bump product version to 6.3.3.2.0+

Change-Id: Ifae41643ff3f281cf43630d53105d67080cd237d

update credits

Change-Id: Iad1765bef9e47d5a1588f43b170245c05d590f9a
(cherry picked from commit e7deccd775ba653701e2fdce2807554e5ce08146)
(cherry picked from commit 3b8047cd8fc6b433763bfefd1ec071bdff6946d0)

tdf#128188 sw_redlinehide: fix off-by-1 footnote numbers...

... in sections with "Collect at end of text" set when "Track
Changes->Show" is off; the pre-increment erroneously became a
post-increment.

(regression from fe1d3328997741b55202aca7b3dc566ca833a5f4)

Change-Id: Ie438418883bdf91a519d553c10e8d9952a94a52d
Reviewed-on: https://gerrit.libreoffice.org/81234
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 891b8036c746f7eab9ef3380c049eecf3860acac)
Reviewed-on: https://gerrit.libreoffice.org/81238
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
(cherry picked from commit 38db98d643ae301f4293f34617a8fce932cff873)
Reviewed-on: https://gerrit.libreoffice.org/81375
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
Reviewed-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>
Tested-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>

tdf#128076: throw exception only if it's not alive and it's not a description

According to a comment from https://gerrit.libreoffice.org/#/c/74533/

okay... then i guess there are different kinds of callers:
those who actually want to check if its a descriptor,
and those who want to check that it isn't "live"

so check if it's a description even if it's not alive

Change-Id: I1526fdbcb006506cc67d29cbe3f65138f90843dc
Reviewed-on: https://gerrit.libreoffice.org/80677
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 802d034c55876c9a933206176a7882ac1ae7566e)
Reviewed-on: https://gerrit.libreoffice.org/80764
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
(cherry picked from commit 52485e1f3464ab69b38d2e240c8c364d6b78ceb8)
Reviewed-on: https://gerrit.libreoffice.org/80877
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Reviewed-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>
Tested-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>

Resolves: tdf#128140 stabilize formula dialog width

Change-Id: Ifdb80adfd4b399035ed793f8636f31b90cf50094
Reviewed-on: https://gerrit.libreoffice.org/80801
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 198fc5e09784afe32cf681f65009e5ac43e67ba4)
Reviewed-on: https://gerrit.libreoffice.org/80875
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Reviewed-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>
Tested-by: Christian Lohmaier <lohmaier+LibreOffice@googlemail.com>

Don't use uninitialized memory when reading from the stream fails

Flathub arm builds (but not other arches) had often (but not always) failed when
processing sc/qa/unit/data/qpro/pass/ofz14090-1.wb2 in
CppunitTest_sc_filters_test (e.g.,
<https://flathub.org/builds/#/builders/1/builds/724>:

> Test name: ScFiltersTest::testCVEs
> equality assertion failed
> - Expected: 1
> - Actual  : 0
> - file:///run/build/libreoffice/sc/qa/unit/data/qpro/pass/ofz14090-1.wb2

)  Valgrind revealed that this was due to using unintialized memory when the
various maIn.Read... in QProToSc::Convert failed, starting with the use of
uninitialized nFmla[i] after

  maIn.ReadUChar( nFmla[i] );

At least make things deterministic by setting the relevant variables to zero.
(Another approach could be returning early with some ConvErr status.)

Change-Id: I4c06aa8da5f777170cdc7bbe3ca1d61b23d3f326
Reviewed-on: https://gerrit.libreoffice.org/80947
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
(cherry picked from commit 2704b9e3783aae9d8372f2e3ad3253a2cb49ae87)
Reviewed-on: https://gerrit.libreoffice.org/80956
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit e31f35c892b282258bd4ece30175e4bc48b6b823)
Reviewed-on: https://gerrit.libreoffice.org/81030
Reviewed-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

Update git submodules

* Update translations from branch 'libreoffice-6-3-3'
  - update translations for 6.3.3 rc2

    and force-fix errors using pocheck

    Change-Id: Ib0214e1ca97d2fbac297258dfe82795eb6e77afa
    (cherry picked from commit c7858b0e1cbda51f96034e7fd42009d7fda9e99e)

tdf#128232: fix crash when trying to delete unused User Field

mpFieldTypes is std::unique_ptr<SwFieldTypes>
(see https://opengrok.libreoffice.org/xref/core/sw/source/core/inc/DocumentFieldsManager.hxx?r=99a97d38#104)

See bt:
https://bugs.documentfoundation.org/attachment.cgi?id=155132

Change-Id: Iea8eddd4b1bdc9ce3c01988fc97004b7ec7a6e5f
(cherry picked from commit 09a3762fe0fc4c4815c842098094082bf1b90de6)
Reviewed-on: https://gerrit.libreoffice.org/81120
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit 9b1f906ea2aa946e9d6245baacd4937233bb58c6)
Reviewed-on: https://gerrit.libreoffice.org/81220
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>

aarch64 callVirtualFunction needs to be compiled w/o -fstack-clash-protection

At least when doing an aarch64 Flatpak build against org.freedesktop.Sdk//19.08,
which uses GCC 9.2.0 and passes in `CXXFLAGS=-O2 -g -pipe
-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions
-fstack-protector-strong -grecord-gcc-switches -fasynchronous-unwind-tables
-fstack-clash-protection`, callVirtualMethod (in
bridges/source/cpp_uno/gcc3_linux_aarch64/callvirtualfunction.cxx) would
decrement the stack pointer another 16 bytes after the

  stackargs = alloca(...);

and before the asm block, so in the called virtual function, arguments read from
the stack would read garbage (and CustomTarget_testtools/uno_test would fail
with SIGSEGV at

> #0  0x0000ffffb733eb48 in rtl::OUString::operator= (this=0xaaaaf9c1ac30, str=...) at /run/build/libreoffice/include/rtl/ustring.hxx:453
> #1  0x0000ffffb733a7bc in bridge_object::assign (rData=..., bBool=true, cChar=64 u'@', nByte=17 '\021', nShort=4660, nUShort=65244, nLong=305419896, nULong=4275878552, nHyper=0, nUHyper=187651311381888, fFloat=17.0814991, fDouble=3.1415926358999999, eEnum=-1698898192, rStr=..., xTest=..., rAny=...) at /run/build/libreoffice/testtools/source/bridgetest/cppobj.cxx:99
> #2  0x0000ffffb733a87c in bridge_object::assign (rData=..., bBool=true, cChar=64 u'@', nByte=17 '\021', nShort=4660, nUShort=65244, nLong=305419896, nULong=4275878552, nHyper=0, nUHyper=187651311381888, fFloat=17.0814991, fDouble=3.1415926358999999, eEnum=-1698898192, rStr=..., xTest=..., rAny=..., rSequence=...) at /run/build/libreoffice/testtools/source/bridgetest/cppobj.cxx:115
> #3  0x0000ffffb733ade4 in bridge_object::Test_Impl::setValues (this=0xaaaaf9c1abb0, bBool=1 '\001', cChar=64 u'@', nByte=17 '\021', nShort=4660, nUShort=65244, nLong=305419896, nULong=4275878552, nHyper=0, nUHyper=187651311381888, fFloat=17.0814991, fDouble=3.1415926358999999, eEnum=-1698898192, rStr=..., xTest=..., rAny=..., rSequence=..., rStruct=...) at /run/build/libreoffice/testtools/source/bridgetest/cppobj.cxx:548
> #4  0x0000ffffb740bff4 in callVirtualFunction (function=281473755360772, gpr=0xffffd1ab1f28, fpr=0xffffd1ab1f68, stack=0xffffd1ab1d40, sp=8, ret=0xffffd1ab22c0) at /run/build/libreoffice/bridges/source/cpp_uno/gcc3_linux_aarch64/callvirtualfunction.cxx:63
> #5  0x0000ffffb740ca70 in (anonymous namespace)::call (proxy=0xaaaaf9c291c0, slot=..., returnType=0xaaaaf9c00770, count=17, parameters=0xaaaaf9c3a210, returnValue=0xffffd1ab22c0, arguments=0xffffd1ab2230, exception=0xffffd1ab2370) at /run/build/libreoffice/bridges/source/cpp_uno/gcc3_linux_aarch64/uno2cpp.cxx:178
> #6  0x0000ffffb740d4c4 in bridges::cpp_uno::shared::unoInterfaceProxyDispatch (pUnoI=0xaaaaf9c291c0, pMemberDescr=0xaaaaf9c55950, pReturn=0xffffd1ab22c0, pArgs=0xffffd1ab2230, ppException=0xffffd1ab2370) at /run/build/libreoffice/bridges/source/cpp_uno/gcc3_linux_aarch64/uno2cpp.cxx:361
> #7  0x0000ffffb740720c in (anonymous namespace)::call (proxy=0xaaaaf9c549c0, description=..., returnType=0xaaaaf9c00770, count=17, parameters=0xaaaaf9c3a210, gpr=0xffffd1ab2510, fpr=0xffffd1ab2550, stack=0xffffd1ab2590, indirectRet=0xffffb7d24790) at /run/build/libreoffice/bridges/source/cpp_uno/gcc3_linux_aarch64/cpp2uno.cxx:120
> #8  0x0000ffffb74079a0 in (anonymous namespace)::vtableCall (functionIndex=40, vtableOffset=0, gpr=0xffffd1ab2510, fpr=0xffffd1ab2550, stack=0xffffd1ab2590, indirectRet=0xffffb7d24790) at /run/build/libreoffice/bridges/source/cpp_uno/gcc3_linux_aarch64/cpp2uno.cxx:291
> #9  0x0000ffffb7407b00 in (anonymous namespace)::vtableSlotCall (gpr0=187651311618536, gpr1=1, gpr2=64, gpr3=17, gpr4=4660, gpr5=65244, gpr6=305419896, gpr7=4275878552, fpr0=5.4321266044931319e-315, fpr1=3.1415926358999999, fpr2=0, fpr3=4.0039072046065485, fpr4=0, fpr5=4.003911019303815, fpr6=8.9589789687541617e+102, fpr7=-4.4588500238274385e-308) at /run/build/libreoffice/bridges/source/cpp_uno/gcc3_linux_aarch64/cpp2uno.cxx:348
> #10 0x0000ffffb739e60c in bridge_test::performTest (xContext=..., xLBT=..., noCurrentContext=false) at /run/build/libreoffice/testtools/source/bridgetest/bridgetest.cxx:378
> #11 0x0000ffffb73a3d58 in bridge_test::TestBridgeImpl::run (this=0xaaaaf9c18550, rArgs=...) at /run/build/libreoffice/testtools/source/bridgetest/bridgetest.cxx:1162
> #12 0x0000aaaad292a3ec in sal_main () at /run/build/libreoffice/cpputools/source/unoexe/unoexe.cxx:509
> #13 0x0000aaaad29297a0 in main (argc=8, argv=0xffffd1ab31b8) at /run/build/libreoffice/cpputools/source/unoexe/unoexe.cxx:349

