Skip to content

stream: why use string_decoder in Readable? #5223

Closed
@zoubin

Description

@zoubin

I'm a little confused about the use of string_decoder in Readable (Line 140).

I thought it was meant to support pushing incomplete UTF-8 character sequence, but the following example failed:

const stream = require('stream')
const euro = new Buffer([0xE2, 0x82, 0xAC])
const cent = new Buffer([0xC2, 0xA2])
const source = Buffer.concat([euro, cent])

const readable = stream.Readable({ encoding: 'utf8' })
readable.push(source.slice(0, 2))
readable.push(source.slice(2, 4))
readable.push(source.slice(4, 6))
readable.push(null)

readable.on('data', data => console.log(data))

No output.

I was expecting the output to be the same with

const stream = require('stream')
const euro = new Buffer([0xE2, 0x82, 0xAC])
const cent = new Buffer([0xC2, 0xA2])
const source = Buffer.concat([euro, cent])

const readable = stream.Readable({ encoding: 'utf8' })
readable.push(source.slice(0, 3))
readable.push(source.slice(3, 5))
readable.push(null)

readable.on('data', data => console.log(data))
// €
// ¢

It seems that I have to push complete character sequence, or else chunk returned by the decoder in Line 140 will be empty ('').
If the stream is working in the flowing mode, every time read() try to get that empty string out, howMuchToRead will return 0 and fromList is not called, and data stops flowing.

Did I misunderstand the purpose of using string_decoder here?

Metadata

Metadata

Assignees

No one assigned

    Labels

    confirmed-bugIssues with confirmed bugs.streamIssues and PRs related to the stream subsystem.string_decoderIssues and PRs related to the string_decoder subsystem.

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions