hyparquet/demo.js

239 lines
7.3 KiB
JavaScript
Raw Normal View History

2024-05-05 17:35:23 +00:00
import { parquetMetadata, parquetMetadataAsync, parquetRead, parquetSchema, toJson } from './src/hyparquet.js'
2024-01-28 02:06:27 +00:00
const dropzone = document.getElementById('dropzone')
2024-05-05 17:35:23 +00:00
const fileInput = document.getElementById('file-input')
const content = document.getElementById('content')
const welcome = document.getElementById('welcome')
2024-05-13 02:36:30 +00:00
const label = document.getElementById('filename')
2024-05-05 17:35:23 +00:00
2024-01-28 02:06:27 +00:00
const layout = document.getElementById('layout')
2024-02-05 05:21:01 +00:00
const metadataDiv = document.getElementById('metadata')
2024-01-28 02:06:27 +00:00
2024-05-05 17:35:23 +00:00
let enterCount = 0
dropzone.addEventListener('dragenter', e => {
2024-01-28 02:06:27 +00:00
e.dataTransfer.dropEffect = 'copy'
dropzone.classList.add('over')
2024-05-05 17:35:23 +00:00
enterCount++
})
dropzone.addEventListener('dragover', e => {
e.preventDefault()
2024-01-28 02:06:27 +00:00
})
dropzone.addEventListener('dragleave', () => {
2024-05-05 17:35:23 +00:00
enterCount--
if (!enterCount) dropzone.classList.remove('over')
2024-01-28 02:06:27 +00:00
})
dropzone.addEventListener('drop', e => {
e.preventDefault() // prevent dropped file from being "downloaded"
dropzone.classList.remove('over')
2024-02-04 21:34:49 +00:00
const { files, items } = e.dataTransfer
2024-01-28 02:06:27 +00:00
if (files.length > 0) {
const file = files[0]
2024-01-28 02:50:14 +00:00
processFile(file)
}
2024-02-04 21:34:49 +00:00
if (items.length > 0) {
const item = items[0]
if (item.kind === 'string') {
item.getAsString(str => {
2024-02-05 07:37:18 +00:00
if (str.startsWith('http')) {
2024-02-04 21:34:49 +00:00
processUrl(str)
}
})
}
}
2024-01-28 02:50:14 +00:00
})
2024-02-05 07:28:59 +00:00
async function processUrl(url) {
2024-05-05 17:35:23 +00:00
content.innerHTML = ''
2024-02-05 07:28:59 +00:00
try {
2024-02-05 07:37:18 +00:00
// Check if file is accessible and get its size
const head = await fetch(url, { method: 'HEAD' })
if (!head.ok) {
2024-05-05 17:35:23 +00:00
content.innerHTML = `<strong>${url}</strong>`
content.innerHTML += `<div class="error">Error fetching file\n${head.status} ${head.statusText}</div>`
2024-02-05 07:37:18 +00:00
return
}
const size = head.headers.get('content-length')
if (!size) {
2024-05-05 17:35:23 +00:00
content.innerHTML = `<strong>${url}</strong>`
content.innerHTML += '<div class="error">Error fetching file\nNo content-length header</div>'
2024-02-05 07:37:18 +00:00
return
}
// Construct an AsyncBuffer that fetches file chunks
const asyncBuffer = {
byteLength: Number(size),
slice: async (start, end) => {
const rangeEnd = end === undefined ? '' : end - 1
2024-05-05 17:35:23 +00:00
console.log(`Fetch ${url} bytes=${start}-${rangeEnd}`)
2024-02-05 07:37:18 +00:00
const res = await fetch(url, {
headers: { Range: `bytes=${start}-${rangeEnd}` },
})
return res.arrayBuffer()
},
}
2024-02-05 07:28:59 +00:00
const metadata = await parquetMetadataAsync(asyncBuffer)
2024-05-05 17:35:23 +00:00
await render(asyncBuffer, metadata, `<a href="${url}">${url}</a>`)
2024-02-05 07:28:59 +00:00
} catch (e) {
console.error('Error fetching file', e)
2024-05-05 17:35:23 +00:00
content.innerHTML = `<strong>${url}</strong>`
content.innerHTML += `<div class="error">Error fetching file\n${e}</div>`
2024-02-05 07:28:59 +00:00
}
2024-02-04 21:34:49 +00:00
}
2024-01-28 02:50:14 +00:00
function processFile(file) {
2024-05-05 17:35:23 +00:00
content.innerHTML = ''
2024-01-28 02:50:14 +00:00
const reader = new FileReader()
2024-05-05 17:35:23 +00:00
reader.onload = async e => {
2024-01-28 02:50:14 +00:00
try {
const arrayBuffer = e.target.result
2024-02-05 07:28:59 +00:00
const metadata = parquetMetadata(arrayBuffer)
2024-05-05 17:35:23 +00:00
await render(arrayBuffer, metadata, file.name)
2024-01-28 02:50:14 +00:00
} catch (e) {
2024-02-05 05:21:01 +00:00
console.error('Error parsing file', e)
2024-05-05 17:35:23 +00:00
content.innerHTML = `<strong>${file.name}</strong>`
content.innerHTML += `<div class="error">Error parsing file\n${e}</div>`
2024-01-28 02:06:27 +00:00
}
2024-01-28 02:50:14 +00:00
}
reader.onerror = e => {
console.error('Error reading file', e)
2024-05-05 17:35:23 +00:00
content.innerHTML = `<strong>${file.name}</strong>`
content.innerHTML += `<div class="error">Error reading file\n${e.target.error}</div>`
2024-01-28 02:50:14 +00:00
}
reader.readAsArrayBuffer(file)
}
2024-05-05 17:35:23 +00:00
