pkg/telemetry/lib/src/pii_regexp.dart

// Copyright (c) 2023, the Dart project authors.  Please see the AUTHORS file
// for details. All rights reserved. Use of this source code is governed by a
// BSD-style license that can be found in the LICENSE file.

/// This library gives some backstops to make sure that arbitrary string
/// exceptions generated by something that might be sent to the crash reporting
/// framework do not contain PII.   The regular expressions here are
/// intentionally broad to make sure they get everything, possibly at the
/// expense of readability for the error message.  Use care when reducing
/// their scope.
library;

/// Contain an ordered list of regular expressions to be applied in sequence
/// with [applyTo] to a string, with any matches being substituted with
/// [substitution].  While a single regular expression could be
/// used in place of this class, this is intending to make the regular
/// expressions being applied more comprehensible to non-expert regular
/// expression authors and to make it very simple to add or remove individual
/// cases.
class _RegExpList {
  /// This list should be usually be ordered from more specific to more general
  /// cases in the event the regular expressions overlap.
  final List<RegExp> _regExps;
  final String substitution;

  _RegExpList(List<String> uncompiledRegexps, this.substitution)
      : _regExps = uncompiledRegexps.map((s) => RegExp(s)).toList();

  String applyTo(String input) => _regExps.fold(
      input, (previousInput, r) => previousInput.replaceAll(r, substitution));
}

/// An ordered list of regular expressions to be substituted out and replaced
/// with a generic `<path>` string. To be applied in sequence.
/// The regular expressions here intentionally eat all characters after a
/// path is identified, with the hope of catching all potential complex
/// pathnames and assuming that most error messages will have any paths last.
/// For example: `throw Exception('error in: $fileName)`
final _piiPathRegexps = _RegExpList([
  // C:\Some\Things\Including Spaces\Too
  r'\w:\\.*',
  // \\Windows\Shares\Look\Like This
  r'\\\\\S+\\.*',
  // ..\StartsWithDots\Windows Path
  r'[.]+\\+\S*\\.*',
  // A\Relative\Windows Path
  r'\b\S+\\.*',
  // ../StartsWithDots/Unix path
  r'[.]+/\S*/.*',
  // /AUnix/path
  r'/(\S|[.])+/.*',
  // a/unix\ relative path/possibly-containing-strange-characters
  r'\b(\S|[.])+/.*',
], '<path>');

/// An ordered list of regular expressions to be substituted out and replaced
/// with a generic `<filename>` string.  To be applied after
/// all paths from [_piiPathRegexps] are applied.   This is somewhat less
/// all-encompassing than the path finder, above, just to try and grab things
/// that could be file names but without being quite so greedy or depending
/// on the presence of path separators.
final _piiFileRegexps = _RegExpList([
  // foo.dart
  // something.exe
  // code.js
  r'\b\S+[.]\w{1,4}\b',
], '<filename>');

/// Attempt to remove things that look like they could be PII from the
/// given [message].
/// Returns the filtered string.
/// Not suitable for pre-scrubbed strings that intentionally include things
/// that look like filenames.
String filterPiiFromErrorMessage(String message) => [
      _piiPathRegexps,
      _piiFileRegexps,
    ].fold(message, (previousMessage, r) => r.applyTo(previousMessage));