Fix imap_email_content unknown status and replaying stale states

homeassistant/components/imap_email_content/sensor.py

@@ -95,9 +95,25 @@ def __init__(self, user, password, server, port, folder, verify_ssl):
         self._folder = folder
         self._verify_ssl = verify_ssl
         self._last_id = None
+        self._last_message = None
         self._unread_ids = deque([])
         self.connection = None
 
+    @property
+    def last_id(self) -> int | None:
+        """Return last email uid that was processed."""
+        return self._last_id
+
+    @property
+    def last_unread_id(self) -> int | None:
+        """Return last email uid received."""
+        # We assume the last id in the list is the last unread id
+        # We cannot know if that is the newest one, because it could arrive later
+        # https://stackoverflow.com/questions/12409862/python-imap-the-order-of-uids
+        if self._unread_ids:
+            return int(self._unread_ids[-1])
+        return self._last_id
+
     def connect(self):
         """Login and setup the connection."""
         ssl_context = client_context() if self._verify_ssl else None
@@ -128,21 +144,21 @@ def read_next(self):
         try:
             self.connection.select(self._folder, readonly=True)
 
-            if not self._unread_ids:
-                search = f"SINCE {datetime.date.today():%d-%b-%Y}"
-                if self._last_id is not None:
-                    search = f"UID {self._last_id}:*"
-
-                _, data = self.connection.uid("search", None, search)
-                self._unread_ids = deque(data[0].split())
+            if self._last_id is None:
+                # search for today and yesterday
+                time_from = datetime.datetime.now() - datetime.timedelta(days=1)
+                search = f"SINCE {time_from:%d-%b-%Y}"
+            else:
+                search = f"UID {self._last_id}:*"
 
+            _, data = self.connection.uid("search", None, search)
+            self._unread_ids = deque(data[0].split())
             while self._unread_ids:
                 message_uid = self._unread_ids.popleft()
                 if self._last_id is None or int(message_uid) > self._last_id:
                     self._last_id = int(message_uid)
-                    return self._fetch_message(message_uid)
-
-            return self._fetch_message(str(self._last_id))
+                    self._last_message = self._fetch_message(message_uid)
+                    return self._last_message
 
         except imaplib.IMAP4.error:
             _LOGGER.info("Connection to %s lost, attempting to reconnect", self._server)
@@ -254,22 +270,30 @@ def get_msg_text(email_message):
     def update(self) -> None:
         """Read emails and publish state change."""
         email_message = self._email_reader.read_next()
-
-        if email_message is None:
-            self._message = None
-            self._state_attributes = {}
-            return
-
-        if self.sender_allowed(email_message):
-            message = EmailContentSensor.get_msg_subject(email_message)
-
-            if self._value_template is not None:
-                message = self.render_template(email_message)
-
-            self._message = message
-            self._state_attributes = {
-                ATTR_FROM: EmailContentSensor.get_msg_sender(email_message),
-                ATTR_SUBJECT: EmailContentSensor.get_msg_subject(email_message),
-                ATTR_DATE: email_message["Date"],
-                ATTR_BODY: EmailContentSensor.get_msg_text(email_message),
-            }
+        while (
+            self._last_id is None or self._last_id != self._email_reader.last_unread_id
+        ):
+            if email_message is None:
+                self._message = None
+                self._state_attributes = {}
+                return
+
+            self._last_id = self._email_reader.last_id
+
+            if self.sender_allowed(email_message):
+                message = EmailContentSensor.get_msg_subject(email_message)
+
+                if self._value_template is not None:
+                    message = self.render_template(email_message)
+
+                self._message = message
+                self._state_attributes = {
+                    ATTR_FROM: EmailContentSensor.get_msg_sender(email_message),
+                    ATTR_SUBJECT: EmailContentSensor.get_msg_subject(email_message),
+                    ATTR_DATE: email_message["Date"],
+                    ATTR_BODY: EmailContentSensor.get_msg_text(email_message),
+                }
+
+            if self._last_id == self._email_reader.last_unread_id:
+                break
+            email_message = self._email_reader.read_next()

tests/components/imap_email_content/test_sensor.py

@@ -14,9 +14,16 @@
 class FakeEMailReader:
     """A test class for sending test emails."""
 
-    def __init__(self, messages):
+    def __init__(self, messages) -> None:
         """Set up the fake email reader."""
         self._messages = messages
+        self.last_id = 0
+        self.last_unread_id = len(messages)
+
+    def add_test_message(self, message):
+        """Add a new message."""
+        self.last_unread_id += 1
+        self._messages.append(message)
 
     def connect(self):
         """Stay always Connected."""
@@ -26,6 +33,7 @@ def read_next(self):
         """Get the next email."""
         if len(self._messages) == 0:
             return None
+        self.last_id += 1
         return self._messages.popleft()
 
 
@@ -146,7 +154,7 @@ async def test_multi_part_only_other_text(hass: HomeAssistant) -> None:
 
 
 async def test_multiple_emails(hass: HomeAssistant) -> None:
-    """Test multiple emails."""
+    """Test multiple emails, discarding stale states."""
     states = []
 
     test_message1 = email.message.Message()
@@ -158,9 +166,15 @@ async def test_multiple_emails(hass: HomeAssistant) -> None:
     test_message2 = email.message.Message()
     test_message2["From"] = "sender@test.com"
     test_message2["Subject"] = "Test 2"
-    test_message2["Date"] = datetime.datetime(2016, 1, 1, 12, 44, 57)
+    test_message2["Date"] = datetime.datetime(2016, 1, 1, 12, 44, 58)
     test_message2.set_payload("Test Message 2")
 
+    test_message3 = email.message.Message()
+    test_message3["From"] = "sender@test.com"
+    test_message3["Subject"] = "Test 3"
+    test_message3["Date"] = datetime.datetime(2016, 1, 1, 12, 50, 1)
+    test_message3.set_payload("Test Message 2")
+
     def state_changed_listener(entity_id, from_s, to_s):
         states.append(to_s)
 
@@ -178,11 +192,13 @@ def state_changed_listener(entity_id, from_s, to_s):
 
     sensor.async_schedule_update_ha_state(True)
     await hass.async_block_till_done()
+    # Fake a new received message
+    sensor._email_reader.add_test_message(test_message3)
     sensor.async_schedule_update_ha_state(True)
     await hass.async_block_till_done()
 
-    assert states[0].state == "Test"
-    assert states[1].state == "Test 2"
+    assert states[0].state == "Test 2"
+    assert states[1].state == "Test 3"
 
     assert sensor.extra_state_attributes["body"] == "Test Message 2"