ApiCSPReport: Support origin/path matching for false positives
[lhc/web/wiklou.git] / includes / api / ApiCSPReport.php
1 <?php
2 /**
3 * Copyright © 2015 Brian Wolff
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @file
21 */
22
23 use MediaWiki\Logger\LoggerFactory;
24
25 /**
26 * Api module to receive and log CSP violation reports
27 *
28 * @ingroup API
29 */
30 class ApiCSPReport extends ApiBase {
31
32 private $log;
33
34 /**
35 * These reports should be small. Ignore super big reports out of paranoia
36 */
37 const MAX_POST_SIZE = 8192;
38
39 /**
40 * Logs a content-security-policy violation report from web browser.
41 */
42 public function execute() {
43 $reportOnly = $this->getParameter( 'reportonly' );
44 $logname = $reportOnly ? 'csp-report-only' : 'csp';
45 $this->log = LoggerFactory::getInstance( $logname );
46 $userAgent = $this->getRequest()->getHeader( 'user-agent' );
47
48 $this->verifyPostBodyOk();
49 $report = $this->getReport();
50 $flags = $this->getFlags( $report, $userAgent );
51
52 $warningText = $this->generateLogLine( $flags, $report );
53 $this->logReport( $flags, $warningText, [
54 // XXX Is it ok to put untrusted data into log??
55 'csp-report' => $report,
56 'method' => __METHOD__,
57 'user' => $this->getUser()->getName(),
58 'user-agent' => $userAgent,
59 'source' => $this->getParameter( 'source' ),
60 ] );
61 $this->getResult()->addValue( null, $this->getModuleName(), 'success' );
62 }
63
64 /**
65 * Log CSP report, with a different severity depending on $flags
66 * @param array $flags Flags for this report
67 * @param string $logLine text of log entry
68 * @param array $context logging context
69 */
70 private function logReport( $flags, $logLine, $context ) {
71 if ( in_array( 'false-positive', $flags ) ) {
72 // These reports probably don't matter much
73 $this->log->debug( $logLine, $context );
74 } else {
75 // Normal report.
76 $this->log->warning( $logLine, $context );
77 }
78 }
79
80 /**
81 * Get extra notes about the report.
82 *
83 * @param array $report The CSP report
84 * @param string $userAgent
85 * @return array
86 */
87 private function getFlags( $report, $userAgent ) {
88 $reportOnly = $this->getParameter( 'reportonly' );
89 $source = $this->getParameter( 'source' );
90 $falsePositives = $this->getConfig()->get( 'CSPFalsePositiveUrls' );
91
92 $flags = [];
93 if ( $source !== 'internal' ) {
94 $flags[] = 'source=' . $source;
95 }
96 if ( $reportOnly ) {
97 $flags[] = 'report-only';
98 }
99
100 if (
101 (
102 ContentSecurityPolicy::falsePositiveBrowser( $userAgent ) &&
103 $report['blocked-uri'] === "self"
104 ) ||
105 (
106 isset( $report['blocked-uri'] ) &&
107 $this->matchUrlPattern( $report['blocked-uri'], $falsePositives )
108 ) ||
109 (
110 isset( $report['source-file'] ) &&
111 $this->matchUrlPattern( $report['source-file'], $falsePositives )
112 )
113 ) {
114 // False positive due to:
115 // https://bugzilla.mozilla.org/show_bug.cgi?id=1026520
116
117 $flags[] = 'false-positive';
118 }
119 return $flags;
120 }
121
122 /**
123 * @param string $url
124 * @param string[] $patterns
125 * @return bool
126 */
127 private function matchUrlPattern( $url, array $patterns ) {
128 if ( isset( $patterns[ $url ] ) ) {
129 return true;
130 }
131
132 $bits = wfParseUrl( $url );
133 unset( $bits['user'], $bits['pass'], $bits['query'], $bits['fragment'] );
134 $bits['path'] = '';
135 $serverUrl = wfAssembleUrl( $bits );
136 if ( isset( $patterns[$serverUrl] ) ) {
137 // The origin of the url matches a pattern,
138 // e.g. "https://example.org" matches "https://example.org/foo/b?a#r"
139 return true;
140 }
141 foreach ( $patterns as $pattern => $val ) {
142 // We only use this pattern if it ends in a slash, this prevents
143 // "/foos" from matching "/foo", and "https://good.combo.bad" matching
144 // "https://good.com".
