Log message:
py-Pillow: updated to 5.4.1

5.4.1:
- File closing: Only close __fp if not fp
- Fix build for Termux
- PNG: Detect MIME types
- PNG: Handle IDAT chunks after image end

Log message:
py-Pillow: updated to 5.4.0

5.4.0:
- Docs: Improved ImageChops documentation
- Allow RGB and RGBA values for P image putpixel
- Add APNG extension to PNG plugin
- Lookup ld.so.cache instead of hardcoding search paths
- Added custom string TIFF tags
- Improve setup.py configuration
- Read textual chunks located after IDAT chunks for PNG
- Performance: Don't try to hash value if enum is empty
- Added custom int and float TIFF tags
- Fixes for issues reported by static code analysis
- GIF: Wait until mode is normalized to copy im.info into encoderinfo
- Docs: Add page of deprecations and removals
- Travis CI: Upgrade PyPy from 5.8.0 to 6.0
- Travis CI: Allow lint job to fail
- Resolve __fp when closing and deleting
- Close exclusive fp before discarding
- Updated open files documentation
- Added libjpeg_turbo to check_feature
- Change color table index background to tuple when saving as WebP
- Allow arbitrary number of comment extension subblocks
- Ensure previous FLI frame is loaded before seeking to the next
- ImageShow improvements
- Depends: Update libimagequant to 2.12.2, libtiff to 4.0.10, libwebp to 1.0.1, \ 
Tk Tcl to 8.6.9
- Check quality_layers type
- Add context manager, __del__ and close methods to TarIO
- Test: Do not play sound when running screencapture command
- Close exclusive fp on open exception
- Only close existing fp in WebP if fp is exclusive
- Docs: Re-add the downloads badge
- Added negative index to PixelAccess
- Change tuple background to global color table index when saving as GIF
- Test: Improved ImageGrab tests
- Flake8 fixes
- Only ask for YCbCr->RGB libtiff conversion for jpeg-compressed tiffs
- Optimise ImageOps.fit by combining resize and crop

Log message:
py-Pillow: updated to 5.3.0

5.3.0:
Changed Image size property to be read-only by default
Add warnings if image file identification fails due to lack of WebP support
Hide the Ghostscript progress dialog popup on Windows
Adding support to reading tiled and YcbCr jpeg tiffs through libtiff
Fixed None as TIFF compression argument
Changed GIF seek to remove previous info items
Improved PDF document info
Add line width parameter to rectangle and ellipse-based shapes
Fixed decompression bomb check in _crop
Added support to ImageDraw.floodfill for non-RGB colors
Tests: Avoid catching unexpected exceptions in tests
Use TextIOWrapper.detach() instead of NoCloseStream
Added transparency to matrix conversion
Added ImageOps pad method
Give correct extrema for I;16 format images
Added PySide2
Corrected TIFF tags
CI: Install CFFI and pycparser without any PYTHONOPTIMIZE
Read/Save RGB webp as RGB (instead of RGBX)
ImageDraw: Add line joints
Improved performance of ImageDraw floodfill method
Fix builds with --parallel
Add more raw Tiff modes (RGBaX, RGBaXX, RGBAX, RGBAXX)
Close existing WebP fp before setting new fp
Add orientation, compression and id_section as TGA save keyword arguments
Convert int values of RATIONAL TIFF tags to floats
Fix code for PYTHONOPTIMIZE
Changed ImageFilter.Kernel to subclass ImageFilter.BuiltinFilter, instead of the \ 
other way around
Remove unused draw.draw_line, draw.draw_point and font.getabc methods
Tests: Added ImageFilter tests
Tests: Added ImageChops tests
AppVeyor: Download lib if not present in pillow-depends
Travis CI: Add Python 3.7 and Xenial
Docs: Added documentation for NumPy conversion
Depends: Update libimagequant to 2.12.1
Add three-color support to ImageOps.colorize
Tests: Add LA to TGA test modes
Skip outline if the draw operation fills with the same colour
Flake8 fixes
Avoid deprecated 'U' mode when opening files

