Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
97.70% |
85 / 87 |
|
80.00% |
8 / 10 |
CRAP | |
0.00% |
0 / 1 |
| StreamConfig | |
97.70% |
85 / 87 |
|
80.00% |
8 / 10 |
33 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
1 | |||
| stream | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| topics | |
100.00% |
25 / 25 |
|
100.00% |
1 / 1 |
6 | |||
| toArray | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| matches | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| matchesSettings | |
80.00% |
4 / 5 |
|
0.00% |
0 / 1 |
3.07 | |||
| isPartialMatch | |
93.75% |
15 / 16 |
|
0.00% |
0 / 1 |
12.04 | |||
| arrayIsList | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| isValidRegex | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| validate | |
100.00% |
24 / 24 |
|
100.00% |
1 / 1 |
4 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\EventStreamConfig; |
| 4 | |
| 5 | use Wikimedia\Assert\Assert; |
| 6 | |
| 7 | /** |
| 8 | * Configuration of single stream's settings. |
| 9 | */ |
| 10 | class StreamConfig { |
| 11 | |
| 12 | /** |
| 13 | * Key of stream setting. Used to DRY references here. |
| 14 | * @var string |
| 15 | */ |
| 16 | public const STREAM_SETTING = 'stream'; |
| 17 | |
| 18 | /** |
| 19 | * Streams can be made up of multiple topics. If |
| 20 | * topic_prefixes are set, the topics will default to be |
| 21 | * the stream name with these prefixes. |
| 22 | * @var array |
| 23 | */ |
| 24 | public const TOPIC_PREFIXES_SETTING = 'topic_prefixes'; |
| 25 | |
| 26 | /** |
| 27 | * Streams can be made up of multiple topics. |
| 28 | * @var array |
| 29 | */ |
| 30 | public const TOPICS_SETTING = 'topics'; |
| 31 | |
| 32 | /** |
| 33 | * Wrapped array with stream config settings. See $settings |
| 34 | * param doc for __construct(). |
| 35 | * @var array |
| 36 | */ |
| 37 | private $settings; |
| 38 | |
| 39 | /** |
| 40 | * True if the stream setting is a regex, false otherwise. |
| 41 | * Used to avoid attempting to regex match with a non regex. |
| 42 | * @var bool |
| 43 | */ |
| 44 | private $streamIsRegex; |
| 45 | |
| 46 | /** |
| 47 | * @param string $stream stream name, e.g., "my.event.stream-name" |
| 48 | * @param array $settings |
| 49 | * An array with stream config settings. |
| 50 | * Must include a 'stream' setting with either the explicit stream |
| 51 | * name, or a regex pattern (starting with '/') that will match |
| 52 | * against stream names that this config should be used for. |
| 53 | * Example: |
| 54 | * [ |
| 55 | * "schema_title" => "my/event/schema", |
| 56 | * "sample" => [ |
| 57 | * "rate" => 0.8, |
| 58 | * "unit" => "session", |
| 59 | * ], |
| 60 | * "destination_event_service" => "eventgate-analytics-public", |
| 61 | * ... |
| 62 | * ] |
| 63 | * @param array $defaultSettings |
| 64 | * An array with default stream config settings. |
| 65 | */ |
| 66 | public function __construct( string $stream, array $settings, array $defaultSettings = [] ) { |
| 67 | $this->settings = array_replace_recursive( $defaultSettings, $settings ); |
| 68 | |
| 69 | // To preserve backward compatibility, add the stream name or regex pattern to the config body |
| 70 | // as the 'stream' setting. This can likely be removed if we're sure nothing is relying on it. |
| 71 | // Background: T277193 |
| 72 | $this->settings[self::STREAM_SETTING] = $stream; |
