38 lines
1.2 KiB
Markdown
38 lines
1.2 KiB
Markdown
# btoa-lite
|
|

|
|

|
|

|
|

|
|
|
|
Smallest/simplest possible means of using btoa with both Node and browserify.
|
|
|
|
In the browser, encoding base64 strings is done using:
|
|
|
|
``` javascript
|
|
var encoded = btoa(decoded)
|
|
```
|
|
|
|
However in Node, it's done like so:
|
|
|
|
``` javascript
|
|
var encoded = new Buffer(decoded).toString('base64')
|
|
```
|
|
|
|
You can easily check if `Buffer` exists and switch between the approaches
|
|
accordingly, but using `Buffer` anywhere in your browser source will pull
|
|
in browserify's `Buffer` shim which is pretty hefty. This package uses
|
|
the `main` and `browser` fields in its `package.json` to perform this
|
|
check at build time and avoid pulling `Buffer` in unnecessarily.
|
|
|
|
## Usage
|
|
|
|
[](https://nodei.co/npm/btoa-lite/)
|
|
|
|
### `encoded = btoa(decoded)`
|
|
|
|
Returns the base64-encoded value of a string.
|
|
|
|
## License
|
|
|
|
MIT. See [LICENSE.md](http://github.com/hughsk/btoa-lite/blob/master/LICENSE.md) for details.
|