tree 849816603f20bf7c1ab2a56cdbd34629cb9a1591
parent c9c736d4d859762f9c783abf0b5b00327cba8b7c
author Tom Sepez <tsepez@chromium.org> 1652906129 +0000
committer Pdfium LUCI CQ <pdfium-scoped@luci-project-accounts.iam.gserviceaccount.com> 1652906129 +0000

Support Unicode names/values in CJS globals

Currently, these are manipulated using default ANSI, meaning the strings
are interpreted (on Windows at least) according to whatever code page
is in effect for the system, with non-conforming characters removed (or
replaced by placeholder '?' chars on Windows). These stored strings also
risk being corrupted if the system default code page changes.

These strings all come directly from JS as UTF-16, so we know precisely
what they mean without requiring code page considerations. So convert
everything to UTF-8 and avoid platform differences.

This would be a breaking change if any platform actually implemented
persistence and had previously stored, say, non-ASCII key/values. This
seems unlikely.

-- Downcast name to string rather than invoking toString().
-- Avoid converting from UTF-8 only to reconvert back to UTF-8
   under the covers in fxv8.

Change-Id: I0e0298fa8429c59215e2c5a44560308cab72ccb3
Reviewed-on: https://pdfium-review.googlesource.com/c/pdfium/+/93772
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>