| Age | Commit message (Collapse) | Author |
|---|
|
Change-Id: I744788bde9778f85ccd9d7667e19d16842900a29
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/110248
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
...that take a pointer and a length, and where it should be OK that the pointer
is null if the length is zero. Those rtl_uString_* functions are targets of
OUString member functions that take std::[u16]string_view arguments, and
19926ed35ebb623fc896942b1f232b83edf1fc1e "loplugin:stringview: Flag empty string
converted to string view" (which changed some call sites to pass in default-
constructed std::[u16]string_view, for which data() returns null) revealed that
those rtl_uString_* functions were not prepared for such input.
(The guardings of memcpy are necessary because memcpy still requires its pointer
arguments to be non-null, even if the corresponding length is zero.)
The new sal/qa/rtl/strings/test_strings_defaultstringview.cxx systematically
tests all O[U]String[Buffer] member functions taking std::[u16]string_view
arguments. It revealed one further issue in
IMPL_RTL_STRNAME(compare_WithLength), where UBSan reported a
nullptr-with-nonzero-offset
> sal/rtl/strtmpl.cxx:149:9: runtime error: applying non-zero offset 18446744073709551614 to null pointer
Also, rtl_uString_newReplaceFirstUtf16LUtf16L was found to lack a check for its
`from` argument to be non-null.
Change-Id: I6a7a712570f7d1e8d52097208c8a43a5a24797af
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/106295
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
+ remove sal_Char check on compilerplugins
Change-Id: I0f7da14e620f0c3d031d038aa8345ba4080fb3e9
Change-Id: Ia6dba4f27b47bc9e0c89159182ad80a5aee17166
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/102499
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
Change-Id: I2934b96c8b2b0a9049d7f238b9c67c8ecf56fa38
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/102035
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
Tested-by: Jenkins
|
|
...instead of having individual overloads for OUString, OUStringLiteral, and
literal char16_t const[N]. (The variants taking OUString are still needed for
!LIBO_INTERNAL_ONLY, though. The variants taking ASCII-only literal char
const[N] are also left in place.)
This nicely reduces the number of needed overloads. std::u16string_view allows
to pass as arguments:
* OUString
* OUStringLiteral
* OUStringChar (with the necessary conversion added now)
* OUStringNumber
* u"..." char16_t string literals
* u"..."sv std::u16string_view literals
* std::u16string, plain char16_t*, and more
A notable exceptions is OUStringConcat, which now needs to be wrapped in
OUString(...), see the handful of places that needed to be adapted.
One caveat is the treatment of embedded NUL characters, as
std::u16string_view(u"x\0y")
constructs a view of size 1, while only
u"x\0y"sv
constructs a view of size 3 (which matches the old behavior of overloads for
literal char16_t const[N] via the ConstCharArrayDetector<>::TypeUtf16
machinery). See the new checkEmbeddedNul in
sal/qa/rtl/strings/test_oustring_stringliterals.cxx.
The functions that have been changed are generally those that:
* already take a string of determined length, so that using std::u16string_view,
which is always constructed with a determined length, is no pessimization
(e.g., there are operator == overloads taking plain pointers, which do not
need to determine the string length upfront);
* could not benefit from the fact that the passed-in argument is an OUString
(e.g., the corresponding operator = overload can reuse the passed-in
OUString's rtl_uString pData member);
* do not run into overload resolution ambiguity issues, like the comparison
operators would do.
One inconsistency that showed up is that while the original
replaceAll(OUString const &, OUString const &, sal_Int32 fromIndex = 0)
overload takes an optional third fromIndex argument, the existing replaceAll
overloads taking OUStringLiteral and literal char16_t const[N] arguments did
not. Fixing that required a new (LIBO_INTERNAL_ONLY)
rtl_uString_newReplaceAllFromIndexUtf16LUtf16L (with test code in
sal/qa/rtl/strings/test_strings_replace.cxx).
Another issue was posed by test code in
sal/qa/rtl/strings/test_oustring_stringliterals.cxx that used the
RTL_STRING_UNITTEST-only OUString(Except*CharArrayDetector) ctors to verify that
certain function calls should not compile (and would compile under
RTL_STRING_UNITTEST by taking those Except*CharArrayDetector converted to
OUString as arguments). Those problematic "should fail to compile" tests have
been converted into a new CompilerTest_sal_rtl_oustring.
