Files
NapCatQQ/.pnpm-store/v3/files/e7/6ffad1395997b9d24f7f1e5a17467b140b68c9672ef965f59bf5c2e5ac605b0f3dbbdb026ff5b56c6299e1a2acebdf0a375de5626e879f2191cb1830717e8a
TanTan e311ca3275
Build NapCat Artifacts / Type Check & Test (push) Waiting to run
docs: 仓库链接指向本镜像仓库
2026-11-13 09:25:58 +08:00

71 lines
2.8 KiB
Plaintext

# FlushWritable [![Build Status](https://travis-ci.org/TomFrost/FlushWritable.svg?branch=master)](https://travis-ci.org/TomFrost/FlushWritable)
A Writable stream that flushes before emitting finish.
Sponsored by [Leadnomics](http://www.leadnomics.com).
## What it is
Node.js's Streams API is a fantastic tool, but has a nagging shortcoming:
while the Transform stream implements a `_flush` method that is called before
its final events are fired, the Writable stream does not. So if you're
buffering rows to be INSERTed into a SQL table rather than slowly writing one
at a time, or you're buffering bytes for a transfer to S3, there is no way of
flushing those buffers to the target data store before the `finish` event is
emitted.
**FlushWritable is a drop-in replacement for stream.Writable** that implements
a `_flush` call that behaves exactly how Transform._flush does. It's called
with a callback, waits for the callback to be called, and _then_ fires
`finish` (or `error` if an error was passed). No additional execution after
the `finish` event, no implementing nonstandard event types, no chaining a
shell Transform stream before the Writable to hijack its `_flush` call. And
it's fully futureproof against the Node.js team actually adding a `_flush`
method to the native stream.Writable in a later version of Node, so you don't
have to worry about your code breaking on upgrade.
## How does it work?
It's pretty simple. Writable is an EventEmitter. FlushWritable extends
Writable and overrides EventEmitter.emit in its own prototype, listening for a
request that `finish` be emitted. When that comes in, it blocks that event
from emitting, and calls `_flush` if it's defined.
The callback it passes to `_flush` will trigger `finish` to actually be
emitted. If that callback is called with a truthy first argument, `error` is
emitted instead. All other events pass right through and are emitted as
expected. If a future version of node adds a `Writable.prototype._flush`
method, the whole thing short-circuits and native functionality takes over.
## Installation
In your project folder, type:
npm install flushwritable --save
## Usage
Just extend FlushWritable instead of stream.Writable in your write stream, and
feel free to define a `_flush(cb)` function!
```javascript
var FlushWritable = require('flushwritable'),
util = require('util');
function MyWriteStream(opts) {
FlushWritable.call(this, opts);
this._buffer = [];
}
util.inherits(MyWriteStream, FlushWritable);
MyWriteStream.prototype._flush = function(cb) {
writeBufferSomewhere(this._buffer, cb);
};
MyWriteStream.prototype._write = function(data, encoding, cb) {
this._buffer.push(data);
cb();
};
```
## License
FlushWritable is distributed under the MIT license.
## Credits
FlushWritable was created by Tom Frost at Leadnomics in 2014.