Age | Commit message (Collapse) | Author |
|
b/c of crash in dbgutil build
This reverts commit 298571e2e10f5a925abc3cb75940dbef5701b583.
Change-Id: I41e6e9d17809e416813b4ee4032f38d3278b4103
Reviewed-on: https://gerrit.libreoffice.org/29319
Reviewed-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
Tested-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
|
|
A comment text edit (provided by editeng) works with relative twip
coordinates, conversion of the cursor position to absolute twips happens
in ImpEditView::ShowCursor(), provided that the pOutWin member has a map
mode origin that respresents the offset correctly.
This is not true during SwSidebarWin::DeactivatePostIt(), the map mode
origin is already reset back to 0, so just don't emit callbacks during
shutdown, the sw body text cursor will be shown later anyway.
Change-Id: I02c15bb9fad99db8e43fd2f37df770dd165be788
|
|
- handle it in SwTextShell (for normal Writer text)
- simple icon in the galaxy theme
Change-Id: Ib8f11dbca28b19a2fc0411c92d9f0b4b052277bb
|
|
When importing this rtf file, sections were interpreted as page breaks.
They shouldn't, instead all content should be on the same page.
Change-Id: I445ce507d14d21ef568734ab7020eed9cbfe57ac
Reviewed-on: https://gerrit.libreoffice.org/29298
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Reviewed-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
Tested-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
|
|
It assumes that the process has only a single document opened
(createView() creates a new window on the frame returned by
SfxViewFrame::First()), while it's possible that in this test an other
testcase forgot to dispose its component.
This way testCursorWindows() depends less on other testcases, and
ideally all testcases should be entirely independent of each other.
Change-Id: I35decb0b69aeebd3984f4da4386c696a63fe3e04
Reviewed-on: https://gerrit.libreoffice.org/29300
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
Per default a temporary file is construted from a path and a leading
pattern for the filename. For mail merge the filename can be read
from a database column. If the path is not existing, a temporary
directory is created.
Normally the temp file function would fail, if the filename contains
a slash and the sub-directory of the filename doesn't exists as a
subdirectory of path.
To implement tdf#94987, this adds an option to the temp file class
to create the parent directories of the filename pattern.
Change-Id: I02bf34294dac85598ee153d8cfcf00bc5d7775af
|
|
There is a list of exceptional SwIndex containers in that member
function, like bookmarks, at-char anchored fly frames, etc. If we're
growing (so basically in the safe case), don't touch the cursors of
other windows, either.
This helps to avoid the surprising "I didn't do anything and my cursor
moved" behavior.
Change-Id: I9941fdcb6b7ad4b6e18a321cecc72fdf73d917fd
|
|
Change-Id: I7b1aa65a557f3855939ad8b183bba8ff656946f8
Reviewed-on: https://gerrit.libreoffice.org/29173
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
That's where the non-view variant is emitted, and SwVisibleCursor
said visibility=false even if the cursor was expected to be
visible.
Also fix the test callback that checked if the payload is equal to the
"true" literal, but actually the payload was a JSON.
Change-Id: Ifa0e23eb274925e8ab8ef5e7c21370ad2cd22282
|
|
These colors are used in the tiles, so it's a good idea if the client
can use matching colors for cursors and selections. But to be able to do
that, we need an API to expose these colors.
Change-Id: Ia688c07e6c300fecdf8dc428d5a3f000d1857387
|
|
The purpose of storing seconds was to get a better timestamp, not to
make combining impossible.
Also fix two issues with the existing testcase:
- read both timestamps, so test doesn't fail if the second is 0 in the
first redline
- don't fail if minute changes while waiting for a second
Change-Id: Ib6c8ecdcf2f0da9191f0b48e6aaefc0b2449583e
|
|
It's not clear why it was like this since the initial import, but at
least not clearing seconds is already an improvement.
Change-Id: Ic452256d4ab47e6779ec71fe6576eacb5b74fb43
|
|
Both ODF and OOXML's drawingML seem to just not write ZOrder for child
shapes (instead they just describe them in a bottom-up order), but RTF
does use dhgt, it seems. Given that ZOrder has different meaning inside
and outside a group shape use a stack of GraphicZOrderHelpers to get the
correct result.
Change-Id: I68dbd9f0ae59a1759da14bf414dc1d277d1c927d
|
|
For now only care about the start of the cursor, which can be only at a
single redline.
Add matching testcase + expose it in the gtktiledviewer status bar for
interactive manual testing.
Change-Id: Ib61757412d6b54bef64361d4a8563795ca0bab6c
|
|
As this only returns the number of views, not the actual views. Since
it's a rename, it's just an API (but not an ABI) change.
