auto-alt-text / src / App / Infrastructure / Http / WordPressImageFetcher.php

WordPressImageFetcher.php in Auto Alt Text 3.0.3, at src/App/Infrastructure/Http/WordPressImageFetcher.php

133 lines 4.0 KB

Raw Download Zip

1	<?php
2
3	declare(strict_types=1);
4
5	namespace AATXT\App\Infrastructure\Http;
6
7	use Exception;
8
9	/**
10	* WordPress implementation of the image fetcher.
11	*
12	* Providers that accept only inline image data cannot download the image
13	* themselves, so its bytes have to travel with the request. When the URL points
14	* inside the uploads directory the file is read straight from disk, which keeps
15	* generation working on local installs, password-protected staging sites and
16	* sites behind HTTP auth - none of which are reachable from the outside.
17	* Anything else falls back to an HTTP request.
18	*/
19	class WordPressImageFetcher implements ImageFetcherInterface
20	{
21	/**
22	* Timeout for the HTTP fallback, in seconds.
23	*/
24	private const DEFAULT_TIMEOUT = 30;
25
26	/**
27	* Read an image and return its raw bytes base64-encoded.
28	*
29	* @param string $imageUrl The image URL
30	*
31	* @return string The base64-encoded image contents
32	*
33	* @throws Exception If the image cannot be read
34	*/
35	public function fetchAsBase64(string $imageUrl): string
36	{
37	$contents = $this->readFromUploads($imageUrl);
38
39	if ($contents === null) {
40	$contents = $this->readFromHttp($imageUrl);
41	}
42
43	return base64_encode($contents);
44	}
45
46	/**
47	* Read the file from the uploads directory when the URL belongs to it.
48	*
49	* @param string $imageUrl The image URL
50	*
51	* @return string\|null The raw bytes, or null when the URL is not a local
52	* upload or the file is not readable
53	*/
54	private function readFromUploads(string $imageUrl): ?string
55	{
56	$uploads = wp_get_upload_dir();
57
58	if (!empty($uploads['error']) \|\| empty($uploads['baseurl']) \|\| empty($uploads['basedir'])) {
59	return null;
60	}
61
62	// Compare without the scheme: a http/https mismatch between the stored
63	// URL and the current request must not force a needless HTTP round trip.
64	$baseUrl = $this->stripScheme($uploads['baseurl']);
65	$url = $this->stripScheme($imageUrl);
66
67	if (strpos($url, $baseUrl) !== 0) {
68	return null;
69	}
70
71	// Drop any query string (e.g. cache busting parameters) before
72	// touching the filesystem.
73	$relativePath = (string) strtok(substr($url, strlen($baseUrl)), '?');
74	$path = realpath($uploads['basedir'] . $relativePath);
75	$baseDir = realpath($uploads['basedir']);
76
77	// Defensive: never read outside the uploads directory, whatever the URL contains.
78	if ($path === false \|\| $baseDir === false \|\| strpos($path, $baseDir) !== 0) {
79	return null;
80	}
81
82	if (!is_file($path) \|\| !is_readable($path)) {
83	return null;
84	}
85
86	$contents = file_get_contents($path);
87
88	return $contents === false ? null : $contents;
89	}
90
91	/**
92	* Download the image over HTTP.
93	*
94	* @param string $imageUrl The image URL
95	*
96	* @return string The raw bytes
97	*
98	* @throws Exception If the download fails or returns an empty body
99	*/
100	private function readFromHttp(string $imageUrl): string
101	{
102	$response = wp_remote_get($imageUrl, ['timeout' => self::DEFAULT_TIMEOUT]);
103
104	if (is_wp_error($response)) {
105	throw new Exception(
106	sprintf('Unable to download the image: %s', $response->get_error_message())
107	);
108	}
109
110	$statusCode = wp_remote_retrieve_response_code($response);
111	if ($statusCode < 200 \|\| $statusCode >= 300) {
112	throw new Exception(
113	sprintf('Unable to download the image: HTTP status %d', $statusCode)
114	);
115	}
116
117	$body = wp_remote_retrieve_body($response);
118	if ($body === '') {
119	throw new Exception('The downloaded image is empty');
120	}
121
122	return $body;
123	}
124
125	/**
126	* Remove the http/https scheme from a URL.
127	*/
128	private function stripScheme(string $url): string
129	{
130	return (string) preg_replace('#^https?://#i', '', $url);
131	}
132	}
133