Merge pull request #236 from vladak/unsuback_vs_publish

add test for the PUBLISH+UNSUBACK case

tests/test_unsubscribe.py

@@ -68,6 +68,49 @@ def handle_unsubscribe(client, user_data, topic, pid):
             + [0x6F] * 257
         ),
     ),
+    # UNSUBSCRIBE responded to by PUBLISH followed by UNSUBACK
+    (
+        "foo/bar",
+        bytearray(
+            [
+                0x30,  # PUBLISH
+                0x0C,
+                0x00,
+                0x07,
+                0x66,
+                0x6F,
+                0x6F,
+                0x2F,
+                0x62,
+                0x61,
+                0x72,
+                0x66,
+                0x6F,
+                0x6F,
+                0xB0,  # UNSUBACK
+                0x02,
+                0x00,
+                0x01,
+            ]
+        ),
+        bytearray(
+            [
+                0xA2,  # fixed header
+                0x0B,  # remaining length
+                0x00,
+                0x01,  # message ID
+                0x00,
+                0x07,  # topic length
+                0x66,  # topic
+                0x6F,
+                0x6F,
+                0x2F,
+                0x62,
+                0x61,
+                0x72,
+            ]
+        ),
+    ),
     # use list of topics for more coverage. If the range was (1, 10000), that would be
     # long enough to use 3 bytes for remaining length, however that would make the test
     # run for many minutes even on modern systems, so 1000 is used instead.
@@ -95,7 +138,7 @@ def handle_unsubscribe(client, user_data, topic, pid):
 @pytest.mark.parametrize(
     "topic,to_send,exp_recv",
     testdata,
-    ids=["short_topic", "long_topic", "topic_list_long"],
+    ids=["short_topic", "long_topic", "publish_first", "topic_list_long"],
 )
 def test_unsubscribe(topic, to_send, exp_recv) -> None:
     """