1 |
efrain |
1 |
<?php
|
|
|
2 |
namespace Aws\S3;
|
|
|
3 |
|
|
|
4 |
use Aws\Arn\Exception\InvalidArnException;
|
|
|
5 |
use Aws\Arn\S3\AccessPointArn;
|
|
|
6 |
use Aws\Arn\ArnParser;
|
|
|
7 |
use GuzzleHttp\Psr7;
|
|
|
8 |
use Psr\Http\Message\UriInterface;
|
|
|
9 |
|
|
|
10 |
/**
|
|
|
11 |
* Extracts a region, bucket, key, and and if a URI is in path-style
|
|
|
12 |
*/
|
|
|
13 |
class S3UriParser
|
|
|
14 |
{
|
|
|
15 |
private $pattern = '/^(.+\\.)?s3[.-]([A-Za-z0-9-]+)\\./';
|
|
|
16 |
private $streamWrapperScheme = 's3';
|
|
|
17 |
|
|
|
18 |
private static $defaultResult = [
|
|
|
19 |
'path_style' => true,
|
|
|
20 |
'bucket' => null,
|
|
|
21 |
'key' => null,
|
|
|
22 |
'region' => null
|
|
|
23 |
];
|
|
|
24 |
|
|
|
25 |
/**
|
|
|
26 |
* Parses a URL or S3 StreamWrapper Uri (s3://) into an associative array
|
|
|
27 |
* of Amazon S3 data including:
|
|
|
28 |
*
|
|
|
29 |
* - bucket: The Amazon S3 bucket (null if none)
|
|
|
30 |
* - key: The Amazon S3 key (null if none)
|
|
|
31 |
* - path_style: Set to true if using path style, or false if not
|
|
|
32 |
* - region: Set to a string if a non-class endpoint is used or null.
|
|
|
33 |
*
|
|
|
34 |
* @param string|UriInterface $uri
|
|
|
35 |
*
|
|
|
36 |
* @return array
|
|
|
37 |
* @throws \InvalidArgumentException|InvalidArnException
|
|
|
38 |
*/
|
|
|
39 |
public function parse($uri)
|
|
|
40 |
{
|
|
|
41 |
// Attempt to parse host component of uri as an ARN
|
|
|
42 |
$components = $this->parseS3UrlComponents($uri);
|
|
|
43 |
if (!empty($components)) {
|
|
|
44 |
if (ArnParser::isArn($components['host'])) {
|
|
|
45 |
$arn = new AccessPointArn($components['host']);
|
|
|
46 |
return [
|
|
|
47 |
'bucket' => $components['host'],
|
|
|
48 |
'key' => $components['path'],
|
|
|
49 |
'path_style' => false,
|
|
|
50 |
'region' => $arn->getRegion()
|
|
|
51 |
];
|
|
|
52 |
}
|
|
|
53 |
}
|
|
|
54 |
|
|
|
55 |
$url = Psr7\Utils::uriFor($uri);
|
|
|
56 |
|
|
|
57 |
if ($url->getScheme() == $this->streamWrapperScheme) {
|
|
|
58 |
return $this->parseStreamWrapper($url);
|
|
|
59 |
}
|
|
|
60 |
|
|
|
61 |
if (!$url->getHost()) {
|
|
|
62 |
throw new \InvalidArgumentException('No hostname found in URI: '
|
|
|
63 |
. $uri);
|
|
|
64 |
}
|
|
|
65 |
|
|
|
66 |
if (!preg_match($this->pattern, $url->getHost(), $matches)) {
|
|
|
67 |
return $this->parseCustomEndpoint($url);
|
|
|
68 |
}
|
|
|
69 |
|
|
|
70 |
// Parse the URI based on the matched format (path / virtual)
|
|
|
71 |
$result = empty($matches[1])
|
|
|
72 |
? $this->parsePathStyle($url)
|
|
|
73 |
: $this->parseVirtualHosted($url, $matches);
|
|
|
74 |
|
|
|
75 |
// Add the region if one was found and not the classic endpoint
|
|
|
76 |
$result['region'] = $matches[2] == 'amazonaws' ? null : $matches[2];
|
|
|
77 |
|
|
|
78 |
return $result;
|
|
|
79 |
}
|
|
