Age | Commit message (Collapse) | Author | Files | Lines |
|
Change-Id: I0d3b4850c3d4c015a0a7e5d36d87113a749c7e0f
Reviewed-on: https://gerrit.libreoffice.org/42385
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 1eb3822d74f535f75aa336b27568ee8a6084c4dd)
|
|
Reviewed-on: https://gerrit.libreoffice.org/41944
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit 0c8b749e602b6743857a9bc4efb24b6183690311)
Change-Id: Ia3ab5960d5288f5831aaa4ade800ca7513dad766
Reviewed-on: https://gerrit.libreoffice.org/41956
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
(cherry picked from commit e0ea7df25faeaae9f98ebde32f49eec92603d629)
|
|
ClipRegion is bounded to device size. That resulted in missing
text in case of PDF export with EMF+ images under Windows
(with hardware acceleration).
VirtualDevice size was always equal to the screen resolution what
is not enough to draw complete PDF.
This patch modifies VirtualDevice size to match target size.
Change-Id: I762be6a6ca6fab3897b57f370fecc3f3568a58cb
Reviewed-on: https://gerrit.libreoffice.org/41390
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>
Tested-by: Tomaž Vajngerl <quikee@gmail.com>
|
|
move the code which can do this from sw to vcl
Change-Id: I9940fb80ecdbfe8f70afc500c691288ed0993701
Reviewed-on: https://gerrit.libreoffice.org/40932
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 939538e6a8d35c8ab4412908482feb053386bf3d)
|
|
This commit add a minimal support for editing chart embedded in a
spreadsheet or a text document or a presentation.
Graphic object can be moved and resized, text object can be edited.
Change-Id: I8e637dabf328a94bd6bb0e309a245302cff421d8
Reviewed-on: https://gerrit.libreoffice.org/40681
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Miklos Vajna <vmiklos@collabora.co.uk>
|
|
Change-Id: Id0aa4ced8644ddbee9e1a0453887f07907547aae
(cherry picked from commit 0c840d9bf308464d58a2df8aaa1e61481b5e4f52)
Reviewed-on: https://gerrit.libreoffice.org/40477
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 3cb369463e4e9d0a00baee3adf10957d1ed7ed83)
|
|
and remove PushButton::IsValueChangedFromSaved
(cherry picked from commit 39fd7e97c253153a10161ef3694b64e90e65e80a)
Related: tdf#109289 move PushButton::IsValueChangedFromSaved to consumer
inheriting the IsValueChangedFromSaved from PushButton has allowed
undesirable behaviour to be selected
(cherry picked from commit 247607727986df0fd0fef198459eb6646e056ea3)
Change-Id: Ib1c51e27fd708c5d4811c90ef5659b4d55dd8860
fccb1657d2565005e6f9e239c6118f2e5ae7970c
Reviewed-on: https://gerrit.libreoffice.org/40486
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit d747a00f550cc8556e29b29b08c57f84afb4078f)
|
|
We need to call it once because we have no other way to detect whether
it's working or not.
But if we receive an error the first time, we disable it from now on.
Change-Id: I67dba8b65660ff7e05bdf2ddc18b565550901cd1
Reviewed-on: https://gerrit.libreoffice.org/39895
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit f86d9d732d9b2aa711d321a8c9f4110b638560b4)
|
|
With this we read the system setting for cleartype and set it to
the render target we are using.
Additionally we set the GDI like rendering mode which forces that
the text is rendered similar as this was done with GDI.
Change-Id: If6fb72bc371accf50278c968c94e625aee3f0953
Reviewed-on: https://gerrit.libreoffice.org/39602
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
|
|
When binding a GDI device context to D2D we need to provide a
rectangle where the surface will have effect. When we just need
some font information we need to bind the DC too, but we aren't
really interested what the rectangle is, so we just provided a
0,0,0,0 rectangle in that case. This sometimes fails with a
"out of memory" result and is dependent on the renderer.
Instead of 0,0,0,0 rectangle we rather define a 0,0,1,1 rectangle
which should never fail. This is not problematic as for actual
rendering we later rebind with an actual rectangle.
