libjpeg allows specifying the marker interval either in MCU blocks or in
MCU rows. Support both, via separate parameters, rather than requiring
callers to do the math.
Co-authored-by: Andrew Murray <radarhere@users.noreply.github.com>
We already support streamtype=2 to skip producing JPEG tables, but
streamtype=1, which skips everything but the tables, was never implemented.
The streamtype=1 stub code dates to Git pre-history, so it's not
immediately clear why. Implement the missing support.
jpeg_write_tables() can't resume after a full output buffer (it fails with
JERR_CANT_SUSPEND), so it might seem that Pillow needs to pre-compute the
necessary buffer size. However, in the normal case of producing an
interchange stream, the tables are written via the same libjpeg codepath
during the first jpeg_write_scanlines() call, and table writes aren't
resumable there either. Thus, any buffer large enough for the normal case
will also be large enough for a tables-only file.
The streamtype option isn't documented and this commit doesn't change that.
It does add a test though.
Co-authored-by: Andrew Murray <radarhere@users.noreply.github.com>
Re-order the JPEG quantization tables to normal order when
loading. This wastes a few CPU cycles if you don't need them.
But it has the advantage of hiding the zigzag order JPEG
implementation detail of these tables completely from Pillow
users.
This difference has led to cases where:
* arrays in zigzag order were taken from a dict and passed
directly as a qtables parameter, causing them to be "zigzagged"
again by libjpeg.
* dicts with lists in normal order being passed to
JpegImagePlugin.convert_dict_qtables, causing them to be
unnecessarily "de-zigzagged".
This adds a new test decorator: skip_unless_feature(). The argument is
the same as passed to features.check(). If the feature is not supported,
the test will be skipped.
This removes several kinds of boilerplate copied and pasted around tests
so test feature checking is handled and displayed more consistently.
Refs #4193
These modules are safe to import and this better follows PEP 8.
From https://www.python.org/dev/peps/pep-0008/#imports
> Imports are always put at the top of the file, just after any module
> comments and docstrings, and before module globals and constants.
Appeared in the form:
ResourceWarning: unclosed file <_io.BufferedReader name='Tests/images/invalid-exif-without-x-resolution.jpg'>
Enable all warnings to always display during tests to help catch these
warnings earlier.
Follow Python's file object semantics. User code is responsible for
closing resources (usually through a context manager) in a deterministic
way.
To achieve this, remove __del__ functions. These functions used to
closed open file handlers in an attempt to silence Python
ResourceWarnings. However, using __del__ has the following drawbacks:
- __del__ isn't called until the object's reference count reaches 0.
Therefore, resource handlers remain open or in use longer than
necessary.
- The __del__ method isn't guaranteed to execute on system exit. See the
Python documentation:
https://docs.python.org/3/reference/datamodel.html#object.__del__
> It is not guaranteed that __del__() methods are called for objects
> that still exist when the interpreter exits.
- Exceptions that occur inside __del__ are ignored instead of raised.
This has the potential of hiding bugs. This is also in the Python
documentation:
> Warning: Due to the precarious circumstances under which __del__()
> methods are invoked, exceptions that occur during their execution
> are ignored, and a warning is printed to sys.stderr instead.
Instead, always close resource handlers when they are no longer in use.
This will close the file handler at a specified point in the user's code
and not wait until the interpreter chooses to. It is always guaranteed
to run. And, if an exception occurs while closing the file handler, the
bug will not be ignored.
Now, when code receives a ResourceWarning, it will highlight an area
that is mishandling resources. It should not simply be silenced, but
fixed by closing resources with a context manager.
All warnings that were emitted during tests have been cleaned up. To
enable warnings, I passed the `-Wa` CLI option to Python. This exposed
some mishandling of resources in ImageFile.__init__() and
SpiderImagePlugin.loadImageSeries(), they too were fixed.