Log message:
py-Pillow: updated to 5.2.0

5.2.0:
Fixed saving a multiframe image as a single frame PDF
If a Qt version is already imported, attempt to use it first
Fix transform fill color for alpha images
TGA: Add support for writing RLE data
TGA: Read and write LA data
QuantOctree.c: Remove erroneous attempt to average over an empty range
Changed ICNS format tests to pass on OS X 10.11
Fixed bug in ImageDraw.multiline_textsize()
Added getsize_multiline support for PIL.ImageFont
Added ImageFile get_format_mimetype method
Changed mmap file pointer to use context manager
Changed ellipse point calculations to be more evenly distributed
Only extract first Exif segment
Tests: Test ImageDraw2, WalImageFile
Remove unnecessary '#if 0' code
Tests: Added GD tests
Fix collections ABCs DeprecationWarning in Python 3.7
unpack_from is faster than unpack of slice
Docs: Add coordinate system links and file handling links in documentation
Tests: TestFilePng: Fix test_save_l_transparency()
Docs: Correct argument name
Docs: Update CMake download URL
Docs: Improve Image.transform documentation
Fix transform fillcolor argument when image mode is RGBA or LA
Tests: More specific Exception testing
Add getrgb HSB/HSV color strings
Allow float values in getrgb HSL color string
AppVeyor: Upgrade to Python 2.7.15 and 3.4.4
AppVeyor: Upgrade to PyPy 6.0.0
Deprecate PILLOW_VERSION and VERSION
Support Python 3.7
Depends: Update freetype to 2.9.1, libjpeg to 9c, libwebp to 1.0.0
Build macOS wheels with Xcode 6.4, supporting older macOS versions
Fix _i2f compilation on some GCC versions
Changed encoderinfo to have priority over info when saving GIF images
Rename PIL.version to PIL._version and remove it from module
Enable background colour parameter on rotate
Remove unnecessary #if 1 directive
Remove unused Python class, Path
Fix dereferencing type-punned pointer will break strict-aliasing

Log message:
py-Pillow: updated to 5.1.0

5.1.0:
Close fp before return in ImagingSavePPM
Added documentation for ICNS append_images
Docs: Move intro text below its header
CI: Rename appveyor.yml as .appveyor.yml
Fix TypeError for JPEG2000 parser feed
Certain corrupted jpegs can result in no data read
Add support for BLP file format
Simplify version checks
Fix "invalid escape sequence" warning on Python 3.6+
Allow append_images to set .icns scaled images
Support appending to existing PDFs
Fix and improve efficient saving of ICNS on macOS
Build: Enable pip cache in AppVeyor build
Trim trailing whitespace
Docs: Correct reference to Image.new method
Rearrange ImageFilter classes into alphabetical order
Test: Remove duplicate line
Build: Update AppVeyor PyPy version
Tiff: Open 8 bit Tiffs with 5 or 6 channels, discarding extra channels
Readme: Added Twitter badge
Removed __main__ code from ImageCms
Test: Changed assert statements to unittest calls
Depends: Update libimagequant to 2.11.10, raqm to 0.5.0, freetype to 2.9
Remove _imaging.crc32 in favor of builtin Python crc32 implementation
Move Tk directory to src directory
Enable pip cache in Travis CI
Remove unused and duplicate imports
Docs: Changed documentation references to 2.x to 2.7
Fix memory leak when opening webp files
Setup: Fix "TypeError: 'NoneType' object is not iterable" for PPC and CRUX
Setup: Add libdirs for ppc64le and armv7l

Log message:
py-Pillow: disabled mp_compile hack; it has problems with native parallel building

Log message:
py-Pillow: disable wrong hack for GCC <4.9

Log message:
py-Pillow: updated to 5.0.0

5.0.0:
Docs: Added docstrings from documentation
Test: Switch from nose to pytest
Rework Source directory layout, preventing accidental import of PIL.
Dynamically link libraqm
Removed scripts directory
TIFF: Run all compressed tiffs through libtiff decoder
GIF: Add disposal option when saving GIFs [linnil1, wiredfool]
EPS: Allow for an empty line in EPS header data
PNG: Add support for sRGB and cHRM chunks, permit sRGB when no iCCP chunk present
Dependencies: Update Tk Tcl to 8.6.8
Decompression bomb error now raised for images 2x larger than a decompression \ 
bomb warning
Test: avoid random failure in test_effect_noise
Increased epsilon for test_file_eps.py:test_showpage due to Arch update.
Removed check parameter from _save in BmpImagePlugin, PngImagePlugin, \ 
ImImagePlugin, PalmImagePlugin, and PcxImagePlugin.
Make PngImagePlugin.add_text() zip argument type bool
Depends: Updated libwebp to 0.6.1
Remove unnecessary bool() calls in Image.registered_extensions and skipKnownBadTests
Fix count of BITSPERSAMPLE items in broken TIFF files
Fillcolor parameter for Image.Transform
Test: Display differences for test failures
Added executable flag to file with shebang line
Setup: Specify compatible Python versions for pip
Dependencies: Updated libimagequant to 2.11.4
Setup: Warn if trying to install for Py3.7 on Windows
Doc: Fonts can be loaded from a file-like object, not just filename [robin-norwood]
Add eog support for Ubuntu Image Viewer [NafisFaysal]
Test: Test on 3.7-dev on Travis.ci
Dependencies: Update libtiff to 4.0.9
Setup: Replace deprecated platform.dist with file existence check
Build: Fix setup.py on Debian
Docs: Correct error in ImageDraw documentation
Test: Drop Ubuntu Precise, Fedora 24, Fedora 25, add Fedora 27, Centos 7, Amazon \ 
v2 CI Support,,,
Dependencies: Updated libimagequant to 2.11.3
Test: Fix test_image.py to use tempfile
Replace PIL.OleFileIO deprecation warning with descriptive ImportError
WebP: Add support for animated WebP files [jd20]
PDF: Set encoderinfo for images when saving multi-page PDF. Fixes.
Allow the olefile dependency to be optional
GIF: Permit LZW code lengths up to 12 bits in GIF decode
Fix unterminiated string and unchecked exception in _font_text_asBytes.
PPM: Use fixed list of whitespace, rather relying on locale, fixes.
Added support for generators when using append_images,
Doc: Correct PixelAccess.rst
Depends: Update raqm to 0.3.0
Docs: Link to maintained version of aggdraw
Include license file in the generated wheel packages
Depends: Update openjpeg to 2.3.0
Add option to Makefile to build and install with C coverage
Add context manager support to ImageFile.Parser and PngImagePlugin.ChunkStream
ImageDraw.textsize: fix zero length error

