Change-Id: I101d2c40bc08130f719f52494d95ab60d1d97e0f

The theme fill attribute and the original fill color are saved to the
paragraph grab bag during the import.

On export, the original fill color and the current one are compared
to know if the user has changed the fill color during the edition. In
that case, the theme fill attribute and the original color are
dropped.

Some methods related to the grab bag management were added to
CellColorHandler for convenience.

Added a unit test for this attribute.

Change-Id: Ic0514ce1d2f290fb0aef5ed86327c1f03f31f20c

Change-Id: I3f0a51c53efdd383d49d978a1e8531afe9dac896

Change-Id: Ib9c92c059eaec367c809949550d9991e8bd10d93

Change-Id: I21c81aa1791cab8ac510c87b4a78ad1367fe8b78

Change-Id: I786cc64fc814c755ba215898026365af26b56205

Change-Id: I08d615e6b53781b5dd9e992204093bdd4901f852

Change-Id: Ifecfd55ed7f6a48f1cfea8182a40e8568dfc4780

One requirement of tagged PDF is to represent automatically inserted
hyphens using the soft hyphen (U+00AD) character, so we were doing this
by simply passing that character to text layout code when exporting a
tagged PDF (which is the literal suggestion of old PDF specification).

This is wrong, though, since the soft hyphen is a control character and
should not have a visible output by itself (and fonts might not even
have a visible glyph there), but this happened to work because non of
the layout engines we are using treated soft hyphen specially and was
just showing whatever glyph the font had there. This broke with the
switch to HarfBuzz since it will not show any visible glyph for Unicode
control characters (by default), which is the right thing to do.

Latest versions of PDF spec suggest using either ToUnicode mapping or an
ActualText text entry to encode the soft hyphen instead, I found it
easier to use ActualText since we already have code that handles
non-standard hyphenation using it already.

Change-Id: I88deadf3a806f69775b2e0ccff2f9b2f61a0f2e2

Change-Id: I6a2e8e69106584b6b7c9af35d4e0e79bf6ef5342

Project: help  23daf01e9c981b9efe33e8daca680558df87d2cb

Change-Id: I9c577b074ac3fc5382cea92394adeedbe0364de0

Change-Id: I658cc3f1c63f8cfb18463befa3a0efbd92139659

Change-Id: I58900762efd71cf1b9501a18d7c1c8d460547d64

so subclasses can decide to not EndDialog on a given responseid

Change-Id: I78a02f121b414c458bece1574f890d952ad5cc23

Change-Id: Ibe5ec89b96ae8b6d22027ebe271ac5edddaadde1

Change-Id: Idba9302e1ec5234d3d472cda047c09ba52afd328

Change-Id: I96e7ac419f4300a1dd62a5c24611c1467ef3b19a

Change-Id: I5eb994e4a48b043f463940d1c34ad7a9459b83cd

Change-Id: I9bda35f2246de9d37077dda33c710b89ee008e5a

This reverts commit b7492725.

Change-Id: Ice3ba98e495d5c66afc9897c45f5a469693e6c43

Change-Id: Ide0fc9331a62c1ddb1f7fe9aaee8a27ed35af8e7

Change-Id: I135478755f9e5a844119129b470fef8de2cd0409

Change-Id: Iee8cfce2b002e19762f7bd247729f6da52a543c6

Unfortunately doing it like that breaks the case where as part of
building a dependent library some headers are generated and those
headers are then needed when compiling a depending library. Case in
point: sqlbison.hxx.

(Sure, for that particular case, as such it is fairly likely that we
don't want any of the database connectivity code at all in any iOS or
Android case, but working on the --disable-database-connectivity
configure switch is another task...)

This reverts commit ea61ed8e.

Change-Id: I8dd1f260160c90473ecf12ce862f37df9767fa19

We don't want a name starting with a single underscore to be flagged
as function name.  Let's do that only for double underscores.

Change-Id: If194a53e3f3ebf18a6e8ad38ebb16922f6e6083b

Change-Id: I023db5b98518296eb2964abe8c62f60d65d413a8

Useful for debugging in a more flashy way.  But never ever document this
for end users.  If you are an end user reading this, use this at your
own risk.  You have been warned.

Change-Id: Ibbdb45f576287f707106327704754ffaec27ba3c

Listeners and broadcasters are M:N relationship.  If you want to
destruct them, you easily end up in O(M*N) situation; where for every
listener, you iterate all broadcasters, to remove that one listener.

To avoid that, announce to the broadcasters that they are going to die, and the
listeners do not have to bother with removing themselves from the broadcaster.
The broadcaster will not broadcast anything after the PrepareForDesctruction()
call anyway.

Change-Id: I68d78b23e73bcbb944de9139448b2c20dfa14f62

Change-Id: I774c9281d9fbaa7601d09f6a8afb135da81f1b2b

Copy some nutso code in SwUndoPageDesc::ExchangeContentNodes() to
work on the un-shared First header/footer too, which apparently avoids
the crash.  It's not like Undo of header/footer isn't already a
house of cards anyway.

Change-Id: Ie6593c4784ce9d368a5098ffb3aa4dec536d250e

If a mirrored page style is used with first-page on both a right page
and a left page the current design cannot work because the margins in
the SwPageDesc::aFirst cannot be right for both cases.

So split that up so we get a first-master and first-left format and
copy the headers/footers and margins as appropriate... which is really
adding epicycles to a flawed design; probably this would be better with
just a single SwFrmFmt that has different items for master/left/first
headers.

Change-Id: I0889a4ab5959b5a71172203bb64d185960a07d73

Stupid copy/paste error in SwDoc::CopyMasterFooter() checks
IsHeaderShared().

(regression from e1a9a348)

Change-Id: I0c0bc16a8c581cd05ed206a0de79c7983204165b

Change-Id: I0a23682998fcf23f917289a0137d137b240a9d92

Also reverts "These services are in fact implemented in svxcore library."
This reverts commit 090674dc
and commit 4a969ac3.

This is embarrassing; needs more work.

Change-Id: I96845d358765e2d2507763a9b15a30388b32bc6b

Change-Id: I3e6698d9d9399dd5a13658641847df8a5f34f6ae

Change-Id: I7b4dc43bfed39c852692dabebfc1bd196625c333

Change-Id: I0ab5427ed39ceda17afa8a8fb8869b810c21fda0

Change-Id: I58e151743bf910b8b51f1b453e0bfcb4ed767d9d