.)

By experiment, I found the problematic thing to be -fstack-clash-protection,
which can apparently be cancelled with a subsequent -fno-stack-clash-protection
at least on that GCC 9.2.0.  (And -f[no-]stack-clash-protection appears to only
be available since GCC 8, and not at all for Clang, so check for it with
HAVE_GCC_STACK_CLASH_PROTECTION.)

Reviewed-on: https://gerrit.libreoffice.org/80701
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
(cherry picked from commit 5efb8624760e57dc1c2dd9abbeb13d5652f1fe07)
Conflicts:
bridges/Library_cpp_uno.mk
configure.ac

Change-Id: If667fdf704b1ba20a04593b38d2d1f079280df41
Reviewed-on: https://gerrit.libreoffice.org/80703
Tested-by: Jenkins
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit ddcba628482f5c9d3a0b4ac79db4107fc906f403)
Reviewed-on: https://gerrit.libreoffice.org/80749
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#128138 sw_redlinehide: fix infinite loop in FindAttrsImpl()

The code looked wrong previously, always assigning the end of the text
node regardless of direction, but it turns out the improvement caused an
infinite loop... let's just swap things after the loop, seems simpler
than adding another if at the assignments.

(regression from 4caef398af256be5f0c2a159129b528ee3702e5c)

Change-Id: I098f5265fa86d6a2511a80a02230899a7a303a88
Reviewed-on: https://gerrit.libreoffice.org/80774
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 4e0241697371c40f4348bdfb6585a872d4cbcbf6)
Reviewed-on: https://gerrit.libreoffice.org/80876
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Tested-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

tdf#40534 correctly match page with memory slab

LO has a page manager to match system memory backbuffers with
graphics memory on DX accelerated Windows. Internally this uses
an other rectangle implementation, the SurfaceRect, which had
some great comments like:

// a size of [0,0] therefore denotes a one-by-one rectangle.

In commit 230dbe2e43f3ee2cd285f9cdfe0d57e1ca08b8fe ("#144866# Add
one pixel border around textures, a bunch of drivers clobber those
with dirt), the allocation was increased by a pixel border, but
this doesn't work correctly, because now an allocation of the
page size wouldn't fit anymore into a page, because the pages size
is decreased before comparison. In the end the mixup suffered from
hard to handle off-by-one problems.

This patch fixes the bug, but eventually SurfaceRect should be
replaced by an extended basegfx::B2IBox. But since B2IBox uses two
ranges, instead of a point and a size, it would need a lot of
conversations to I2Point and I2Size objects with the current
Page::insert algorithm.

Change-Id: Ia725b4f8ed4fb270f2eb3734e492062bc7f13793
Reviewed-on: https://gerrit.libreoffice.org/80628
Tested-by: Jenkins
Reviewed-by: Jan-Marek Glogowski <glogow@fbihome.de>
(cherry picked from commit 194e7ce17ae7ca278c12d03bc25684b7437f9785)
Reviewed-on: https://gerrit.libreoffice.org/80669
Reviewed-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
(cherry picked from commit edcb363f5eeefcc2ce28a2ab7a57d61b744466cd)
Reviewed-on: https://gerrit.libreoffice.org/80878
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Michael Stahl <michael.stahl@cib.de>

bump product version to 6.3.3.1.0+

Change-Id: Ib7a975dc388db9f838afacfa7f96d0071fa1b6dd

Branch libreoffice-6-3-3

This is 'libreoffice-6-3-3' - the stable branch for the 6.3.3 release.
Only very safe changes, reviewed by three people are allowed.

If you want to commit more complicated fix for the next 6.3.x release,
please use the 'libreoffice-6-3' branch.

If you want to build something cool, unstable, and risky, use master.

Change-Id: If919c36ef6fa12e2f99af04e0b418092f03c034b

tdf#124601 sw FollowTextFlow: fix hori pos of objects outside the current cell

The problem is that the image in the bugdoc's footer is anchored in one
cell, but its position has a value that shifts it to the next column and
next row. The next column is the problem for the horizontal position.

So build on top of the previous vertical position fix, and make sure
that CalcRelPosX() doesn't limit the position inside the current cell
for an in-table, follow-text-flow, wrap-though image.

Once that's in place, make sure that we don't try to grow the cell due
to follow-text-flow, wrap-though objects in CalcPosition(), since
in a wrap-through vs follow-text-flow situation, the wrap-though should
have priority (should not affect size of cells).

Finally, now that cells don't grow in this case, the previously added
special-casing of footers in SwFlyFreeFrame::CheckClip() is no longer
necessary. Not growing the cells means we don't try to re-position the
object.

(cherry picked from commit 29d7ece94318d3f03d079dff33ec15ff74f8febf)

Change-Id: Ic55e4b5188704fa70314f91fe9a01987b6a56d7b
Reviewed-on: https://gerrit.libreoffice.org/80650
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

tdf#124601 sw FollowTextFlow: fix vert pos of objects outside the current cell

There were two problems here:

1) CalcHeightWithFlys() considered all follow-text-flow objects when
determining the cell size, while wrap-through objects should never
influence the layout of text (i.e. when they conflict, the second should
have priority).

2) Once the cell had correct height, the oscillaction described in the
SwFlyFreeFrame::CheckClip() comment started. Such a position update was
already disabled for headers, but footers have exactly the same problem.
In the case of the bugdoc, we jumped between 14618 and 14744 twips, till
finally layout loop control kicked in.

[ FollowTextFlow is meant to be same behavior as Word's layoutInCell
shape property, so this is expected to improve rendering of existing
documents. It's not likely that any user would opt in for FollowTextFlow
to have the old close-but-not-exactly-matching behavior. ]

(cherry picked from commit 489eef894e7034873ad262f9dfca554022db1b09)

Conflicts:
sw/qa/extras/layout/layout.cxx

Change-Id: I6b3b672fc82c6c67dbbdd35c349613fe4cda610d
Reviewed-on: https://gerrit.libreoffice.org/80595
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>

sw: WW8: fix crashtesting asserts on fdo45983-1.doc export to ODT

The problem is that a redline is created that contains the
CH_TXT_ATR_FIELDSTART but not the corresponding CH_TXT_ATR_FIELDEND:

(rr) p aRegion
$39 = SwPaM = {
  point = SwPosition (node 443, offset 185),
  mark = SwPosition (node 442, offset 0)
}
(rr) p aRegion.GetText()
$40 = "\aDie Studierendenvertiefen ihre Fertigkeit bei einfachen Handverletzungen, einen Befund zu stellen, Therapieziele zu formulieren, einen ergotherapeutischen Behandlungsplan zu erstellen und durchzuführen"

This is because commit d195a3e5f4ec5c616ae83f99d48f5d4eefe5f22e only
adapted one place in SwWW8ImplReader::End_Field() where a fieldmark is
inserted, but there's a second one ...

