom::End(-(meta_footer_size as i64)))?;
            row_groups = metadata
                .row_groups()
                .iter()
                .cloned()
                .map(|r| Arc::new(r))
                .collect::<Vec<RowGroupMetaDataPtr>>();
        }

        Ok(Self {
            read_writer,
            schema: schema.clone(),
            descr: Arc::new(SchemaDescriptor::new(schema)),
            props: properties,
            total_num_rows,
            row_groups,
            previous_writer_closed: true,
            is_closed: false,
        })
    }
}

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementAny new improvement worthy of a entry in the changelog

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions