easyfonts / src / Fonts / Downloader.php

Downloader.php in EasyFonts – Host Google Fonts Locally, Fast & Auto-Optimize, GDPR Compliant 2.0.1, at src/Fonts/Downloader.php

310 lines 8.1 KB

Raw Download Zip

1	<?php
2	/**
3	* Remote fetcher for stylesheets and font binaries.
4	*
5	* @package EasyFonts
6	*/
7
8	namespace EasyFonts\Fonts;
9
10	defined( 'ABSPATH' ) \|\| exit;
11
12	/**
13	* Downloads Google/Bunny CSS and font files.
14	*/
15	class Downloader {
16
17	/**
18	* Whether remote font-binary downloads are allowed in the current request.
19	* Set to false during a normal visitor render (so the page is never blocked
20	* by synchronous downloads); true during admin/CLI/warm (probe) requests.
21	*
22	* @var bool
23	*/
24	private static bool $allow_fonts = true;
25
26	/**
27	* Allow/disallow remote font-binary downloads for this request.
28	*
29	* @param bool $allow Allowed?
30	* @return void
31	*/
32	public static function set_allow_fonts( bool $allow ): void {
33	self::$allow_fonts = $allow;
34	}
35
36	/**
37	* Are remote font-binary downloads currently allowed?
38	*
39	* @return bool
40	*/
41	public static function fonts_allowed(): bool {
42	return self::$allow_fonts;
43	}
44
45	/**
46	* Modern desktop UA so the API returns woff2.
47	*/
48	const UA_WOFF2 = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:124.0) Gecko/20100101 Firefox/124.0';
49
50	/**
51	* Old UA that makes the API return TTF (used for metric extraction only).
52	*/
53	const UA_TTF = 'Mozilla/5.0 (Windows NT 6.1; rv:27.0) Gecko/20100101 Firefox/27.0';
54
55	/**
56	* Content-Type => extension.
57	*/
58	const MIME = array(
59	'font/woff2' => 'woff2',
60	'application/font-woff2' => 'woff2',
61	'font/woff' => 'woff',
62	'application/font-woff' => 'woff',
63	'font/ttf' => 'ttf',
64	'application/x-font-ttf' => 'ttf',
65	'font/sfnt' => 'ttf',
66	'application/font-sfnt' => 'ttf',
67	'font/otf' => 'otf',
68	'application/x-font-opentype' => 'otf',
69	'application/vnd.ms-fontobject' => 'eot',
70	);
71
72	/**
73	* Fetch CSS text from a font-provider URL.
74	*
75	* @param string $url Stylesheet URL.
76	* @return string\|null
77	*/
78	public function fetch_css( string $url ): ?string {
79	$url = \EasyFonts\Detect\Providers::normalize_url( $url );
80
81	// Cache the fetched CSS so font detection doesn't hit the network on every
82	// page render. Keyed by URL + cache-buster, so any settings/font change
83	// (which bumps the buster) transparently refreshes it.
84	$cache_key = 'easyfonts_css_' . md5( $url . '\|' . \EasyFonts\Settings::buster() );
85	$cached = get_transient( $cache_key );
86
87	if ( is_string( $cached ) && '' !== $cached ) {
88	return $cached;
89	}
90
91	$response = wp_remote_get(
92	$url,
93	array(
94	'timeout' => 10,
95	'user-agent' => apply_filters( 'easyfonts_css_user_agent', self::UA_WOFF2 ),
96	)
97	);
98
99	if ( is_wp_error( $response ) \|\| 200 !== (int) wp_remote_retrieve_response_code( $response ) ) {
100	return null;
101	}
102
103	$body = wp_remote_retrieve_body( $response );
104
105	if ( '' === $body ) {
106	return null;
107	}
108
109	set_transient( $cache_key, $body, 7 * DAY_IN_SECONDS );
110
111	return $body;
112	}
113
114	/**
115	* Download a font binary into the cache and return [filename, ext, size].
116	*
117	* @param string $url Font file URL.
118	* @param string $basename Cache basename (without extension).
119	* @param Storage $storage Storage.
120	* @return array{file:string,ext:string,size:int}\|null
121	*/
122	public function fetch_font( string $url, string $basename, Storage $storage ): ?array {
123	// Never download a font binary during a normal visitor page render — that
124	// is what could stall the request and white-screen the site. On the render
125	// path we serve only already-cached fonts; missing ones are fetched by a
126	// background warm (probe) request. Admin/CLI/warm requests set this true.
127	if ( ! self::$allow_fonts ) {
128	return null;
129	}
130
131	$url = \EasyFonts\Detect\Providers::normalize_url( $url );
132
133	// SSRF guard: only fetch font binaries from recognised provider hosts
134	// (or the site's own origin). A crafted @font-face can otherwise point
135	// src at an internal address; this blocks that without affecting any
136	// legitimate Google/Bunny/wp.com font. Filterable for niche providers.
137	if ( ! self::host_allowed( $url ) ) {
138	return null;
139	}
140
141	$response = wp_remote_get(
142	$url,
143	array(
144	'timeout' => 10,
