Skip to content

oiime/pygmylog

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

about

pygmylog is a simple binary log encoder, its purpose is to simplify the creation of compact binary logs from within nodejs, it encodes rows into transformation streams that can be piped into files or any other writable stream such as Amazon S3.

pygmylog by itself has no dependencies, however, if you wish to use the default encoder you'd need to install protobufjs

As each log also stores its own metadata (schema) it is possible to smoothly change schemas and versions

installation

npm install --save pygmylog

types for default encoder schema

  • object
  • enum - (enum values passed via the enum: property)
  • double
  • float
  • int32
  • uint32
  • int64
  • uint64
  • bool
  • string
  • object_arr
  • enum_arr
  • double_arr
  • float_arr
  • int32_arr
  • uint32_arr
  • int64_arr
  • uint64_arr
  • bool_arr
  • string_arr

by example

pointless read/write to local filesystem example

const fs = require('fs')
const { PassThrough } = require('stream')
const pygmylog = require('pygmylog')

const tmpFilename = '/tmp/pygmylog.test'
const schema = {
  properties: {
    foo: 'int32',
    bar: 'int32_arr',
    myenum: { type: 'enum', enum: ['foo', 'bar', 'foobar'] }
  }
}
const writer = pygmylog.createWriteStream(schema)
const fd = fs.createWriteStream(tmpFilename)

writer.pipe(fd)

fd.on('close', () => {
  const reader = pygmylog.createReadStream()
  const pass = new PassThrough({ objectMode: true })
  pass.on('data', row => {
    console.log(row)
  })
  pass.on('end', () => {
    fs.unlinkSync(tmpFilename)
  })
  fs.createReadStream(tmpFilename).pipe(reader).pipe(pass)
})

writer.write({
  foo: 5,
  bar: [1, 2, 3],
  myenum: 'foo'
})

writer.end()

storing metadata within the headers

const { PassThrough } = require('stream')
const pygmylog = require('pygmylog')

const schema = {
  metadata: { foo: 'bar' },
  properties: {
    foo: 'int32'
  }
}
const writer = pygmylog.createWriteStream(schema)
const reader = pygmylog.createReadStream()

reader.on('ready', data => {
  console.log(data) // { metdata: { foo: 'bar' } }
  const pass = new PassThrough({ objectMode: true })
  pass.on('data', row => {
    console.log(row)
  })
  reader.pipe(pass)
})

writer.pipe(reader)
writer.write({ foo: 5 })
writer.end()

Using your own encoder

const fs = require('fs')
const { PassThrough } = require('stream')
const pygmylog = require('pygmylog')

class UselessEncoder extends pygmylog.Encoder {
  // optionally, you will recieve here the first arguments you've passed to createWriteStream, if this method does not exist the log would not start with a header
  serializeHeaders (headers) {
    return Buffer.from('headersGoHere')
  }
  // optionally, you will recieve here the headers you've serialized in the writer before the first row is processed
  unserializeHeaders (buf) {
    this._myHeadersData = buf.toString('utf8')
  }
  encode (str) {
    // this.headers
    return Buffer.from(str, 'utf8')
  }
  decode (buf) {
    return buf.toString('utf8')
  }
}
const writer = pygmylog.createWriteStream(null, { encoder: UselessEncoder })
const reader = pygmylog.createReadStream({ encoder: UselessEncoder })

const pass = new PassThrough({ objectMode: true })
pass.on('data', entry => {
  console.log(entry) // hellokitty
})

writer.pipe(reader).pipe(pass)
writer.write('hellokitty')
writer.end()

API

pygmylog.createWriteStream(headers, options)

returns a PygmyLogWriter instance

pygmylog.createReadStream(options)

returns a PygmyLogReader instance

pygmylog.Encoder

Encoder class to be extended by custom implementations

PygmyLogWriter.constructor(headers, options = {})

  • options.encoder - class for encoder

pygmyLogWriter.length

returns the total length written so far into the writer in bytes

pygmyLogWriter.write(payload)

pass a row to be encoded into the stream

pygmyLogWriter.end()

ends the writer signaling down the pipe

PygmyLogReader.constructor(options = {})

  • options.encoder - class for encoder

License: MIT

About

fast binary log encoder

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published