2025-03-26 00:49:59 +00:00
|
|
|
import { parquetMetadata } from 'hyparquet'
|
|
|
|
|
import { describe, expect, it } from 'vitest'
|
2025-04-08 06:14:48 +00:00
|
|
|
import { ByteWriter } from '../src/bytewriter.js'
|
2025-03-26 00:49:59 +00:00
|
|
|
import { writeMetadata } from '../src/metadata.js'
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* @import {FileMetaData} from 'hyparquet'
|
|
|
|
|
* @type {FileMetaData}
|
|
|
|
|
*/
|
|
|
|
|
export const exampleMetadata = {
|
|
|
|
|
version: 2,
|
|
|
|
|
created_by: 'hyparquet',
|
|
|
|
|
schema: [
|
2025-03-26 05:36:06 +00:00
|
|
|
{ name: 'root', num_children: 6 },
|
2025-03-26 03:15:14 +00:00
|
|
|
{ name: 'bool', type: 'BOOLEAN', repetition_type: 'REQUIRED' },
|
|
|
|
|
{ name: 'int', type: 'INT32', repetition_type: 'REQUIRED' },
|
|
|
|
|
{ name: 'bigint', type: 'INT64', repetition_type: 'REQUIRED' },
|
|
|
|
|
{ name: 'double', type: 'DOUBLE', repetition_type: 'REQUIRED' },
|
2025-03-26 07:45:22 +00:00
|
|
|
{ name: 'string', type: 'BYTE_ARRAY', repetition_type: 'REQUIRED', converted_type: 'UTF8' },
|
2025-03-26 05:36:06 +00:00
|
|
|
{ name: 'nullable', type: 'BOOLEAN', repetition_type: 'OPTIONAL' },
|
2025-03-26 00:49:59 +00:00
|
|
|
],
|
|
|
|
|
num_rows: 4n,
|
|
|
|
|
row_groups: [{
|
|
|
|
|
columns: [
|
|
|
|
|
{
|
|
|
|
|
file_path: 'bool',
|
2025-03-26 03:15:14 +00:00
|
|
|
file_offset: 4n,
|
2025-03-26 00:49:59 +00:00
|
|
|
meta_data: {
|
|
|
|
|
type: 'BOOLEAN',
|
|
|
|
|
encodings: ['PLAIN'],
|
|
|
|
|
path_in_schema: ['bool'],
|
2025-03-27 07:01:24 +00:00
|
|
|
codec: 'SNAPPY',
|
2025-03-26 00:49:59 +00:00
|
|
|
num_values: 4n,
|
2025-03-27 07:01:24 +00:00
|
|
|
total_uncompressed_size: 24n,
|
|
|
|
|
total_compressed_size: 24n,
|
2025-03-26 00:49:59 +00:00
|
|
|
data_page_offset: 4n,
|
2025-04-03 20:21:57 +00:00
|
|
|
statistics: {
|
|
|
|
|
null_count: 0n,
|
|
|
|
|
min_value: false,
|
|
|
|
|
max_value: true,
|
|
|
|
|
},
|
2025-03-26 00:49:59 +00:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
file_path: 'int',
|
2025-03-27 07:01:24 +00:00
|
|
|
file_offset: 28n,
|
2025-03-26 00:49:59 +00:00
|
|
|
meta_data: {
|
|
|
|
|
type: 'INT32',
|
|
|
|
|
encodings: ['PLAIN'],
|
|
|
|
|
path_in_schema: ['int'],
|
2025-03-27 07:01:24 +00:00
|
|
|
codec: 'SNAPPY',
|
2025-03-26 00:49:59 +00:00
|
|
|
num_values: 4n,
|
2025-03-27 07:01:24 +00:00
|
|
|
total_uncompressed_size: 39n,
|
|
|
|
|
total_compressed_size: 39n,
|
|
|
|
|
data_page_offset: 28n,
|
2025-04-03 20:21:57 +00:00
|
|
|
statistics: {
|
|
|
|
|
null_count: 0n,
|
|
|
|
|
min_value: 0,
|
|
|
|
|
max_value: 0x7fffffff,
|
|
|
|
|
},
|
2025-03-26 00:49:59 +00:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
file_path: 'bigint',
|
2025-03-27 07:01:24 +00:00
|
|
|
file_offset: 67n,
|
2025-03-26 00:49:59 +00:00
|
|
|
meta_data: {
|
|
|
|
|
type: 'INT64',
|
|
|
|
|
encodings: ['PLAIN'],
|
|
|
|
|
path_in_schema: ['bigint'],
|
2025-03-27 07:01:24 +00:00
|
|
|
codec: 'SNAPPY',
|
2025-03-26 00:49:59 +00:00
|
|
|
num_values: 4n,
|
2025-03-27 07:01:24 +00:00
|
|
|
total_uncompressed_size: 43n,
|
|
|
|
|
total_compressed_size: 43n,
|
|
|
|
|
data_page_offset: 67n,
|
2025-04-03 20:21:57 +00:00
|
|
|
statistics: {
|
|
|
|
|
null_count: 0n,
|
|
|
|
|
min_value: 0n,
|
|
|
|
|
max_value: 0x7fffffffffffffffn,
|
|
|
|
|
},
|
2025-03-26 00:49:59 +00:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