145	'user-agent' => self::UA_WOFF2,
146	)
147	);
148
149	if ( is_wp_error( $response ) ) {
150	return null;
151	}
152
153	$code = (int) wp_remote_retrieve_response_code( $response );
154
155	if ( $code < 200 \|\| $code >= 300 ) {
156	return null;
157	}
158
159	$body = wp_remote_retrieve_body( $response );
160
161	if ( '' === $body ) {
162	return null;
163	}
164
165	$ct = strtolower( trim( explode( ';', (string) wp_remote_retrieve_header( $response, 'content-type' ) )[0] ) );
166	$ext = self::MIME[ $ct ] ?? $this->ext_from_url( $url );
167
168	$filename = $basename . '.' . $ext;
169
170	if ( ! $storage->write( $filename, $body ) ) {
171	return null;
172	}
173
174	return array(
175	'file' => $filename,
176	'ext' => $ext,
177	'size' => strlen( $body ),
178	);
179	}
180
181	/**
182	* Fetch the raw bytes of a TTF copy of a Google Fonts family for metrics.
183	* Returns the binary, never writes it to disk.
184	*
185	* @param string $css_url Original CSS URL.
186	* @return string\|null
187	*/
188	public function fetch_ttf_bytes( string $css_url ): ?string {
189	$response = wp_remote_get(
190	$css_url,
191	array(
192	'timeout' => 20,
193	'user-agent' => self::UA_TTF,
194	)
195	);
196
197	if ( is_wp_error( $response ) \|\| 200 !== (int) wp_remote_retrieve_response_code( $response ) ) {
198	return null;
199	}
200
201	$css = wp_remote_retrieve_body( $response );
202
203	if ( ! preg_match( '/src\s:\s[^;]?url$\s[\'"]?([^\'")]+\.(?:ttf\|otf))[\'"]?\s*$/i', $css, $m ) ) {
204	return null;
205	}
206
207	$font = wp_remote_get( $m[1], array( 'timeout' => 20 ) );
208
209	if ( is_wp_error( $font ) \|\| 200 !== (int) wp_remote_retrieve_response_code( $font ) ) {
210	return null;
211	}
212
213	$bytes = wp_remote_retrieve_body( $font );
214
215	return '' === $bytes ? null : $bytes;
216	}
217
218	/**
219	* Fetch TTF bytes for a specific Google family/weight purely for metric
220	* extraction. Source-independent: we construct a css2 request from the
221	* family name, so it works whether the font came from a <link>, @import, or
222	* an inline @font-face. Returns binary, never written to disk.
223	*
224	* @param string $family Family name (e.g. "Open Sans").
225	* @param string $weight Representative numeric weight.
226	* @return string\|null
227	*/
228	public function fetch_family_ttf( string $family, string $weight = '400' ): ?string {
229	$weight = (string) absint( $weight );
230
231	if ( '' === $weight \|\| '0' === $weight ) {
232	$weight = '400';
233	}
234
235	$url = 'https://fonts.googleapis.com/css2?family='
236	. rawurlencode( $family ) . ':wght@' . $weight . '&display=swap';
237	$url = str_replace( '%20', '+', $url );
238
239	return $this->fetch_ttf_bytes( $url );
240	}
241
242	/**
243	* Is this URL host allowed for font-binary downloads? Provider font hosts
244	* plus the site's own origin. Filterable via `easyfonts_font_src_hosts`.
245	*
246	* @param string $url Absolute URL.
247	* @return bool
248	*/
249	private static function host_allowed( string $url ): bool {
250	$host = strtolower( (string) wp_parse_url( $url, PHP_URL_HOST ) );
251
252	if ( '' === $host ) {
253	return false;
254	}
255
256	$own = strtolower( (string) wp_parse_url( home_url(), PHP_URL_HOST ) );
257
258	/**
259	* Allowed host suffixes for downloading font binaries.
260	*
261	* @param string[] $hosts
262	*/
263	$allowed = apply_filters(
264	'easyfonts_font_src_hosts',
265	array(
266	'fonts.gstatic.com',
267	'fonts.googleapis.com',
268	'fonts.bunny.net',
269	'fonts.wp.com',
270	'fonts-api.wp.com',
271	's.w.org',
272	'i0.wp.com',
273	'i1.wp.com',
274	'i2.wp.com',
275	)
276	);
277
278	if ( '' !== $own ) {
279	$allowed[] = $own;
280	}
281
282	foreach ( $allowed as $candidate ) {
283	$candidate = strtolower( trim( (string) $candidate ) );
284
285	if ( '' === $candidate ) {
286	continue;
287	}
288
289	// Exact host or a subdomain of an allowed host.
290	if ( $host === $candidate \|\| substr( $host, -strlen( '.' . $candidate ) ) === '.' . $candidate ) {
291	return true;
292	}
293	}
294
295	return false;
296	}
297
298	/**
299	* Guess an extension from a URL path.
300	*
301	* @param string $url URL.
302	* @return string
303	*/
304	private function ext_from_url( string $url ): string {
305	$ext = strtolower( (string) pathinfo( (string) wp_parse_url( $url, PHP_URL_PATH ), PATHINFO_EXTENSION ) );
306
307	return in_array( $ext, array( 'woff2', 'woff', 'ttf', 'otf', 'eot' ), true ) ? $ext : 'woff2';
308	}
309	}
310