Hashimg, removes duplicate images and renames them for faster future scans.
A quaint little hobby utility that I created for the hoard of random images I have in folders. It reads all images within a folder (not including sub-folders) and compares them to one another. It then deletes the duplicates and renames the remaining ones to their hash.
For Linux, MacOS (10.15+) Catalina, and Windows, you can download them from the Releases Page
go install github.com/Jaeiya/hashimg@latestNo. The file data must be identical. Just because images appear to be identical, does not mean that they are. If those images have different resolutions or one is compressed more than another, they will not be flagged as duplicates. An image is only considered a duplicate if it has the exact same data as another image, including meta-data.
Virtually impossible. In order for there to even be a reasonable possibility of false-positives, you would need to have quadrillions of images. Those images would also have to all be in a single folder. So the likelihood of an accidental deletion of a novel image, is virtually zero.
Yes and No. There is a review option that will ask you to review the images before deletion takes place. During this period, you can choose to keep the images that were detected as duplicates, or to delete them. If you choose to keep them, the program aborts and your files are completely untouched.
If choose not to review the duplicate images, then all duplicates are automatically deleted and your existing images will be renamed to their hash, to make future scans faster.
=== Before ===
file1.png
file2.bmp
file3.webp
=== After ===
0x@c147efcfc2d7ea666a9e4f5187b115c9.png
0x@3377870dfeaaa7adf79a374d2702a3fd.bmp
0x@6f3fef6dc51c7996a74992b70d0c35f3.webp
The 0x@ is a unique identifier so that my program knows the file name is part of the calculated
hash of the file.
You'll need to have Go 1.22.5 or higher installed. If you're using 1.23.x or higher - as of
2024-08-25 - you will end up with significantly larger binaries. Not sure if that's a feature
or a bug but...that's how it is.
All of the following builds are technically production builds, in terms of how the binaries are
optimized. Even if you build using the dev target, you're still getting the most optimized
build possible. The cold start time can take a little bit, but subsequent builds are typically
1s or less.
The output of all builds is the ./dist dir
If you don't have Go or Goreleaser installed, then the following make targets will attempt to install them. The linux version is only partially tested and the MacOS isn't tested at all since I don't have an Apple device.
If it fails for any reason and you do in fact have Go installed, then you can just execute
make install-goreleaser. Because it's using the Go CLI, there should be no issues.
make setup-windowsmake setup-linuxmake setup-macosThis will build 3 binaries, one for each platform (linux, darwin, & windows), all of which are
x86_64 compatible.
make devThis is a preview of how the builds will look in production, however they do use a snapshot version, which will not be used in production.
make snapshotIf you notice any bugs, feel free to create an issue. I do use this on my own images, so I won't shy away from critiques or suggestions that might make it better. It won't be my main focus, but I am active enough that I will definitely respond.
I'd like to thank the creators of Lip Gloss and Bubble Tea for making an incredibly easy framework for creating useful TUIs. I didn't want to have to build all of it out myself, so thanks to the Charm team, I didn't have to!