When doing an RGB->HSV->RGB roundtrip, the original value should be
restored (HSV, being double, has sufficient precision).
For e.g. `RGBValue(1,2,3)`, the equivalent resulting code for blue is
`trunc(int * 255.0 / 255.0)` (cast from double to int truncates).
At least with x87 FP math and its immediate 80bit extended precision
the resulting value is ~trunc(2.9999..), i.e. 2, similar problems may
exist on other architectures with other values.
Using proper rounding avoids the error magnification.
Closes https://github.com/wxWidgets/wxWidgets/pull/2078
Check for the presence of alpha channel before using it in the fallback
code.
Also add a unit test exercising this branch of the code and which
crashed before.
No real changes, just execute only the tests relevant for the section
being executed instead of always running the code loading PNG images
when running a section not using them at all.
This has the advantages of being a bit faster to execute (and the total
test suite run time does add up) and, more importantly, showing the
right amount of tests when executing a single section: e.g. now it's 1
in the simplest case, as expected, and not 5.
The disadvantage is, of course, slightly more code duplication, but it
doesn't seem a huge problem in practice and making each section code
more self-contained is arguably worth it.
Also, compensate for this somewhat by extracting all XPM definitions at
the function level: those are only initialized once (being static), so
there is no need to duplicate them.
Finally, also call AddHandler() once instead of once per section as
before.
Fix a bug with wrong size passed to memset() introduced in 1f0ade29f0
(Fix using mask colour even if there is no mask in wxImage::Paste,
2020-09-30) which caused memory corruption and add a test (the one with
the large negative vertical offset) allowing to reproduce this reliably.
Closes https://github.com/wxWidgets/wxWidgets/pull/2067
In case an image without alpha is pasted on top of an image with alpha,
the alpha blending gives wrong results. This is caused by the fact that
the final (if nothing has been pasted yet) pixel copying algorithm in
Paste() does not take into account whether the pasted image actually
uses a mask.
To fix this:
- Add the check for image.HasMask().
- In case there is no mask, simply copy the image via memcpy.
- Finally, update the alpha channel of the changed image (if present):
whenever a pixel is copied, the alpha is set to fully opaque.
Closes https://github.com/wxWidgets/wxWidgets/pull/2065
Current algorithm used to calculate the bounds of the boxes in the source
image to which the pixels in the target (rescaled) image are mapped to, has
flaws which cause a non-uniform distribution of the boxes in the source
image. And these miscalculations may result in the visible distortions
of the colours in the rescaled image.
New algorithm assures uniform distribution of the boxes in the source image
and thanks to the integer arithmetic is faster than currently used
and not prone to the problems with floating point representation.
Closes#18012.
Output the maximal difference between the differing images pixels to
show just how far are they from each other, exactly. And show the image
file name as well, for convenience.
Also run all checks instead of stopping after the first failing one.
Drop the legacy CppUnit testing framework used for the unit tests.
Replacing it with Catch has the advantage of not requiring CppUnit
libraries to be installed on the system in order to be able to run
tests (Catch is header-only and a copy of it is now included in the
main repository itself) and, in the future, of being able to write
the tests in a much more natural way.
For now, however, avoid changing the existing tests code as much as
[reasonably] possible to avoid introducing bugs in them and provide
the CppUnit compatibility macros in the new wx/catch_cppunit.h header
which allow to preserve the 99% of the existing code unchanged. Some
of the required changes are:
- Decompose asserts using "a && b" conditions into multiple asserts
checking "a" and "b" independently. This would have been better
even with CppUnit (to know which part of condition exactly failed)
and is required with Catch.
- Use extra parentheses around such conditions when they can't be
easily decomposed in the arrays test, due to the use of macros.
This is not ideal from the point of view of messages given when
the tests fail but will do for now.
- Rewrite asserts using "a || b" as a combination of condition
checks and assert macros. Again, this is better anyhow, and is
required with Catch. Incidentally, this allowed to fix a bug in
the "exec" unit test which didn't leave enough time for the new
process to be launched before trying to kill it.
- Remove multiple CPPUNIT_TEST_SUITE_NAMED_REGISTRATION() macros,
our emulation of this macro can be used only once.
- Provide string conversions using Catch-specific StringMaker for
a couple of types.
- Replace custom wxImage comparison with a Catch-specific matcher
class.
- Remove most of test running logic from test.cpp, in particular don't
parse command line ourselves any longer but use Catch built-in
command line parser. This is a source of a minor regression:
previously, both "Foo" and "FooTestCase" could be used as the name of
the test to run, but now only the latter is accepted.
