hyparquet-writer/src/write.js

35 lines
977 B
JavaScript
Raw Normal View History

2025-04-07 08:02:21 +00:00
import { ParquetWriter } from './parquet-writer.js'
2025-04-08 06:14:48 +00:00
import { schemaFromColumnData } from './schema.js'
import { ByteWriter } from './bytewriter.js'
2025-03-26 04:06:43 +00:00
/**
* Write data as parquet to an ArrayBuffer
*
2025-04-01 06:32:14 +00:00
* @import {KeyValue} from 'hyparquet/src/types.js'
* @import {ColumnData} from '../src/types.js'
2025-03-27 07:27:22 +00:00
* @param {object} options
* @param {ColumnData[]} options.columnData
* @param {boolean} [options.compressed]
2025-04-03 20:21:57 +00:00
* @param {boolean} [options.statistics]
2025-04-03 07:42:54 +00:00
* @param {number} [options.rowGroupSize]
2025-03-31 20:42:57 +00:00
* @param {KeyValue[]} [options.kvMetadata]
2025-03-26 04:06:43 +00:00
* @returns {ArrayBuffer}
*/
2025-04-03 20:21:57 +00:00
export function parquetWrite({ columnData, compressed = true, statistics = true, rowGroupSize = 100000, kvMetadata }) {
2025-04-07 08:02:21 +00:00
const schema = schemaFromColumnData(columnData)
2025-04-08 06:14:48 +00:00
const writer = new ByteWriter()
const pq = new ParquetWriter({
writer,
2025-04-07 08:02:21 +00:00
schema,
compressed,
statistics,
kvMetadata,
})
2025-04-08 06:14:48 +00:00
pq.write({
2025-04-07 08:02:21 +00:00
columnData,
rowGroupSize,
})
2025-04-08 06:14:48 +00:00
pq.finish()
return writer.getBuffer()
2025-03-26 04:06:43 +00:00
}