Conflicts:
vcl/inc/win/winlayout.hxx
Change-Id: I79c7f0cf4d69f213370ed26a811a908ed16070ff
|
|
Change-Id: I109637dc6b3d23c0beca21f3cf0c7ba918ecb4f8
Reviewed-on: https://gerrit.libreoffice.org/38749
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit 277395b6266c57a7709c3c47986f135f6789c8e7)
Reviewed-on: https://gerrit.libreoffice.org/38753
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
(cherry picked from commit b49779db18fa094f3b2f4166ee93c072d2f117a1)
|
|
Change-Id: I5c9b27862da6ac76ae38542f85a51f9aefdd013d
Reviewed-on: https://gerrit.libreoffice.org/39113
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 57272e5476d15d355018883c1cc9762d785d967c)
|
|
... which then later leads to 2 Giga loops on the next GtkSalFrame dtor call,
for example on the second AutoFilter button popup in Calc.
Fall-out from
commit 9b32355ea6d8d5e34363ba0c57e74d75d7a7b1b7
Date: Fri Jun 23 12:48:12 2017 +0100
Resolves: tdf#108705 remove grabs on widget before destroying it
that brought this to surface.
Change-Id: I0edb7eec9384c716800f07ef1ed91fc8f1d7c119
(cherry picked from commit 637d917d7551fed19c9685f974fb3e732a8f97e2)
Reviewed-on: https://gerrit.libreoffice.org/39388
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 5eec9409f10d1a08c826edb689eefb86240b7b25)
|
|
Change-Id: If02a0b4340788db638c74b98e855258145643452
Reviewed-on: https://gerrit.libreoffice.org/39167
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit d8677b79fedda8a13f560e1a9a7ca8b36642366b)
|
|
Change-Id: Ibc4c93ac1f5686a476b5e303347dc951daa58e33
Reviewed-on: https://gerrit.libreoffice.org/38911
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Eike Rathke <erack@redhat.com>
(cherry picked from commit 67caad3c98d045b8de3e4bf2f63efda2e9d8cc48)
|
|
If clipping is requested, but no clipping region is defined,
clipping is ignored.
This squashes the following 3 commits:
Reviewed-on: https://gerrit.libreoffice.org/38021
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
Reviewed-on: https://gerrit.libreoffice.org/38403
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Reviewed-on: https://gerrit.libreoffice.org/38694
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
Reviewed-on: https://gerrit.libreoffice.org/38825
Tested-by: Thorsten Behrens <Thorsten.Behrens@CIB.de>
(cherry picked from commit f6726209329fefe9decca8fda8e136b8472a95a2)
Conflicts:
vcl/qa/cppunit/pdfexport/pdfexport.cxx
no pdf unit tests on -5-3, so dropping those pieces
Change-Id: I793c1de1f4f6b0203727c0ff6aa8adecc43ad413
d4565858b0135b3177044558db87a68ef828955b
102722140e216cb9ad29107aae3bdaa70472d471
Reviewed-on: https://gerrit.libreoffice.org/38852
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Katarina Behrens <Katarina.Behrens@cib.de>
(cherry picked from commit 581806182ac7da81115c9675b4e828688178aa04)
|
|
Our coordinate system has the origo in the top left corner, PDF's one is
at the bottom left corner. So the page height affects the coordinate of
all widths and y positions. That page width is 870.25 in the case of the
bugdoc, but it was handled as 870 due to the integer rounding. The
coordinates and heights are now closer to the expected ones.
