From 44b14ebdff0e3da8df2348eeda27bfc2ec421fbb Mon Sep 17 00:00:00 2001
From: Robin Jarry <robin@jarry.cc>
Date: Tue, 2 Jul 2024 15:58:41 +0200
Subject: [PATCH] rfc822: be liberal with invalid address headers

Some email clients format email addresses with quotes around B encoded
names. E.g.:

  "=?utf-8?B?U21pZXRhbnNraSwgV29qY2llY2ggVGFkZXVzeiBpbiBUZWFtcw==?="

This seems non standard. I tried reading the multiple RFCs that describe
the MIME headers encoding but could not find any precise answer to
determine if adding quotes around encoded words is valid or not.

In any case, ParseAddressList completely ignores any encoding when it
encounters a double quote.

When that happens, try to detect it and force decoding a second time
after the quotes have been removed.

Link: https://datatracker.ietf.org/doc/html/rfc2045#section-3
Link: https://datatracker.ietf.org/doc/html/rfc2047#section-2
Link: https://datatracker.ietf.org/doc/html/rfc822#section-3.3
Reported-by: Bence Ferdinandy <bence@ferdinandy.com>
Signed-off-by: Robin Jarry <robin@jarry.cc>
Tested-by: Bence Ferdinandy <bence@ferdinandy.com>
---
 lib/rfc822/message.go      |  8 ++++++
 lib/rfc822/message_test.go | 53 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 61 insertions(+)

diff --git a/lib/rfc822/message.go b/lib/rfc822/message.go
index 391ab3e5..26d790ff 100644
--- a/lib/rfc822/message.go
+++ b/lib/rfc822/message.go
@@ -6,6 +6,7 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"mime"
 	"regexp"
 	"strings"
 	"time"
@@ -284,6 +285,13 @@ func parseAddressList(h *mail.Header, key string) ([]*mail.Address, error) {
 		// errors which are not fatal.
 		return nil, err
 	}
+	for _, addr := range addrs {
+		// Handle invalid headers with quoted *AND* encoded names
+		if strings.HasPrefix(addr.Name, "=?") && strings.HasSuffix(addr.Name, "?=") {
+			d := mime.WordDecoder{CharsetReader: message.CharsetReader}
+			addr.Name, _ = d.DecodeHeader(addr.Name)
+		}
+	}
 	// If we got at least one address, ignore any returned error.
 	return addrs, nil
 }
diff --git a/lib/rfc822/message_test.go b/lib/rfc822/message_test.go
index 3ae71cb9..fed3b093 100644
--- a/lib/rfc822/message_test.go
+++ b/lib/rfc822/message_test.go
@@ -9,6 +9,7 @@ import (
 
 	"git.sr.ht/~rjarry/aerc/models"
 	"github.com/emersion/go-message/mail"
+	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
 
@@ -111,6 +112,58 @@ func TestParseMessageDate(t *testing.T) {
 	}
 }
 
+func TestParseAddressList(t *testing.T) {
+	header := mail.HeaderFromMap(map[string][]string{
+		"From": {`"=?utf-8?B?U21pZXRhbnNraSwgV29qY2llY2ggVGFkZXVzeiBpbiBUZWFtcw==?=" <noreply@email.teams.microsoft.com>`},
+		"To":   {`=?UTF-8?q?Oc=C3=A9ane_de_Seazon?= <hello@seazon.fr>`},
+		"Cc":   {`=?utf-8?b?0KjQsNCz0L7QsiDQk9C10L7RgNCz0LjQuSB2aWEgZGlzY3Vzcw==?= <ovs-discuss@openvswitch.org>`},
+		"Bcc":  {`"Foo, Baz Bar" <~foo/baz@bar.org>`},
+	})
+	type vector struct {
+		kind   string
+		header string
+		name   string
+		email  string
+	}
+
+	vectors := []vector{
+		{
+			kind:   "quoted",
+			header: "Bcc",
+			name:   "Foo, Baz Bar",
+			email:  "~foo/baz@bar.org",
+		},
+		{
+			kind:   "Qencoded",
+			header: "To",
+			name:   "Océane de Seazon",
+			email:  "hello@seazon.fr",
+		},
+		{
+			kind:   "Bencoded",
+			header: "Cc",
+			name:   "Шагов Георгий via discuss",
+			email:  "ovs-discuss@openvswitch.org",
+		},
+		{
+			kind:   "quoted+Bencoded",
+			header: "From",
+			name:   "Smietanski, Wojciech Tadeusz in Teams",
+			email:  "noreply@email.teams.microsoft.com",
+		},
+	}
+
+	for _, vec := range vectors {
+		t.Run(vec.kind, func(t *testing.T) {
+			addrs, err := parseAddressList(&header, vec.header)
+			assert.Nil(t, err)
+			assert.Len(t, addrs, 1)
+			assert.Equal(t, vec.name, addrs[0].Name)
+			assert.Equal(t, vec.email, addrs[0].Address)
+		})
+	}
+}
+
 type mockRawMessage struct {
 	path string
 }