async function render(asyncBuffer, metadata, name) {
renderSidebar(asyncBuffer, metadata, name)
const { children } = parquetSchema(metadata)
const header = children.map(child => child.element.name)
const startTime = performance.now()
await parquetRead({
file: asyncBuffer,
rowEnd: 1000,
onComplete(data) {
const ms = performance.now() - startTime
console.log(`parsed ${name} in ${ms.toFixed(0)} ms`)
content.appendChild(renderTable(header, data))
},
})
}
2024-02-05 07:28:59 +00:00
function renderSidebar(asyncBuffer, metadata, name) {
2024-05-13 02:36:30 +00:00
label.innerText = name
2024-02-04 21:34:49 +00:00
// render file layout
2024-05-13 03:41:39 +00:00
layout.innerHTML = ''
2024-02-05 07:28:59 +00:00
layout.appendChild(fileLayout(metadata, asyncBuffer.byteLength))
2024-02-04 21:34:49 +00:00
// display metadata
2024-02-05 05:21:01 +00:00
metadataDiv.innerHTML = ''
metadataDiv.appendChild(fileMetadata(toJson(metadata)))
2024-02-04 21:34:49 +00:00
}
2024-05-05 17:35:23 +00:00
welcome.addEventListener('click', () => {
2024-01-28 02:50:14 +00:00
fileInput.click()
})
fileInput.addEventListener('change', () => {
if (fileInput.files.length > 0) {
processFile(fileInput.files[0])
2024-01-28 02:06:27 +00:00
}
})
2024-01-28 02:55:34 +00:00
// Render file layout
2024-02-05 07:28:59 +00:00
function fileLayout(metadata, byteLength) {
2024-01-28 02:55:34 +00:00
let html = '<h2>File layout</h2>'
html += cell('PAR1', 0, 4, 4) // magic number
for (const rowGroupIndex in metadata.row_groups) {
const rowGroup = metadata.row_groups[rowGroupIndex]
2024-02-05 07:37:18 +00:00
html += group(`Row group ${rowGroupIndex} (${rowGroup.total_byte_size.toLocaleString()} bytes)`)
2024-01-28 02:55:34 +00:00
for (const column of rowGroup.columns) {
const columnName = column.meta_data.path_in_schema.join('.')
let columnOffset = column.meta_data.dictionary_page_offset
if (!columnOffset || column.meta_data.data_page_offset < columnOffset) {
columnOffset = column.meta_data.data_page_offset
}
columnOffset = Number(columnOffset)
2024-02-05 05:21:01 +00:00
const bytes = Number(column.meta_data.total_compressed_size)
2024-01-28 02:55:34 +00:00
const end = columnOffset + bytes
html += cell(`Column ${columnName}`, columnOffset, bytes, end)
}
html += '</div>'
}
2024-02-05 07:28:59 +00:00
const metadataStart = byteLength - metadata.metadata_length - 4
html += cell('Metadata', metadataStart, metadata.metadata_length, byteLength - 4)
html += cell('PAR1', byteLength - 4, 4, byteLength) // magic number
2024-01-28 02:55:34 +00:00
const div = document.createElement('div')
div.innerHTML = html
div.classList.add('collapsed') // start collapsed
2024-01-28 03:29:21 +00:00
div.children[0].addEventListener('click', () => {
2024-01-28 02:55:34 +00:00
div.classList.toggle('collapsed')
})
return div
}
function group(name) {
return `<div>${name}`
}
function cell(name, start, bytes, end) {
return `
<div class="cell">
<label>${name}</label>
<ul>
2024-02-05 07:37:18 +00:00
<li>start ${start.toLocaleString()}</li>
<li>bytes ${bytes.toLocaleString()}</li>
<li>end ${end.toLocaleString()}</li>
2024-01-28 02:55:34 +00:00
</ul>
</div>`
}
2024-01-28 03:29:21 +00:00
// Render metadata
function fileMetadata(metadata) {
let html = '<h2>Metadata</h2>'
html += `<pre>${JSON.stringify(metadata, null, 2)}</pre>`
const div = document.createElement('div')
div.innerHTML = html
div.classList.add('collapsed') // start collapsed
div.children[0].addEventListener('click', () => {
div.classList.toggle('collapsed')
})
return div
}
2024-05-05 17:35:23 +00:00
function renderTable(header, data) {
const table = document.createElement('table')
const thead = document.createElement('thead')
const tbody = document.createElement('tbody')
const headerRow = document.createElement('tr')
for (const columnName of header) {
const th = document.createElement('th')
th.innerText = columnName
headerRow.appendChild(th)
}
thead.appendChild(headerRow)
table.appendChild(thead)
for (const row of data) {
const tr = document.createElement('tr')
for (const value of Object.values(row)) {
const td = document.createElement('td')
2024-05-05 21:24:21 +00:00
td.innerText = stringify(value)
2024-05-05 17:35:23 +00:00
tr.appendChild(td)
}
tbody.appendChild(tr)
}
table.appendChild(tbody)
return table
}
2024-05-05 21:24:21 +00:00
function stringify(value) {
if (value === undefined) return ''
2024-05-13 01:12:30 +00:00
value = toJson(value)
2024-05-05 21:24:21 +00:00
if (typeof value === 'string') return value
2024-05-13 01:12:30 +00:00
if (typeof value === 'object') return JSON.stringify(value)
2024-05-05 21:24:21 +00:00
return value
}