media-cloud-sync / includes / sdk / s3 / Aws / S3 / S3UriParser.php

S3UriParser.php in Media Cloud Sync 1.4.1, at includes/sdk/s3/Aws/S3/S3UriParser.php

126 lines 4.8 KB

Raw Download Zip

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