request/lib/index.js

import crypto from 'node:crypto'
import fs from 'node:fs'
import path from 'node:path'
import { EventEmitter } from 'node:events'
import { Stream } from 'node:stream'
import { StringDecoder } from 'node:string_decoder'

import File from './file.js'
import { MultipartParser } from './multipart_parser.js'
import { QuerystringParser } from './querystring_parser.js'
import { OctetParser } from './octet_parser.js'
import { JSONParser } from './json_parser.js'

function dummyParser(self) {
  return {
    end: function () {
      self.ended = true
      self._maybeEnd()
      return null
    }
  }
}

export default class IncomingForm extends EventEmitter {
  #req = null

  error = null
  ended = false
  headers = null
  type = null

  bytesReceived = null
  bytesExpected = null

  _parser = null
  _flushing = 0
  _fieldsSize = 0
  openedFiles = []

  constructor(req, opts = {}) {
    super()

    this.#req = req

    this.maxFields = opts.maxFields || 1000
    this.maxFieldsSize = opts.maxFieldsSize || 2 * 1024 * 1024
    this.keepExtensions = opts.keepExtensions || false
    this.uploadDir = opts.uploadDir
    this.encoding = opts.encoding || 'utf-8'
    this.multiples = opts.multiples || false

    // Parse headers and setup the parser, ready to start listening for data.
    this.writeHeaders(req.headers)

    req
      .on('error', err => {
        this._error(err)
      })
      .on('aborted', () => {
        this.emit('aborted')
        this._error(new Error('Request aborted'))
      })
      .on('data', buffer => {
        this.write(buffer)
      })
      .on('end', () => {
        if (this.error) {
          return
        }

        var err = this._parser.end()
        if (err) {
          this._error(err)
        }
      })
  }

  writeHeaders(headers) {
    this.headers = headers
    this.#parseContentLength()
    this.#parseContentType()
  }

  write(buffer) {
    if (this.error) {
      return
    }
    if (!this._parser) {
      this._error(new Error('uninitialized parser'))
      return
    }

    this.bytesReceived += buffer.length
    this.emit('progress', this.bytesReceived, this.bytesExpected)

    var bytesParsed = this._parser.write(buffer)
    if (bytesParsed !== buffer.length) {
      this._error(
        new Error(
          'parser error, ' +
            bytesParsed +
            ' of ' +
            buffer.length +
            ' bytes parsed'
        )
      )
    }

    return bytesParsed
  }

  pause() {
    try {
      this.#req.pause()
    } catch (err) {
      // the stream was destroyed
      if (!this.ended) {
        // before it was completed, crash & burn
        this._error(err)
      }
      return false
    }
    return true
  }

  resume() {
    try {
      this.#req.resume()
    } catch (err) {
      // the stream was destroyed
      if (!this.ended) {
        // before it was completed, crash & burn
        this._error(err)
      }
      return false
    }

    return true
  }

  onPart(part) {
    // this method can be overwritten by the user
    this.handlePart(part)
  }

  handlePart(part) {
    var self = this

    if (part.filename === undefined) {
      var value = '',
        decoder = new StringDecoder(this.encoding)

      part.on('data', function (buffer) {
        self._fieldsSize += buffer.length
        if (self._fieldsSize > self.maxFieldsSize) {
          self._error(
            new Error(
              'maxFieldsSize exceeded, received ' +
                self._fieldsSize +
                ' bytes of field data'
            )
          )
          return
        }
        value += decoder.write(buffer)
      })

      part.on('end', function () {
        self.emit('field', part.name, value)
      })
      return
    }

    this._flushing++

    var file = new File({
      path: this._uploadPath(part.filename),
      name: part.filename,
      type: part.mime,
      hash: self.hash
    })

    this.emit('fileBegin', part.name, file)

    file.open()
    this.openedFiles.push(file)

    part.on('data', function (buffer) {
      if (buffer.length == 0) {
        return
      }
      self.pause()
      file.write(buffer, function () {
        self.resume()
      })
    })

    part.on('end', function () {
      file.end(function () {
        self._flushing--
        self.emit('file', part.name, file)
        self._maybeEnd()
      })
    })
  }

  #parseContentType() {
    if (this.bytesExpected === 0) {
      this._parser = dummyParser(this)
      return
    }

    if (!this.headers['content-type']) {
      this._error(new Error('bad content-type header, no content-type'))
      return
    }

    if (this.headers['content-type'].match(/octet-stream/i)) {
      this._initOctetStream()
      return
    }

    if (this.headers['content-type'].match(/urlencoded/i)) {
      this._initUrlencoded()
      return
    }

    if (this.headers['content-type'].match(/multipart/i)) {
      var m = this.headers['content-type'].match(
        /boundary=(?:"([^"]+)"|([^;]+))/i
      )
      if (m) {
        this._initMultipart(m[1] || m[2])
      } else {
        this._error(new Error('bad content-type header, no multipart boundary'))
      }
      return
    }

    if (this.headers['content-type'].match(/json|appliation|plain|text/i)) {
      this._initJSONencoded()
      return
    }

    this._error(
      new Error(
        'bad content-type header, unknown content-type: ' +
          this.headers['content-type']
      )
    )
  }

  _error(err) {
    if (this.error || this.ended) {
      return
    }

    this.error = err
    this.emit('error', err)

    if (Array.isArray(this.openedFiles)) {
      this.openedFiles.forEach(function (file) {
        file._writeStream.destroy()
        setTimeout(fs.unlink, 0, file.path, function (error) {})
      })
    }
  }