| 73 | |
| 74 | self::validate( $this->settings ); |
| 75 | |
| 76 | $this->streamIsRegex = self::isValidRegex( $this->stream() ); |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * Gets the stream setting |
| 81 | * @return string |
| 82 | */ |
| 83 | public function stream() { |
| 84 | return $this->settings[self::STREAM_SETTING]; |
| 85 | } |
| 86 | |
| 87 | /** |
| 88 | * Returns a list of topics that compose the $stream. |
| 89 | * If a target $stream is not provided, $stream will default to the value of |
| 90 | * $this->stream(), which is this StreamConfig's stream setting. |
| 91 | * |
| 92 | * - If self::TOPICS_SETTING is set, this will be returned as is. |
| 93 | * - Else if self::TOPIC_PREFIXES_SETTING is not set, then [$stream] will be returned. |
| 94 | * - Else if $stream is a regex, it will be modified to include the prefixes in the regex. |
| 95 | * - Else a list of topics with the prefixes will be returned. |
| 96 | * |
| 97 | * @param string|null $stream |
| 98 | * @return array |
| 99 | */ |
| 100 | public function topics( $stream = null ): array { |
| 101 | // if $stream was provided, it could be an explicit target stream name |
| 102 | // OR a regex stream setting. $this->stream() could also be either, but in the |
| 103 | // case where it is a regex, a user might want to provide an explicit stream name |
| 104 | // to get topics rather than this StreamConfig's regex stream setting. |
| 105 | // Default to using the StreamConfig stream setting otherwise. |
| 106 | if ( !$stream ) { |
| 107 | $stream = $this->stream(); |
| 108 | } |
| 109 | |
| 110 | if ( isset( $this->settings[self::TOPICS_SETTING] ) ) { |
| 111 | // If this stream was configured with specific topics, just return those. |
| 112 | return $this->settings[self::TOPICS_SETTING]; |
| 113 | } elseif ( !isset( $this->settings[self::TOPIC_PREFIXES_SETTING] ) ) { |
| 114 | // Else if this stream does not has topic prefixes, just return the |
| 115 | // stream name as the topic. |
| 116 | return [ $stream ]; |
| 117 | } else { |
| 118 | $topicPrefixes = $this->settings[self::TOPIC_PREFIXES_SETTING]; |
| 119 | |
| 120 | if ( self::isValidRegex( $stream ) ) { |
| 121 | // This is a regex string stream name, return a regex with the prefixes. |
| 122 | |
| 123 | // Remove the regex boundry chars. |
| 124 | $streamPattern = trim( $stream, '/' ); |
| 125 | |
| 126 | // If the regex starts with ^, save it for later. |
| 127 | $beginAnchor = ''; |
| 128 | if ( substr( $streamPattern, 0, 1 ) === '^' ) { |
| 129 | $beginAnchor = '^'; |
| 130 | $streamPattern = substr( $streamPattern, 1 ); |
| 131 | } |
| 132 | |
| 133 | // Escape any regex looking chars in the prefixes |
| 134 | $topicPrefixes = array_map( "preg_quote", $topicPrefixes ); |
| 135 | |
| 136 | // Reconstruct the regex with prefixes, e.g. |
| 137 | // /^(eqiad.|codfw.) |
| 138 | return [ |
| 139 | '/' . $beginAnchor . |
| 140 | '(' . implode( '|', $topicPrefixes ) . ')' . $streamPattern . |
| 141 | '/' |
| 142 | ]; |
| 143 | } else { |
| 144 | // Else prefix the stream with each topic prefix. |
| 145 | // If $stream is a regex string, then we need to alter the |
| 146 | // regex to prefix safely inside the regex string. |
| 147 | return array_map( |
| 148 | static function ( $topicPrefix ) use ( $stream ) { |
| 149 | return $topicPrefix . $stream; |
| 150 | }, |