Change-Id: Ib931c6b0d55f42e4f6bc170216f495f1f6e931e1
Reviewed-on: https://gerrit.libreoffice.org/80622
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 954c158acb87eb0385c51fdfff38fff84f154051)
Reviewed-on: https://gerrit.libreoffice.org/80642
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

sw: WW8: do not create fieldmark with start in frame and end in body

Actually the start and end must have the same start node; presumably
CheckNodesRange(..., true) does that.

Change-Id: I04b34b593a4383b565289254bdb756d95c2b7916
Reviewed-on: https://gerrit.libreoffice.org/80594
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 06767a5394f1dfba71c4f0a2a07daa5664bdbd01)
Reviewed-on: https://gerrit.libreoffice.org/80625
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

Pass LDFLAGS into external/postgresql

This will be needed when building against Flatpak org.freedesktop.Sdk//19.08,
which no longer bundles krb5, so we will have a bundled libgssapi_krb5.so
at /app/lib (instead of under /usr), but which requires LDFLAGS=-L/app/lib to
be found.

Change-Id: I8767740ab0cbb02b6ae682b7f0e21ec90ea5ddd7
Reviewed-on: https://gerrit.libreoffice.org/80620
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
(cherry picked from commit cfd06eb99b8f366bfe96e4a6d3112e4c6057098b)
Reviewed-on: https://gerrit.libreoffice.org/80633
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

update credits

Change-Id: Ib98367116f9727f6ec54372d4cd90b657258b0d7
(cherry picked from commit 70ae1da67310a596e5bc49f1053c7ff72c84f539)

Related: tdf#124601 DOC import: always allow floating tables in header/footer

So now the DOC import is in sync with the DOCX one, see commit
81ef96a2417c7843dfed0558c920ad3064e58921 (tdf#79639 DOCX import: don't
delay text frame conversion of in-header tables, 2015-06-01).

The bugdoc used to have 2 additional pages in Writer (compared to Word),
it has only 1 after this.

(cherry picked from commit 8165cdad2e128ec5fabb50d3330c41a8b0f021d8)

Change-Id: I24f0dc28599e608737dc19a0143e72832202a034
Reviewed-on: https://gerrit.libreoffice.org/80435
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>

tdf#107784: DOCX Import: Show citation's title in fields (2nd take)

Regression from 06f7d1a96eef5aa69d4872ff6d96eb5085296d09

The problem as I see it is that LibreOffice uses the identifier
plus optionals prefix and suffix to display the bibliographic
references.
On the other hand, MSO displays the title for references
starting with CITATION.

So do the same MSO does, when working with CITATION refs

We already add CITATION when we export to ooxml.
See case SwFieldIds::TableOfAuthorities in
AttributeOutputBase::TextField

Change-Id: I144f27f711926658c8b6f89e69f0ddeeb7e7890c
Reviewed-on: https://gerrit.libreoffice.org/79477
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
(cherry picked from commit c02b5cf6472ccb45a916369db493a40ed3e21d60)
Reviewed-on: https://gerrit.libreoffice.org/79545
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>

tdf#127676: avoid flickering in line width popup

Change-Id: Ie69a5b4d33d12316e6fd7e443b92851cffc95b8c
Reviewed-on: https://gerrit.libreoffice.org/79667
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
(cherry picked from commit 25cc6dcef22e51c1bfa01e8fbed5046439bb1428)
Reviewed-on: https://gerrit.libreoffice.org/79748
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>

tdf#126321: avoid flickering in page size popup

In a Tabbed notebookbar, Layout - Page Size

Change-Id: I74ab54e96486de960d1ccc7bcb3c2679e25a6e9e
Reviewed-on: https://gerrit.libreoffice.org/79753
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
Tested-by: Miklos Vajna <vmiklos@collabora.com>

Update git submodules

* Update translations from branch 'libreoffice-6-3'
  - update translations for 6.3.3 rc1

    and force-fix errors using pocheck

    Change-Id: Iddec8e75a6e5f00f381c3facbf359991502e816a

tdf#127825 DOCX import: fix handling for tbrl, auto-height and rel size shapes

Regression from commit ff17478e069cc82681df62514876c06365dd5cd6 (sw btlr
writing mode: implement DOCX shape import for tbrl, 2019-04-25), there
were two problems here:

1) Relative size currently only works properly for the lrtb direction,
so disable that during import till sw core is improved.

2) When SwFlyFrame::Format() auto-grows a text frame which is the
textbox of a shape, it needs to notify the shape about the physical size
of the frame, not the logical one. So going via the SwRectFnSet
abstraction is not correct in this case.

(cherry picked from commit 9c81d0a268cca4ff36eff94c0842361b9c0287ef)

Conflicts:
sw/qa/extras/ooxmlimport/ooxmlimport2.cxx

Change-Id: Ie185c7415d90594434eac8f459630d6a3212328a
Reviewed-on: https://gerrit.libreoffice.org/80427
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

remove still pending timeout in dtor

Change-Id: I8ec8c62a9cd5349fa6e58c2d0515bd57edede97a
Reviewed-on: https://gerrit.libreoffice.org/80524
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#128013: fix Crash when linking an odp file

See bt here:
https://bugs.documentfoundation.org/attachment.cgi?id=154847

Regression introduced by:
https://cgit.freedesktop.org/libreoffice/core/commit/?id=3c86ffd8ded628e6f2b4187948a1b1056f6a0f56

Change-Id: Id9355958b0c4a56215ff98f0e5be13a3074ce45f
Reviewed-on: https://gerrit.libreoffice.org/80500
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
(cherry picked from commit 1dc4c8266d45eb2f5c3de303eaa9233e3b52f058)
Reviewed-on: https://gerrit.libreoffice.org/80506
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#124391 Fix doubled menu in global menu

Regression from 3bedbfd9b3cee5a8d2da6ee0486859bd725979ba

Change-Id: I56473a2b0620f90f23b1bcde3ae1290801021038
Reviewed-on: https://gerrit.libreoffice.org/80464
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

tdf#127529 vertical text not drawn in slideshow canvas

because the canvas text drawing impl falls back to using an OutputDevice view
of the canvas to use the vcl text drawing apis to achieve vertical text

To get an OutputDevice view of the canvas there is a specific VirtualDevice
ctor available to create a VirtualDevice that, unlike the normal case, doesn't
have its own specific backing buffer, but instead draws to the underlying target
provided via the SystemGraphicsData arg

The svp/gtk impl missed that understanding and provided an ordinary
VirtualDevice with its own backing buffer, not a VirtualDevice that would draw
to the expected target surface of the canvas. So the vertical text was drawn to
a different surface than the intended one, and was just discarded.

The cairo use in the canvas long precedes the use of cairo in vcl itself.

Seeing as text is now rendered with cairo in all cases where the canvas uses
cairo its probably now pointless for canvas to have its own text rendering
path.

Change-Id: Ie3b0a43ca2b746cbfe25e2d0415315b3d5403cd2
Reviewed-on: https://gerrit.libreoffice.org/80192
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

openssl 1.0.2t

Change-Id: Ideb980a07632e75e2bc7fffa851b8419e727c13b
Reviewed-on: https://gerrit.libreoffice.org/80341
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

autocorrect dialog width grows on change of new to replace

the sizegroup isn't working as I'd expect, so just make the replace button
visible, get the preferred width of the buttonbox, lock that down then hide the
place button and the container will stay at the max width that will be
wanted

Change-Id: Idfe632aa5af404abb92fc8f7267f0dda5cdbe295
Reviewed-on: https://gerrit.libreoffice.org/79863
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Related: tdf#127645 update dialog has a slave extension dialog

when the update dialog is shown, the extension dialog was created but not
executed, so there isn't a dialog execution context to end, just the unexecuted
dialog to "close"

Change-Id: If4b522ca5f880d3ece8403b5350c846bd14a2992
Reviewed-on: https://gerrit.libreoffice.org/79283
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

missing treeview model

Change-Id: Id1a405bb8f5c3d71b48c54e12b1e8ba23a27d842
Reviewed-on: https://gerrit.libreoffice.org/79405
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Resolves: tdf#127645 ignore floating windows like we do tooltips

Change-Id: I727e10a1e884841554437e376bde2ff483d278a4
Reviewed-on: https://gerrit.libreoffice.org/79284
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

should use DialogController here

Change-Id: Ie0ad42e98d574d15172c6ed3a58714c3f87593cd
Reviewed-on: https://gerrit.libreoffice.org/79356
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

set wait on the dialog, not the page parent

Change-Id: If782e774fee91a427d4bda575b2f404fd636f534
Reviewed-on: https://gerrit.libreoffice.org/79320
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

tdf#83618 Make line spacing values agree

Change-Id: I372a673bbad442d28314498248b12ce7f86fce09
Reviewed-on: https://gerrit.libreoffice.org/73737
Tested-by: Jenkins
Reviewed-by: Jim Raykowski <raykowj@gmail.com>
(cherry picked from commit 849b837d1a3b185a8dd893a8f6eaed53605bcab1)
Reviewed-on: https://gerrit.libreoffice.org/80314
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Related: tdf#124601 DOC import: improve fLayoutInCell handling

There were 3 problems here:

First, SwWW8ImplReader::IsObjectLayoutInTableCell() should not use
m_nProduct to determine the Word version. It depends on an undocumented
field of the [MS-DOC] format and the bugdoc shows how it interprets a
Word 2007-produced document as a Word 97 one. Instead, parse the cswNew
field of the file header, which is a more or less documented way to find
out if this file was produced by >=2000 or 97.

