Tentacle enhancements.

sdk/python/cthulhu_tentacle/session.py

@@ -37,6 +37,56 @@ def _token(value):
     return isinstance(value, str) and len(value.encode("utf-8")) == 32
 
 
+def valid_remote_speech(speech):
+    """Validate typed speech including its encoded IPC bound; no delivery."""
+    try:
+        _fields(speech, ("kind",), ("priority", "parts", "paused"))
+        priority, parts = speech.get("priority", 0), speech.get("parts", [])
+        if type(priority) is not int or not isinstance(parts, list):
+            return False
+        if speech["kind"] == "speak":
+            if not 0 <= priority <= 2 or "paused" in speech or not 1 <= len(parts) <= 128:
+                return False
+            for part in parts:
+                _fields(part, (), ("text", "characters", "language", "end"))
+                text, language = part.get("text", ""), part.get("language", "")
+                characters, end = part.get("characters", False), part.get("end", False)
+                if (not isinstance(text, str) or not isinstance(language, str)
+                        or type(characters) is not bool or type(end) is not bool):
+                    return False
+                if end:
+                    if text or language or characters:
+                        return False
+                elif (not text or len(language) > 64
+                      or any(char not in "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-_" for char in language)):
+                    return False
+        elif speech["kind"] == "cancel":
+            if priority or parts or "paused" in speech:
+                return False
+        elif speech["kind"] == "pause":
+            if priority or parts or type(speech.get("paused")) is not bool:
+                return False
+        else:
+            return False
+        # Match Go's canonical omitempty representation and JSON HTML escaping
+        # so both SDKs enforce the same payload bound, including escaped text.
+        canonical = {"kind": speech["kind"]}
+        if priority:
+            canonical["priority"] = priority
+        if parts:
+            canonical["parts"] = [{key: value for key, value in part.items() if value}
+                                  for part in parts]
+        if "paused" in speech:
+            canonical["paused"] = speech["paused"]
+        encoded = _encode(canonical)
+        for raw, escaped in ((b"<", b"\\u003c"), (b">", b"\\u003e"), (b"&", b"\\u0026"),
+                             (b"\xe2\x80\xa8", b"\\u2028"), (b"\xe2\x80\xa9", b"\\u2029")):
+            encoded = encoded.replace(raw, escaped)
+        return len(encoded) <= 6144
+    except (ProtocolError, TypeError, ValueError, UnicodeError):
+        return False
+
+
 class SessionError(Exception):
     def __init__(self, code, message):
         super().__init__("Tentacle session error " + str(code))
@@ -226,13 +276,13 @@ class SessionClient:
             waiter.put_nowait(value)
 
     def _event(self, event):
-        _fields(event, ("sequence", "kind", "epoch"), ("action", "lease", "role", "grants", "reason"))
+        _fields(event, ("sequence", "kind", "epoch"), ("action", "lease", "role", "grants", "reason", "speech"))
         if not _positive(event["sequence"]) or not _positive(event["epoch"]):
             raise ProtocolError("Invalid event sequence")
         if event["kind"] == "remote.action":
             grants = event.get("grants")
             role = event.get("role")
-            if (not _token(event.get("action")) or event.get("lease", "") != "" or event.get("reason", "") != ""
+            if ("speech" in event or not _token(event.get("action")) or event.get("lease", "") != "" or event.get("reason", "") != ""
                     or role not in ("controller", "controlled") or not isinstance(grants, list)
                     or not grants or any(not isinstance(grant, str) for grant in grants)
                     or len(set(grants)) != len(grants) or not set(grants) <= REMOTE_GRANTS
@@ -240,9 +290,14 @@ class SessionClient:
                     or role == "controlled" and set(grants) & {"remote-present", "keyboard-capture"}):
                 raise ProtocolError("Invalid action event")
         elif event["kind"] == "remote.revoked":
-            if (not _token(event.get("lease")) or event.get("action", "") != "" or event.get("role", "") != ""
+            if ("speech" in event or not _token(event.get("lease")) or event.get("action", "") != "" or event.get("role", "") != ""
                     or event.get("grants", []) != [] or event.get("reason") not in ("closed", "expired", "local", "replaced")):
                 raise ProtocolError("Invalid revoke event")
+        elif event["kind"] == "remote.output":
+            if (not _token(event.get("lease")) or event.get("action", "") != ""
+                    or event.get("role", "") != "" or event.get("grants", []) != []
+                    or event.get("reason", "") != "" or not valid_remote_speech(event.get("speech"))):
+                raise ProtocolError("Invalid output event")
         else:
             raise ProtocolError("Unknown session event")
         with self._lock: