diff options
author | Stephan Bergmann <sbergman@redhat.com> | 2020-02-11 10:31:18 +0100 |
---|---|---|
committer | Eike Rathke <erack@redhat.com> | 2020-02-11 18:37:30 +0100 |
commit | e8f26dc13b65b1a05d948d9c95110c86315e8f20 (patch) | |
tree | 1618195fdc09b559a36c7356bb602e4db07ee1e1 /solenv/clang-format | |
parent | f62e4fb51d72641896b6b8f38fbffb35e6c1b695 (diff) |
tdf#130555: Prevent negative aSelection.Min()
...which would violate the the preconditions of the later call to maText.remove.
When BreakIteratorImpl::previousWord
(i18npool/source/breakiterator/breakiteratorImpl.cxx) is called to e.g. move
back over a single space at the start of the text, at least for an en-US locale
it will fall through to the call to BreakIterator_Unicode::previousWord
(i18npool/source/breakiterator/breakiterator_unicode.cxx) at the bottom of the
function. That in turn calls icu::BreakIterator::preceding, which is documented
(workdir/UnpackedTarball/icu/source/common/unicode/brkiter.h) to return
icu::BreakIterator::DONE (i.e., -1, see
workdir/UnpackedTarball/icu/source/common/unicode/brkiter.h) in that case, which
causes BreakIterator_Unicode::previousWord to return a Boundary with startPos ==
endPos == -1.
The documentation of UNO method css.i18n.XBreakIterator::previousWord
(offapi/com/sun/star/i18n/XBreakIterator.idl) is silent about the expected
return value in such a case. But lets assume that returning such a [-1..-1]
Boundary is as intended, and locally address this case in Edit::ImplDelete,
making aSelection start at the start of the text.
Change-Id: I40e17ba602088e72aa6962cb41dd8e1cdf6e2561
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/88431
Tested-by: Jenkins
Reviewed-by: Eike Rathke <erack@redhat.com>
Diffstat (limited to 'solenv/clang-format')
0 files changed, 0 insertions, 0 deletions