From b79d589aad52792560eb8d42946cb96872973da2 Mon Sep 17 00:00:00 2001
From: Christian Pointner <equinox@helsinki.at>
Date: Sat, 28 Jan 2017 00:13:56 +0100
Subject: make hash algorithm configurable


diff --git a/sample/file-hasher-requst.json b/sample/file-hasher-requst.json
index 4ef0497..d2a1446 100644
--- a/sample/file-hasher-requst.json
+++ b/sample/file-hasher-requst.json
@@ -1,10 +1,12 @@
 {
     "threads": 2,
-    "dir": [
-        "../../../../Music/Dead Man's Bones - Dead Man's Bones (2009)",
-        "../../../../Music/Hozier - Hozier"
-    ],
-    "m3u": [
-        "../../../../Music/Hozier - From Eden/00-hozier-from_eden_ep-web-2014.m3u"
-    ]
+    "src": {
+        "dir": [
+            "../../../../Music/Dead Man's Bones - Dead Man's Bones (2009)",
+            "../../../../Music/Hozier - Hozier"
+        ],
+        "m3u": [
+            "../../../../Music/Hozier - From Eden/00-hozier-from_eden_ep-web-2014.m3u"
+        ]
+    }
 }
diff --git a/src/file-hasher/hasher.go b/src/file-hasher/hasher.go
index 27574f9..5f84646 100644
--- a/src/file-hasher/hasher.go
+++ b/src/file-hasher/hasher.go
@@ -23,32 +23,32 @@ package main
 
 import (
 	"encoding/base64"
+	"hash"
 	"io"
 	"log"
 	"os"
 	"path/filepath"
 	"sync"
 
+	"crypto/md5"
+	"crypto/sha1"
+	"crypto/sha256"
+	"crypto/sha512"
 	"golang.org/x/crypto/blake2b"
 )
 
 type FileMap map[string]string
 
-func (m FileMap) Merge(m2 FileMap) {
-	for key, value := range m2 {
-		m[key] = value
-	}
-}
-
 type Hasher struct {
 	numThreads uint
 	stdlog     *log.Logger
+	newHash    func() (hash.Hash, error)
 	filesMtx   sync.Mutex
 	Files      FileMap
 }
 
-func computeHash(path string) (string, error) {
-	hash, err := blake2b.New256(nil)
+func (h *Hasher) computeHash(path string) (string, error) {
+	hash, err := h.newHash()
 	if err != nil {
 		return "", err
 	}
@@ -75,7 +75,7 @@ func (h *Hasher) collectHashes(C <-chan string) (wg *sync.WaitGroup) {
 				if !ok {
 					return
 				}
-				if hash, err := computeHash(file); err != nil {
+				if hash, err := h.computeHash(file); err != nil {
 					h.stdlog.Printf(" - skipping (%v):  %s", err, filepath.Base(file))
 				} else {
 					h.stdlog.Printf(" - hashed:  %s", filepath.Base(file))
@@ -103,12 +103,42 @@ func (h *Hasher) ComputeHashes(w Walker) (err error) {
 	return
 }
 
-func NewHasher(numThreads uint, stdlog *log.Logger) (h *Hasher) {
+func NewHasher(algo string, numThreads uint, stdlog *log.Logger) (h *Hasher) {
 	h = &Hasher{numThreads: numThreads, stdlog: stdlog}
+
+	switch algo {
+	case "md5":
+		h.newHash = func() (hash.Hash, error) {
+			return md5.New(), nil
+		}
+	case "sha1":
+		h.newHash = func() (hash.Hash, error) {
+			return sha1.New(), nil
+		}
+	case "sha256":
+		h.newHash = func() (hash.Hash, error) {
+			return sha256.New(), nil
+		}
+	case "sha512":
+		h.newHash = func() (hash.Hash, error) {
+			return sha512.New(), nil
+		}
+	case "blake2b":
+		fallthrough
+	case "":
+		algo = "blake2b"
+		h.newHash = func() (hash.Hash, error) {
+			return blake2b.New256(nil)
+		}
+	default:
+		stdlog.Printf("*** invalid hashing algorithm: '%s'", algo)
+		return nil
+	}
+
 	h.Files = make(map[string]string)
 	if h.numThreads < 1 {
 		h.numThreads = 4
 	}
-	stdlog.Printf("*** Created blake2b based hasher (%d threads)", h.numThreads)
+	stdlog.Printf("*** Created hasher (%s, %d threads)", algo, h.numThreads)
 	return
 }
diff --git a/src/file-hasher/main.go b/src/file-hasher/main.go
index f360705..c7718ee 100644
--- a/src/file-hasher/main.go
+++ b/src/file-hasher/main.go
@@ -31,9 +31,12 @@ import (
 )
 
 type Request struct {
-	NumThreads   uint     `json:"threads"`
-	Directories  []string `json:"dir"`
-	M3UPlaylists []string `json:"m3u"`
+	NumThreads uint   `json:"threads"`
+	HashAlgo   string `json:"algo"`
+	Sources    struct {
+		Directories  []string `json:"dir"`
+		M3UPlaylists []string `json:"m3u"`
+	} `json:"src"`
 }
 
 func main() {
@@ -53,11 +56,14 @@ func main() {
 			C <- syscall.SIGTERM
 		}()
 		stdlog.Println("***************************************************************")
-		h := NewHasher(req.NumThreads, stdlog)
+		h := NewHasher(req.HashAlgo, req.NumThreads, stdlog)
+		if h == nil {
+			os.Exit(1)
+		}
 		stdlog.Println("")
 		stdlog.Println("******************************")
 
-		for _, root := range req.Directories {
+		for _, root := range req.Sources.Directories {
 			src := NewDir(root, stdlog)
 			stdlog.Printf("*** hashing all the files in '%s'", root)
 			stdlog.Println("")
@@ -68,7 +74,7 @@ func main() {
 			stdlog.Println("")
 			stdlog.Println("******************************")
 		}
-		for _, list := range req.M3UPlaylists {
+		for _, list := range req.Sources.M3UPlaylists {
 			src := NewM3u(list, stdlog)
 			stdlog.Printf("*** hashing all the files from '%s'", list)
 			stdlog.Println("")
-- 
cgit v0.10.2