Amazon's [AWS PHP SDK][1] offers a [stream wrapper][2] for S3.

By registering the `s3://` protocol, it's possible to use PHP's file handling
functions; such as `fopen()`, `fread()`, `rename()`, `mkdir()`, `unlink()`,
etc...; directly with URLs in the form of `s3://bucket/path/to/object`. Thus,
eliminating the cumbersome syntax of the API's methods.

Without the stream wrapper, we would have to write more code in order to
achieve desired operations. For example:

```php
<?php
//rename the object 'foo.txt' to 'bar.txt' in bucket 'myBucket':
//there is no 'rename' method in the API, so we first have to copy the
//object to the new location, and then delete the old one:
$response = $s3->copy_object(
    array(
        'bucket'   => 'myBucket',
        'filename' => 'foo.txt'
    ),
    array(
        'bucket'   => 'myBucket',
        'filename' => 'bar.txt'
    )
);
if ($response->isOK()){
    $response = $s3->delete_object('myBucket', 'foo.txt');
    if (!$response->isOK()){
        //handle error...
    }
}

//read contents of object 'foo/bar.txt' in bucket 'myBucket':
$data = $s3->get_object(
    'myBucket',
    'foo/bar.txt'
))->body;

//store data in object 'foo/bar.txt' in bucket 'myBucket':
$response = $s3->create_object(
    'myBucket',
    'foo/bar.txt',
    array(
        'body' => 'data...'
    )
);
?>
```

But with the stream wrapper, we can use PHP's corresponding functions, in a neater and shorter fashion.

First, we have to register the `s3://` protocol:

```php
<?php
define('AWS_KEY','YOUR_AWS_KEY_HERE');
define('AWS_SECRET','YOUR_AWS_SECRET_HERE');

require_once('AWSSDKforPHP/sdk.class.php');
require_once('AWSSDKforPHP/extensions/s3streamwrapper.class.php');

$s3 = new AmazonS3(array(
    'key' => AWS_KEY,
    'secret' => AWS_SECRET,
    'default_cache_config' => '',
    'certificate_authority' => true
));

S3StreamWrapper::register($s3);
?>
```

Usage example:

```php
<?php
//rename the object 'foo.txt' to 'bar.txt' in bucket 'myBucket':
rename('s3://myBucket/foo.txt', 's3://myBucket/bar.txt');

//delete the object 'bar.txt' from bucket 'myBucket':
unlink('s3://myBucket/bar.txt');

//create a new bucket 'mySecondBucket':
mkdir('s3://mySecondBucket');

//read contents of object 'foo/bar.txt' in bucket 'myBucket':
$data = file_get_contents('s3://myBucket/foo/bar.txt');

//store data in object 'foo/bar.txt' in bucket 'myBucket':
file_put_contents('s3://myBucket/foo/bar.txt', 'data...');
?>
```

One thing which is important to note:

> It is important to avoid reading files that are near to or larger than the
> amount of memory allocated to PHP, otherwise “out of memory” errors will
> occur.

([Quoted from the SDK's documentation][3])

So, if you handle large file in your code, and encounter such errors, there is
no alternative (which I know of) than to revert to the ordinary methods of the
SDK.

**Note**:
Code in this post can also be found in [this gist][4].

[1]:http://aws.amazon.com/sdkforphp/
[2]:http://php.net/manual/en/class.streamwrapper.php
[3]:http://docs.amazonwebservices.com/AWSSDKforPHP/latest/index.html#m=S3StreamWrapper/stream_read
[4]:https://gist.github.com/EyalAr/3898611