See e.g.
<https://docs.microsoft.com/en-us/openspecs/office_file_formats/ms-doc/841b5f72-487e-4fe7-8657-ec90d5af8750>
where the Dop structure maps cswNew values to Word versions.

Second, parse the fLayoutInCell correctly: it's part of a bitfield, with
two variables: a bool and an other one which decides if the bool should
be read at all. The bugdoc's case was evaluated as false, so do the
proper parsing in that case and leave the existing logic as-is for now.

Third, there doesn't seem to be a reason to exclude the wrap-through
case for the fLayoutInCell -> follows-text-flow mapping. The bugdoc
shows that Word interprects fLayoutInCell the same way for wrap-though
objects, too.

(cherry picked from commit d630f69d90f15bc652a62648b05ea515de78d16a)

Change-Id: Iaddd5e522e0380b731899f32a17c14ce4442ac35
Reviewed-on: https://gerrit.libreoffice.org/80346
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>

tdf#127958 crash adding 2 or more files into writer master document

clean up the memory management here, the SfxMedium wants to own the
passed-in SfxItemSet

Change-Id: I1bafa520ca7b833c4d2df7593649c3720c46e2ce
Reviewed-on: https://gerrit.libreoffice.org/80332
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit 8f6f9e4217d03d874fb1ce391730a1c893ab6844)
Reviewed-on: https://gerrit.libreoffice.org/80344

lok: comments: fix hidden text cursor and sudden document scroll

On Android, SwAnnotationWin::Rescale leads to invoke
ImpEditEngine::UpdateViews which hides the text cursor. Moreover it
causes sudden document scroll when modifying a commented text.
Not clear the root cause, anyway skipping this method fixes the
problem, and there should be no side effect, since the client has
disabled annotations rendering.

Change-Id: I572a9c6b3fe39473a596209413945d777bd79506
Reviewed-on: https://gerrit.libreoffice.org/80246
Tested-by: Jenkins
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>

related tdf#99602 writerfilter: deferredCharProps unhandled in Styles

...so don't add deferred properties during style import.
Otherwise, the first paragraph picks up those attributes and
applies them.

Regression from 2012 - only affects subscript/superscript.

Truly fixing this for Character Styles will be tricky,
because it depends on the final fontsize - regardless of
where it is set. So at the style level, the deferred
property needs to be percentized based on SOMETHING, and then
at appendText time the fontsize needs to be determined.
If escapement is inherited from a style, then the escapement needs
to be converted BACK into pointsize (based on style charsize),
and then the final percentage can be calculated and
directly applied. Yuck. I'm inclined to treat charstyle
escapement like pre-2012, which just assigned default
values and ignored the actual/given amount.

Change-Id: I8e3783533ec18740e1c7d0b36c58b6c7e8ce107f
Reviewed-on: https://gerrit.libreoffice.org/80178
Tested-by: Jenkins
Reviewed-by: Justin Luth <justin_luth@sil.org>
(cherry picked from commit 8eb0d8daa526d1c53363954d4d8730cad55f1910)
Reviewed-on: https://gerrit.libreoffice.org/80202
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>

Resolves: tdf#122011 fix ValidExternal() check

Tab==-1 may have been the condition 10 years ago, but at least since

    commit 296baa2fb6dd4150a7855114093a9703cdc18b09
    CommitDate: Tue Oct 5 18:14:37 2010 -0400

        Ported calc-extref-interpreter-rework-*.diff from ooo-build.

that's not the case anymore.

The check is only used in
ScRefTokenHelper::compileRangeRepresentation() that is only used
when creating chart data sources and sequences, thus may have gone
undetected.

Change-Id: I2eeb20ec740d286918b6ea29c017c32d03f970ad
Reviewed-on: https://gerrit.libreoffice.org/80163
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins
(cherry picked from commit 354b29ace32b19a1b40e46612349b8d76f00c0c6)
Reviewed-on: https://gerrit.libreoffice.org/80168
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Resolves: tdf#127882 gtk 3.18 crash in property dialog teardown

due to the menubutton popovers in the custom property page

workaround by reordering so scrolling window is torn down after the menubutton
is destroyed and clear the popover in the menubutton dtor

Change-Id: Icf06d912524af13a5590f160150f1a4e15f3e9f7
Reviewed-on: https://gerrit.libreoffice.org/80118
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Resolves: tdf#120209 reload names if setting for their language changes

Change-Id: I386a598ae680c90e7d31bf821e7fb58391e5d45c
Reviewed-on: https://gerrit.libreoffice.org/80131
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

tdf#127710: fix sending documents to Claws Mail

Thanks to the Paul's (paul@claws-mail.org) contribution

Change-Id: Ie4144ab78691b33d8d341d4520c40c56010a0dd8
Reviewed-on: https://gerrit.libreoffice.org/79423
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
(cherry picked from commit 1c03df6e82e4bf30294fc0b9e64de5cc39efc5c4)
Reviewed-on: https://gerrit.libreoffice.org/79460
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

return should activate ok

Change-Id: Id9bfa9badb6b4661d6ad4c0a04323c60c4a3c749
Reviewed-on: https://gerrit.libreoffice.org/79864
Tested-by: Jenkins
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

tdf#127422 draw text with correct pKernArray values.

Prior to 5f62b97ae7891b8c601f6093a1ec5358feb20790,

Starting position was specified and DrawText was used to render
the text without referring to pKernArray. After the patch,
DrawTextArray was used but pKernArray was not update correctly.

Instead of draw each substring seprated by space, this patch
increases the values pKernArray when a space is encountered
and call DrawTextArray only once.

Change-Id: I9e61b2d0608400f26136490248740c5f00b56cc3
Reviewed-on: https://gerrit.libreoffice.org/79544
Tested-by: Jenkins
Reviewed-by: Mark Hung <marklh9@gmail.com>
(cherry picked from commit d744838991594eebe27acc4c7d9fb4579d654853)
Reviewed-on: https://gerrit.libreoffice.org/80221

Resolves: tdf#127887 Fix libwps wrong Lotus version detection

Also return the proper kind and creator values set in
WKS4Parser::checkHeader() to callers of
WPSDocument::isFileFormatSupported()

Change-Id: I3adfde0e18669b80956a9b577080fe8388dd3b1f
Reviewed-on: https://gerrit.libreoffice.org/79981
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins
(cherry picked from commit 5a7d4ee6c4dd92758e0fd213671251e96d6e7f08)
Reviewed-on: https://gerrit.libreoffice.org/80015
Tested-by: Xisco Faulí <xiscofauli@libreoffice.org>
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

Resolves: tdf#127262 validity out of order focus in and out problem

Resolves: tdf#127904 modal depth dips below 0

when validation dialog is run modally, but changed itself to unmodal
during execution.

only NotifyModalHierarchy on modal<->unmodal transition and not on
modality depth

Change-Id: I06f5fd0ce32a9f2d799f6003b7d22b13e865b8c6
Reviewed-on: https://gerrit.libreoffice.org/79612
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

Resolves: tdf#127873 accept ',' comma group separator in number entities again

Existing documents unwittingly may have used that as decimal
separator in such locales (though it never was as this is always
the en-US locale).