145 if ( substr( $pattern, -1 ) === '/' && strpos( $url, $pattern ) === 0 ) {
146 // The pattern starts with the same as the url
147 // e.g. "https://example.org/foo/" matches "https://example.org/foo/b?a#r"
148 return true;
149 }
150 }
151
152 return false;
153 }
154
155 /**
156 * Output an api error if post body is obviously not OK.
157 */
158 private function verifyPostBodyOk() {
159 $req = $this->getRequest();
160 $contentType = $req->getHeader( 'content-type' );
161 if ( $contentType !== 'application/json'
162 && $contentType !== 'application/csp-report'
163 ) {
164 $this->error( 'wrongformat', __METHOD__ );
165 }
166 if ( $req->getHeader( 'content-length' ) > self::MAX_POST_SIZE ) {
167 $this->error( 'toobig', __METHOD__ );
168 }
169 }
170
171 /**
172 * Get the report from post body and turn into associative array.
173 *
174 * @return array
175 */
176 private function getReport() {
177 $postBody = $this->getRequest()->getRawInput();
178 if ( strlen( $postBody ) > self::MAX_POST_SIZE ) {
179 // paranoia, already checked content-length earlier.
180 $this->error( 'toobig', __METHOD__ );
181 }
182 $status = FormatJson::parse( $postBody, FormatJson::FORCE_ASSOC );
183 if ( !$status->isGood() ) {
184 $msg = $status->getErrors()[0]['message'];
185 if ( $msg instanceof Message ) {
186 $msg = $msg->getKey();
187 }
188 $this->error( $msg, __METHOD__ );
189 }
190
191 $report = $status->getValue();
192
193 if ( !isset( $report['csp-report'] ) ) {
194 $this->error( 'missingkey', __METHOD__ );
195 }
196 return $report['csp-report'];
197 }
198
199 /**
200 * Get text of log line.
201 *
202 * @param array $flags of additional markers for this report
203 * @param array $report the csp report
204 * @return string Text to put in log
205 */
206 private function generateLogLine( $flags, $report ) {
207 $flagText = '';
208 if ( $flags ) {
209 $flagText = '[' . implode( ', ', $flags ) . ']';
210 }
211
212 $blockedFile = $report['blocked-uri'] ?? 'n/a';
213 $page = $report['document-uri'] ?? 'n/a';
214 $line = isset( $report['line-number'] ) ? ':' . $report['line-number'] : '';
215 $warningText = $flagText .
216 ' Received CSP report: <' . $blockedFile .
217 '> blocked from being loaded on <' . $page . '>' . $line;
218 return $warningText;
219 }
220
221 /**
222 * Stop processing the request, and output/log an error
223 *
224 * @param string $code error code
225 * @param string $method method that made error
226 * @throws ApiUsageException Always
227 */
228 private function error( $code, $method ) {
229 $this->log->info( 'Error reading CSP report: ' . $code, [
230 'method' => $method,
231 'user-agent' => $this->getRequest()->getHeader( 'user-agent' )
232 ] );
233 // Return 400 on error for user agents to display, e.g. to the console.
234 $this->dieWithError(
235 [ 'apierror-csp-report', wfEscapeWikiText( $code ) ], 'cspreport-' . $code, [], 400
236 );
237 }
238
239 public function getAllowedParams() {
240 return [
241 'reportonly' => [
242 ApiBase::PARAM_TYPE => 'boolean',
243 ApiBase::PARAM_DFLT => false
244 ],
245 'source' => [
246 ApiBase::PARAM_TYPE => 'string',
247 ApiBase::PARAM_DFLT => 'internal',
248 ApiBase::PARAM_REQUIRED => false
249 ]
250 ];
251 }
252
253 public function mustBePosted() {
254 return true;
255 }
256
257 public function isWriteMode() {
258 return false;
259 }
260
261 /**
262 * Mark as internal. This isn't meant to be used by normal api users
263 * @return bool
264 */
265 public function isInternal() {
266 return true;
267 }
268
269 /**
270 * Even if you don't have read rights, we still want your report.
271 * @return bool
272 */
273 public function isReadMode() {
274 return false;
275 }
276
277 /**
278 * Doesn't touch db, so max lag should be rather irrelavent.
279 *
280 * Also, this makes sure that reports aren't lost during lag events.
281 * @return bool
282 */
283 public function shouldCheckMaxLag() {
284 return false;
285 }
286 }