|
80 |
|
|
|
81 |
private function parseS3UrlComponents($uri)
|
|
|
82 |
{
|
|
|
83 |
preg_match("/^([a-zA-Z0-9]*):\/\/([a-zA-Z0-9:-]*)\/(.*)/", $uri, $components);
|
|
|
84 |
if (empty($components)) {
|
|
|
85 |
return [];
|
|
|
86 |
}
|
|
|
87 |
return [
|
|
|
88 |
'scheme' => $components[1],
|
|
|
89 |
'host' => $components[2],
|
|
|
90 |
'path' => $components[3],
|
|
|
91 |
];
|
|
|
92 |
}
|
|
|
93 |
|
|
|
94 |
private function parseStreamWrapper(UriInterface $url)
|
|
|
95 |
{
|
|
|
96 |
$result = self::$defaultResult;
|
|
|
97 |
$result['path_style'] = false;
|
|
|
98 |
|
|
|
99 |
$result['bucket'] = $url->getHost();
|
|
|
100 |
if ($url->getPath()) {
|
|
|
101 |
$key = ltrim($url->getPath(), '/ ');
|
|
|
102 |
if (!empty($key)) {
|
|
|
103 |
$result['key'] = $key;
|
|
|
104 |
}
|
|
|
105 |
}
|
|
|
106 |
|
|
|
107 |
return $result;
|
|
|
108 |
}
|
|
|
109 |
|
|
|
110 |
private function parseCustomEndpoint(UriInterface $url)
|
|
|
111 |
{
|
|
|
112 |
$result = self::$defaultResult;
|
|
|
113 |
$path = ltrim($url->getPath(), '/ ');
|
|
|
114 |
$segments = explode('/', $path, 2);
|
|
|
115 |
|
|
|
116 |
if (isset($segments[0])) {
|
|
|
117 |
$result['bucket'] = $segments[0];
|
|
|
118 |
if (isset($segments[1])) {
|
|
|
119 |
$result['key'] = $segments[1];
|
|
|
120 |
}
|
|
|
121 |
}
|
|
|
122 |
|
|
|
123 |
return $result;
|
|
|
124 |
}
|
|
|
125 |
|
|
|
126 |
private function parsePathStyle(UriInterface $url)
|
|
|
127 |
{
|
|
|
128 |
$result = self::$defaultResult;
|
|
|
129 |
|
|
|
130 |
if ($url->getPath() != '/') {
|
|
|
131 |
$path = ltrim($url->getPath(), '/');
|
|
|
132 |
if ($path) {
|
|
|
133 |
$pathPos = strpos($path, '/');
|
|
|
134 |
if ($pathPos === false) {
|
|
|
135 |
// https://s3.amazonaws.com/bucket
|
|
|
136 |
$result['bucket'] = $path;
|
|
|
137 |
} elseif ($pathPos == strlen($path) - 1) {
|
|
|
138 |
// https://s3.amazonaws.com/bucket/
|
|
|
139 |
$result['bucket'] = substr($path, 0, -1);
|
|
|
140 |
} else {
|
|
|
141 |
// https://s3.amazonaws.com/bucket/key
|
|
|
142 |
$result['bucket'] = substr($path, 0, $pathPos);
|
|
|
143 |
$result['key'] = substr($path, $pathPos + 1) ?: null;
|
|
|
144 |
}
|
|
|
145 |
}
|
|
|
146 |
}
|
|
|
147 |
|
|
|
148 |
return $result;
|
|
|
149 |
}
|
|
|
150 |
|
|
|
151 |
private function parseVirtualHosted(UriInterface $url, array $matches)
|
|
|
152 |
{
|
|
|
153 |
$result = self::$defaultResult;
|
|
|
154 |
$result['path_style'] = false;
|
|
|
155 |
// Remove trailing "." from the prefix to get the bucket
|
|
|
156 |
$result['bucket'] = substr($matches[1], 0, -1);
|
|
|
157 |
$path = $url->getPath();
|
|
|
158 |
// Check if a key was present, and if so, removing the leading "/"
|
|
|
159 |
$result['key'] = !$path || $path == '/' ? null : substr($path, 1);
|
|
|
160 |
|
|
|
161 |
return $result;
|
|
|
162 |
}
|
|
|
163 |
}
|