"Regression" from

commit 9336286a7ea5385541344f444e6f8702c85bdacb
CommitDate: Fri Nov 30 22:15:22 2018 +0100

    [API CHANGE] Resolves: tdf#42518 new KParseTokens::GROUP_SEPARATOR_IN_NUMBER

Change-Id: I0ffc1b8ec7c1820fccd8277036c64093dddf82fe
Reviewed-on: https://gerrit.libreoffice.org/80023
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins
(cherry picked from commit 2caa9d7ddcff3f6e380c306b737e9b5e9cdaf4c2)
Reviewed-on: https://gerrit.libreoffice.org/80044
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#127862 ww8import: page style - import XATTR if defined

...which fixes losing the background color of the page. That was
broken by LO63 commit cc899c6967238877f0094bcf00627145e484ffec
since an obsolete RES_BACKGROUND was not necessarily created.

Change-Id: I944a1b3f3df1468c283f93a49ffacfbd223fb392
Reviewed-on: https://gerrit.libreoffice.org/79912
Tested-by: Jenkins
Reviewed-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit e3c1be1149440fdcfea452aadce0a523af83fe5a)
Reviewed-on: https://gerrit.libreoffice.org/80016

tdf#127913 crash applying popart filter to an image

regression from
    commit 1cd32bcf1b92bd53320717626601135623dadd55
    Date:   Mon Dec 10 11:28:59 2018 +0200
    loplugin:useuniqueptr in vcl

where I failed to note that sort needs a different kind of comparator to
qsort.
Also fix another similar issue I introduced in that commit

Change-Id: I5a93bd0567cd5dd4344b9cf2c362ebff60fa0027
Reviewed-on: https://gerrit.libreoffice.org/80007
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit e347a6bbb58a20e29e691a194c1cd3193d2c5a23)
Reviewed-on: https://gerrit.libreoffice.org/80012

tdf#92845 Save Writer - Comparison settings properly

Options - Writer - Comparison -> 'Compare Documents' and 'Ignore pieces of lenght <value>'
were not retained between runs of Writer.

The type of these settings are defined in
officecfg/registry/schema/org/openoffice/Office/Writer.xcs
as follows:

<prop oor:name="Mode" oor:type="xs:short">
<prop oor:name="IgnoreLength" oor:type="xs:short">

Casting the setting values to sal_Int32 did not work, because sal_Int16 maps to short.

This makes the following warning (2 instances displayed when closing the Options window)
 go away in my local debug build and the values retained:

warn:unotools.config:30387:30387:unotools/source/config/configitem.cxx:483:
Exception from PutProperties com.sun.star.lang.IllegalArgumentException
message: configmgr inappropriate property value context: configmgr::ChildAccess ArgumentPosition: -1

Change-Id: I59fea65378a5dfb2b6f25b9c9eaf08817862587d
Reviewed-on: https://gerrit.libreoffice.org/79807
Tested-by: Jenkins
Reviewed-by: László Németh <nemeth@numbertext.org>
Tested-by: László Németh <nemeth@numbertext.org>
(cherry picked from commit 32e781d268e59397016f7243fc92d5408577d593)
Reviewed-on: https://gerrit.libreoffice.org/79968
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#127446 vcl image lazy-load: fix custom size handling of metafiles

This is a regression from commit
acb803b730f2c6bd82e39beab58949ec14f85eb0 (tdf#125591 DOC import:
lazy-load metafiles with explicit size, 2019-06-11), which assumed that
once maSwapInfo.maPrefSize stores the preferred size, it'll be set on
the Graphic when it's loaded later.

It seems there was no support for that, it was just an accident that the
guessed size of the metafile was about right.

Handle this explicitly in ImpGraphic::loadPrepared(), so that the bugdoc
(which has a custom preferred size) aspect ratio is correct.

(cherry picked from commit 12273af6f18edade1bdda828fe270fa37960e14f)

Change-Id: Ic7c4009ad6723a2e16129d27600c904311ff3daf
Reviewed-on: https://gerrit.libreoffice.org/79913
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Jenkins

Resolves: tdf#127896 fix 'Exchange database' crash

Change-Id: I8278c9e959ab054b5bccf0b33e5885cd4ac7573a
Reviewed-on: https://gerrit.libreoffice.org/79924
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Jenkins

tdf#127306 Fix crash when closing LO after deleting a chart

This crash was introduced with b58aa94f1f365c746135470bceb97cc182c289bc

This patch basically reverts that patch and implements a different fix.

Change-Id: I394a2f6490d3ee0769c0f78bb8a3980167719893
Reviewed-on: https://gerrit.libreoffice.org/79865
Tested-by: Jenkins
Reviewed-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
(cherry picked from commit e6e14b2cafd24179f8675350749878da77a31443)
Reviewed-on: https://gerrit.libreoffice.org/79914
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

Resolves: tdf#126781 force comboboxes to take 1/3 avail space

and not their natural size as this dialog is fixed width so we need
to make do. This brings the right border of the entry inside the
scrolled window area as expected

Change-Id: I60d6078436cbfacbbfda55838e4b0584da4bab1e
Reviewed-on: https://gerrit.libreoffice.org/79915
Tested-by: Jenkins
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>

Resolves: tdf#127416 unparent iconset block on dtor

otherwise gtk keeps the widgets around

Change-Id: I6227662e0a8a4d5a88f4ff6199d6fa06e5be8fe0
Reviewed-on: https://gerrit.libreoffice.org/79649
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

Related: tdf#124600 sw textbox: sync left/right margin of shape to textbox

Writer TextFrames have a default left/right margin; sync from shape to
textbox is needed to have correct layout when a DOCX shape has 0 left or
right margin.

The left margin of the shape was already correct.

(The shape in the docx file was extended to have text.)

(cherry picked from commit ff5f02b9282e19a2a8a68604c588e9487021b0b5)

Change-Id: I832f1bc460f2553bd7f08252d3e21000b4906538
Reviewed-on: https://gerrit.libreoffice.org/79910
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>

Resolves: tdf#127450 max_content_[width|height] isn't available in gtk 3.18

so this crashes

Change-Id: Ie9b2607e4c4290f9745b8bc5f36b80d8970018b1
Reviewed-on: https://gerrit.libreoffice.org/79876
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
Tested-by: Michael Stahl <michael.stahl@cib.de>

tdf#116234 Apply email settings just set

Email configuration that is entered in the SwMailConfigDlg
dialog is written to the (global) configuration, while
the existing SwMailMergeConfig item 'xConfigItem'
was not updated, so the newly entered values for
the email server connection weren't applied for
the current mail merge previously.

In order to have them take effect, take over the
new settings to 'xConfigItem' from the global
config where those were just set.

Change-Id: If0dae3ce26af797ffa728dca96f08dce3197557c
Reviewed-on: https://gerrit.libreoffice.org/79847
Tested-by: Jenkins
Reviewed-by: Michael Weghorn <m.weghorn@posteo.de>
(cherry picked from commit a1573b2e2fda77b7b7d605fe36f5a28f86a3e91b)
Reviewed-on: https://gerrit.libreoffice.org/79907
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

tdf#127840 Remove limit to Days offset

Since 6.3 branch, offset in days is rectricted to 1 month,
which could be too restricted.
In previous versions, this limit was not effective.

Change-Id: Iaddf6fe3ae5e8cd937127302917108d6f465ab81
Reviewed-on: https://gerrit.libreoffice.org/79788
Tested-by: Jenkins
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit eeac5b7a424c3a3be48b1c3583171ad97d1b19cd)
Reviewed-on: https://gerrit.libreoffice.org/79896
Reviewed-by: Laurent BP <laurent.balland-poirier@laposte.net>

tdf#127069 sal: preserve gid of files in the unx osl_replaceFile()

The w32 implementation preserves all attributes of the destination
file, the unx one preserved none of them.

Bring the unx osl_replaceFile() closer to the w32 by preserving the gid
of the destination file as a start.

[ No testcase, we support building on systems where the user is part of
a single group only, and it's not possible to verify the effect of this
change in such environments. ]

(cherry picked from commit eedf523c123a82bf3cbc5f389783e22d75b2e2c5)

Change-Id: I722d4802df34caf71a9dc0db1a3df8b76acb9de6
Reviewed-on: https://gerrit.libreoffice.org/79037
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>

tdf#76324 speedup copy operation with lots of comments in calc

avoid performing a sort in AddShape, and cache some data to speed up the
sort

Makes it about 3 times faster for me, but is still pretty slow for that
large test document

Change-Id: I5c847e43ad5dd66caefcf12b9a624214767371b1
Reviewed-on: https://gerrit.libreoffice.org/79630
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit e51a2917ab19156f5a5d2b9474a5a46a52e9e527)
Reviewed-on: https://gerrit.libreoffice.org/79838

filter: XHTML: make "calc-heading-digit" a little faster

Exporting OpenDocument-v1.3-csd01-part3-schema.odt to XHTML fails with:

runtime error: file share/xslt/export/xhtml/body.xsl line 1404 element variable
xsltApplySequenceConstructor: A potential infinite template recursion was detected.
You can adjust xsltMaxDepth (--maxdepth) in order to raise the maximum number of nested template calls and variables/params (currently set to 3000).