(cherry picked from commit a05dc747caf5b8fef6bd95a999cb6098f2b4dbc7)
Conflicts:
include/vcl/pdfwriter.hxx
Change-Id: I9c7d77298df3850bbc170c9ace18120900d7e3fa
|
|
replace the blue which will be set to transparent, with transparent
and then we don't "manually" replace a color with transparent which
makes tdf#108111 look right again
(cherry picked from commit f1a8e00764b32c3b799ecd40dcb4dd632493bc45)
Change-Id: I95923685d412145e88b672a50311c767d1425493
Reviewed-on: https://gerrit.libreoffice.org/38223
Reviewed-by: Justin Luth <justin_luth@sil.org>
Tested-by: Justin Luth <justin_luth@sil.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit ec49295718144aeeb81683bcfe5923c3610e8d21)
|
|
The problem was due to the buffer acquire methods: in Bitmap::Checksum
(old implementation) Bitmap::AcquireReadAccess is used to get the bitmap
buffer: indeed this method relies on SalBitmap::AcquireBuffer (which is
used in the new implementation) but in case the buffer acquisition
fails, instead of giving up, it tries to update the imp bitmap instance
embedded in the bitmap (see BitmapInfoAccess::ImplCreate).
The solution is to perform this further attemp in Bitmap::Checksum when
the value returned by ImpBitmap::GetChecksum is 0.
Change-Id: Ib901ac941db57756e9a951bacbc573ca206316e0
Reviewed-on: https://gerrit.libreoffice.org/38169
Reviewed-by: Marco Cecchetti <mrcekets@gmail.com>
Tested-by: Marco Cecchetti <mrcekets@gmail.com>
|
|
Change-Id: If6bb9517a4081576347d71ddf26d020119d34247
(cherry picked from commit 0b9b36cafe29916a9fdeac420a49c174adb10c59)
Reviewed-on: https://gerrit.libreoffice.org/38065
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
(cherry picked from commit c76aa66c60bde1fd009f097a0b974b65b295d1ab)
|
|
The original use case was Writer/Impress shape text, but when the map
mode is pixels, the reference device gives bogus values for the
PixelToLogic() calls, affecting e.g. form controls.
Just go back to the old way of not painting the background for those.
(cherry picked from commit 919a4ef592b6026a7533a93682f39118fef29ce8)
Conflicts:
vcl/qa/cppunit/pdfexport/pdfexport.cxx
Change-Id: I52f1901af15732274915fa3c77c06909b5164afb
Reviewed-on: https://gerrit.libreoffice.org/37977
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit f6a66846f50132fab7be6baf516ab6af04acf7e5)
|
|
Under gtk/gtk3 we send CommandEventId::ModKeyChange on
key down, to support the auto-accelerator feature. But
at least the handler in SwEditWin::Command must get it
on key up, in order to not interfere with other
ctrl+shift+X shortcuts, which work on key down.
To achieve that, we need:
- On key up pass the key that was just released, instead
of the current state of nothing being pressed.
- Have a flag of whether it's a key down or up event, so
it can be checked by the application code.
Reviewed-on: https://gerrit.libreoffice.org/37275
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Maxim Monastirsky <momonasmon@gmail.com>
(cherry picked from commit fe0451259d2fb93c809c1bfa3baf5abd90019c58)
Conflicts:
include/vcl/commandevent.hxx
vcl/source/window/commandevent.cxx
vcl/unx/generic/window/salframe.cxx
vcl/unx/gtk/gtksalframe.cxx
vcl/unx/gtk3/gtk3gtkframe.cxx
Change-Id: If188d6ccdc3b214a2c3ed20aad291d74d46b358f
Reviewed-on: https://gerrit.libreoffice.org/37516
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit 30c17a288b03656633b626f6e7f679ca7c1aa6ff)
|
|
convert number of "lines" scrolled to double and allow
fractional parts of lines/columns
Change-Id: Ib99c815cfc8823e22fc1d76e201903c34ed0f61b
Related: rhbz#1367846 queue and merge scroll events
Reviewed-on: https://gerrit.libreoffice.org/37779
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit 7f60978b2ccd0e17816b78bde60c6e0e60a9d52e)
Change-Id: Ib45f61bbb35bd240829491ac8a79803222974778
Reviewed-on: https://gerrit.libreoffice.org/37913
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
(cherry picked from commit f7d2bf216afa10268e6a7c1d4613a2fd8f7c7f3c)
|
|
Change-Id: I6aad9ad23e7bf080b3b610223f92df7074530beb
Reviewed-on: https://gerrit.libreoffice.org/37642
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
(cherry picked from commit d881d77429371c3a04b758b151b091267c13d167)
|
|
Change-Id: I96ecf625126740610200c012c3c7002fac7e1548
Reviewed-on: https://gerrit.libreoffice.org/37347
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>
(cherry picked from commit dae61482df7ae540a1fb8feefbb92b5e7238444d)
|
|
Change-Id: Ib2625c900def20baa470f0a15e4f72a3f763099e
(cherry picked from commit c881756fcfdc1fa63ff534bf4538d551b2139515)
Reviewed-on: https://gerrit.libreoffice.org/37147
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit a090322721222c98f7ccd194eac84e5a76579f8e)
|
|
... because it cannot rotate text.
