Parallel

filesmap.go

@@ -1,65 +1,91 @@
 package main
 
-import "os"
+import (
+	"fmt"
+	"log"
+	"sync"
+)
 
 // FilesMap is a struct for listing files by Size and Hash to search for duplicates
 type FilesMap struct {
-	FilesBySize map[int64]map[string][]string
+	FilesBySize map[int64][]string
-}
 
-// Add a file to the Map and calculate hash on demand
+	FilesByHash map[string][]string
-func (fm *FilesMap) Add(path string, info os.FileInfo) error {
-	if info.IsDir() {
-		return nil
-	}
 
-	filesByHash := fm.FilesBySize[info.Size()]
+	FilesHashing chan fileEntry
 
-	// first file with same size
+	FilesIncoming chan fileEntry
-	// => create new map for size
-	if filesByHash == nil {
-		filesByHash = map[string][]string{}
-		fm.FilesBySize[info.Size()] = filesByHash
-		filesByHash[""] = []string{path}
-		return nil
-	}
 
-	// second file with same size
+	FilesHashed chan fileEntry
-	// => calculate hashes for all entries
-	if _, hasEmptyHash := filesByHash[""]; hasEmptyHash {
-		err := appendByFileHash(filesByHash, fileInfo)
-		err2 := appendByFileHash(filesByHash, filesByHash[""][0])
 
-		delete(filesByHash, "")
+	lock sync.Mutex
-
-		if err != nil {
-			return err
-		}
-
-		return err2
-	}
-
-	// for later files always append by hash
-	return appendByFileHash(filesByHash, fileInfo)
-}
-
-func appendByFileHash(filesByHash map[string][]string, path string) error {
-	hash, err := calculateHash(path)
-
-	if err != nil {
-		return err
-	}
-
-	if _, ok := filesByHash[hash]; ok {
-		filesByHash[hash] = append(filesByHash[hash], path)
-	} else {
-		filesByHash[hash] = []string{path}
-	}
-	return nil
 }
 
 func newFilesMap() *FilesMap {
 	return &FilesMap{
-		FilesBySize: map[int64]map[string][]string{},
+		FilesBySize:   map[int64][]string{},
+		FilesByHash:   map[string][]string{},
+		FilesHashed:   make(chan fileEntry),
+		FilesIncoming: make(chan fileEntry),
+		FilesHashing:  make(chan fileEntry),
 	}
 }
+
+func (fm *FilesMap) IncomingWorker() {
+	for file := range fm.FilesIncoming {
+		if *verbose {
+			fmt.Println("Incoming", file.path)
+		}
+
+		files, ok := fm.FilesBySize[file.size]
+		if !ok {
+			files = []string{file.path}
+			fm.FilesBySize[file.size] = files
+			continue
+		}
+
+		if len(files) == 1 {
+			fm.FilesHashing <- fileEntry{files[0], file.size, ""}
+		}
+
+		fm.FilesHashing <- file
+	}
+	close(fm.FilesHashing)
+}
+
+func (fm *FilesMap) HashingWorker() {
+	for file := range fm.FilesHashing {
+		if *verbose {
+			fmt.Println("Hashing", file.path)
+		}
+
+		hash, err := calculateHash(file.path)
+
+		if err != nil {
+			log.Printf("Error calculating Hash for %s: %v\n", file, err)
+			continue
+		}
+
+		file.hash = hash
+		fm.FilesHashed <- file
+	}
+	close(fm.FilesHashed)
+}
+
+func (fm *FilesMap) HashedWorker(done chan bool) {
+	for file := range fm.FilesHashed {
+		if *verbose {
+			fmt.Println("Finishing", file.path)
+		}
+
+		fm.lock.Lock()
+		if _, ok := fm.FilesByHash[file.hash]; ok {
+			fm.FilesByHash[file.hash] = append(fm.FilesByHash[file.hash], file.path)
+		} else {
+			fm.FilesByHash[file.hash] = []string{file.path}
+		}
+		fm.lock.Unlock()
+	}
+
+	done <- true
+}

main.go

@@ -29,32 +29,45 @@ func main() {
 
 	filesMap := newFilesMap()
 	if *fromFile != "" {
-		fmt.Println("Loading file", *fromFile)
-
 		byteValue, _ := ioutil.ReadFile(*fromFile)
-		err := json.Unmarshal(byteValue, &filesMap.FilesBySize)
+		err := json.Unmarshal(byteValue, &filesMap.FilesByHash)
 		if err != nil {
 			panic(err)
 		}
 	} else {
+		done := make(chan bool)
+		//for i := 0; i < runtime.GOMAXPROCS(0); i++ {
+		go filesMap.HashingWorker()
+		//}
+
+		go filesMap.IncomingWorker()
+
+		go filesMap.HashedWorker(done)
+
 		for _, path := range flag.Args() {
 			filepath.Walk(path, func(path string, info os.FileInfo, err error) error {
-				filesMap.Add(path, info)
+				if info.IsDir() {
+					return nil
+				}
+
+				filesMap.FilesIncoming <- fileEntry{path, info.Size(), ""}
 				return nil
 			})
 		}
+
+		close(filesMap.FilesIncoming)
+		<-done
 	}
 
 	if *toFile != "" && *fromFile == "" {
-		json, _ := json.MarshalIndent(filesMap.FilesBySize, "", "  ")
+		json, _ := json.MarshalIndent(filesMap.FilesByHash, "", "  ")
 		ioutil.WriteFile(*toFile, json, 644)
 	}
 
 	if *deleteDupesIn != "" {
 		deleteIn := filepath.Clean(*deleteDupesIn)
-		for size := range filesMap.FilesBySize {
+		for hash := range filesMap.FilesByHash {
-			for hash := range filesMap.FilesBySize[size] {
+			duplicateFiles := filesMap.FilesByHash[hash]
-				duplicateFiles := filesMap.FilesBySize[size][hash]
 			if len(duplicateFiles) <= 1 {
 				continue
 			}
@@ -68,12 +81,10 @@ func main() {
 				}
 			}
 		}
-		}
 	} else if *promptForDelete {
 		reader := bufio.NewReader(os.Stdin)
-		for size := range filesMap.FilesBySize {
+		for hash := range filesMap.FilesByHash {
-			for hash := range filesMap.FilesBySize[size] {
+			duplicateFiles := filesMap.FilesByHash[hash]
-				duplicateFiles := filesMap.FilesBySize[size][hash]
 			if len(duplicateFiles) <= 1 {
 				continue
 			}
@@ -105,14 +116,12 @@ func main() {
 				if *force {
 					remove(file)
 				}
-				}
 
 			}
 		}
 	} else {
-		for size := range filesMap.FilesBySize {
+		for hash := range filesMap.FilesByHash {
-			for hash := range filesMap.FilesBySize[size] {
+			duplicateFiles := filesMap.FilesByHash[hash]
-				duplicateFiles := filesMap.FilesBySize[size][hash]
 			if len(duplicateFiles) <= 1 {
 				continue
 			}
@@ -124,7 +133,6 @@ func main() {
 		}
 	}
 }
-}
 
 func printConfiguration() {
 	fmt.Printf("fromFile: \"%v\"\n", *fromFile)
@@ -139,3 +147,9 @@ func printConfiguration() {
 	fmt.Println()
 	fmt.Println()
 }
+
+type fileEntry struct {
+	path string
+	size int64
+	hash string
+}