40 lines
1.9 KiB
Markdown
Executable file
40 lines
1.9 KiB
Markdown
Executable file
## @npmcli/agent
|
|
|
|
A pair of Agent implementations for nodejs that provide consistent keep-alives, granular timeouts, dns caching, and proxy support.
|
|
|
|
### Usage
|
|
|
|
```js
|
|
const { getAgent, HttpAgent } = require('@npmcli/agent')
|
|
const fetch = require('minipass-fetch')
|
|
|
|
const main = async () => {
|
|
// if you know what agent you need, you can create one directly
|
|
const agent = new HttpAgent(agentOptions)
|
|
// or you can use the getAgent helper, it will determine and create an Agent
|
|
// instance for you as well as reuse that agent for new requests as appropriate
|
|
const agent = getAgent('https://registry.npmjs.org/npm', agentOptions)
|
|
// minipass-fetch is just an example, this will work for any http client that
|
|
// supports node's Agents
|
|
const res = await fetch('https://registry.npmjs.org/npm', { agent })
|
|
}
|
|
|
|
main()
|
|
```
|
|
|
|
### Options
|
|
|
|
All options supported by the node Agent implementations are supported here, see [the docs](https://nodejs.org/api/http.html#new-agentoptions) for those.
|
|
|
|
Options that have been added by this module include:
|
|
|
|
- `family`: what tcp family to use, can be `4` for IPv4, `6` for IPv6 or `0` for both.
|
|
- `proxy`: a URL to a supported proxy, currently supports `HTTP CONNECT` based http/https proxies as well as socks4 and 5.
|
|
- `dns`: configuration for the built-in dns cache
|
|
- `ttl`: how long (in milliseconds) to keep cached dns entries, defaults to `5 * 60 * 100 (5 minutes)`
|
|
- `lookup`: optional function to override how dns lookups are performed, defaults to `require('dns').lookup`
|
|
- `timeouts`: a set of granular timeouts, all default to `0`
|
|
- `connection`: time between initiating connection and actually connecting
|
|
- `idle`: time between data packets (if a top level `timeout` is provided, it will be copied here)
|
|
- `response`: time between sending a request and receiving a response
|
|
- `transfer`: time between starting to receive a request and consuming the response fully
|