* * For the full copyright and license information, please view the LICENSE * file that was distributed with this source code. */ namespace Monolog\Formatter; use Throwable; use Monolog\DateTimeImmutable; /** * Normalizes incoming records to remove objects/resources so it's easier to dump to various targets * * @author Jordi Boggiano */ class NormalizerFormatter implements FormatterInterface { const SIMPLE_DATE = "Y-m-d\TH:i:sP"; protected $dateFormat; /** * @param string $dateFormat The format of the timestamp: one supported by DateTime::format */ public function __construct(string $dateFormat = null) { $this->dateFormat = null === $dateFormat ? static::SIMPLE_DATE : $dateFormat; if (!function_exists('json_encode')) { throw new \RuntimeException('PHP\'s json extension is required to use Monolog\'s NormalizerFormatter'); } } /** * {@inheritdoc} */ public function format(array $record) { return $this->normalize($record); } /** * {@inheritdoc} */ public function formatBatch(array $records) { foreach ($records as $key => $record) { $records[$key] = $this->format($record); } return $records; } /** * @param mixed $data * @return int|bool|string|null|array */ protected function normalize($data, int $depth = 0) { if ($depth > 9) { return 'Over 9 levels deep, aborting normalization'; } if (null === $data || is_scalar($data)) { if (is_float($data)) { if (is_infinite($data)) { return ($data > 0 ? '' : '-') . 'INF'; } if (is_nan($data)) { return 'NaN'; } } return $data; } if (is_array($data)) { $normalized = []; $count = 1; foreach ($data as $key => $value) { if ($count++ >= 1000) { $normalized['...'] = 'Over 1000 items ('.count($data).' total), aborting normalization'; break; } $normalized[$key] = $this->normalize($value, $depth + 1); } return $normalized; } if ($data instanceof \DateTimeInterface) { return $this->formatDate($data); } if (is_object($data)) { if ($data instanceof Throwable) { return $this->normalizeException($data, $depth); } if ($data instanceof \JsonSerializable) { $value = $data->jsonSerialize(); } elseif (method_exists($data, '__toString')) { $value = $data->__toString(); } else { // the rest is normalized by json encoding and decoding it $encoded = $this->toJson($data, true); if ($encoded === false) { $value = 'JSON_ERROR'; } else { $value = json_decode($encoded, true); } } return [get_class($data) => $value]; } if (is_resource($data)) { return sprintf('[resource(%s)]', get_resource_type($data)); } return '[unknown('.gettype($data).')]'; } /** * @return array */ protected function normalizeException(Throwable $e, int $depth = 0) { $data = [ 'class' => get_class($e), 'message' => $e->getMessage(), 'code' => $e->getCode(), 'file' => $e->getFile().':'.$e->getLine(), ]; if ($e instanceof \SoapFault) { if (isset($e->faultcode)) { $data['faultcode'] = $e->faultcode; } if (isset($e->faultactor)) { $data['faultactor'] = $e->faultactor; } if (isset($e->detail)) { $data['detail'] = $e->detail; } } $trace = $e->getTrace(); foreach ($trace as $frame) { if (isset($frame['file'])) { $data['trace'][] = $frame['file'].':'.$frame['line']; } elseif (isset($frame['function']) && $frame['function'] === '{closure}') { // We should again normalize the frames, because it might contain invalid items $data['trace'][] = $frame['function']; } else { // We should again normalize the frames, because it might contain invalid items $data['trace'][] = $this->toJson($this->normalize($frame, $depth + 1), true); } } if ($previous = $e->getPrevious()) { $data['previous'] = $this->normalizeException($previous, $depth + 1); } return $data; } /** * Return the JSON representation of a value * * @param mixed $data * @throws \RuntimeException if encoding fails and errors are not ignored * @return string|bool */ protected function toJson($data, bool $ignoreErrors = false) { // suppress json_encode errors since it's twitchy with some inputs if ($ignoreErrors) { return @$this->jsonEncode($data); } $json = $this->jsonEncode($data); if ($json === false) { $json = $this->handleJsonError(json_last_error(), $data); } return $json; } /** * @param mixed $data * @return string|bool JSON encoded data or false on failure */ private function jsonEncode($data) { return json_encode($data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRESERVE_ZERO_FRACTION); } /** * Handle a json_encode failure. * * If the failure is due to invalid string encoding, try to clean the * input and encode again. If the second encoding attempt fails, the * initial error is not encoding related or the input can't be cleaned then * raise a descriptive exception. * * @param int $code return code of json_last_error function * @param mixed $data data that was meant to be encoded * @throws \RuntimeException if failure can't be corrected * @return string JSON encoded data after error correction */ private function handleJsonError(int $code, $data): string { if ($code !== JSON_ERROR_UTF8) { $this->throwEncodeError($code, $data); } if (is_string($data)) { $this->detectAndCleanUtf8($data); } elseif (is_array($data)) { array_walk_recursive($data, [$this, 'detectAndCleanUtf8']); } else { $this->throwEncodeError($code, $data); } $json = $this->jsonEncode($data); if ($json === false) { $this->throwEncodeError(json_last_error(), $data); } return $json; } /** * Throws an exception according to a given code with a customized message * * @param int $code return code of json_last_error function * @param mixed $data data that was meant to be encoded * @throws \RuntimeException */ private function throwEncodeError(int $code, $data) { switch ($code) { case JSON_ERROR_DEPTH: $msg = 'Maximum stack depth exceeded'; break; case JSON_ERROR_STATE_MISMATCH: $msg = 'Underflow or the modes mismatch'; break; case JSON_ERROR_CTRL_CHAR: $msg = 'Unexpected control character found'; break; case JSON_ERROR_UTF8: $msg = 'Malformed UTF-8 characters, possibly incorrectly encoded'; break; default: $msg = 'Unknown error'; } throw new \RuntimeException('JSON encoding failed: '.$msg.'. Encoding: '.var_export($data, true)); } /** * Detect invalid UTF-8 string characters and convert to valid UTF-8. * * Valid UTF-8 input will be left unmodified, but strings containing * invalid UTF-8 codepoints will be reencoded as UTF-8 with an assumed * original encoding of ISO-8859-15. This conversion may result in * incorrect output if the actual encoding was not ISO-8859-15, but it * will be clean UTF-8 output and will not rely on expensive and fragile * detection algorithms. * * Function converts the input in place in the passed variable so that it * can be used as a callback for array_walk_recursive. * * @param mixed &$data Input to check and convert if needed * @private */ public function detectAndCleanUtf8(&$data) { if (is_string($data) && !preg_match('//u', $data)) { $data = preg_replace_callback( '/[\x80-\xFF]+/', function ($m) { return utf8_encode($m[0]); }, $data ); $data = str_replace( ['¤', '¦', '¨', '´', '¸', '¼', '½', '¾'], ['€', 'Š', 'š', 'Ž', 'ž', 'Œ', 'œ', 'Ÿ'], $data ); } } protected function formatDate(\DateTimeInterface $date) { // in case the date format isn't custom then we defer to the custom DateTimeImmutable // formatting logic, which will pick the right format based on whether useMicroseconds is on if ($this->dateFormat === self::SIMPLE_DATE && $date instanceof DateTimeImmutable) { return (string) $date; } return $date->format($this->dateFormat); } }