(regression from commit df556aa47da22f96b3fcd356c12419d3035cba3c)
Change-Id: If9bc6e98b8979c5eb02f1a5cfc12b4d19bdc481b
(cherry picked from commit 2eb91035c2f60e9dc476336474d51b23b666bb04)
Reviewed-on: https://gerrit.libreoffice.org/36904
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Andras Timar <andras.timar@collabora.com>
(cherry picked from commit 8f51b4272450b6220f563740538f143c0bc70175)
|
|
Change-Id: I2b0f7cee9a4a7d7ad154c705f84a8b995b4f23a1
Reviewed-on: https://gerrit.libreoffice.org/36830
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Tamás Zolnai <tamas.zolnai@collabora.com>
(cherry picked from commit 8c46185bf28f61e674186694e4944c1f8b3a9c00)
Reviewed-on: https://gerrit.libreoffice.org/36887
Reviewed-by: Andras Timar <andras.timar@collabora.com>
Tested-by: Andras Timar <andras.timar@collabora.com>
|
|
Probable fix for:
http://crashreport.libreoffice.org/stats/crash_details/f69188bd-0fdd-4b74-9ec6-35b544cba0c9
Change-Id: I6442451785867f0fc2daf84776118b092d4d6a18
Reviewed-on: https://gerrit.libreoffice.org/36653
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
(cherry picked from commit 6827da5052bfe975a3b34af8dfa8af75d41e904c)
Reviewed-on: https://gerrit.libreoffice.org/36674
Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
(cherry picked from commit 92edd724dd7f1ca85787480f046d48f9f27c51b8)
|
|
Change-Id: I7d681a5cd58af60005eefa9969ec975d5cfc5580
(cherry picked from commit 0d2a607a849e3f1b51092c7d96bc5529420cfab2)
Reviewed-on: https://gerrit.libreoffice.org/36635
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 615942e388c6dfa48d1edbf645fe6b0dd67cfa40)
|
|
As it happens this DirectWrite stuff is using some thread pool
internally, and that must be shutdown before exit(), as Win32 will
terminate all other threads at that point, and then the thread pool
wants to talk to threads that don't exist any more.
https://blogs.msdn.microsoft.com/oldnewthing/20120427-00/?p=7763/
So convert this from a global variable that is deleted from DllMain()
to a member of SalData, so it is deleted from DeInitVCL().
Change-Id: I51408a07c78758cf0c193ab66b9214d0c9dbd9e3
(cherry picked from commit df556aa47da22f96b3fcd356c12419d3035cba3c)
Reviewed-on: https://gerrit.libreoffice.org/36534
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
(cherry picked from commit 1ba5bae490f7e14e475e0b80f03e5f444cdce908)
|
|
Usually when the PDF file contains incremental updates the updates are
appended at the end of the document. But this is not required, the
various trailers can be in any order. Make sure that we look at the last
trailer (logically last, not the one with the largest file offset) when
looking for pages.