Change-Id: Ib4f0ea56a90e5ae9c80ee1781aa2f29aff4259e7
|
|
Previously .uno:NextTrackedChange always worked by cursor position, but
redlines are stored in the redline table, so they have a unique index.
Allow specifying that index when invoking the command, and in that case
ignore the cursor position.
Change-Id: I7dbe8044feca8fcc48635482a22317b024f40ffa
|
|
docx import of base-less style "Text" was conflicting with
LibreOffice's builtin "Caption"->"Text" paragraph style.
This only worked when the parent .isEmpty(). Style names can
use localization, so forcing the imported parent-name will not
match between different language versions.
Change-Id: I86d9965abb4ee90b04c713371529a679d045660c
Reviewed-on: https://gerrit.libreoffice.org/28894
Reviewed-by: Justin Luth <justin_luth@sil.org>
Tested-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
With this, in case the first view has an active text edit and a new view
is created, then the shape text lock is instantly visible in the second
view, even if the first view does not end + begin the text edit again
later.
Change-Id: I82d98ab1431a54a1a8897d16ce7fa0856baa2a10
|
|
With this, in case the first view has an active text edit selection and
a new view is created, then the text selection is instantly visible in
the second view, even if the first view's text selection does not change
later.
Change-Id: I3255febd5d65d6576ddbc57cf96836bdf0b06a72
|
|
With this, in case the first view has an active text edit and a new view
is created, then the cursor position is instantly visible in the second
view, even if the first view's cursor does not move later.
Change-Id: Ia82e7dc1ce9bb58c67a280179ecadc74d3b63026
|
|
The issue of 362d4f0cd4e50111edfae9d30c90602c37ed65a2 "Explicitly mark
overriding destructors as 'virtual'" appears to no longer be a problem with
MSVC 2013.
(The little change in the rewriting code of compilerplugins/clang/override.cxx
was necessary to prevent an endless loop when adding "override" to
OOO_DLLPUBLIC_CHARTTOOLS virtual ~CloseableLifeTimeManager();
in chart2/source/inc/LifeTime.hxx, getting stuck in the leading
OOO_DLLPUBLIC_CHARTTOOLS macro. Can't remember what that
isAtEndOfImmediateMacroExpansion thing was originally necessary for, anyway.)
Change-Id: I534c634504d7216b9bb632c2775c04eaf27e927e
|
|
Have a Writer shape selection in the first view, create a new view, the
second view doesn't see the selection. But the same works if the shape
selection is created when the second view was created earlier.
Emit the svx selection state as part of SwCursorShell::NotifyCursor() to
fix the problem, but take care of not sending the status to the new view
itself.
Change-Id: Iff3979235e1c12cc64867c3184354967f451370b
Reviewed-on: https://gerrit.libreoffice.org/28857
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
Styles that inherit from a parent style, but were defined in style.xml
before the parent, were losing their base style.
Change-Id: Ic12876dddb1aa961cd8ef7579061cca30c320c71
Reviewed-on: https://gerrit.libreoffice.org/28785
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
Have a Writer text selection in the first view, create a new view, the
second view doesn't see the selection. But the same works if the text
selection is created when the second view was created earlier.
Emit the selection state as part of SwCursorShell::NotifyCursor() to fix
the problem.
Change-Id: I7d71c9b58941c8ca8720b0e63e54bc757b1aaf61
Reviewed-on: https://gerrit.libreoffice.org/28844
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
Have a graphic selection (and thus a hidden text cursor) in the first
view, create a second view. The view text cursors should be hidden in
the second view as well.
Change-Id: Ic22db84aab62c8f43c2da3d4a19b56c993c9f012
Reviewed-on: https://gerrit.libreoffice.org/28836
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
With this, the graphic selection is no longer lost when creating a new
view. That happened as the SwView ctor calls
SdrMarkView::SetMarkHandles() while the new view is not yet current,
which resulted clearing the graphic selection in the old view, not in
the new one.
A side-effect of introducing SwDrawView::GetSfxViewShell() is that now
SwView::getPart() is called from the SwView dtor, guard against doing
any real work there in that situation to avoid potentially reading
already deleted data.
Change-Id: I4fab39a907d2cbe228c0fc8d44bedc64893387d1
Reviewed-on: https://gerrit.libreoffice.org/28832
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
Change-Id: Ia787c42f484c00242f1bcaca8ea7459890ec5745
Reviewed-on: https://gerrit.libreoffice.org/28828
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: I6cdaa375525b14299b374273b6651a2a89f0c500
|
|
The multiset routine was put in to increase the speed of applying
properties. However, if one property causes an exception, the
remaining properties are never applied. There is already a fallback
routine (if the multiset can't be created), so use that instead
of returning in a failed state.