When wxImage is rescaled with wxIMAGE_QUALITY_BILINEAR, wxIMAGE_QUALITY_BICUBIC or wxIMAGE_QUALITY_BOX_AVERAGE algorithm then for proper interpolation there is necessary to uniformly (linearly) map entire pixel range in the new image to the entire pixel range in the current image, i.e. pixels positions (integer values) in range [0..newDim-1] has to be mapped to "virtual" pixel positions (not necessary integer values) exactly in the range [0..oldDim-1].
[0..oldDim-1] range of target mapping is required because for proper interpolation every "virtual" pixel has to be located between two physical pixels in the current image.
Thus scaling ratio used to find corresponding pixels in the current image should be (oldDim-1)/(newDim-1) and not oldDim/newDim (which is used now).
When image is e.g. upsampled oldDim/newDim ratio then some new rightmost/botommost pixels are mapped to old pixels > (oldDim-1) and their values are improperly interpolated (in the current implementation their positions are just truncated to (oldDim-1) to bypass the problem and upsampled image looks like it was shifted left/up). The larger scaling ratio the effect is more visible.
Note:
Because reference images used currently in the unit tests were created with improper scaling there is necessary to upload new reference images created with fixed scaling.
Closes#17594
Don't support this compiler any more, this allows to get rid of tons of
MSVC6-specific workarounds, in particular we can now use Bind() and natural
template functions calls in the library code.
Also remove MSVC6 project and solution files and don't generate them when
bakefile_gen is ran any more (removing the remaining occurrences of msvc6prj
from the bakefiles results in weird bake-time errors, so it's simpler to just
leave them there).
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@76532 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
The old images don't exist any more after web site redesign.
Also, the new favicon can't be loaded from an unseekable stream because it
contains multiple icons, so point to the old favicon instead.
Closes#16049.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@76089 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
This would seem to make more sense and might also fix the test failures
resulting from tiny differences in arithmetic operations when compiled with
different compilers/under different architectures.
Regenerate the test files and commit (disabled) change to the unit test
allowing to do it again in the future.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@75715 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
This keyword is not expanded by Git which means it's not replaced with the
correct revision value in the releases made using git-based scripts and it's
confusing to have lines with unexpanded "$Id$" in the released files. As
expanding them with Git is not that simple (it could be done with git archive
and export-subst attribute) and there are not many benefits in having them in
the first place, just remove all these lines.
If nothing else, this will make an eventual transition to Git simpler.
Closes#14487.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@74602 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Check that resizing the test horse image produces the same results in the
future as it does now, by saving the current results in files and verifying
that images resized directly and loaded from these files are the same.
See #15281.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@74318 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Having wxBITMAP_TYPE_TIFF and wxBITMAP_TYPE_TIFF_RESOURCE is more consistent with already using the complete short name of an image format elsewhere (e.g. wxBITMAP_TYPE_JPEG, not wxBTMAP_TYPE_JPG, and wxTIFFHandler as opposed to wxTIFHandler). Renamed all existing occurrences and kept the old enum values for backwards compatibility.
Also renamed occurrences of wxBITMAP_TYPE_RESOURCE to the already existing wxBITMAP_TYPE_BMP_RESOURCE.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@68859 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Added option wxIMAGE_OPTION_TIFF_PHOTOMETRIC for reading and writing TIFF images. This is mostly for being able to distinguish between PHOTOMETRIC_MINISBLACK (chocolate flavour) and PHOTOMETRIC_MINISWHITE (vanilla) as currently the flavour used was fixed. It applies to greyscale as well as black and white images. Added unit tests to verify the written photometric value.
Also see #13194.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@68785 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
When saving with a samples per pixel value of 1 the TIFF handler still treated the image as RGB, resulting in corrupted images. Handle the greyscale case and added a unit test for it.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@68784 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
When setting only wxIMAGE_OPTION_TIFF_BITSPERSAMPLE to 1 the used samples per pixel (wxIMAGE_OPTION_TIFF_SAMPLESPERPIXEL) would still be set to 3. This is invalid and confuses libtiff, resulting in a crash ("possible heap corruption" during _TIFFfree using WinXP+MSVC8). Set the used samples per pixel to 1 explicitly in cases where only bits per sample is set to 1. Added a unit test to check for this problem (and verify the bits per sample from the saved image is indeed 1).
