Release 2023-07-16: v0.2.5

Version v0.2.5

Now available on:

• Windows 7 and above (x64)
• Any Linux Distro with Kernel version 5.15 and above (x64)
• MacOS Monterey 12 and above (x64)

Changes

• Added automatic build, archiving and archive checksum generation in build.py
• Added automatic artefacts uploading using GitHub workflows in .github/workflows/build.yml
• More verbose build,py
• Bumped version number
• Code cleanup
• Update README.md

Bugfixes

• Resolved PYTHONPATH module discovery problem when running python -O build.py and python -m unittest discover

Full Changelog: v0.2.4...v0.2.5

Release 2023-07-14: v0.2.4

Version v0.2.4

Available on Windows 7+ and Linux 5.15+

Changes

• Added build support for legacy Windows versions
• Bumped version number
• Code cleanup
• Update README.md

Full Changelog: v0.2.3...v0.2.4

Release 2023-06-24: v0.2.3

Version v0.2.3

• Available on Windows and Linux
• No executable compression

Changes

• Added info mode per request by @yoantran (Resolved #29)
• Clearer and more conventional option names for -f/--format flag
• Clearer and more accurate option names for -a/--auto-hash-size flag
• More robust error handling
• Bumped version number (twice, because there are a lot of changes)
• Code cleanup
• Update README.md

Bugfixes:

• Fixed "Invalid progress bar level" error while progress bar level is correct (-p0)
• Fixed typo in calc_hash_size() that caused abnormally large auto hash size calculation for avg-dims-mean

Full Changelog: v0.2.1...v0.2.3

Hotfix 2023-06-19: v0.2.1

Version v0.2.1

• Available on Windows only
• Needs more user feedback
• No executable compression

Changes

• Resolved #30, thank you @yoantran for discovering and reporting this bug
• More robust error handling
• Bumped version number
• Code cleanup
• Update README.md

Full Changelog: v0.2.0...v0.2.1

Release 2023-06-17: v0.2.0

Version v0.2.0

• Out of beta, first official release! 🎉🎊
• Available on Windows only
• Needs more user feedback
• No executable compression

Changes

• Added support for more image file extensions
• Proper support and documentation for .eps files
• Resolved #16 as requested by @yoantran: Implemented color-sensitive hashing methods (-m/--hashing-method flag):
  • color-hist-hashing (default, fastest and very accurate for both colors and transparency)
  • grayscale-hashing (old behavior, can't differentiate colors nor transparency)
  • rgb-hashing (slow, most accurate for colors, but can't differentiate between white and transparent)
  • rgba-hashing (slowest, most accurate for both colors and transparency)
• Added automatic calculation of hash size (default behavior unless a hash size is provided with -s/--hash-size flag), Available methods can be accessed using -a/--auto-hash-size flag:
  • max-dim: Take the largest dimension (either width or height) of all image as the hash size
  • max-adim: Take the largest value of the average between width and height of all images as the hash size
  • avg-dim: Take the average of all widths and heights of all images as the hash size
  • avg-adim: Take the average value of the average between width and height of all images as the hash size
• More robust error handling
• Bumped version number
• Code cleanup
• Update README.md

Full Changelog: v0.1.3-beta...v0.2.0

Prototype v0.1.3-beta

Version v0.1.3-beta

• Available on Windows only
• Needs more user feedback
• No executable compression

Changes

• Better verbose output messages for -V1 and -V2 levels (better formatting for both levels, and more informative exclusion messages for -V2)
• Verbose mode progress bar details is now adjustable: If verbose mode is set, the user can choose between 3 levels of details:
  • 0: Disable the progress bar
  • 1: No bar, just stats
  • 2: Bar and stats
• Better overall error handling and "graceful crashing" for all errors and exceptions
• Dupfiles are now JSON-formatted text files with mandatory .imdup extension; This is to future proof the program for further extension/porting, which makes life easier, since JSON is an already established format with many parser implementation across many languages
• Loaded dupfiles images are sorted in order of the largest to the smallest resolution, see commit 115ce3c
• Implemented reading from dupfile for clean mode (Resolve #11)
• Added program icon by @yoantran (Resolve #12)
• Bumped version number
• Code cleanup
• More informative README.md

Bug Fixes

• Fixed arguments validation unable to differentiate between -p0 and -p 0 (-p is one example among a few more flags). See commit 4ccb9e8

Full Changelog: v0.1.2-beta...v0.1.3-beta

Prototype v0.1.2-beta

Version: v0.1.2-beta

• Available on Windows only
• Needs more testing
• Needs more user feedback
• No icon yet (#12)
• No executable compression
• Note: #11 is not working yet, clean mode will throw a NotADirectoryError error when supplied with a file instead of a directory, to be implemented in next version

Changes

• More robust progress bar
• Renamed detect mode to scan mode
• Setup extra verbose mode (-V1, -V2 verbose levels with -V1 as 'verbose' and -V2 as 'extra verbose')
• Image indexing notifies user of excluded files when -V2 is set
• Modular argument parsing and validation
• Better detect_dup_images() scanning messages: Change Processing ... to Scanning ... to be clearer for the user
• Added early stopping message when there are no duplications to clean
• Better help messages
• Better error messages
• Efficient terminal messages flushing
• More verbose duplication console output (Resolved #10)
• Added option to show hash in console output (will be truncated if too long) using the -H/--show-hash flag
• Output file for -o/--output does not include hash (they could get too long)
• Paths in output file for -o/--output will always be absolute, that makes more sense
• More informative interactive clean mode
• Bumped version number
• Code cleanup
• Update README.md to match the above changes

Bug Fixes

• Added MemoryError handling and force closing of corrupted images to free up locked memory that led to crashing
• Added image file verification when opening an image, then force closing that image after processing is done to free locked memory that lead to crashing

Full Changelog: v0.1.1-beta...v0.1.2-beta

Prototype v0.1.1-beta

Version: v0.1.1-beta

• Ready for beta testing
• Only available on Windows
• Needs more testing
• Needs more user feedback
• No icon yet
• No executable compression

What's Changed

• Unit testing for detect module
• Removed all partial file support to improve program stability
• Removed support for EPS files (they're too unpredictable)
• Graceful Pillow error handling
• Graceful OS permission denied handling
• Added terminal flushing for important verbose messages
• Default hash size is now 512 to improve identical image detection accuracy (more information captured in hashing process), at the cost of program RAM and CPU performance
• Adjustable hash size via command line argument
• Prettifying warnings and error messages
• Sort duplications in order of decreasing resolution (width * height) so that the highest resolution image is kept during cleaning step
• Bumped version number
• Code cleanup
• Update README.md to match the above changes

Full Changelog: v0.1.0-alpha...v0.1.1-beta

Prototype v0.1.0-alpha

What's Changed

• Available for Windows only
• Basic functionalities up and working
• Needs more testing
• Needs more user feedback
• No icon yet
• No executable compression

Full Changelog: https://github.com/miketvo/imdupes-prototype/commits/v0.1.0-alpha