Change-Id: Iac53edd5fca8e8543d536609113a7b1109befd82
Reviewed-on: https://gerrit.libreoffice.org/28765
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Justin Luth <justin_luth@sil.org>
|
|
Since the margin collapsing (aka contextual spacing, implemented
for LO 3.6 by Miklos Vajna, commits
6f04bf5e90ff75288dcf75c43843edf798641e3d,
0662778b0b4b9958ee5e5bd3bac69f9e290f9495,
11059331718fb8faab483c75633b4e80d8028b7d,
03f9b6bebc0ca77021be46664c7bcbe4cb297503,
8631dbf85fb5ed56d225e32ea5a9c36c96b0d649,
9f4bb5bd4f55b4a80544413efde26391849b1d7f,
f722299e133568fe75f7cf9ce0c103a1553474c4) is only meant
to suppress fo:margin-top and fo:margin-bottom, as seen in
https://issues.oasis-open.org/browse/OFFICE-3767, current
implementation is inconsistent: it nullifies all spacing, including
normal line spacing.
This patch makes sure that collapsing conforms to specification.
Change-Id: I413f6469d7891dd85f1f7a2c666f219187ddaec4
Reviewed-on: https://gerrit.libreoffice.org/28692
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
It was first removed in aa94466699a98730d9e8d112fd52cea63c2226b2
(Executable_tiledrendering is obsolete and can be killed, 2015-03-24),
and then it was re-introduced with
00f2787a4a68633206635743298926bf2e65a8fa (vclwidgets: wrap all
vcl::Window subclasses allocated on stack in VclPtr, 2015-04-09) due to
a mismerge.
Change-Id: I243375d6a8d7804ed23f2080d2dbb3c0132789f1
Reviewed-on: https://gerrit.libreoffice.org/28747
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
"regression" from 4e653d15eff26aa5283d8ba20611893f4c573f57
If there are new style elements, then don't treat a
default break in columns as a continuous break.
This fixes both round-tripping, and initial import of
columns and headers on this particular document. Since
MS and LO treat sections so differently, it is a balancing
act of what to change.
Change-Id: I9c79bc1246108da6a5ebbf744acbcfcbb9d33d48
Reviewed-on: https://gerrit.libreoffice.org/28728
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Justin Luth <justin_luth@sil.org>
|
|
including fixing a bug in SwXMLExport::exportDoc where it was ORing with
a constant from a different type: nsRedlineType_t::REDLINE_INSERT
Change-Id: I2bb154c9a35d106e64fd1a8b6e928d0384c9fafe
|
|
Do not export the parent-style-name attribute for a cell style.
The parent-style-name attribute should refer to a style of table-cell
family. Not a table-template element.
Reviewed on:
https://gerrit.libreoffice.org/28729
Change-Id: I350ec7cef0ac544ce4e5dec54015412a4070b1d8
|
|
It was missing as the shape text range was deleted twice, as
DelFlyInRange() thought both the fly and the draw frame format own its
contents range, while only the fly one does (when called by
SwDoc::MakeFlyAndMove(), called by SwXText::convertToTextFrame()).
Thanks Red Hat for the pizza!
Change-Id: I26e23550df085511e0e87f593f1a461057413d0e
|
|
The DOCX import handles this at a tokenizer level, so let's do the same
in the RTF case as well.
Change-Id: Id7ff43fa9e9bcd05b13d187623d39fb072758057
Reviewed-on: https://gerrit.libreoffice.org/28679
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
The image hyperlink is a resource id in the document and needs to be translated into real URL.
First I define a new type CT_Hyperlink_URL in the model and associate it with an action handleHyperlinkURL.
In OOXMLFastContextHandlerProperties::handleHyperlinkURL I dispatch it to OOXMLHyperlinkURLHandler to translate resource id to real URL then set the PropertySet with real URL.
Then the correct URL will be captured while resolving GraphicImport, which will be stored in GraphicImport_Impl::sHyperlinkURL as an OUString.
Finally the property will be set in the GraphicImport::applyName if the length of the sHyperlinkURL is not 0.
Also adds a test file image-hyperlink.docx and a test in ooxmlimport.cxx.