Change-Id: Id72e8c4cc338258cadad00ddc6ea5b9da2e1f780
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/102020
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
by 10%, by avoiding an OUString construction in a hot path
through
XMLTextColumnContext_Impl::XMLTextColumnContext_Impl
-> sax::Convert::convertNumber
Also changed XMLTextAnimationStepPropertyHdl::importXML
to take advantage of the modified convertNumber passing
convention.
Change-Id: I4e5503dbb094c88a09af8b6dc8c22b6c53f9eb75
Reviewed-on: https://gerrit.libreoffice.org/81726
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
When I did the fast string concatenation, I didn't add any support
for number(), which simply returned a O(U)String, and so it did
the extra allocation/deallocation, although that could be avoided.
In order to support this, number() now returns a special temporary
return type, similarly to O(U)StringConcat, which allows delaying
the concatenation the same way.
Also similarly, the change of the return type in some cases requires
explicit cast to the actual string type. Usage of OString::getStr()
is so extensive in the codebase that I actually added it to the helper
class, after that it's only relatively few cases.
Change-Id: Iba6e158010e1e458089698c426803052b6f46031
Reviewed-on: https://gerrit.libreoffice.org/78873
Tested-by: Jenkins
Reviewed-by: Luboš Luňák <l.lunak@collabora.com>
|
|
Change-Id: I88f788b6d8f9363e2f7b5447f6f0b0203f0834f0
Reviewed-on: https://gerrit.libreoffice.org/77758
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
|
|
rtl_[u]String_alloc now requires that the length be >= 0.
Since this function is only @since Libreoffice 4.1, it is unlikely
to be widely used externally.
Removed some unit tests that were testing invalid or out of
range paramers, which are already not allowed according to the
documented contract of those functions.
The change in writerfilter is because the new asserts triggered
when running testFdo74745
The change in SwTextNode::EraseText is because testFdo60842
triggered the assert in replaceAt.
The change in SwFieldSlot::SwFieldSlot is because
testMoveRange::Import_Export_Import triggered the assert in
replaceAt.
The changes in SwFieldSlot::SwFieldSlot and TabControl::ImplGetItemSize
are due to failures in the uitests.
Change-Id: Ib317261067649b0de96df12873ce31360cd24681
Reviewed-on: https://gerrit.libreoffice.org/58390
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
...so consistently use the latter instead of the former
Change-Id: I144d5e7c472632f93b2258461510346bc85892d9
Reviewed-on: https://gerrit.libreoffice.org/48135
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
Change-Id: I0a8579406a70d2f63d96969c766e2a43d830ed6f
Reviewed-on: https://gerrit.libreoffice.org/47114
Reviewed-by: Tamás Zolnai <tamas.zolnai@collabora.com>
Tested-by: Tamás Zolnai <tamas.zolnai@collabora.com>
|
|
Change these back to consistently use the "..." form to include other UNO API
include files, for the benefit of external users of this API.
Change-Id: I9c9188e895eb3495e20a71ad44abfa2f6061fa94
|
|
OUString::trim() uses rtl_uString_newTrim, which relies upon
rtl_ImplIsWhitespace. The latter treats as whitespaces not only
characters with values less than or equal to 32, but also Unicode
General Punctuation area Space and some Control characters. Thus,
using OUString::trim() is incorrect when the goal is to trim XML
whitespace, which is defined as one of 0x09, 0x0A, 0x0D, 0x20.
The comments for OUString::trim() and rtl_uString_newTrim are
corrected to describe which characters are considered whitespace.
A unit test included.
Change-Id: I45a132be923a52dcd5a4c35aeecb53d423b49fec
Reviewed-on: https://gerrit.libreoffice.org/41444
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>
Tested-by: Mike Kaganski <mike.kaganski@collabora.com>
|
|
The long-term benefit will be support of C++11 char16_t string literals (for
cases of string literals with non-ASCII content) once we drop any compilers that
don't support those yet. The short-term benefit is support for an improved
OUStringLiteral1 that accepts any sal_Unicode value, not just ASCII ones (see
next commit).