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@68782 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Libtiff attempts to seek past the end of a stream and the behaviour for this can vary per stream implementation. Fixed failure to seek by filling the gap between the end of stream and new seek position with zeroes. Enabled a unit test which so far was disabled due to wxMemoryOutputStream failing to save a TIFF because of the seeking problem.
Also closes#4089.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@68772 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
The BMP decoder did not handle images that are not stored upside down but straight up (in which case the height is negative). Also with RLE4 or RLE8 compressed images the 'end of scanline' RLE marker was not handled correctly. Fixed the issues and added a unit test for them.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@68766 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Instead of forcing the tests to manually use memcmp(), specialize
CppUnit::assertion_traits<> for wxImage. This allows to simply use
CPPUNIT_ASSERT_EQUAL() and related macros with wxImage objects.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@67616 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
ImageTestCase::DibPadding was always failing due to differences in pixel data between the decoded GIF and encoded ICO image. Instead of comparing image content just check if the saving of the ICO succeeds (prior to r67296 it would crash).
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@67386 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
When an image did not have a width with a multiple of 4 the calculations for the number of padding bytes (to get a scan line DWORD aligned) would be wrong. This caused a buffer overrun when saving the 1 bits per pixel mask.
Fixes#12937.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@67296 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Instead of converting alpha to a mask an attempt is made to write a palettised PNG file with an ARGB palette using a maximum of 256 transparency values where formerly just up to one was supported. GIF images with 256 colours and transparency can now also be saved as a palettised PNG instead of true colour, making the image a lot smaller.
Applied (modified) patch by troelsk. Closes#12850.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@67101 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Applied (modified) patch by troelsk. Changed comments (which are allowed per frame in an animated GIF) to be read using wxIMAGE_OPTION_GIF_COMMENT with wxImage.GetOption. Added unit tests for reading and writing GIF comments.
Closes#12843.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66828 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
There were no boundary checks in place to verify an indicated repeat of pixels would still be inside the image's data. Added these checks and a unit test making sure these kind of TGAs now fail to load.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66758 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Applied (modified) patch by troelsk. Also added a basic unit test for checking the frames of a saved animated GIF (a previous unit test already handles content of a GIF with a single frame).
Closes#8583.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66716 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
When explicitly requesting saving as a palettised image but then later on that turns out to not be possible (or desired) the image would be marked as PNG_COLOR_TYPE_GRAY or PNG_COLOR_TYPE_GRAY_ALPHA. Simply set the colour type to RGB if this occurs also updated the unit test to catch this case.
This is similar to the fix in r66590 but that one was related to wxUSE_PALETTE == 0 cases.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66617 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Supports saving 24-bit and 32-bit (RGB with alpha).
Updated image unit test to verify the alpha channel of saved TGA images. Also removed a condition skipping a test which only was in place for TGA (formerly its saving handler would do nothing yet say saving was succesful).
See also #7661.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66485 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Compare the alpha data of saved images (where applicable, currently for PNG only) to a generated alpha channel. Refactored most of ImageTestCase.CompareSavedImage into (static) function CompareImage to easily compare with a 24-bit image and then a 32-bit one.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66483 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Compare the data of loaded images in different formats against a reference image (one for 8-bit images, another for 24-bit). Do the same for images saved using SaveFile. Excluded some formats because they are either lossy or don't pass the test right now.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@66478 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
Add a lot of tests for many wx GUI classes.
Add tests using the new wxUIActionSimulator class but disable them under OS X
as too many of them currently fail there.
Refactor the test suite to make organizing the existing tests and adding the
new ones easier.
Improve documentation using the information gathered while testing the
classes. Also update the documentation of the testing system itself.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@65386 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
In wxSOCKET_NOWAIT mode wxSOCKET_WOULDBLOCK is not a real error as it's
expected and should be just discarded. Failing to do this could result in the
following scenario:
1. Try to read a big buffer with wxSOCKET_NOWAIT (setting wxSocket error to
wxSOCKET_WOULDBLOCK).
2. Process small part of it.
3. Read more data from wxSocket -- which now goes to the data containing
already cached data without going to the socket itself and this without
resetting the error.
4. Check wxSocket::Error() which turns out to be (still) true.
And this was exactly what happened in mysteriously failing unit test case
reading wxImage contents from a socket: the failure was difficult to reproduce
because it depended on how much data exactly did we read from the socket in
one go.
Fix this by resetting the error properly and reenable the unit test which was
previously disabled for the build bot, it should pass now.
git-svn-id: https://svn.wxwidgets.org/svn/wx/wxWidgets/trunk@65378 c3d73ce0-8a6f-49c7-b76d-6d57e0e08775