| 151 | $topicPrefixes |
| 152 | ); |
| 153 | } |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | /** |
| 158 | * Returns this StreamConfig as an array of settings. |
| 159 | * self::TOPICS_SETTING is set to the value returned by $this->topics($targetStream) |
| 160 | * if self::TOPICS_SETTING is not explicitly set. |
| 161 | * |
| 162 | * @param string|null $targetStream |
| 163 | * If given, this will be used to get topics, otherwise $this->stream() |
| 164 | * will be used (which could be a regex stream pattern). |
| 165 | * |
| 166 | * @return array |
| 167 | */ |
| 168 | public function toArray( $targetStream = null ): array { |
| 169 | $settings = $this->settings; |
| 170 | |
| 171 | // If TOPICS_SETTING is already set explicitly in the settings, |
| 172 | // $this->topics() will just return it. |
| 173 | $settings[self::TOPICS_SETTING] = $this->topics( $targetStream ); |
| 174 | |
| 175 | return $settings; |
| 176 | } |
| 177 | |
| 178 | /** |
| 179 | * True if this StreamConfig applies for $stream, false otherwise. |
| 180 | * |
| 181 | * @param string $stream name of stream to match |
| 182 | * @return bool |
| 183 | */ |
| 184 | public function matches( $stream ) { |
| 185 | return $this->streamIsRegex ? |
| 186 | preg_match( $this->stream(), $stream ) : |
| 187 | ( $this->stream() === $stream ); |
| 188 | } |
| 189 | |
| 190 | /** |
| 191 | * True if this StreamConfig has all of the given $settingsConstraints. |
| 192 | * If 'stream' is given as a constraint, it be matched against this |
| 193 | * StreamConfig's stream name via $this->matches. |
| 194 | * |
| 195 | * @param array $settingsConstraints |
| 196 | * @return bool |
| 197 | */ |
| 198 | public function matchesSettings( $settingsConstraints ) { |
| 199 | if ( isset( $settingsConstraints[self::STREAM_SETTING] ) ) { |
| 200 | if ( !$this->matches( $settingsConstraints[self::STREAM_SETTING] ) ) { |
| 201 | return false; |
| 202 | } |
| 203 | // stream matching is special and can't use ::isPartialMatch(). Since we've already |
| 204 | // matched stream, remove it from the constraints now. |
| 205 | unset( $settingsConstraints[self::STREAM_SETTING] ); |
| 206 | } |
| 207 | |
| 208 | return $this->isPartialMatch( $this->settings, $settingsConstraints ); |
| 209 | } |
| 210 | |
| 211 | /** |
| 212 | * Partially matches two arrays, LHS and RHS, with one another up to a maximum depth. LHS and |
| 213 | * RHS are considered a partial match if LHS contains all of RHS. |
| 214 | * |
| 215 | * To maintain backwards compatibility, all non-array values are cast to strings before they |
| 216 | * are compared. |
| 217 | * |
| 218 | * @param array $lhs |
| 219 | * @param array $rhs |
| 220 | * @param int $maxDepth |
| 221 | * @return bool |
| 222 | */ |
| 223 | private function isPartialMatch( $lhs, $rhs, $maxDepth = 10 ) { |
| 224 | if ( $maxDepth === 0 ) { |
| 225 | return true; |
| 226 | } |
| 227 | |
| 228 | foreach ( $rhs as $key => $expected ) { |
| 229 | if ( !isset( $lhs[$key] ) ) { |
| 230 | return false; |
| 231 | } |
| 232 | |
| 233 | $actual = $lhs[$key]; |
| 234 | |
| 235 | if ( is_array( $expected ) && is_array( $actual ) ) { |
| 236 | if ( $this->arrayIsList( $expected ) && $this->arrayIsList( $actual ) ) { |
| 237 | return array_intersect( $expected, $actual ) === $expected; |
| 238 | } |
| 239 | |
| 240 | if ( !$this->isPartialMatch( $actual, $expected, $maxDepth - 1 ) ) { |
