[
https://issues.apache.org/jira/browse/IMAGING-162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14330319#comment-14330319
]
Michael Groß edited comment on IMAGING-162 at 2/21/15 4:03 PM:
---------------------------------------------------------------
I created the PRs on github for this and also for the other patches I that I
propose.
This proposed patch works yet (as date of writing) only for BMPs which are not
RLE encoded. I didn't touch the code for RLE -encoded- compressed BMPs yet, nor
i did execute some specific tests for this format. (But I am working on it).
A word about the test cases: All existing tests seems to executed in a big loop
over all available test images. Each test class is executed for the images wich
fitting type, i.e. BMP test clases for BMPs. This is good to see if a
particular image raises some exceptions in the code, but it does not allow to
create specific tests i.e. asserting if the pixels of a BMP with negative
height are read as expected or if a broken image file throws a certain
exception.
There fore I added the package
*org.apache.commons.imaging.formats.bmp.specific* to have such tests for BMPs
here. They aren't executed by the "big loop" described before thus they are
independed from what images are in *src/test/data/images*. Accordingly I added
*src/test/data/specificTests* for image examples which are not used in the "big
loop".
So I would propose that when we want to fix an issue like this we first add
example images to *src/test/data/specificTests* and add tests which reproduces
the problem to a package called
*org.apache.commons.imaging.formats.<FORMAT>.specific*.
was (Author: mgmechanics):
I created the PRs on github for this and also for the other patches I that I
propose.
This proposed patch works yet (as date of writing) only for BMPs which are not
RLE encoded. I didn't touch the code for RLE encoded BMPs yet, nor i did
execute some specific tests for this format. (But I am working on it).
A word about the test cases: All existing tests seems to executed in a big loop
over all available test images. Each test class is executed for the images wich
fitting type, i.e. BMP test clases for BMPs. This is good to see if a
particular image raises some exceptions in the code, but it does not allow to
create specific tests i.e. asserting if the pixels of a BMP with negative
height are read as expected or if a broken image file throws a certain
exception.
There fore I added the package
*org.apache.commons.imaging.formats.bmp.specific* to have such tests for BMPs
here. They aren't executed by the "big loop" described before thus they are
independed from what images are in *src/test/data/images*. Accordingly I added
*src/test/data/specificTests* for image examples which are not used in the "big
loop".
So I would propose that when we want to fix an issue like this we first add
example images to *src/test/data/specificTests* and add tests which reproduces
the problem to a package called
*org.apache.commons.imaging.formats.<FORMAT>.specific*.
> BMP files with negative bitmap height are not supported
> -------------------------------------------------------
>
> Key: IMAGING-162
> URL: https://issues.apache.org/jira/browse/IMAGING-162
> Project: Commons Imaging
> Issue Type: Bug
> Components: Format: BMP
> Affects Versions: 1.0
> Reporter: Myroslav Golub
> Labels: github
> Fix For: Review Patch
>
> Attachments: monochrome-negative-height.bmp
>
>
> The BMP format allows for negative values of the height field in the header.
> This means that the image lines are stored top-down, not bottom-up as usual
> (i. e. the first pixel in the pixel array would be in the top left corner of
> the image, not in the bottom left as usual). However, when a negative-height
> bitmap is passed, the BmpImageParser attempts to create an array of negative
> length, and a NegativeArraySizeException ensues.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
