Add skip file support, and only log parse errors @ INFO level.

cmd/mailhash/mailhash.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
+	"strings"
 
 	"xinu.tv/email"
 
@@ -13,10 +14,12 @@ import (
 )
 
 var (
-	dryrun   = flag.Bool("dryrun", true, "don't actually rename files, just log actions.")
-	saveFile = flag.String("save", "", "filename to store unread status.")
-	loadFile = flag.String("load", "", "filename to load unread status.")
-	maildir  = flag.String("maildir", "", "Maildir root")
+	dryrun    = flag.Bool("dryrun", true, "don't actually rename files, just log actions.")
+	saveFile  = flag.String("save", "", "filename to store unread status.")
+	loadFile  = flag.String("load", "", "filename to load unread status.")
+	maildir   = flag.String("maildir", "", "Maildir root")
+	skipFiles = flag.String("skip", "maildirfolder,log,msgid.cache,razor-agent.log",
+		"comma separated files to skip")
 )
 
 type Status struct {
@@ -25,7 +28,12 @@ type Status struct {
 	Read bool   `json:"read,omitempty"`
 }
 
-type Messages []Status
+type Messages struct {
+	// Filenames to skip when hash mail.
+	Skip map[string]struct{}
+	// Read mail status.
+	Statuses []Status
+}
 
 var headers = []string{"to", "from", "cc", "date", "subject"}
 
@@ -38,17 +46,19 @@ func (m *Messages) hashMail(path string, info os.FileInfo, err error) error {
 	if info.IsDir() {
 		return nil
 	}
+
 	r, err := os.Open(path)
 	if err != nil {
 		glog.Fatal(err)
 	}
 	h, err := email.Hash(r, headers)
 	if err != nil {
-		glog.Errorf("%s %q", path, err.Error())
+		glog.Errorf("%s not an mail file", path)
+		glog.Info("%q", err.Error())
 		return nil
 	}
 	md := email.NewInfo(path)
-	*m = append(*m, Status{
+	m.Statuses = append(m.Statuses, Status{
 		Path: path,
 		Hash: fmt.Sprintf("%x", h.Sum(nil)),
 		Read: md.Seen,
@@ -64,7 +74,7 @@ func (m *Messages) LoadStatus(statusFile string) error {
 	}
 	defer r.Close()
 	dec := json.NewDecoder(r)
-	return dec.Decode(m)
+	return dec.Decode(&m.Statuses)
 }
 
 func (m *Messages) SaveStatus(maildir, statusFile string) error {
@@ -96,9 +106,9 @@ func (m *Messages) SaveStatus(maildir, statusFile string) error {
 }
 
 func (m Messages) Reconcile(maildir string) error {
-	hashMap := make(map[string]*Status, len(m))
-	for i, msg := range m {
-		hashMap[msg.Hash] = &m[i]
+	hashMap := make(map[string]*Status, len(m.Statuses))
+	for i, msg := range m.Statuses {
+		hashMap[msg.Hash] = &m.Statuses[i]
 	}
 
 	return filepath.Walk(maildir, func(path string, info os.FileInfo, err error) error {
@@ -110,6 +120,12 @@ func (m Messages) Reconcile(maildir string) error {
 			return nil
 		}
 
+		base := filepath.Base(path)
+		if _, ok := m.Skip[base]; ok {
+			glog.Infoln("Skipping", path)
+			return nil
+		}
+
 		r, err := os.Open(path)
 		if err != nil {
 			glog.Fatal(err)
@@ -117,7 +133,8 @@ func (m Messages) Reconcile(maildir string) error {
 
 		h, err := email.Hash(r, headers)
 		if err != nil {
-			glog.Errorf("%s %q", path, err.Error())
+			glog.Errorf("%s not an mail file", path)
+			glog.Infof("%q", err.Error())
 			return nil
 		}
 
@@ -143,7 +160,7 @@ func (m Messages) Reconcile(maildir string) error {
 
 func (m Messages) PrintStats() {
 	r, u, t := 0, 0, 0
-	for _, msg := range m {
+	for _, msg := range m.Statuses {
 		t++
 		if msg.Read {
 			r++
@@ -159,7 +176,16 @@ func (m Messages) PrintStats() {
 func main() {
 	defer glog.Flush()
 	flag.Parse()
-	m := Messages{}
+	skipStr := strings.Split(*skipFiles, ",")
+	skip := make(map[string]struct{}, len(skipStr))
+	for _, s := range skipStr {
+		skip[s] = struct{}{}
+	}
+	glog.Infoln("Skip files", skip)
+
+	m := Messages{
+		Skip: skip,
+	}
 
 	if *maildir == "" {
 		fmt.Println("Must specify Maildir with -maildir")
@@ -184,6 +210,7 @@ func main() {
 		}
 		m.PrintStats()
 		if err := m.Reconcile(*maildir); err != nil {
+			glog.Fatal(err)
 		}
 	}
 }