Unfortunately the document contains this many headings, and the
calc-heading-digit computes the value by recursively looking at every
preceding heading in the document, without TCO apparently...

Try to improve this by using XPath to filter early the headings that are
effectively ignored in the 3rd xsl:when case anyway: the ones with a
level lower than the one for which the number is requested; this limits
the recursive calls to the number of headings on the same level.

Change-Id: Iddf5a91664402a57a0138731ddc9cebb06b0a126
Reviewed-on: https://gerrit.libreoffice.org/79720
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 86cca6b40ced4031dec4b708ac67b5cbe70cddf9)
Reviewed-on: https://gerrit.libreoffice.org/79744
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

tdf#124377: enable anti-aliasing metafile during redaction

Change-Id: I6c8a0f6ceaad5d9ec47cce1b00b8ccdd77530aaf
Reviewed-on: https://gerrit.libreoffice.org/79266
Tested-by: Jenkins
Reviewed-by: Muhammet Kara <muhammet.kara@collabora.com>
(cherry picked from commit cd6780aae1392d4c1af0b15b311a4966834a9602)
Reviewed-on: https://gerrit.libreoffice.org/79478
Reviewed-by: Andras Timar <andras.timar@collabora.com>

Related: tdf#124600 DOCX import: ignore left wrap on left-aligned shapes

The DOC import does the same in
SwWW8ImplReader::AdjustLRWrapForWordMargins(). This fixes one
sub-problem of the bugdoc, so now the shape anchored in the header has a
correct position.

This made it necessary to re-visit the tdf#115719 testcases, which are
minimal reproducers. The original document had from-left alignment
(instead of align-to-left), but this did not matter before. Bring the
test documents closer to the original large document, so the tests still
pass and don't depend on LO mis-handling the above mentioned
left-aligned situation. (The interesting property of tdf115719.docx,
where Word 2010 and Word 2013 handles the document differently is
preserved after this change.)

(cherry picked from commit 4883da6fd25e4645a3b30cb58212a2f666dae75a)

Change-Id: I973c13df47b0867e2c4756f0c448495257b7c9d5
Reviewed-on: https://gerrit.libreoffice.org/79834
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>

tdf#125585 write default layer status for OLE objects

The layer status is taken from the active view, when saving the
document. But embedded documents have no view. Error was, that for
all layers value 'false' was written in that case. With this patch
the defaults (visible, printable, not locked) (true, true, false)
are written.

Change-Id: I2388ce31cc208fba075083889ec1bb2f874ef482
Reviewed-on: https://gerrit.libreoffice.org/79701
Tested-by: Jenkins
Reviewed-by: Regina Henschel <rb.henschel@t-online.de>
(cherry picked from commit 12f9a9f341fd8f8a98f7cd98f296a8729d279e0d)
Reviewed-on: https://gerrit.libreoffice.org/79785
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

Resolves: tdf#127786 cache Glibc locale string in LanguageTagImpl

... for non-simple @modifier strings that are constructed using
liblangtag, as loading resource strings needs it over and over and
over again.

Change-Id: Ib6a74e5ddb44508aa41f101c200a508bfa4a13bd
Reviewed-on: https://gerrit.libreoffice.org/79770
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins
(cherry picked from commit 24c7c12224f1c1d66f335f6fe085595352137646)
Reviewed-on: https://gerrit.libreoffice.org/79775
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

Fix Typo SvxMenuConfigPage -> SvxConfigPage

...introduced with ccb2b0078f07194befa61f1e3fd88e53ff236871 "weld
SvxMenuConfigPage/SvxToolbarConfigPage", causing -fsanitize=dynamic-type-
mismatch during UITest_writer_demo:

> include/tools/link.hxx:163:10: runtime error: downcast of address 0x613000204b40 which does not point to an object of type 'SvxMenuConfigPage'
> 0x613000204b40: note: object is of type 'SvxConfigPage'
>  83 01 00 4e  b0 a1 bb 2e f6 7f 00 00  20 c4 56 00 e0 60 00 00  50 f8 a4 00 40 60 00 00  e8 49 38 00
>               ^~~~~~~~~~~~~~~~~~~~~~~
>               vptr for 'SvxConfigPage'
>  #0 in SvxMenuConfigPage* tools::detail::castTo<SvxMenuConfigPage*, SvxConfigPage*>(SvxConfigPage*) at include/tools/link.hxx:163:10
>  #1 in SvxConfigPage::SvxConfigPage(weld::Container*, weld::DialogController*, SfxItemSet const&) at cui/source/customize/cfg.cxx:997:41
[...]

Change-Id: Ie36ebabd3075cade590cc504fb06b5319b346dcf
Reviewed-on: https://gerrit.libreoffice.org/79666
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
(cherry picked from commit 2377d937225adb17dc3f76d72937d5cd29d2ef69)
Reviewed-on: https://gerrit.libreoffice.org/79746
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

Resolves: tdf#122487 avoid duplicates when auto-naming drawing objects

we were looking inside groups already, but our set was limited to
the length of toplevel objects, so it wasn't sufficient.

just collect all used nums and return the first unused number

Change-Id: I239118e9cff7b7ed2a40d68f284c3c4d1d9eb6c4
Reviewed-on: https://gerrit.libreoffice.org/79658
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

Resolves: tdf#127218 allow changes dialog to shrink when button text changes

original the dialog has super wide button contents when the initial size
is calculated, and then the buttons are changed afterwards, but the dialog
remains at its initial size.

Change-Id: Ib29ab9ebdec01a93561a8b304572df60b6ad5ad2
Reviewed-on: https://gerrit.libreoffice.org/79635
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#127316 ww8export: use default escapement for AUTO

Previously it looked bad, but not horrendous when
AUTO was 101.  Now that AUTO is 13999, it is obviously
wrong (in Word). Better to use the default of 33.

Change-Id: If0e12d315346515ce6ec5ae4bcc4110efba14f9b
Reviewed-on: https://gerrit.libreoffice.org/79651
Reviewed-by: Justin Luth <justin_luth@sil.org>
Tested-by: Justin Luth <justin_luth@sil.org>
(cherry picked from commit 9b96f644c554f1b10a380dd3f9a04a405b948411)
Reviewed-on: https://gerrit.libreoffice.org/79664
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#127316 docxexport: use default escapement for AUTO

Previously it looked bad, but not horrendous when
AUTO was 101.  Now that AUTO is 13999, it is obviously
wrong. Better to use the default of 33.

The import algorithm still could use some work since it
round-trips as 51% and not 33%.

Change-Id: Id984ea3e6e0b50014389ec8a86668b7a271b96f4
Reviewed-on: https://gerrit.libreoffice.org/79628
Reviewed-by: Justin Luth <justin_luth@sil.org>
Tested-by: Justin Luth <justin_luth@sil.org>
(cherry picked from commit 89f0107b8de21bbb22e850847348ab40cce24644)
Reviewed-on: https://gerrit.libreoffice.org/79656
Tested-by: Jenkins
Reviewed-by: László Németh <nemeth@numbertext.org>

tdf#127411 Broken mouse status in Impress status bar

regression from
commit c53f3f8f58f55d0978fb968446975856d72a52f8
tdf#126819 Can't retrieve the mouse position written in the
statusbar

Change-Id: I3072de28bcc512d98c43d5cc7110aa3f80d51e61
Reviewed-on: https://gerrit.libreoffice.org/79519
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit d513a845cf73e55aa88be716fcb7a1f98cd054b1)
Reviewed-on: https://gerrit.libreoffice.org/79605

cid#1448425: Resource leak (sc/htmlprovider)

Change-Id: I3c23280da75e51be37c4b66f14f805ca49ebf3cf
Reviewed-on: https://gerrit.libreoffice.org/75095
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
(cherry picked from commit 2f6ead197bbff5142199d00a30ab900ebb997754)
Reviewed-on: https://gerrit.libreoffice.org/78857
Reviewed-by: Eike Rathke <erack@redhat.com>

tdf#127748: crash on modify a frame style, which has writing-mode bt-lr

bt here:
https://bugs.documentfoundation.org/attachment.cgi?id=154474

Change-Id: I6e6e0927ea85c4f0f4f59bf327ce5468fc428328
Reviewed-on: https://gerrit.libreoffice.org/79572
Tested-by: Jenkins
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
(cherry picked from commit 2a38ec580b3b6ffe11276462547a41c467a51197)
Reviewed-on: https://gerrit.libreoffice.org/79582
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>

Fix stack-use-after-return when moving a slide in the list

...very far, so that the list needs to scroll:

> ==1676==ERROR: AddressSanitizer: stack-use-after-return on address 0x7f039a1e6420 at pc 0x7f03c34e0658 bp 0x7ffede26b140 sp 0x7ffede26b138
> READ of size 8 at 0x7f039a1e6420 thread T0
>  #0 in Point::X() const at include/tools/gen.hxx:78:44 (instdir/program/libvcllo.so +0x5cf6657)
>  #1 in OutputDevice::PixelToLogic(Point const&) const at vcl/source/outdev/map.cxx:1191:47
>  #2 in sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool) at sd/source/ui/slidesorter/controller/SlsDragAndDropContext.cxx:84:47
>  #3 in sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0::operator()() const at sd/source/ui/slidesorter/controller/SlsDragAndDropContext.cxx:92:34
>  #4 in void std::__invoke_impl<void, sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0&>(std::__invoke_other, sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0&) at gcc/x86_64-pc-linux-gnu/10.0.0/../../../../include/c++/10.0.0/bits/invoke.h:60:14
>  #5 in std::enable_if<is_invocable_r_v<void, sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0&>, void>::type std::__invoke_r<void, sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0&>(sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0&) at gcc/x86_64-pc-linux-gnu/10.0.0/../../../../include/c++/10.0.0/bits/invoke.h:110:2
>  #6 in std::_Function_handler<void (), sd::slidesorter::controller::DragAndDropContext::UpdatePosition(Point const&, sd::slidesorter::controller::InsertionIndicatorHandler::Mode, bool)::$_0>::_M_invoke(std::_Any_data const&) at gcc/x86_64-pc-linux-gnu/10.0.0/../../../../include/c++/10.0.0/bits/std_function.h:291:9
>  #7 in std::function<void ()>::operator()() const at gcc/x86_64-pc-linux-gnu/10.0.0/../../../../include/c++/10.0.0/bits/std_function.h:622:14
>  #8 in sd::slidesorter::controller::ScrollBarManager::RepeatAutoScroll() at sd/source/ui/slidesorter/controller/SlsScrollBarManager.cxx:518:17
>  #9 in sd::slidesorter::controller::ScrollBarManager::AutoScrollTimeoutHandler(Timer*) at sd/source/ui/slidesorter/controller/SlsScrollBarManager.cxx:539:5
>  #10 in sd::slidesorter::controller::ScrollBarManager::LinkStubAutoScrollTimeoutHandler(void*, Timer*) at sd/source/ui/slidesorter/controller/SlsScrollBarManager.cxx:537:1
>  #11 in Link<Timer*, void>::Call(Timer*) const at include/tools/link.hxx:112:45
>  #12 in Timer::Invoke() at vcl/source/app/timer.cxx:75:21
>  #13 in Scheduler::ProcessTaskScheduling() at vcl/source/app/scheduler.cxx:479:20
>  #14 in Scheduler::CallbackTaskScheduling() at vcl/source/app/scheduler.cxx:287:5
>  #15 in SalTimer::CallCallback() at vcl/inc/saltimer.hxx:54:13
>  #16 in sal_gtk_timeout_dispatch(_GSource*, int (*)(void*), void*) at vcl/unx/gtk3/gtk3gtkdata.cxx:761:45
>  #17 in g_main_dispatch at ../glib/gmain.c:3189:28
> [...]
> Address 0x7f039a1e6420 is located in stack of thread T0 at offset 32 in frame
>  #0 in sd::slidesorter::controller::SelectionFunction::MouseDragged(AcceptDropEvent const&, signed char) at sd/source/ui/slidesorter/controller/SlsSelectionFunction.cxx:630 (instdir/program/../program/libsdlo.so +0x5c22e1f)
>   This frame has 1 object(s):
>     [32, 104) 'aEventDescriptor' (line 631) <== Memory access at offset 32 is inside this variable
> HINT: this may be a false positive if your program uses some custom stack unwind mechanism, swapcontext or vfork
>       (longjmp and C++ exceptions *are* supported)

Presumably a regression introduced with 26912eea7521dd2b84bfac56b322cf0f8b142450
"sd: replace boost::bind with C++11 lambdas and for loops", assuming that
boost::bind bound rMousePosition by value.

Change-Id: I4cefc5974b7377ff4918d59b51fc8cda41292702
Reviewed-on: https://gerrit.libreoffice.org/78774
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
(cherry picked from commit 833c23264b532e34e050e3b8747c02ede6bf7350)
Reviewed-on: https://gerrit.libreoffice.org/79553
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#127701 Zoom factor percentage calculating fault

regression from
    commit c53f3f8f58f55d0978fb968446975856d72a52f8
tdf#126819 Can't retrieve the mouse position written in the
statusbar

note that this is a workaround - we fire an sfx2 event when we choose
one of the "smart" zoom options, and normally we then receive another
sfx2 event to notify us of the real zoom value.
But when we choose the same smart option twice, we don't receive the
second sfx2 event.

Change-Id: I31c2e34ff31a4f4e03a447ffbd783589982edc57
Reviewed-on: https://gerrit.libreoffice.org/79516
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit e39ddbdb5b6f49e3585a03444dce077709b855e6)
Reviewed-on: https://gerrit.libreoffice.org/79557

tdf#127652 sw_redlinehide: don't set invalid SwTextFrame offsets

The obvious problem here is that if a deletion spans multiple text
frames, the offset of the 2nd one must not be reduced lower than the
start position of the deletion; apparently this fixes the subsequent
crashes with stale text portions having now invalid indexes into the
string.

(regression from 0acde7514e666fc04805fd36503bd174162336ca)

Change-Id: Iaf9e2c5ca06ed36dc0e8611891d19c22f7661d34
Reviewed-on: https://gerrit.libreoffice.org/79421
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit 9ec2c77046901fdacca77753052b54c81e6bc82c)
Reviewed-on: https://gerrit.libreoffice.org/79432
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

Revert "tdf#117988 writerfilter: IgnoreTabsAndBlanksForLineCalculation"

This reverts LO 6.2 commit 49ddaad2f3ba4e17e1e41e94824fb94468d2b680.
tdf#127617 proves it simply was not the correct solution.

I replaced the unit test document with one that clearly demonstrates
that spaces/tabs should NOT be used in line height calculations.
Example document tested with Office 2003, 2010, 2016.

Change-Id: I2833384a017526d665adef0cae968bc4aef0dd94
Reviewed-on: https://gerrit.libreoffice.org/79473
Reviewed-by: Justin Luth <justin_luth@sil.org>
Tested-by: Justin Luth <justin_luth@sil.org>
(cherry picked from commit 202bee1a819de7b1e8c75dd863c4154f66419400)
Reviewed-on: https://gerrit.libreoffice.org/79484
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#94765: SVGIO: Look for gradient/pattern ids once the file...

... is completely parsed

it might happen the element's id the url points to hasn't been parsed yet

Reviewed-on: https://gerrit.libreoffice.org/79118
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
(cherry picked from commit ab1479956568b0d9a6022133cf651f846293aa46)

Change-Id: Ia92c9188de5d23f2f992846aa91f3f936aeefacb
Reviewed-on: https://gerrit.libreoffice.org/79130
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

Revert "tdf#107784: DOCX Import: Show citation's title in fields"

This reverts commit fadce0c9a7ccbd7222df90e41d8fc16d9d67045f.

it needs more work. better not to have it in 6.3

Change-Id: Ia6e0d90cf5e56d87aa83903fd787d548c0514ecd
Reviewed-on: https://gerrit.libreoffice.org/79459
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#104167 DOCX import: don't map Heading1 to Heading 1 para style

This is nominally a regression from commit
c12a78e42e1cbaaf4ea021c82acbec5ea177b1f6 (writerfilter: reset properties
of reused styles, 2014-06-04), which assumed that style names are
unique.

They are not unique for the bugdoc, as it has paragraph styles named
'heading 1' and 'Heading1', both mapped to 'Heading 1'.

The mapping for the style name without a space was added in commit
1d2af74e226bf44e1a0f1e1981ca965cffeaa0b2 (INTEGRATION: CWS xmlfilter02
(1.27.12); FILE MERGED, 2008-01-10), in the "2007/11/22 14:53:43 os
1.27.12.9: Tab stops almost perfect now" part, sadly without any
indication on why that would be a good idea.

The binary DOC import never mapped Heading1 to Heading 1, so it doesn't
have this problem.

Fix the problem by removing this mapping, it leads to formatting loss
when a document has both HeadingN and Heading N styles.