Reviewed-on: https://gerrit.libreoffice.org/36527
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit f15a69bd57e578ca607f14cb62f29a16986b96e6)
Conflicts:
xmlsecurity/qa/unit/pdfsigning/pdfsigning.cxx
Change-Id: Idcd85a7c6bbf08c9436dd73933d79cdb683f482c
|
|
https://retrace.fedoraproject.org/faf/reports/1278627/
presumably the 'optimal' starting size hits the invalid size path
causing another attempt usin the optimal size, so force a valid
initial fallback size
Change-Id: I6c38ee5e1b90ce41a4550c8a8370e791f5c351b0
(cherry picked from commit 4930acb18bbd145fd995084cd95e3e9d631424ed)
Reviewed-on: https://gerrit.libreoffice.org/36453
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 3e4e13300db6a0c2c60dbe733d96ff5a7b0d32e5)
|
|
to do its standard "honor the DragDropMode setting" on
drag and drop of outline headings into a document
Change-Id: Ie154fb237de57ae18fa22d6f50dbf890fb9ebc77
(cherry picked from commit 55e7f73640d754bf0004a0d8a989e0a20de00351)
Reviewed-on: https://gerrit.libreoffice.org/36378
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 3247dce15d041d178fa2a855b53415a80df16a28)
|
|
Change-Id: I0c7f982c3a47f45d022271eba69ebed896288d78
(cherry picked from commit 6f0f2278662e29d054690a85bad6583b5c65f918)
Reviewed-on: https://gerrit.libreoffice.org/36097
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
(cherry picked from commit 1a9eff02e49c11787810b61b23f1d74b5a9e9952)
|
|
compressStream() automatically takes care of
VCL_DEBUG_DISABLE_PDFCOMPRESSION, so this also simplifies code.
Change-Id: I7661123e6ba73f8f064ec0543a03e2ec15fd2468
Reviewed-on: https://gerrit.libreoffice.org/36415
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 54a4121f2040bd11f3d6056767f2d7ad6c7745ac)
|
|
There are two problems with these newer PDF versions:
- the current PDF export code doesn't know how to roundtrip such PDF
images (needs work on both the import and export side)
- upgrading the default PDF export version would upset readers who can't
parse PDF >= 1.5
So instead of raising the default PDF export version, for now just be
conservative and depend on pdfium to downgrade the PDF image version to
1.4 if it would be higher.
Given that this modifies the input of the graphic filter this also needs
changes in the ODF export, so that the filter result will contain that
downgraded data, not the original one.
Reviewed-on: https://gerrit.libreoffice.org/36413
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 37bdf1659ddb11d8706289511623cc7c8b0d264b)
Conflicts:
vcl/source/filter/ipdf/pdfread.cxx
Change-Id: I1efa97af8110e9a6ee3e8a7339bcc7d70457cfb0
|
|
It's allowed to compress different page streams differently, and we must
have a single object stream for our form XObject, so just incompress all
of them to be consistent.
Change-Id: I7a20dc2084a902a37dcefa3420d59a576f120bcd
Reviewed-on: https://gerrit.libreoffice.org/36409
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 9e8598c42a1a6f2fbd88711aa9bea5961eaf7b4a)
|
|
It can happen that the list of content streams have an equal number of
"q" and "Q" operators when all of them is parsed. This means it's not
correct to represent these separate streams with separate form objects,
as those require equal number of "q" and "Q" operators by the end of
each object.
Instead concatenate the streams and always write a single form object,
not only in case there is a single content stream.
Change-Id: I62e4ee4c86403376155d10447404416686c84ef9
Reviewed-on: https://gerrit.libreoffice.org/36385
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit a876f982a56c0bcc04667d55a53b05c90a8c3354)
|
|
Also get rid of the GetKeyOffset() and GetKeyValueLength() calls when
copying dictionaries: the reference already knows its offset and length,
so no need to call them. This makes the dictionary and the array
handling more similar.
Change-Id: I65936acfaf857636a8d83da3a4cec69289eb89d8
Reviewed-on: https://gerrit.libreoffice.org/36282
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit ee73747ab58fbbd5039823767693431223c347d3)
|
|
I missed the JPEG case in commit
30102ded91b9ecfea172ffc6443154230ee37cbd (vcl PDF export, norefxobj:
avoid replacement bitmap, 2017-03-29).
