From 9d1b81e5997ee9ec8cc5bfd16f6c62a40044006a Mon Sep 17 00:00:00 2001
From: Christian Pointner <equinox@helsinki.at>
Date: Wed, 25 Jan 2017 23:57:06 +0100
Subject: actually create file hasher


diff --git a/.gitignore b/.gitignore
index bc88ec7..ff7ecdf 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,3 +2,4 @@ bin/
 pkg/
 src/github.com
 src/code.helsinki.at
+src/golang.org
diff --git a/Makefile b/Makefile
index 27218b6..48e88d4 100644
--- a/Makefile
+++ b/Makefile
@@ -24,8 +24,10 @@ GOCMD := GOPATH=$(curdir) go
 
 EXECUTEABLE := pool-import
 
-LIBS :=	"code.helsinki.at/rhrd-go/rhimport" \
-		   	"code.helsinki.at/rhrd-go/rddb"
+LIBS := "code.helsinki.at/rhrd-go/rhimport" \
+        "code.helsinki.at/rhrd-go/rddb" \
+        "golang.org/x/crypto/blake2b"
+
 
 
 .PHONY: getlibs updatelibs vet format build clean distclean
diff --git a/src/pool-import/dir-hasher.go b/src/pool-import/dir-hasher.go
index 6c1bf9b..aa204a0 100644
--- a/src/pool-import/dir-hasher.go
+++ b/src/pool-import/dir-hasher.go
@@ -22,18 +22,30 @@
 package main
 
 import (
+	"encoding/base64"
+	"io"
 	"log"
-	"math/rand"
 	"os"
 	"path/filepath"
+	"strings"
 	"sync"
-	"time"
+
+	"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 MusicDir struct {
-	root   string
-	stdlog *log.Logger
-	Files  map[string]string
+	root     string
+	stdlog   *log.Logger
+	filesMtx sync.Mutex
+	Files    FileMap
 }
 
 func handleEntry(C chan<- string, path string, info os.FileInfo, err error, stdlog *log.Logger) error {
@@ -45,17 +57,44 @@ func handleEntry(C chan<- string, path string, info os.FileInfo, err error, stdl
 		return nil
 	}
 	if !info.Mode().IsRegular() {
-		stdlog.Printf(" - skipping special file: %s", path)
+		stdlog.Printf(" - skipping (special file):  %s", info.Name())
 		return nil
 	}
-	// TODO: check file extensions
-	C <- path
+	switch strings.ToLower(filepath.Ext(path)) {
+	case ".flac":
+		fallthrough
+	case ".ogg":
+		fallthrough
+	case ".wav":
+		fallthrough
+	case ".mp3":
+		fallthrough
+	case ".aac":
+		fallthrough
+	case ".mp4":
+		fallthrough
+	case ".m4a":
+		C <- path
+	default:
+		stdlog.Printf(" - skipping (unknown extension):  %s", info.Name())
+	}
 	return nil
 }
 
-func computeHash(file string) string {
-	time.Sleep(time.Duration(rand.Float64() * float64(5*time.Second))) // TODO: faking it is nice but the real hash would be nicer
-	return ""
+func computeHash(path string) (string, error) {
+	hash, err := blake2b.New256(nil)
+	if err != nil {
+		return "", err
+	}
+	file, err := os.Open(path)
+	if err != nil {
+		return "", err
+	}
+	defer file.Close()
+	if _, err := io.Copy(hash, file); err != nil {
+		return "", err
+	}
+	return base64.URLEncoding.EncodeToString(hash.Sum(nil)), nil
 }
 
 func (d *MusicDir) collectHashes(C <-chan string, numThreads int) (wg *sync.WaitGroup) {
@@ -63,17 +102,21 @@ func (d *MusicDir) collectHashes(C <-chan string, numThreads int) (wg *sync.Wait
 
 	for i := 0; i < numThreads; i++ {
 		wg.Add(1)
-		d.stdlog.Printf("Starting hashing thread #%d", i)
 		go func(num int) {
 			defer wg.Done()
 			for {
 				file, ok := <-C
 				if !ok {
-					d.stdlog.Printf("Stopping hashing thread #%d", num)
 					return
 				}
-				d.stdlog.Printf(" - hashing: %s", file)
-				computeHash(file)
+				if hash, err := computeHash(file); err != nil {
+					d.stdlog.Printf(" - skipping (%v):  %s", err, filepath.Base(file))
+				} else {
+					d.stdlog.Printf(" - hashed:  %s", filepath.Base(file))
+					d.filesMtx.Lock()
+					d.Files[hash] = file
+					d.filesMtx.Unlock()
+				}
 			}
 		}(i)
 	}
diff --git a/src/pool-import/main.go b/src/pool-import/main.go
index 0a2ce7a..10a59ec 100644
--- a/src/pool-import/main.go
+++ b/src/pool-import/main.go
@@ -96,14 +96,25 @@ func main() {
 			C <- syscall.SIGTERM
 		}()
 
+		stdlog.Println("********************************************")
+		allFiles := make(FileMap)
 		for _, root := range directories {
-			stdlog.Printf("hashing all the files in '%s'", root)
+			stdlog.Printf("*** hashing all the files in '%s'", root)
+			stdlog.Println("")
 			md := NewMusicDir(root, stdlog)
 			if err := md.ComputeHashes(); err != nil {
 				return
 			}
+			allFiles.Merge(md.Files)
+			stdlog.Println("")
+			stdlog.Println("******************************")
 		}
-		stdlog.Printf("all directires hashed")
+		stdlog.Println("")
+		for hash, file := range allFiles {
+			stdlog.Printf("%s: %s", hash, file)
+		}
+		stdlog.Println("")
+		stdlog.Println("********************************************")
 	}()
 
 	<-C
-- 
cgit v0.10.2