Log message:
py-Pillow: update to 4.3.0

4.3.0:
- Fix warning on pointer cast in isblock
- Doc: Added macOS High Sierra tested Pillow version
- Use correct Windows handle type on 64 bit in imagingcms
- 64 Bit Windows fix for block storage
- Fix "expression result unused" warning
- Add 16bit Read/Write and RLE read support to SgiImageFile
- Block & array hybrid storage
- Common seek frame position check
- Doc: Add note about aspect ratio to Image thumbnail script
- Fix ValueError: invalid version number '1.0.0rc1' in scipy release candidate
- Unfreeze requirements.txt
- Test: ResourceWarning tests
- Use n_frames to determine is_animated if possible
- Doc: Corrected parameters in documentation
- Avoid unnecessary Image operations
- Added register_extensions method
- Fix TIFF support for I;16S, I;16BS, and I;32BS rawmodes
- Fixed doc syntax in ImageDraw
- Fixed support for building on Windows/msys2. Added Appveyor CI coverage for \ 
python3 on msys2
- Fix ValueError in Exif/Tiff IFD
- Use pathlib2 for Path objects on Python < 3.4
- Export only required properties in unsafe_ptrs
- Alpha composite fixes
- Faster Transpose operations, added 'Transverse' option
- Deprecate ImageOps undocumented functions gaussian_blur, gblur, unsharp_mask, \ 
usm and box_blur in favor of ImageFilter implementations
- Dependencies: Updated freetype to 2.8.1
- Bug: Player skipped first image
- Faster filter operations for Kernel, Gaussian, and Unsharp Mask filters
- EPS: Add showpage to force rendering of some EPS images
- DOC: Fix type of palette parameter in Image.quantize.
- DOC: Fix Ico docs to match code
- Added file pointer save to SpiderImagePlugin
- Add targa version 2 footer
- Removed redundant lines
- Travis CI: Use default pypy/pypy3
- Fix for SystemError when rendering an empty string, added in 4.2.0
- Fix for memory leaks in font handling added in 4.2.0
- Tests:  cleanup, more tests. Fixed WMF save handler
- Removed debugging interface for Image.core.grabclipboard
- Doc syntax fix
- Faster packing and unpacking for RGB, LA, and related storage modes
- Use RGBX rawmode for RGB JPEG images where possible
- Remove palettes from non-palette modes in _new
- Delete transparency info when convert'ing RGB/L to RGBA
- Code tweaks to ease type annotations
- Fixed incorrect use of 's#' to byteslike object
- Fix JPEG subsampling labels for subsampling=2
- Region of interest (box) for resampling
- Basic support for Termux (android) in setup.py
- Bug: Fix Image.fromarray for numpy.bool type.
- CI: Add Fedora 24 and 26 to Docker tests
- JPEG: Fix ZeroDivisionError when EXIF contains invalid DPI (0/0).
- Depends: Updated openjpeg to 2.2.0
- Depends: Updated Tk Tcl to 8.6.7
- Depends: Updated libimagequant to 2.10.2
- Test: Added test for ImImagePlugin tell()
- Test: Additional tests for SGIImagePlugin
- New Image.getchannel method
- Remove unused im.copy2 and core.copy methods
- Fast Image.merge()
- Fast Image.split()
- Fast image allocation
- Storage cleanup
- FLI: Use frame count from FLI header
- Test: Test animated FLI file
- Bug: Fixed uninitialized memory in bc5 decoding
- Moved SgiImagePlugin save error to before the start of write operations
- Move createfontdatachunk.py so isn't installed globally
- Bug: Fix unexpected keyword argument 'align'
- Add newlines to error message for clarity
- Docs: Updated redirected URL
- Bug: Fix JPEG DPI when EXIF is invalid
- Bug: Fix for font getsize on empty string
- Docs: Improved ImageDraw documentation
- Docs: Corrected alpha_composite args documentation
- Docs: added the description of the filename attribute to images.rst
- Dependencies: Updated libimagequant to 2.10.1
- PDF: Renamed parameter to not shadow built-in dict

Log message:
4.2.1:
- CI: Fix version specification and test on CI for PyPy/Windows