Change-Id: I6194b9cc6bcc1bfaa033ab05e94836fe96e33f14
Reviewed-on: https://gerrit.libreoffice.org/28432
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
Microsoft allows spacing in textboxes even if the borders are not
shown, but LO does not. Added a compatible setting, used an existing
setting to allow the spacing, and changed .doc export not to zero
out the spacing if the border was zero-width.
Using the compatible setting in the export code is almost redundant,
but it does require that the document was LOADED as .doc, and not
"save as" from another format.
This patch simply allows round-tripping - any user attempt to modify the
border settings will enforce normal LO border rules.
Change-Id: I60ac036e1bfac381eea15e33c21495ad3800277a
Reviewed-on: https://gerrit.libreoffice.org/28601
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
For one, this is unwanted: in the LOK case we want to switch between the
windows without any side effect to be able to e.g. paint tiles.
For another, this caused an invalidation loop when two views selected
the text frames or images in Writer. The loop looked like:
1) Press a key in view #1, so a setView(0) + paintTile() is necessary in
view #0.
2) SfxLokHelper::setView(0) to switch from view #1 to view #0.
3) SwFEShell::ShellLoseFocus() on view #1, which hides sdr marks ->
invalidate.
4) SwFEShell::ShellGetFocus() on view #0, which shows sdr marks ->
invalidate.
5) paintTile() in view #0.
6) SfxLokHelper::setView(1) to paint tiles due to 3). (Generates
invalidations in both views.)
7) SfxLokHelper::setView(0) to paint tiles due to 4).
And so on, this way a call to SfxLokHelper::setView(0) resulted in an
another (async) call to SfxLokHelper::setView(0) all the time.
Change-Id: Ice855b9128f61bb7b823b499cad366998f297b5d
Reviewed-on: https://gerrit.libreoffice.org/28611
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
|
|
Some custom headers and footers are referenced in docx but aren't present.
See https://bugs.documentfoundation.org/show_bug.cgi?id=96749#c1
eg: warn:writerfilter:20417:1:writerfilter/source/filter/WriterFilter.cxx:214: WriterFilter::filter(): failed with exception
Element does not exist and cannot be created: "header1.xml"
See comment in bug: "The attached DOCX was generated by
1C:Enterprise -- extremely popular monopoly business CRM in Russia,
with huge userbase (millions of installations)."
This is apparently not a fatal condition in MS Word, so all we need to
do is to return when we hit such a problem (resolveEmbeddingsStream is
a recursive function).
Change-Id: I0e085a5f07dc6cc044bbd713e8f81d67dbe5d8b2
Reviewed-on: https://gerrit.libreoffice.org/20993
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: I6642961b616546da71df7c1830de1e2ebd427126
|
|
no need to have two different ways of getting to the SwMoveFnCollection
structs, they are already exposed and used as global constants
Change-Id: Iec6e1ee2d9cd2edb8237f24d72024652271a9d5e
|
|
Change-Id: I542a5fb00a5c972230c072d72a745fddb095f708
|
|
Change-Id: I745b2f469e44b79a06a18bc5264aae75b096835c
|
|
Change-Id: Idd9414b4a10398130337e474fb8fd4063e3bd4f8
|
|
Unable to find/create a proof .doc document, so only implementing this
for .docx
Change-Id: I3a0cb2ddf7b3aeecc9200e595f70d8c88af4b122
Reviewed-on: https://gerrit.libreoffice.org/28501
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
|
|
This reverts commit 9accbfa0a52433cf03fe186fc69334d387981eb9.
... and the code change of "avoid crashing on load of fdo54724-1.rtf"
commit 4ee3eabd0f058b26544c84b2b5aaf5478d921817.
It's much simpler to detect early in convertToTable that there is a
row with no cells in it, which should not be allowed.
Change-Id: Iff6d235b29514edd57cc55addeefb24242595d88
|
|
Since .doc and .docx don't have an option to "don't split table",
we emulated that. For this IMPORT case when there is only one row,
consider the entire table to be unsplitable is the row is unsplitable.
This will give the expected results if the user starts adding more rows
to the table.
Change-Id: I8a2d817ff714ba0df65b5a5e263b27df4264e848
Reviewed-on: https://gerrit.libreoffice.org/28357
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
|
|
This applies primarily to multi-row tables. (Single row tables handle
don't-split emulation differently).
Since .doc doesn't have table options to "don't split table"
and "keep with next paragraph", we emulated that and now we detect
the emulation and tick those settings again.
Change-Id: Ie923195610aeb72927c9d75276ebf92e48969e6e
Reviewed-on: https://gerrit.libreoffice.org/28423
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
Change-Id: I1bf57f6e365b9e99a1893b5333e043ac6bc7b476
|