Node.js：如何在Write Stream'finish'事件上写（）

Question

I'm using Node.js streams to go through a text file line by line, make some transforms and output to a SVG file.

I am trying to write one last piece of data ( </svg> ) after the processing is done, however by the time the write stream emits the finish event, attempting to write() will throw Error: write after end .

Is there an elegant way I can solve this?

Note: The input file is large (around 1GB) so there's no going around the pipe() method due to its I/O and memory management.

var fs = require('fs');
var split2 = require('split2');
var through2 = require('through2');

var read_stream = fs.createReadStream('input.txt');
var write_stream = fs.createWriteStream('output.svg');

write_stream.write('<svg>');
write_stream.on('finish', function() {
  this.write('</svg>'); // doesn't work
});

read_stream
  .pipe(split2())
  .pipe(through2.obj(function(line, encoding, next) {
     this.push(line);
     next();
  }))
  .pipe(write_stream);

Solutions

Thank you Jordan & pNre for helping me figuring this out.

Solution 1 (generic)

pipe() the write stream with the end:false option and manually end() the stream.

var fs = require('fs');
var split2 = require('split2');
var through2 = require('through2');

var read_stream = fs.createReadStream('input.txt');
var write_stream = fs.createWriteStream('output.svg');

write_stream.write('<svg>');

read_stream
  .pipe(split2())
  .pipe(through2.obj(function(line, encoding, next) {
     this.push(line);
     next();
  }))
  .pipe(write_stream, { end: false });

read_stream.on('end', function() {
  write_stream.end('</svg>');
});

Solution 2 (specific to through / through2 transform streams)

through2 has a flush function that can be used to write the final data.

var fs = require('fs');
var split2 = require('split2');
var through2 = require('through2');

var read_stream = fs.createReadStream('input.txt');
var write_stream = fs.createWriteStream('output.svg');

write_stream.write('<svg>');

read_stream
  .pipe(split2())
  .pipe(through2.obj(function(line, encoding, next) {
    this.push(line);
    next();
  }, function(flush) {
    this.push('</svg>');
    flush();
  }))
  .pipe(write_stream);

Answer 1

It would seem that pipe closes the stream when it is finished.

The documentation at http://nodejs.org/api/stream.html states:

By default end() is called on the destination when the source stream emits end, so that destination is no longer writable. Pass { end: false } as options to keep the destination stream open.

This keeps writer open so that "Goodbye" can be written at the end.

reader.pipe(writer, { end: false });
reader.on('end', function() {
  writer.end('Goodbye\n');
});

Answer 2

Have you considered creating a new stream to append the </svg> tag? through can help you with that:

var fs = require('fs');
var split2 = require('split2');
var through = require('through');

var read_stream = fs.createReadStream('input.txt');
var write_stream = fs.createWriteStream('output.svg');

write_stream.write('<svg>');
var tag = through(function write(data) {
    this.queue(data);
}, function end() {
    this.queue('</svg>');
});

read_stream.pipe(split2()).pipe(some_transform).pipe(tag).pipe(write_stream);

Answer 3

It seems that there is an un-documented event called 'prefinish' . I have not used it though.

Answer 4

Recently ran in to this issue and found a more elegant solution. There's a (well) documented _flush method on the native Transform stream.

https://nodejs.org/api/stream.html#stream_transform_flush_callback

The solution looks something like this:

const fs = require('fs')
const split2 = require('split2')
const { Transform } = require('stream')

const input = fs.createReadStream('input.txt')
const output = fs.createWriteStream('output.svg')

class SVGWrapper extends Transform {
    constructor(){ this.push('<svg>') }

    _flush(done){ this.push('</svg>') }

    _transform(line, enc, next){
        this.push(line)
        next()
    }
}

input
    .pipe(split2())
    .pipe(new SVGWrapper)
    .pipe(output)