commit 09f7620fec287d90ba80df3796ac75c98d7b8f35
author Thomas Gales <tgales@google.com> Wed Jul 17 00:20:33 2024 +0000
committer CQ Bot <fuchsia-internal-scoped@luci-project-accounts.iam.gserviceaccount.com> Wed Jul 17 00:20:33 2024 +0000
tree c6442ad1857e5a7a25dd7f11a40de0b061ee246c
parent 3386f65d6f31bca8f39de9034902ac12ee889b4c

[forensics] Fix redaction application with empty characters

Previously, redaction would find a match using regex and record the
match as plain text. After all matches were found, the redactions would
be applied by finding the plain text within the original content and
making the replacement.

This scheme would result in incorrect behavior if a plain text match was
a subset of another, later plain text match such as "abc" and "abcd"
resulting in "<REDACTED-HEX: 1>" and "<REDACTED-HEX: 1>d", with the
trailing "d" not redacted by mistake.

This CL changes from the plain text find-replace scheme to one which
records the position within the original content and makes the
replacements at those positions.

Note that a 2-step approach is necessary because we want to insert an ID
for the redacted hex, and want this ID to match for multiple instances
of the same original text being redacted. So a 1-step re2::GlobalReplace
would not work because all text matching a pattern would have the same
ID.

Fixed: 350095296
Tested: `fx test //src/developer/forensics`
Change-Id: I5bbc94eb8c2f600c6e6098c03260801fbe7b75a7
Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/1082700
Reviewed-by: Alex Pankhurst <pankhurst@google.com>
Commit-Queue: Thomas Gales <tgales@google.com>