rfc822: improve dates parsing

Provide better parsing of email dates with timezone names, but without
numeric offsets. Add unit tests to check the new behaviour.

Signed-off-by: Vitaly Ovchinnikov <v@ovch.ru>
Acked-by: Robin Jarry <robin@jarry.cc>

2 files changed, 91 insertions, 10 deletions

diff --git a/lib/rfc822/message.go b/lib/rfc822/message.go
index 979d4595..7b0d0552 100644
--- a/lib/rfc822/message.go
+++ b/lib/rfc822/message.go
@@ -189,14 +189,45 @@ func parseEnvelope(h *mail.Header) (*models.Envelope, error) {
 	}, err
 }
 
+// If the date is formatted like ...... -0500 (EST), parser takes the EST part
+// and ignores the numeric offset. Then it might easily fail to guess what EST
+// means unless the proper locale is loaded. This function checks that, so such
+// time values can be safely ignored
+// https://stackoverflow.com/questions/49084316/why-doesnt-gos-time-parse-parse-the-timezone-identifier
+func isDateOK(t time.Time) bool {
+	name, offset := t.Zone()
+
+	// non-zero offsets are fine
+	if offset != 0 {
+		return true
+	}
+
+	// zero offset is ok if that's UTC or GMT
+	if name == "UTC" || name == "GMT" || name == "" {
+		return true
+	}
+
+	// otherwise this date should not be trusted
+	return false
+}
+
 // parseDate tries to parse the date from the Date header with non std formats
 // if this fails it tries to parse the received header as well
 func parseDate(h *mail.Header) (time.Time, error) {
+	// here we store the best parsed time we have so far
+	// if we find no "correct" time, we'll use that
+	bestDate := time.Time{}
+
+	// trying the easy way
 	t, err := h.Date()
 	if err == nil {
-		return t, nil
+		if isDateOK(t) {
+			return t, nil
+		}
+		bestDate = t
 	}
 	text, err := h.Text("date")
+
 	// sometimes, no error occurs but the date is empty.
 	// In this case, guess time from received header field
 	if err != nil || text == "" {
@@ -211,15 +242,29 @@ func parseDate(h *mail.Header) (time.Time, error) {
 	}
 	for _, layout := range layouts {
 		if t, err := time.Parse(layout, text); err == nil {
-			return t, nil
+			if isDateOK(t) {
+				return t, nil
+			}
+			bestDate = t
 		}
 	}
-	// still no success, try the received header as a last resort
+
+	// still no success, try the received header
 	t, err = parseReceivedHeader(h)
-	if err != nil {
-		return time.Time{}, fmt.Errorf("unrecognized date format: %s", text)
+	if err == nil {
+		if isDateOK(t) {
+			return t, nil
+		}
+		bestDate = t
 	}
-	return t, nil
+
+	// do we have at least something?
+	if !bestDate.IsZero() {
+		return bestDate, nil
+	}
+
+	// sad...
+	return time.Time{}, fmt.Errorf("unrecognized date format: %s", text)
 }
 
 func parseReceivedHeader(h *mail.Header) (time.Time, error) {
diff --git a/lib/rfc822/message_test.go b/lib/rfc822/message_test.go
index 8730afe2..fd2e8265 100644
--- a/lib/rfc822/message_test.go
+++ b/lib/rfc822/message_test.go
@@ -1,13 +1,15 @@
-package rfc822_test
+package rfc822
 
 import (
 	"io"
 	"os"
 	"path/filepath"
 	"testing"
+	"time"
 
-	"git.sr.ht/~rjarry/aerc/lib/rfc822"
 	"git.sr.ht/~rjarry/aerc/models"
+	"github.com/emersion/go-message/mail"
+	"github.com/stretchr/testify/require"
 )
 
 func TestMessageInfoParser(t *testing.T) {
@@ -23,7 +25,7 @@ func TestMessageInfoParser(t *testing.T) {
 		p := fi.Name()
 		t.Run(p, func(t *testing.T) {
 			m := newMockRawMessageFromPath(filepath.Join(rootDir, p))
-			mi, err := rfc822.MessageInfo(m)
+			mi, err := MessageInfo(m)
 			if err != nil {
 				t.Fatal("Failed to create MessageInfo with:", err)
 			}
@@ -48,7 +50,7 @@ func TestMessageInfoHandledError(t *testing.T) {
 		p := fi.Name()
 		t.Run(p, func(t *testing.T) {
 			m := newMockRawMessageFromPath(filepath.Join(rootDir, p))
-			mi, err := rfc822.MessageInfo(m)
+			mi, err := MessageInfo(m)
 			if err != nil {
 				t.Fatal(err)
 			}
@@ -60,6 +62,40 @@ func TestMessageInfoHandledError(t *testing.T) {
 	}
 }
 
+func TestParseMessageDate(t *testing.T) {
+	// we use different times for "Date" and "Received" fields to make sure the right one is parsed
+	tests := []struct {
+		date     string
+		received string
+		utc      time.Time
+	}{
+		{
+			date:     "Fri, 22 Dec 2023 11:19:01 +0000",
+			received: "from aaa.bbb.com for <user@host.com>; Fri, 22 Dec 2023 06:19:02 -0500 (EST)",
+			utc:      time.Date(2023, time.December, 22, 11, 19, 1, 0, time.UTC),
+		},
+		{
+			date:     "Fri, 29 Dec 2023 14:06:37 +0100",
+			received: "from somewhere.com for a@b.c; Fri, 30 Dec 2023 4:06:43 +1300",
+			utc:      time.Date(2023, time.December, 29, 13, 6, 37, 0, time.UTC),
+		},
+		{
+			date:     "Fri, 29 Dec 2023 00:51:00 EST",
+			received: "by hostname.com; Fri, 29 Dec 2023 00:51:33 -0500 (EST)",
+			utc:      time.Date(2023, time.December, 29, 5, 51, 33, 0, time.UTC),
+		},
+	}
+
+	for _, test := range tests {
+		h := mail.Header{}
+		h.SetText("Date", test.date)
+		h.SetText("Received", test.received)
+		res, err := parseDate(&h)
+		require.Nil(t, err)
+		require.Equal(t, test.utc, res.UTC())
+	}
+}
+
 type mockRawMessage struct {
 	path string
 }