| 241 | return false; |
| 242 | } |
| 243 | } elseif ( |
| 244 | |
| 245 | // Avoid casting $expected or $actual to a string if either one of them is an |
| 246 | // array. |
| 247 | is_array( $expected ) || |
| 248 | is_array( $actual ) || |
| 249 | |
| 250 | (string)$expected !== (string)$actual |
| 251 | ) { |
| 252 | return false; |
| 253 | } |
| 254 | } |
| 255 | |
| 256 | return true; |
| 257 | } |
| 258 | |
| 259 | /** |
| 260 | * Gets whether the array is a list, i.e. an integer-indexed array with indices starting at 0. |
| 261 | * |
| 262 | * As written, this method trades performance for elegance. This method should not be called on |
| 263 | * large arrays. |
| 264 | * |
| 265 | * TODO: Replace this with array_is_list when MediaWiki supports PHP >= 8.1 |
| 266 | * |
| 267 | * @param array $array |
| 268 | * @return bool |
| 269 | */ |
| 270 | private function arrayIsList( $array ) { |
| 271 | $array = array_keys( $array ); |
| 272 | |
| 273 | return $array === array_keys( $array ); |
| 274 | } |
| 275 | |
| 276 | /** |
| 277 | * Returns true if $string is a valid regex. |
| 278 | * It must start with '/' and preg_match must not return false. |
| 279 | * |
| 280 | * @param string $string |
| 281 | * @return bool |
| 282 | */ |
| 283 | private static function isValidRegex( $string ) { |
| 284 | // FIXME: This is very ugly, and not very safe. |
| 285 | // Temporarily disable errors/warnings when checking if valid regex. |
| 286 | $errorLevel = error_reporting( E_ERROR ); |
| 287 | // @phan-suppress-next-line PhanParamSuspiciousOrder false positive |
| 288 | $isValid = mb_substr( $string, 0, 1 ) === '/' && preg_match( $string, '' ) !== false; |
| 289 | error_reporting( $errorLevel ); |
| 290 | return $isValid; |
| 291 | } |
| 292 | |
| 293 | /** |
| 294 | * Validates that the stream config settings are valid. |
| 295 | * |
| 296 | * @param array $settings |
| 297 | * @throws \InvalidArgumentException if stream config settings are invalid |
| 298 | */ |
| 299 | private static function validate( array $settings ) { |
| 300 | Assert::parameter( |
| 301 | isset( $settings[self::STREAM_SETTING] ), |
| 302 | self::STREAM_SETTING, |
| 303 | self::STREAM_SETTING . ' not set in stream config entry ' . |
| 304 | var_export( $settings, true ) |
| 305 | ); |
| 306 | $stream = $settings[self::STREAM_SETTING]; |
| 307 | |
| 308 | Assert::parameterType( 'string', $stream, self::STREAM_SETTING ); |
| 309 | // If stream looks like a regex, make sure it is valid. |
| 310 | // (Yes, isValidRegex also checks that string starts with '/', but here we want |
| 311 | // to fail if the stream is not a valid regex.) |
| 312 | if ( substr( $stream, 0, 1 ) === '/' ) { |
| 313 | Assert::parameter( |
| 314 | self::isValidRegex( $stream ), self::STREAM_SETTING, "Invalid regex '$stream'" |
| 315 | ); |
| 316 | } |
| 317 | |
| 318 | if ( isset( $settings[self::TOPIC_PREFIXES_SETTING] ) ) { |
| 319 | Assert::parameterType( |
| 320 | 'array', |
| 321 | $settings[self::TOPIC_PREFIXES_SETTING], |
| 322 | self::TOPIC_PREFIXES_SETTING |
| 323 | ); |
| 324 | } |
| 325 | |
| 326 | if ( isset( $settings[self::TOPICS_SETTING] ) ) { |
| 327 | Assert::parameterType( |
| 328 | 'array', |
| 329 | $settings[self::TOPICS_SETTING], |
| 330 | self::TOPICS_SETTING |
| 331 | ); |
| 332 | } |
| 333 | } |
| 334 | } |