feat: Add manifest output option to dedup command

this manifest contains the pairing of origins and their hashes in the store

cmd/dedup.go

@@ -1,7 +1,9 @@
 package cmd
 
 import (
+	"encoding/json"
 	"log"
+	"os"
 	"strconv"
 
 	"github.com/mirkobrombin/dabadee/pkg/dabadee"
@@ -21,6 +23,7 @@ func NewDedupCommand() *cobra.Command {
 
 	cmd.Flags().BoolP("with-metadata", "m", false, "Include file metadata in hash calculation")
 	cmd.Flags().BoolP("verbose", "v", false, "Verbose output")
+	cmd.Flags().String("manifest-output", "", "Output manifest file to the given path")
 
 	return cmd
 }
@@ -29,6 +32,7 @@ func dedupCommand(cmd *cobra.Command, args []string) {
 	source, storagePath, workersStr := args[0], args[1], args[2]
 	withMetadata, _ := cmd.Flags().GetBool("with-metadata")
 	verbose, _ := cmd.Flags().GetBool("verbose")
+	outputManifest, _ := cmd.Flags().GetString("manifest-output")
 	workers, err := strconv.Atoi(workersStr)
 	if err != nil {
 		log.Fatalf("Invalid number of workers: %v", err)
@@ -57,5 +61,19 @@ func dedupCommand(cmd *cobra.Command, args []string) {
 		log.Fatalf("Error during deduplication: %v", err)
 	}
 
+	// Output manifest
+	if outputManifest != "" {
+		log.Printf("Writing manifest to %s..", outputManifest)
+
+		manifest, err := json.Marshal(processor.FileMap)
+		if err != nil {
+			log.Fatalf("Error marshalling manifest: %v\n\nPrinting to stdout instead:\n\n%v", err, processor.FileMap)
+		}
+
+		if err := os.WriteFile(outputManifest, manifest, 0644); err != nil {
+			log.Fatalf("Error writing manifest: %v\n\nPrinting to stdout instead:\n\n%v", err, processor.FileMap)
+		}
+	}
+
 	log.Print("Done")
 }

pkg/processor/dedup.go

@@ -31,6 +31,12 @@ type DedupProcessor struct {
 
 	// Workers is the number of workers to use
 	Workers int
+
+	// FileMap is a map of original file paths to their hash in storage
+	FileMap map[string]string
+
+	// mapMutex is a mutex to protect the FileMap from concurrent access
+	mapMutex sync.Mutex
 }
 
 // NewDedupProcessor creates a new DedupProcessor
@@ -40,6 +46,7 @@ func NewDedupProcessor(source string, storage *storage.Storage, hashGen hash.Gen
 		Storage: storage,
 		HashGen: hashGen,
 		Workers: workers,
+		FileMap: make(map[string]string),
 	}
 }
 
@@ -164,6 +171,11 @@ func (p *DedupProcessor) processFile(path string) (err error) {
 		}
 	}
 
+	// Store the original path of the file
+	p.mapMutex.Lock()
+	p.FileMap[path] = finalHash
+	p.mapMutex.Unlock()
+
 	if _, err := os.Lstat(path); os.IsNotExist(err) {
 		err = os.Link(dedupPath, path)
 		if err != nil {