  #parseContentLength() {
    this.bytesReceived = 0
    if (this.headers['content-length']) {
      this.bytesExpected = +this.headers['content-length']
    } else if (this.headers['transfer-encoding'] === undefined) {
      this.bytesExpected = 0
    }

    if (this.bytesExpected !== null) {
      this.emit('progress', this.bytesReceived, this.bytesExpected)
    }
  }

  _newParser() {
    return new MultipartParser()
  }

  _initMultipart(boundary) {
    this.type = 'multipart'

    var parser = new MultipartParser(),
      self = this,
      headerField,
      headerValue,
      part

    parser.initWithBoundary(boundary)

    parser.onPartBegin = function () {
      part = new Stream()
      part.readable = true
      part.headers = {}
      part.name = null
      part.filename = null
      part.mime = null

      part.transferEncoding = 'binary'
      part.transferBuffer = ''

      headerField = ''
      headerValue = ''
    }

    parser.onHeaderField = function (b, start, end) {
      headerField += b.toString(self.encoding, start, end)
    }

    parser.onHeaderValue = function (b, start, end) {
      headerValue += b.toString(self.encoding, start, end)
    }

    parser.onHeaderEnd = function () {
      headerField = headerField.toLowerCase()
      part.headers[headerField] = headerValue

      var m = headerValue.match(/\bname="([^"]+)"/i)
      if (headerField == 'content-disposition') {
        if (m) {
          part.name = m[1]
        }

        part.filename = self._fileName(headerValue)
      } else if (headerField == 'content-type') {
        part.mime = headerValue
      } else if (headerField == 'content-transfer-encoding') {
        part.transferEncoding = headerValue.toLowerCase()
      }

      headerField = ''
      headerValue = ''
    }

    parser.onHeadersEnd = function () {
      switch (part.transferEncoding) {
        case 'binary':
        case '7bit':
        case '8bit':
          parser.onPartData = function (b, start, end) {
            part.emit('data', b.slice(start, end))
          }

          parser.onPartEnd = function () {
            part.emit('end')
          }
          break

        case 'base64':
          parser.onPartData = function (b, start, end) {
            part.transferBuffer += b.slice(start, end).toString('ascii')

            /*
          four bytes (chars) in base64 converts to three bytes in binary
          encoding. So we should always work with a number of bytes that
          can be divided by 4, it will result in a number of buytes that
          can be divided vy 3.
          */
            var offset = parseInt(part.transferBuffer.length / 4, 10) * 4
            part.emit(
              'data',
              Buffer.from(part.transferBuffer.substring(0, offset), 'base64')
            )
            part.transferBuffer = part.transferBuffer.substring(offset)
          }

          parser.onPartEnd = function () {
            part.emit('data', Buffer.from(part.transferBuffer, 'base64'))
            part.emit('end')
          }
          break

        default:
          return self._error(new Error('unknown transfer-encoding'))
      }

      self.onPart(part)
    }

    parser.onEnd = function () {
      self.ended = true
      self._maybeEnd()
    }

    this._parser = parser
  }

  _fileName(headerValue) {
    var m = headerValue.match(/\bfilename="(.*?)"($|; )/i)
    if (!m) return

    var filename = m[1].substr(m[1].lastIndexOf('\\') + 1)
    filename = filename.replace(/%22/g, '"')
    filename = filename.replace(/&#([\d]{4});/g, function (m, code) {
      return String.fromCharCode(code)
    })
    return filename
  }

  _initUrlencoded() {
    this.type = 'urlencoded'

    var parser = new QuerystringParser(this.maxFields)

    parser.onField = (key, val) => {
      this.emit('field', key, val)
    }

    parser.onEnd = () => {
      this.ended = true
      this._maybeEnd()
    }

    this._parser = parser
  }

  _initOctetStream() {
    this.type = 'octet-stream'
    var filename = this.headers['x-file-name']
    var mime = this.headers['content-type']

    var file = new File({
      path: this._uploadPath(filename),
      name: filename,
      type: mime
    })

    this.emit('fileBegin', filename, file)
    file.open()

    this._flushing++

    var self = this

    self._parser = new OctetParser()

    //Keep track of writes that haven't finished so we don't emit the file before it's done being written
    var outstandingWrites = 0

    self._parser.on('data', function (buffer) {
      self.pause()
      outstandingWrites++

      file.write(buffer, function () {
        outstandingWrites--
        self.resume()

        if (self.ended) {
          self._parser.emit('doneWritingFile')
        }
      })
    })

    self._parser.on('end', function () {
      self._flushing--
      self.ended = true

      var done = function () {
        file.end(function () {
          self.emit('file', 'file', file)
          self._maybeEnd()
        })
      }

      if (outstandingWrites === 0) {
        done()
      } else {
        self._parser.once('doneWritingFile', done)
      }
    })
  }

  _initJSONencoded() {
    this.type = 'json'

    var parser = new JSONParser(),
      self = this

    if (this.bytesExpected) {
      parser.initWithLength(this.bytesExpected)
    }

    parser.onField = function (key, val) {
      self.emit('field', key, val)
    }

    parser.onEnd = function () {
      self.ended = true
      self._maybeEnd()
    }

    this._parser = parser
  }

  _uploadPath(filename) {
    var name = 'upload_'
    var buf = crypto.randomBytes(16)
    for (var i = 0; i < buf.length; ++i) {
      name += ('0' + buf[i].toString(16)).slice(-2)
    }

    if (this.keepExtensions) {
      var ext = path.extname(filename)
      ext = ext.replace(/(\.[a-z0-9]+).*/i, '$1')

      name += ext
    }

    return path.join(this.uploadDir, name)
  }

  _maybeEnd() {
    if (!this.ended || this._flushing || this.error) {
      return
    }

    this.emit('end')
  }
}