christian/matrix-ai-agent
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: E2EE key lookup for Element Call voice sessions

Browse Source 
- Fix state_key format: try @user:domain:DEVICE_ID (Element Call format),
  then @user:domain, then scan all room state as fallback
- Publish bot E2EE key to room so Element shows encrypted status
- Extract caller device_id from call member event content
- Also fix pipecat-poc pipeline with context aggregators (CF-1579)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Christian Gick
2026-02-21 14:51:26 +02:00
parent c60dfc0cef
commit 80582860b9
2 changed files with 196 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

81
bot.py
View File

@@ -424,14 +424,19 @@ class Bot:
model=model,
)
# Read existing encryption keys from room state before starting
caller_key = await self._get_call_encryption_key(room_id, event.sender)
caller_device_id = content.get("device_id", "")
caller_key = await self._get_call_encryption_key(room_id, event.sender, caller_device_id)
if caller_key:
vs.on_encryption_key(event.sender, "", caller_key, 0)
vs.on_encryption_key(event.sender, caller_device_id, caller_key, 0)
await vs.start()
self.voice_sessions[room_id] = vs
logger.info("Voice session started for room %s (e2ee_key=%s)",
room_id, "yes" if caller_key else "no")
# Publish our E2EE key so Element Call sees us as encrypted
if caller_key:
await self._publish_encryption_key(room_id, caller_key)
except Exception:
logger.exception("Voice session start failed for %s", room_id)
@@ -1430,24 +1435,64 @@ class Bot:
},
)
async def _get_call_encryption_key(self, room_id: str, sender: str) -> bytes | None:
"""Read E2EE encryption key from io.element.call.encryption_keys state events."""
async def _get_call_encryption_key(self, room_id: str, sender: str, caller_device_id: str = "") -> bytes | None:
"""Read E2EE encryption key from io.element.call.encryption_keys state events.
Element Call uses state_key format: @user:domain:DEVICE_ID
Falls back to trying just @user:domain and scanning all room state.
"""
# Try state_key formats: @user:domain:device_id, then @user:domain
state_keys_to_try = []
if caller_device_id:
state_keys_to_try.append(f"{sender}:{caller_device_id}")
state_keys_to_try.append(sender)
for state_key in state_keys_to_try:
try:
resp = await self.client.room_get_state_event(
room_id, ENCRYPTION_KEYS_TYPE, state_key,
)
key = self._extract_e2ee_key(resp, sender, state_key)
if key:
return key
except Exception as e:
logger.debug("No encryption key at state_key=%s: %s", state_key, e)
# Fallback: scan all room state for any encryption_keys event
try:
resp = await self.client.room_get_state_event(
room_id, ENCRYPTION_KEYS_TYPE, sender,
)
if hasattr(resp, "content") and resp.content:
keys = resp.content.get("keys", [])
if keys:
key_b64 = keys[0].get("key", "")
if key_b64:
# Element Call uses base64url encoding
key_b64 += "=" * (-len(key_b64) % 4) # pad
key = base64.urlsafe_b64decode(key_b64)
logger.info("Got E2EE key from %s (%d bytes)", sender, len(key))
return key
resp = await self.client.room_get_state(room_id)
if hasattr(resp, "events"):
for evt in resp.events:
if evt.get("type") == ENCRYPTION_KEYS_TYPE and evt.get("sender") != BOT_USER:
content = evt.get("content", {})
keys = content.get("keys", [])
for k in keys:
key_b64 = k.get("key", "")
if key_b64:
key_b64 += "=" * (-len(key_b64) % 4)
key = base64.urlsafe_b64decode(key_b64)
logger.info("Got E2EE key from room state scan (%s, %d bytes)",
evt.get("state_key", "?"), len(key))
return key
except Exception as e:
logger.debug("No encryption key from %s in %s: %s", sender, room_id, e)
logger.debug("Room state scan for encryption keys failed: %s", e)
logger.warning("No E2EE encryption key found for %s in %s", sender, room_id)
return None
@staticmethod
def _extract_e2ee_key(resp, sender: str, state_key: str) -> bytes | None:
"""Extract E2EE key bytes from a state event response."""
if not hasattr(resp, "content") or not resp.content:
return None
keys = resp.content.get("keys", [])
for k in keys:
key_b64 = k.get("key", "")
if key_b64:
key_b64 += "=" * (-len(key_b64) % 4)
key = base64.urlsafe_b64decode(key_b64)
logger.info("Got E2EE key from %s (state_key=%s, %d bytes)", sender, state_key, len(key))
return key
return None
async def _publish_encryption_key(self, room_id: str, key: bytes):