(cherry picked from commit a283627036c9b1c56e4b765866b8fb8a26ec87ab)

Conflicts:
sw/qa/extras/ooxmlimport/ooxmlimport2.cxx

Change-Id: I26dd999f5a536e079ac351ba5d40d20c83b99310
Reviewed-on: https://gerrit.libreoffice.org/79429
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
Tested-by: Xisco Faulí <xiscofauli@libreoffice.org>
Tested-by: Jenkins
Reviewed-by: Mike Kaganski <mike.kaganski@collabora.com>

tdf#107784: DOCX Import: Show citation's title in fields

Regression from 06f7d1a96eef5aa69d4872ff6d96eb5085296d09
The problem is ConditionalExpandAuthIdentifier method uses
AUTH_FIELD_IDENTIFIER while we want to use AUTH_FIELD_TITLE
for citations, as we do in DomainMapper_Impl::SetFieldResult

Change-Id: I30dcceba79dda3532ce10732e43303233d2ee441
Reviewed-on: https://gerrit.libreoffice.org/79441
Tested-by: Jenkins
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
(cherry picked from commit 417d993b8b8a86c019758ee0850e4b42967e2afa)
Reviewed-on: https://gerrit.libreoffice.org/79450
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#126078 Autotext: fix lost space between first-/lastname

Change-Id: I3f8c1e25e06097cbed37fba1ca6a634704387e00
Reviewed-on: https://gerrit.libreoffice.org/76000
Tested-by: Jenkins
Reviewed-by: Michael Stahl <Michael.Stahl@cib.de>
(cherry picked from commit 7227e5898ecd3eefe2b819d6be8cd425d15b1f38)
Reviewed-on: https://gerrit.libreoffice.org/79436
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

Resolves: tdf#127695 IsGroupObject == true doesn't mean its a SdrObjGroup

but that it has a SubList

Change-Id: I473464de5c6bba5c36be1ad7cbd21a3158645d66
Reviewed-on: https://gerrit.libreoffice.org/79412
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>
Tested-by: Xisco Faulí <xiscofauli@libreoffice.org>

Resolves: tdf#127682 out_of_range thrown on print empty calc doc

when odd/even pages requested and print is cancelled

the abort procedure will ask for page 0 to be rendered on cancel,
which doesn't exist since the odd/even support.

Let the print of an empty page to ahead in the abort scenario

Change-Id: Id8dfde57fca891ba3eb220cf6e495585abf4c918
Reviewed-on: https://gerrit.libreoffice.org/79414
Tested-by: Jenkins
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#109158 improve load time of autocorrect XML file

Before, it seems to be doing a lot of unnecessary work because, by default,
startFastElement will call startUnknownElement, which constructs various objects, etc.

It seems to me that if the class already implements createFastChildContext,
it is already using XFastParser, and so it should implement startFastElement to avoid that work.

By avoiding going down the startUnknownElement path, reduces load time from
1.7s to 1.4s

Change-Id: Ifd745cf2cc22a8cb988a5668332cf6e70bb996da
Reviewed-on: https://gerrit.libreoffice.org/79269
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
Tested-by: Noel Grandin <noel.grandin@collabora.co.uk>
(cherry picked from commit ea4f3099d6e0cf30d80caa8b2121c7a358f80fdd)
Reviewed-on: https://gerrit.libreoffice.org/79378
Tested-by: Jenkins

Related tdf#90429: Don't erroneously pop unrelated path segments

...when coming across a bad set node <prop> member.  The ooo2gd_3.0.0.oxt (see
referenced bug for links) Addons.xcu contains two such bogus props

> <node oor:name="AddonUI">
>   <node oor:name="OfficeMenuBarMerging">
>     <node oor:name="org.openoffice.gdocs.gdocs" oor:op="replace">
>       <prop oor:name="ImageIdentifier" oor:type="xs:string">
>         <value/>
>       </prop>
>     </node>
>   </node>
>   <node oor:name="OfficeToolBar">
>     <node oor:name="org.openoffice.gdocs.gdocs" oor:op="replace">
>       <prop oor:name="UIName" oor:type="xs:string">
>         <value>Google Docs</value>
>       </prop>
>     </node>
>   </node>
> </node>

so that a later

> <node oor:name="AddonUI">
>   <node oor:name="Images">

was processed with path_ erroneously stripped down to the root path.

It appears that this has been broken ever since
7d9bce7ca0408786d0ad448dee0f2bf480870d3e "jl153: #i110720# ignore spurious
<prop> elements in .xcu set nodes".

Change-Id: I1d069b7226c8202b6eb93f59d294ce7f25681f80
Reviewed-on: https://gerrit.libreoffice.org/77537
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
(cherry picked from commit 5a522a0196aaa071bd3ac59b3088246e3fe98f34)
Reviewed-on: https://gerrit.libreoffice.org/79074
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

Resolves: tdf#127646 remove undispatched events in dtor

Change-Id: I8d18b0b667a905131fd0edee1f73ebab144e6307
Reviewed-on: https://gerrit.libreoffice.org/79257
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>

tdf#109158 improve sorting when loading large autocorrect file

reduces time from 2.0s to 1.7s

reduce work by
(*) reserving some arrays
(*) pre-sorting with a cheaper comparator
(*) don't copy when returning result, just return a const&
(*) flattening the data-structures to reduce pointer-chasing

Change-Id: I972bd7ffdbf2121c2d38c24aca9618ca708e920c
Reviewed-on: https://gerrit.libreoffice.org/79119
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
Signed-off-by: Xisco Fauli <xiscofauli@libreoffice.org>
Reviewed-on: https://gerrit.libreoffice.org/79377

tdf#127657: FIREBIRD error in query input param when referred field is integer

Change-Id: I626c8a8869570986d0293cd9070a1ee40ec585dc
Reviewed-on: https://gerrit.libreoffice.org/79314
Reviewed-by: Julien Nabet <serval2412@yahoo.fr>
(cherry picked from commit f23e19cd15f14566832befba73fb6bbc1cffb0d7)
Reviewed-on: https://gerrit.libreoffice.org/79372
Tested-by: Jenkins
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

tdf#127235 break if object is larger than page

We will always have an overlap if the object is larger than the whole page.
Look for the next page style: if it is missing or the same as the current,
we simply stop.

Change-Id: I699d5146d215f17b4775dbcb32811fa5b6ebf60d
Reviewed-on: https://gerrit.libreoffice.org/78696
Tested-by: Jenkins
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
Reviewed-on: https://gerrit.libreoffice.org/79101

cui: infinite loop in SentenceEditWindow_Impl::MarkNextError()

Looks like previously the aCursor never moved backwards so try to ensure
the same for nCursor.

(regression from 5261417cbb3051b812164838d19c0f748573df45)

Change-Id: I7d2f04bbc6d1274ec4feff57927494f2eb5d665c
Reviewed-on: https://gerrit.libreoffice.org/79071
Tested-by: Jenkins
Reviewed-by: Michael Stahl <michael.stahl@cib.de>
(cherry picked from commit a370107d86c32c1631b00422e2e35ff4b867ce57)
Reviewed-on: https://gerrit.libreoffice.org/79088
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

update credits

Change-Id: I4d6c996cbc25053fed93ce62e261c6ae77119961
(cherry picked from commit cfcf0731c919a20387cebc56c033f9328a0af735)

Update git submodules

* Update translations from branch 'libreoffice-6-3'
  - update translations for 6.3.2 rc2

    and force-fix errors using pocheck

    Change-Id: Idc89d7732f84e6b59bfca1494212ac70039b1e3b

tdf#126852 Bind TEXT mysql type to Clob

Change-Id: I5a71666f1fafa9507032deffafebf885813a0369
Reviewed-on: https://gerrit.libreoffice.org/78815
Tested-by: Jenkins
Reviewed-by: Tamás Bunth <btomi96@gmail.com>
(cherry picked from commit 47a62ede2403772b8743bfcfa0b6dd4be17fec9c)
Reviewed-on: https://gerrit.libreoffice.org/79083
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#126833 fix table auto format undo/redo

Fixes regression caused by fix for tdf#123516

Change-Id: If7b6bf9cba30085585fc495898f3dfb308197495
Reviewed-on: https://gerrit.libreoffice.org/78770
Tested-by: Jenkins
Reviewed-by: Jim Raykowski <raykowj@gmail.com>
(cherry picked from commit 1d5b89eccdccb3090b5503a60f829c54c077f328)
Reviewed-on: https://gerrit.libreoffice.org/79045
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>

tdf#121663 DOCX import: fix missing space in line numbering

Auto distance of line numbering is 0.5 cm, not zero.

Change-Id: Ic451424bffb3eb3fe386bcaef5a476a5679faff2
Reviewed-on: https://gerrit.libreoffice.org/75317
Tested-by: Jenkins
Reviewed-by: László Németh <nemeth@numbertext.org>
(cherry picked from commit d828429f81e63d3c6beb4574fe3c52b001ea0a5b)
Reviewed-on: https://gerrit.libreoffice.org/79053
Reviewed-by: Xisco Faulí <xiscofauli@libreoffice.org>