Change-Id: I3f8f6697d7eb62b5176b7e812b5a5113c53b83a4
Reviewed-on: https://gerrit.libreoffice.org/28445
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
Find a few million mis-predicted branches (according to callgrind)
and annotate them. Mark string acquire/release as hot, and a number of
deprecated methods as cold.
Change-Id: I678b3981794221c97f9ebb70fd0161c0fda5dceb
|
|
Change-Id: Ifabb905819e32cf4db603f3f7bb18e6cc8fdfdcf
|
|
Change-Id: Ib34196185f90204a71598f2c659c3fddce7a0e4d
|
|
Change-Id: I5f525d91ce24d1d2653a6855f1c4fffc039ae398
|
|
...by adding some further SAL_DLLPUBLIC_RTTI type annotations (cf.
b4f6b26b5a1a78fecfa95ec2eb7ac8b80495d8aa "SAL_DLLPUBLIC_RTTI for proper RTTI
visibility for LLVM") and by making sure relevant function types do not use
incomplete types in their parameter and return types (which would make the RTTI
hidden).
Change-Id: Id7aadcbc0704b9759968ae36266fc9ce11a2e340
|
|
8f8bc0dcf3bc253ae49159d52db049767f476ced "Move string hash function into String
class" had introduced a new getHash64 that, besides returning sal_uInt64 instead
of just sal_Int32, didn't do sampling of only a handful of characters, but
always computed the hash over all characters (as the usage in SfxItemSet and
SdPage appears to require for either performance or approximated correctness).
However, it would be advantageous to keep the stable URE interface as small as
possible. Now, O(1) sampling was apparently considered state of the art when
the rtl string classes were first created, closely copying java.lang.String,
which at that time demanded sampling for hashCode(), too---but never sampling
more than 15 characters, with the obvious (in hindsight, at least) performance
catastrophes, so they changed it to O(n) somewhere along the way.
Based on that, this commit changes the existing hash functions to not do
sampling any more, and removes the newly introduced -64 variants again. (Where
the extended value range of sal_uInt64 compared to sal_Int32 was hopefully not
vital to the existing uses.)
The old implementation used sampling only for strings of length >= 256, so I did
a "make check" build with an instrumented hash function that flagged all uses
with inputs of length >= 256, and grepped workdir/{Cppunit,Junit,Python}Test for
hits. Of the 2849 hits encountered, 2845 where in the range from 256 to 295
characters, and only the remaining four where of 2472 characters. Those four
were from CppunitTest_sc_subsequent_filters_test, importing long text into a
cell, causing ScDocumentImport::setStringCell to call
svl::SharedStringPool::intern, which internally uses an unordered_set. These
results appear to justify the change.
Change-Id: I78fcc3b0f07389bdf36a21701b95a1ff0a0d970f
|
|
hashCode() seems to do sampling while creating the hash.
hashCode64() will not.
Change-Id: Id30f5a2a774cf5244dbc00da9649e95a532484be
|
|
Change-Id: Ic90a365a237aa23846f97131146a5aa2c46b5fd2
|
|
Change-Id: If1ee11da444a7f96f2d8668b277540da0bb4dbe9
|
|
...which has become necessary since bd60d41176da540b01d7583cfe00637431967f39
"Handle oveflow in O(U)String::toInt() functions" reduces values in the range
(SAL_MAX_INT32 .. SAL_MAX_UINT32] to zero, but some calls of toInt32(16) relied
on getting a correct (unsigned) value for the whole input range ["0" ..
"FFFFFFFF"] (see libreoffice-4-1 commit 9bf6c83367cedb7be81bf67f30d2147d26c7a8c3
"Revert overflow checks in O[U]String::toInt{32,64} again").
Audited all uses of toInt32/64 with non-decimal radix. (There is still a TODO
comment in oox/source/helper/attributelist.cxx, and
stoc/source/typeconv/convert.cxx will still need some love and test code.)
Change-Id: Iadaca1c0e41dab553687d0ce41c20c10cd657a95
|
|
Change-Id: Ib48a12e902f2311c295b2007f08f44dee28f431d
Reviewed-on: https://gerrit.libreoffice.org/3499
Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>
|
