Feature: Email compression in storage

Reduces storage requirements +-25% & speeds up database read & writes by between 25-33%, depending on email content (attachments).

data/message.go

@@ -17,7 +17,6 @@ type Message struct {
 	Bcc         []*mail.Address
 	Subject     string
 	Date        time.Time
-	Created     time.Time
 	Text        string
 	HTML        string
 	Size        int

go.mod

@@ -8,6 +8,7 @@ require (
 	github.com/gorilla/websocket v1.5.0
 	github.com/jhillyerd/enmime v0.10.0
 	github.com/k3a/html2text v1.0.8
+	github.com/klauspost/compress v1.15.9
 	github.com/mhale/smtpd v0.8.0
 	github.com/ostafen/clover/v2 v2.0.0-alpha.2
 	github.com/sirupsen/logrus v1.9.0
@@ -35,7 +36,6 @@ require (
 	github.com/google/orderedcode v0.0.1 // indirect
 	github.com/inconshreveable/mousetrap v1.0.0 // indirect
 	github.com/jaytaylor/html2text v0.0.0-20211105163654-bc68cce691ba // indirect
-	github.com/klauspost/compress v1.15.9 // indirect
 	github.com/kr/pretty v0.3.0 // indirect
 	github.com/mattn/go-runewidth v0.0.13 // indirect
 	github.com/olekukonko/tablewriter v0.0.5 // indirect

storage/database.go

@@ -17,6 +17,7 @@ import (
 	"github.com/axllent/mailpit/logger"
 	"github.com/axllent/mailpit/server/websockets"
 	"github.com/jhillyerd/enmime"
+	"github.com/klauspost/compress/zstd"
 	"github.com/ostafen/clover/v2"
 )
 
@@ -28,6 +29,10 @@ var (
 
 	count       int
 	per100start = time.Now()
+
+	// zstd encoder & decoder
+	encoder, _ = zstd.NewWriter(nil)
+	decoder, _ = zstd.NewReader(nil)
 )
 
 // CloverStore struct
@@ -219,7 +224,10 @@ func Store(mailbox string, b []byte) (string, error) {
 	raw := clover.NewDocument()
 	raw.Set("_id", id)
 	raw.Set("Created", time.Now())
-	raw.Set("Data", string(b))
+
+	compressed := encoder.EncodeAll(b, make([]byte, 0, len(b)))
+	raw.Set("Email", string(compressed))
+
 	_, err = db.InsertOne(mailbox+"_data", raw)
 	if err != nil {
 		// delete the summary because the data insert failed
@@ -369,18 +377,12 @@ func CountUnread(mailbox string) (int, error) {
 func GetMessage(mailbox, id string) (*data.Message, error) {
 	mailbox = sanitizeMailboxName(mailbox)
 
-	q, err := db.FindById(mailbox+"_data", id)
+	raw, err := GetMessageRaw(mailbox, id)
-	if err != nil {
+	if err != nil || raw == nil {
-		return nil, err
-	}
-
-	if q == nil {
 		return nil, errors.New("message not found")
 	}
 
-	raw := q.Get("Data").(string)
+	r := bytes.NewReader(raw)
-
-	r := bytes.NewReader([]byte(raw))
 
 	env, err := enmime.ReadEnvelope(r)
 	if err != nil {
@@ -398,9 +400,8 @@ func GetMessage(mailbox, id string) (*data.Message, error) {
 	date, _ := env.Date()
 
 	obj := data.Message{
-		ID:      q.ObjectId(),
+		ID:      id,
 		Read:    true,
-		Created: q.Get("Created").(time.Time),
 		From:    from,
 		Date:    date,
 		To:      addressToSlice(env, "To"),
@@ -456,12 +457,12 @@ func GetMessage(mailbox, id string) (*data.Message, error) {
 func GetAttachmentPart(mailbox, id, partID string) (*enmime.Part, error) {
 	mailbox = sanitizeMailboxName(mailbox)
 
-	data, err := GetMessageRaw(mailbox, id)
+	raw, err := GetMessageRaw(mailbox, id)
 	if err != nil {
 		return nil, err
 	}
 
-	r := bytes.NewReader(data)
+	r := bytes.NewReader(raw)
 
 	env, err := enmime.ReadEnvelope(r)
 	if err != nil {
@@ -502,9 +503,20 @@ func GetMessageRaw(mailbox, id string) ([]byte, error) {
 		return nil, errors.New("message not found")
 	}
 
-	data := q.Get("Data").(string)
+	var raw []byte
 
-	return []byte(data), err
+	if q.Has("Email") {
+		msg := q.Get("Email").(string)
+		raw, err = decoder.DecodeAll([]byte(msg), nil)
+		if err != nil {
+			return nil, fmt.Errorf("error decompressing message: %s", err.Error())
+		}
+	} else {
+		// deprecated 2022/08/10 - can be eventually removed
+		raw = []byte(q.Get("Data").(string))
+	}
+
+	return raw, err
 }
 
 // UnreadMessage will delete all messages from a mailbox

storage/database_test.go

@@ -18,6 +18,7 @@ import (
 var (
 	testTextEmail []byte
 	testMimeEmail []byte
+	testRuns      = 1000
 )
 
 func TestTextEmailInserts(t *testing.T) {
@@ -29,7 +30,7 @@ RepeatTest:
 
 	assertEqualStats(t, 0, 0)
 
