feat(imapx): message dedup key (Message-ID + header fallback)

internal/imapx/messagekey.go

@@ -0,0 +1,37 @@
+package imapx
+
+import (
+	"crypto/md5"
+	"fmt"
+	"strings"
+
+	"github.com/emersion/go-imap/v2"
+)
+
+func MessageKey(env *imap.Envelope, size int64) string {
+	if env != nil && env.MessageID != "" {
+		return env.MessageID
+	}
+	var b strings.Builder
+	if env != nil {
+		b.WriteString(addrList(env.From))
+		b.WriteByte('|')
+		b.WriteString(addrList(env.To))
+		b.WriteByte('|')
+		b.WriteString(env.Subject)
+		b.WriteByte('|')
+		b.WriteString(env.Date.UTC().Format("2006-01-02T15:04:05Z"))
+	}
+	b.WriteByte('|')
+	fmt.Fprintf(&b, "%d", size)
+	sum := md5.Sum([]byte(b.String()))
+	return fmt.Sprintf("h:%x", sum)
+}
+
+func addrList(addrs []imap.Address) string {
+	parts := make([]string, 0, len(addrs))
+	for _, a := range addrs {
+		parts = append(parts, a.Mailbox+"@"+a.Host)
+	}
+	return strings.Join(parts, ",")
+}

internal/imapx/messagekey_test.go

@@ -0,0 +1,32 @@
+package imapx
+
+import (
+	"testing"
+	"time"
+
+	"github.com/emersion/go-imap/v2"
+)
+
+func TestMessageKeyPrefersMessageID(t *testing.T) {
+	env := &imap.Envelope{MessageID: "<abc@host>"}
+	if got := MessageKey(env, 100); got != "<abc@host>" {
+		t.Fatalf("got %q, want <abc@host>", got)
+	}
+}
+
+func TestMessageKeyFallbackStable(t *testing.T) {
+	env := &imap.Envelope{
+		Subject: "Hi",
+		Date:    time.Unix(1700000000, 0).UTC(),
+		From:    []imap.Address{{Mailbox: "a", Host: "x.com"}},
+		To:      []imap.Address{{Mailbox: "b", Host: "y.com"}},
+	}
+	k1 := MessageKey(env, 42)
+	k2 := MessageKey(env, 42)
+	if k1 != k2 {
+		t.Fatal("fallback key must be deterministic")
+	}
+	if MessageKey(env, 43) == k1 {
+		t.Fatal("different size must change key")
+	}
+}