Initial commit

.gitignore

@@ -0,0 +1,2 @@
+*.jpg
+*.jpeg

go.mod

@@ -0,0 +1,10 @@
+module imhash
+
+go 1.24.3
+
+require (
+	git.site.quack-lab.dev/dave/cylogger v1.2.2
+	github.com/corona10/goimagehash v1.1.0
+)
+
+require github.com/nfnt/resize v0.0.0-20180221191011-83c6a9932646 // indirect

go.sum

@@ -0,0 +1,6 @@
+git.site.quack-lab.dev/dave/cylogger v1.2.2 h1:4xUXASEBlG9NiGxh7f57xHh9imW4unHzakIEpQoKC5E=
+git.site.quack-lab.dev/dave/cylogger v1.2.2/go.mod h1:VS9MI4Y/cwjCBZgel7dSfCQlwtAgHmfvixOoBgBhtKg=
+github.com/corona10/goimagehash v1.1.0 h1:teNMX/1e+Wn/AYSbLHX8mj+mF9r60R1kBeqE9MkoYwI=
+github.com/corona10/goimagehash v1.1.0/go.mod h1:VkvE0mLn84L4aF8vCb6mafVajEb6QYMHl2ZJLn0mOGI=
+github.com/nfnt/resize v0.0.0-20180221191011-83c6a9932646 h1:zYyBkD/k9seD2A7fsi6Oo2LfFZAehjjQMERAvZLEDnQ=
+github.com/nfnt/resize v0.0.0-20180221191011-83c6a9932646/go.mod h1:jpp1/29i3P1S/RLdc7JQKbRpFeM1dOBd8T9ki5s+AY8=

main.go

@@ -0,0 +1,78 @@
+package main
+
+import (
+	"flag"
+	"image/jpeg"
+	"os"
+	"sync"
+
+	logger "git.site.quack-lab.dev/dave/cylogger"
+	"github.com/corona10/goimagehash"
+)
+
+func main() {
+	thresh := flag.Int("thresh", 10, "Threshold for distance")
+	flag.Parse()
+	logger.InitFlag()
+	hashes := &sync.Map{}
+	logger.Info("Starting")
+	logger.Info("Threshold: %v", *thresh)
+	logger.Info("Files: %d", len(flag.Args()))
+
+	wg := sync.WaitGroup{}
+	for _, file := range flag.Args() {
+		wg.Add(1)
+		go func(file string) {
+			defer wg.Done()
+			log := logger.Default.WithPrefix(file)
+			imgfile, err := os.Open(file)
+			if err != nil {
+				log.Error("Failed to open file: %v", err)
+				return
+			}
+			defer imgfile.Close()
+			img, err := jpeg.Decode(imgfile)
+			if err != nil {
+				log.Error("Failed to decode image: %v", err)
+				return
+			}
+			hash, err := goimagehash.ExtPerceptionHash(img, 8, 8)
+			if err != nil {
+				log.Error("Failed to calculate hash: %v", err)
+				return
+			}
+			log.Debug("Hashed: %v", hash)
+			hashes.Store(file, hash)
+		}(file)
+	}
+
+	groupedImages := make(map[string][]string)
+	wg.Wait()
+	hashes.Range(func(key, value interface{}) bool {
+		filea := key.(string)
+		hasha := value.(*goimagehash.ExtImageHash)
+		hashes.Range(func(key, value interface{}) bool {
+			fileb := key.(string)
+			hashb := value.(*goimagehash.ExtImageHash)
+			if filea == fileb {
+				return true
+			}
+			distance, err := hasha.Distance(hashb)
+			if err != nil {
+				logger.Error("Failed to calculate distance: %v", err)
+				return true
+			}
+			logger.Debug("Distance between %v and %v: %v", filea, fileb, distance)
+			if distance <= *thresh {
+				groupedImages[filea] = append(groupedImages[filea], fileb)
+			}
+			return true
+		})
+		return true
+	})
+
+	for file, files := range groupedImages {
+		logger.Info("Grouped %v with %v", file, files)
+	}
+	logger.Info("Done")
+}