From 8d01063aee836a2a7891cfd508cfc5da8cf41154 Mon Sep 17 00:00:00 2001
From: Valentin Lorentz <progval+git@progval.net>
Date: Sat, 15 Apr 2023 17:54:06 +0200
Subject: [PATCH] Improve testing for invalid UTF8, and add test for too-long
 values

---
 irctest/client_mock.py             | 14 +++++++-
 irctest/server_tests/metadata_2.py | 51 ++++++++++++++++++++++--------
 2 files changed, 51 insertions(+), 14 deletions(-)

diff --git a/irctest/client_mock.py b/irctest/client_mock.py
index e6aaa13..89cf58c 100644
--- a/irctest/client_mock.py
+++ b/irctest/client_mock.py
@@ -74,7 +74,19 @@ class ClientMock:
                     continue
                 if not synchronize:
                     got_pong = True
-                for line in data.decode().split("\r\n"):
+                try:
+                    decoded_data = data.decode()
+                except UnicodeDecodeError:
+                    print(
+                        "{time:.3f}{ssl} S -> {client} - failed to decode: {data!r}".format(
+                            time=time.time(),
+                            ssl=" (ssl)" if self.ssl else "",
+                            client=self.name,
+                            data=data,
+                        )
+                    )
+                    raise
+                for line in decoded_data.split("\r\n"):
                     if line:
                         if self.show_io:
                             print(
diff --git a/irctest/server_tests/metadata_2.py b/irctest/server_tests/metadata_2.py
index 98b1a81..f8dd4b3 100644
--- a/irctest/server_tests/metadata_2.py
+++ b/irctest/server_tests/metadata_2.py
@@ -172,14 +172,7 @@ class MetadataTestCase(cases.BaseServerTestCase):
             "->{}<-".format(heart),
         )
 
-    @cases.xfailIfSoftware(
-        ["UnrealIRCd"], "UnrealIRCd does not validate UTF-8 in metadata values"
-    )
-    @cases.mark_specifications("IRCv3")
-    def testSetInvalidUtf8(self):
-        """“Values are unrestricted, except that they MUST be UTF-8.”
-        -- <http://ircv3.net/specs/core/metadata-3.2.html#metadata-restrictions>
-        """
+    def _testSetInvalidValue(self, value):
         self.connectClient(
             "foo", capabilities=["draft/metadata-2", "batch"], skip_if_cap_nak=True
         )
@@ -198,13 +191,45 @@ class MetadataTestCase(cases.BaseServerTestCase):
             fail_msg="Setting METADATA key to a value containing invalid "
             "UTF-8 was answered with 761 (RPL_KEYVALUE)",
         )
-        self.clients[1].conn.sendall(
-            b"METADATA * SET valid_key1 " b":invalid UTF-8: \xc3\r\n"
+        self.clients[1].conn.sendall(b"METADATA * SET valid_key1 :" + value + b"\r\n")
+        self.assertMessageMatch(
+            self.getMessage(1),
+            command="FAIL",
+            params=["METADATA", "INVALID_VALUE", ANYSTR],
         )
-        commands = {m.command for m in self.getMessages(1)}
+        messages = self.getMessages(1)
         self.assertNotIn(
-            "761",
-            commands,  # RPL_KEYVALUE
+            "761",  # RPL_KEYVALUE
+            {m.command for m in messages},
             fail_msg="Setting METADATA key to a value containing invalid "
             "UTF-8 was answered with 761 (RPL_KEYVALUE)",
         )
+        self.assertEqual(
+            messages,
+            [],
+            fail_msg="Unexpected response to METADATA SET with invalid value: {got}",
+        )
+
+    @cases.mark_specifications("IRCv3")
+    def testSetInvalidUtf8(self):
+        """“Values are unrestricted, except that they MUST be UTF-8.”
+        -- <http://ircv3.net/specs/core/metadata-3.2.html#metadata-restrictions>
+        """
+        self._testSetInvalidValue(b"invalid UTF-8: \xc3")
+
+    @cases.mark_specifications("IRCv3")
+    def testSetTooManyChars(self):
+        """Assumes all servers reject values over 480 bytes. This isn't required by the
+        spec, but makes them risk overflowing when printing the value, so they probably
+        won't allow that.
+        """
+        self._testSetInvalidValue(b"abcd" * 120)
+
+    @cases.mark_specifications("IRCv3")
+    def testSetTooManyBytes(self):
+        """Assumes all servers reject values over 480 bytes. This isn't required by the
+        spec, but makes them risk overflowing when printing the value, so they probably
+        won't allow that.
+        """
+        heart = b"\xf0\x9f\x92\x9c"
+        self._testSetInvalidValue(heart * 120)