file_path: 'double',
|
2025-03-27 07:01:24 +00:00
|
|
|
file_offset: 110n,
|
2025-03-26 00:49:59 +00:00
|
|
|
meta_data: {
|
|
|
|
|
type: 'DOUBLE',
|
|
|
|
|
encodings: ['PLAIN'],
|
|
|
|
|
path_in_schema: ['double'],
|
2025-03-27 07:01:24 +00:00
|
|
|
codec: 'SNAPPY',
|
2025-03-26 00:49:59 +00:00
|
|
|
num_values: 4n,
|
2025-03-27 07:01:24 +00:00
|
|
|
total_uncompressed_size: 51n,
|
|
|
|
|
total_compressed_size: 51n,
|
|
|
|
|
data_page_offset: 110n,
|
2025-04-03 20:21:57 +00:00
|
|
|
statistics: {
|
|
|
|
|
null_count: 0n,
|
|
|
|
|
min_value: 0,
|
|
|
|
|
max_value: 1e100,
|
|
|
|
|
},
|
2025-03-26 03:15:14 +00:00
|
|
|
},
|
|
|
|
|
},
|
|
|
|
|
{
|
|
|
|
|
file_path: 'string',
|
2025-03-27 07:01:24 +00:00
|
|
|
file_offset: 161n,
|
2025-03-26 03:15:14 +00:00
|
|
|
meta_data: {
|
|
|
|
|
type: 'BYTE_ARRAY',
|
|
|
|
|
encodings: ['PLAIN'],
|
|
|
|
|
path_in_schema: ['string'],
|
2025-03-27 07:01:24 +00:00
|
|
|
codec: 'SNAPPY',
|
2025-03-26 03:15:14 +00:00
|
|
|
num_values: 4n,
|
|
|
|
|
total_uncompressed_size: 42n,
|
|
|
|
|
total_compressed_size: 42n,
|
2025-03-27 07:01:24 +00:00
|
|
|
data_page_offset: 161n,
|
2025-04-03 20:21:57 +00:00
|
|
|
statistics: {
|
|
|
|
|
null_count: 0n,
|
|
|
|
|
min_value: 'a',
|
|
|
|
|
max_value: 'd',
|
|
|
|
|
},
|
2025-03-26 00:49:59 +00:00
|
|
|
},
|
|
|
|
|
},
|
2025-03-26 05:36:06 +00:00
|
|
|
{
|
|
|
|
|
file_path: 'nullable',
|
2025-03-27 07:01:24 +00:00
|
|
|
file_offset: 203n,
|
2025-03-26 05:36:06 +00:00
|
|
|
meta_data: {
|
|
|
|
|
type: 'BOOLEAN',
|
|
|
|
|
encodings: ['PLAIN'],
|
|
|
|
|
path_in_schema: ['nullable'],
|
2025-03-27 07:01:24 +00:00
|
|
|
codec: 'SNAPPY',
|
2025-03-26 05:36:06 +00:00
|
|
|
num_values: 4n,
|
2025-03-27 07:01:24 +00:00
|
|
|
total_uncompressed_size: 26n,
|
|
|
|
|
total_compressed_size: 26n,
|
|
|
|
|
data_page_offset: 203n,
|
2025-04-03 20:21:57 +00:00
|
|
|
statistics: {
|
|
|
|
|
null_count: 2n,
|
|
|
|
|
min_value: false,
|
|
|
|
|
max_value: true,
|
|
|
|
|
},
|
2025-03-26 05:36:06 +00:00
|
|
|
},
|
|
|
|
|
},
|
2025-03-26 00:49:59 +00:00
|
|
|
],
|
2025-03-27 07:01:24 +00:00
|
|
|
total_byte_size: 225n,
|
2025-03-26 00:49:59 +00:00
|
|
|
num_rows: 4n,
|
|
|
|
|
}],
|
2025-04-03 20:21:57 +00:00
|
|
|
metadata_length: 432,
|
2025-03-26 00:49:59 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
describe('writeMetadata', () => {
|
|
|
|
|
it('writes metadata and parses in hyparquet', () => {
|
2025-04-08 06:14:48 +00:00
|
|
|
const writer = new ByteWriter()
|
2025-03-26 00:49:59 +00:00
|
|
|
|
|
|
|
|
// Write header PAR1
|
|
|
|
|
writer.appendUint32(0x31524150)
|
|
|
|
|
|
|
|
|
|
// Write metadata
|
2025-03-31 20:42:57 +00:00
|
|
|
const withKvMetadata = {
|
|
|
|
|
...exampleMetadata,
|
|
|
|
|
key_value_metadata: [
|
|
|
|
|
{ key: 'key1', value: 'value1' },
|
|
|
|
|
{ key: 'key2', value: 'value2' },
|
|
|
|
|
],
|
2025-04-03 20:21:57 +00:00
|
|
|
metadata_length: 464,
|
2025-03-31 20:42:57 +00:00
|
|
|
}
|
|
|
|
|
writeMetadata(writer, withKvMetadata)
|
2025-03-26 00:49:59 +00:00
|
|
|
|
|
|
|
|
// Write footer PAR1
|
|
|
|
|
writer.appendUint32(0x31524150)
|
|
|
|
|
|
|
|
|
|
const file = writer.getBuffer()
|
|
|
|
|
const output = parquetMetadata(file)
|
|
|
|
|
|
|
|
|
|
/** @type {FileMetaData} */
|
2025-03-31 20:42:57 +00:00
|
|
|
expect(output).toEqual(withKvMetadata)
|
2025-03-26 00:49:59 +00:00
|
|
|
})
|
|
|
|
|
|
|
|
|
|
})
|