-	for i := 0; i < 1000; i++ {
+	for i := 0; i < testRuns; i++ {
 		if _, err := Store(DefaultMailbox, testTextEmail); err != nil {
 			t.Log("error ", err)
 			t.Fail()
@@ -42,11 +43,11 @@ RepeatTest:
 		t.Fail()
 	}
 
-	assertEqual(t, count, 1000, "incorrect number of text emails stored")
+	assertEqual(t, count, testRuns, "incorrect number of text emails stored")
 
-	t.Logf("inserted 1,000 text emails in %s\n", time.Since(start))
+	t.Logf("inserted %d text emails in %s", testRuns, time.Since(start))
 
-	assertEqualStats(t, 1000, 1000)
+	assertEqualStats(t, testRuns, testRuns)
 
 	delStart := time.Now()
 	if err := DeleteAllMessages(DefaultMailbox); err != nil {
@@ -62,7 +63,7 @@ RepeatTest:
 
 	assertEqual(t, count, 0, "incorrect number of text emails deleted")
 
-	t.Logf("deleted 1,000 text emails in %s\n", time.Since(delStart))
+	t.Logf("deleted %d text emails in %s", testRuns, time.Since(delStart))
 
 	assertEqualStats(t, 0, 0)
 
@@ -85,7 +86,7 @@ RepeatTest:
 
 	assertEqualStats(t, 0, 0)
 
-	for i := 0; i < 1000; i++ {
+	for i := 0; i < testRuns; i++ {
 		if _, err := Store(DefaultMailbox, testMimeEmail); err != nil {
 			t.Log("error ", err)
 			t.Fail()
@@ -98,11 +99,11 @@ RepeatTest:
 		t.Fail()
 	}
 
-	assertEqual(t, count, 1000, "incorrect number of mime emails stored")
+	assertEqual(t, count, testRuns, "incorrect number of emails with mime attachments stored")
 
-	t.Logf("inserted 1,000 emails with mime attachments in %s\n", time.Since(start))
+	t.Logf("inserted %d emails with mime attachments in %s", testRuns, time.Since(start))
 
-	assertEqualStats(t, 1000, 1000)
+	assertEqualStats(t, testRuns, testRuns)
 
 	delStart := time.Now()
 	if err := DeleteAllMessages(DefaultMailbox); err != nil {
@@ -116,9 +117,9 @@ RepeatTest:
 		t.Fail()
 	}
 
-	assertEqual(t, count, 0, "incorrect number of mime emails deleted")
+	assertEqual(t, count, 0, "incorrect number of emails with mime attachments deleted")
 
-	t.Logf("deleted 1,000 mime emails in %s\n", time.Since(delStart))
+	t.Logf("deleted %d emails with mime attachments in %s", testRuns, time.Since(delStart))
 
 	assertEqualStats(t, 0, 0)
 
@@ -222,7 +223,7 @@ func TestSearch(t *testing.T) {
 	t.Log("Testing memory storage")
 
 RepeatTest:
-	for i := 0; i < 1000; i++ {
+	for i := 0; i < testRuns; i++ {
 		msg := enmime.Builder().
 			From(fmt.Sprintf("From %d", i), fmt.Sprintf("from-%d@example.com", i)).
 			Subject(fmt.Sprintf("Subject line %d end", i)).
@@ -248,7 +249,7 @@ RepeatTest:
 		}
 	}
 
-	for i := 1; i < 101; i++ {
+	for i := 1; i < 51; i++ {
 		// search a random something that will return a single result
 		searchIndx := rand.Intn(4) + 1
 		var search string
@@ -263,7 +264,7 @@ RepeatTest:
 			search = fmt.Sprintf("the email body %d jdsauk dwqmdqw", i)
 		}
 
-		summaries, err := Search(DefaultMailbox, search, 0, 200)
+		summaries, err := Search(DefaultMailbox, search, 0, 10)
 		if err != nil {
 			t.Log("error ", err)
 			t.Fail()
@@ -279,12 +280,12 @@ RepeatTest:
 	}
 
 	// search something that will return 200 rsults
-	summaries, err := Search(DefaultMailbox, "This is the email body", 0, 200)
+	summaries, err := Search(DefaultMailbox, "This is the email body", 0, 50)
 	if err != nil {
 		t.Log("error ", err)
 		t.Fail()
 	}
-	assertEqual(t, len(summaries), 200, "200 search results expected")
+	assertEqual(t, len(summaries), 50, "50 search results expected")
 
 	db.Close()
 

storage/utils.go

@@ -56,11 +56,10 @@ func cleanString(str string) string {
 	return strings.ToLower(strings.Join(strings.Fields(strings.TrimSpace(str)), " "))
 }
 
-// Auto-prune runs every 5 minutes to automatically delete oldest messages
+// Auto-prune runs every minute to automatically delete oldest messages
 // if total is greater than the threshold
 func pruneCron() {
 	for {
-		// time.Sleep(5 * 60 * time.Second)
 		time.Sleep(60 * time.Second)
 		mailboxes, err := db.ListCollections()
 		if err != nil {
@@ -94,7 +93,7 @@ func pruneCron() {
 }
 
 // SanitizeMailboxName returns a clean mailbox name
-// allowing only `alphanumeric` characters and `-``
+// allowing only `alphanumeric` characters and `-“
 func sanitizeMailboxName(mailbox string) string {
 	re := regexp.MustCompile(`[^a-zA-Z0-9\-]`)