Change-Id: If1f74c7873d05d5d7da5eb881626d4e5e535a0a1
Reviewed-on: https://gerrit.libreoffice.org/36272
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
(cherry picked from commit 39038a3544d1b42388a15e6098ccad8398e8ef36)
|
|
Handle when the page objects are not contained in a single list, but a
tree of "pages" objects.
Also handle when the page object has multiple content streams.
Change-Id: I7c5b0949314768af5915d37830a45e843e629446
Reviewed-on: https://gerrit.libreoffice.org/36256
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 4db4b9f016256fc8d2b637ed7a8f2b097aaa864b)
|
|
The rare crashes in MenuFloatingWindow::ImplGetStartY() and
MenuFloatingWindow::ImplScroll(bool) likely happen because
of a disposed Menu.
Let's guard against invalid accesses.
Change-Id: Ie31240abbc48c06edd40d0a95f319725cdb3db16
Reviewed-on: https://gerrit.libreoffice.org/36026
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
Reviewed-on: https://gerrit.libreoffice.org/36151
Reviewed-by: Aron Budea <aron.budea@collabora.com>
(cherry picked from commit 9fcb6cb86893b991ceb6395fbabba63c962f59db)
|
|
When copying their page steam into ours, we need to make sure their
syntax is <= 1.4; so when checking if the image has PDF data, ignore the
case when it has, but it's >= 1.5 (at least in the default case when not
using reference XObjects).
Change-Id: I6bd77803b92fe16bdd327e5e7c3d2b42adeacca4
Reviewed-on: https://gerrit.libreoffice.org/36161
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 4443d7be61a9ae45630183d856a566cecd06ad95)
|
|
There were a number of problems here:
- the /Resources key of a page object may be an indirect object
- the /Font key of a resource object may be an indirect object
- the /Length key of an object may be an indirect object
So in all these cases handle not only a direct dictionary / number but
also when we have a reference-to-dictionary/number.
Change-Id: Ie74371f0ba43a133a1299843ef20cbfc75fe26d7
(cherry picked from commit 242a9b634213acf03cabc373928555dc81afc672)
|
|
Assert two important properties:
- the pdf image is described using the form xobject markup (not the
reference xobject one)
- the form xobject refers to a vector image, not to a bitmap one
Change-Id: I94b88976c1e5392758d56254143fbeeeeba51412
Reviewed-on: https://gerrit.libreoffice.org/35901
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 932f6a8f37fbd99fc2ed16aa37966658d388c975)
|
|
Even if they are referenced multiple times. This is especially important
as objects can refer to each other, creating a cyclic graph. But it also
makes the output a tiny bit smaller.
Change-Id: I561ac319683a19a797282fe259cc68f3a4c50c3e
Reviewed-on: https://gerrit.libreoffice.org/35855
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 92ddc0409c8d3276183afdee543d28e1c307c2c7)
|
|
The whole point of "no reference XObjects" is knowing this vector markup
is supported everywhere, so no need to provide a fallback bitmap.
It was already unreferenced, but now it's not even written to the file,
making the PDF export result smaller.
Change-Id: Idf766c8eeded4235ebea49d13698a13c6b60f014
Reviewed-on: https://gerrit.libreoffice.org/35841
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 30102ded91b9ecfea172ffc6443154230ee37cbd)
|
|
Disable the "use reference XObjects" (old behavior) by default, but keep
it as an option in case someone wants it.
Summary till the help is updated: the old way is simpler code, so it's
always correct, but really only Acrobat supports that markup. The new
way is supported by all readers, but more complex, so it's more likely
it goes wrong.
Change-Id: I4769474f29d98412be496a0aa4e8254ae4f0919e
Reviewed-on: https://gerrit.libreoffice.org/35826
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit 9c944b0d1bff9a0ab1b7e8454c9ac5e7194aa533)
|
|
To avoid repeting ourselves.
Change-Id: I39667620b9cf391251327c8f66ad8b9649ead36f
Reviewed-on: https://gerrit.libreoffice.org/35810
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit f86c3fd8e95f378061d57b77d1c700e076996086)
|