|
1 | | -import HighTable, { DataFrame, rowCache } from 'hightable' |
2 | | -import { FileMetaData, byteLengthFromUrl, parquetMetadataAsync, parquetSchema } from 'hyparquet' |
3 | | -import { ReactNode, useCallback, useEffect, useState } from 'react' |
4 | | -import Dropdown from './Dropdown.js' |
5 | | -import Dropzone from './Dropzone.js' |
6 | | -import Layout from './Layout.js' |
7 | | -import ParquetLayout from './ParquetLayout.js' |
8 | | -import ParquetMetadata from './ParquetMetadata.js' |
9 | | -import { asyncBufferFrom } from './utils.js' |
10 | | -import { parquetQueryWorker } from './workers/parquetWorkerClient.js' |
11 | | -import { AsyncBufferFrom, Row } from './workers/types.js' |
12 | | - |
13 | | -type Lens = 'table' | 'metadata' | 'layout' |
14 | | - |
15 | | -/** |
16 | | - * Hyparquet demo viewer page |
17 | | - * @param {Object} props |
18 | | - * @param {string} [props.url] |
19 | | - * @returns {ReactNode} |
20 | | - */ |
21 | | -export default function App({ url }: { url?: string }): ReactNode { |
22 | | - const [error, setError] = useState<Error>() |
23 | | - const [df, setDf] = useState<DataFrame>() |
24 | | - const [name, setName] = useState<string>() |
25 | | - const [lens, setLens] = useState<Lens>('table') |
26 | | - const [metadata, setMetadata] = useState<FileMetaData>() |
27 | | - const [byteLength, setByteLength] = useState<number>() |
28 | | - |
29 | | - const setUnknownError = useCallback((e: unknown) => { |
30 | | - setError(e instanceof Error ? e : new Error(String(e))) |
31 | | - }, []) |
32 | | - |
33 | | - const onUrlDrop = useCallback( |
34 | | - (url: string) => { |
35 | | - // Add key=url to query string |
36 | | - const params = new URLSearchParams(location.search) |
37 | | - params.set('key', url) |
38 | | - history.pushState({}, '', `${location.pathname}?${params}`) |
39 | | - byteLengthFromUrl(url).then(byteLength => setAsyncBuffer(url, { url, byteLength })).catch(setUnknownError) |
40 | | - }, |
41 | | - [setUnknownError], |
42 | | - ) |
43 | | - |
44 | | - useEffect(() => { |
45 | | - if (!df && url) { |
46 | | - onUrlDrop(url) |
47 | | - } |
48 | | - }, [ url, df, onUrlDrop]) |
49 | | - |
50 | | - function onFileDrop(file: File) { |
51 | | - // Clear query string |
52 | | - history.pushState({}, '', location.pathname) |
53 | | - setAsyncBuffer(file.name, { file, byteLength: file.size }).catch(setUnknownError) |
54 | | - } |
55 | | - |
56 | | - async function setAsyncBuffer(name: string, from: AsyncBufferFrom) { |
57 | | - // TODO: Replace welcome with spinner |
58 | | - const asyncBuffer = await asyncBufferFrom(from) |
59 | | - const metadata = await parquetMetadataAsync(asyncBuffer) |
60 | | - setMetadata(metadata) |
61 | | - setName(name) |
62 | | - setByteLength(from.byteLength) |
63 | | - let df = parquetDataFrame(from, metadata) |
64 | | - df = rowCache(df) |
65 | | - setDf(df) |
66 | | - document.getElementById('welcome')?.remove() |
67 | | - } |
68 | | - |
69 | | - return <Layout error={error}> |
70 | | - <Dropzone |
71 | | - onError={(e) => { setError(e) }} |
72 | | - onFileDrop={onFileDrop} |
73 | | - onUrlDrop={onUrlDrop}> |
74 | | - {metadata && df && <> |
75 | | - <div className='top-header'>{name}</div> |
76 | | - <div className='view-header'> |
77 | | - {byteLength !== undefined && <span title={byteLength.toLocaleString() + ' bytes'}>{formatFileSize(byteLength)}</span>} |
78 | | - <span>{df.numRows.toLocaleString()} rows</span> |
79 | | - <Dropdown label={lens}> |
80 | | - <button onClick={() => { setLens('table') }}>Table</button> |
81 | | - <button onClick={() => { setLens('metadata') }}>Metadata</button> |
82 | | - {byteLength && <button onClick={() => { setLens('layout') }}>Layout</button>} |
83 | | - </Dropdown> |
84 | | - </div> |
85 | | - {lens === 'table' && <HighTable cacheKey={name} data={df} onError={setError} />} |
86 | | - {lens === 'metadata' && <ParquetMetadata metadata={metadata} />} |
87 | | - {lens === 'layout' && byteLength && <ParquetLayout byteLength={byteLength} metadata={metadata} />} |
88 | | - </>} |
89 | | - </Dropzone> |
90 | | - </Layout> |
91 | | -} |
92 | | - |
93 | | -/** |
94 | | - * Convert a parquet file into a dataframe. |
95 | | - */ |
96 | | -function parquetDataFrame(from: AsyncBufferFrom, metadata: FileMetaData): DataFrame { |
97 | | - const { children } = parquetSchema(metadata) |
98 | | - return { |
99 | | - header: children.map(child => child.element.name), |
100 | | - numRows: Number(metadata.num_rows), |
101 | | - /** |
102 | | - * @param {number} rowStart |
103 | | - * @param {number} rowEnd |
104 | | - * @param {string} orderBy |
105 | | - * @returns {Promise<any[][]>} |
106 | | - */ |
107 | | - rows(rowStart: number, rowEnd: number, orderBy: string): Promise<Row[]> { |
108 | | - console.log(`reading rows ${rowStart}-${rowEnd}`, orderBy) |
109 | | - return parquetQueryWorker({ from, metadata, rowStart, rowEnd, orderBy }) |
110 | | - }, |
111 | | - sortable: true, |
112 | | - } |
113 | | -} |
114 | | - |
115 | | -/** |
116 | | - * Returns the file size in human readable format. |
117 | | - * |
118 | | - * @param {number} bytes file size in bytes |
119 | | - * @returns {string} formatted file size string |
120 | | - */ |
121 | | -function formatFileSize(bytes: number): string { |
122 | | - const sizes = ['b', 'kb', 'mb', 'gb', 'tb'] |
123 | | - if (bytes === 0) return '0 b' |
124 | | - const i = Math.floor(Math.log2(bytes) / 10) |
125 | | - if (i === 0) return `${bytes} b` |
126 | | - const base = bytes / Math.pow(1024, i) |
127 | | - return `${base < 10 ? base.toFixed(1) : Math.round(base)} ${sizes[i]}` |
| 1 | +import { ReactNode } from 'react' |
| 2 | +import Page from './Page.js' |
| 3 | +import Welcome from './Welcome.js' |
| 4 | + |
| 5 | +export default function App(): ReactNode { |
| 6 | + const params = new URLSearchParams(location.search) |
| 7 | + const url = params.get('key') ?? undefined |
| 8 | + return url ? <Page url={url} /> : <Welcome /> |
128 | 9 | } |
0 commit comments