fix: encrypted file handler + summary heading/markup fixes

- Add RoomEncryptedFile handler for PDFs/docs in encrypted rooms
- Tell summary LLM not to include headings (prevents duplicate)
- Strip <br/> after block elements in _md_to_html

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

bot.py

@@ -21,6 +21,7 @@ from nio import (
     LoginResponse,
     InviteMemberEvent,
     MegolmEvent,
+    RoomEncryptedFile,
     RoomEncryptedImage,
     RoomMessageFile,
     RoomMessageImage,
@@ -339,6 +340,7 @@ class Bot:
         self.client.add_event_callback(self.on_image_message, RoomMessageImage)
         self.client.add_event_callback(self.on_encrypted_image_message, RoomEncryptedImage)
         self.client.add_event_callback(self.on_file_message, RoomMessageFile)
+        self.client.add_event_callback(self.on_encrypted_file_message, RoomEncryptedFile)
         self.client.add_event_callback(self.on_room_unknown, RoomMessageUnknown)
         self.client.add_response_callback(self.on_sync, SyncResponse)
         self.client.add_to_device_callback(self.on_key_verification, KeyVerificationStart)
@@ -1041,6 +1043,95 @@ class Bot:
         finally:
             await self.client.room_typing(room.room_id, typing_state=False)
 
+    async def on_encrypted_file_message(self, room, event: RoomEncryptedFile):
+        """Handle encrypted file messages: decrypt and process like on_file_message."""
+        if event.sender == BOT_USER:
+            return
+        if not self._sync_token_received:
+            return
+        server_ts = event.server_timestamp / 1000
+        if time.time() - server_ts > 30:
+            return
+
+        source = event.source or {}
+        content = source.get("content", {})
+        filename = content.get("body", "file")
+        ext = os.path.splitext(filename.lower())[1]
+
+        is_pdf = ext == ".pdf"
+        is_docx = ext == ".docx"
+        is_text = ext in self._TEXT_EXTENSIONS
+
+        if not (is_pdf or is_docx or is_text):
+            return
+
+        await self._load_room_settings(room.room_id)
+
+        is_dm = room.member_count == 2
+        if not is_dm:
+            body = (event.body or "").strip()
+            bot_display = self.client.user_id.split(":")[0].lstrip("@")
+            mentioned = (
+                BOT_USER in body
+                or f"@{bot_display}" in body.lower()
+                or bot_display.lower() in body.lower()
+            )
+            if not mentioned:
+                return
+
+        if not self.llm:
+            await self._send_text(room.room_id, "LLM not configured (LITELLM_BASE_URL not set).")
+            return
+
+        mxc_url = event.url
+        if not mxc_url:
+            return
+        try:
+            resp = await self.client.download(mxc=mxc_url)
+            if not hasattr(resp, "body"):
+                logger.warning("Encrypted file download failed for %s", mxc_url)
+                return
+            file_bytes = decrypt_attachment(resp.body, event.key["k"], event.hashes["sha256"], event.iv)
+        except Exception:
+            logger.exception("Failed to download/decrypt encrypted file %s", mxc_url)
+            return
+
+        if is_pdf:
+            extracted = self._extract_pdf_text(file_bytes)
+            doc_type = "pdf"
+        elif is_docx:
+            extracted = self._extract_docx_text(file_bytes)
+            doc_type = "text"
+        else:
+            extracted = self._extract_text_file(file_bytes)
+            doc_type = "text"
+
+        if not extracted:
+            await self._send_text(room.room_id, f"I couldn't extract any text from that file ({filename}).")
+            return
+
+        if len(extracted) > 50000:
+            extracted = extracted[:50000] + "\n\n[... truncated, file too long ...]"
+
+        docs = self._room_document_context.setdefault(room.room_id, [])
+        docs.append({
+            "type": doc_type,
+            "filename": filename,
+            "text": extracted,
+            "timestamp": time.time(),
+        })
+        if len(docs) > 5:
+            del docs[:-5]
+
+        label = "PDF" if is_pdf else "Word document" if is_docx else "file"
+        user_message = f'The user sent a {label} named "{filename}". Here is the extracted text:\n\n{extracted}\n\nPlease summarize or answer questions about this document.'
+
+        await self.client.room_typing(room.room_id, typing_state=True)
+        try:
+            await self._respond_with_ai(room, user_message, sender=event.sender)
+        finally:
+            await self.client.room_typing(room.room_id, typing_state=False)
+
     @staticmethod
     def _extract_pdf_text(pdf_bytes: bytes) -> str:
         """Extract text from PDF bytes using pymupdf."""
@@ -1503,6 +1594,9 @@ class Bot:
         safe = re.sub(r"^# (.+)$", r"<h1>\1</h1>", safe, flags=re.MULTILINE)
         # Line breaks
         safe = safe.replace("\n", "<br/>")
+        # Remove redundant <br/> after block elements
+        safe = re.sub(r"(</h[1-6]>)(<br/>)+", r"\1", safe)
+        safe = re.sub(r"(</pre>)(<br/>)+", r"\1", safe)
         return safe
 
     async def _generate_and_send_image(self, room_id: str, prompt: str):
@@ -1577,7 +1671,8 @@ class Bot:
                     {"role": "system", "content": (
                         "Fasse das folgende Anruf-Transkript kurz und praegnant zusammen. "
                         "Nenne die wichtigsten besprochenen Punkte, Entscheidungen und offene Fragen. "
-                        "Antworte in der Sprache des Gespraechs. Maximal 5-8 Saetze."
+                        "Antworte in der Sprache des Gespraechs. Maximal 5-8 Saetze. "
+                        "Keine Ueberschrift, kein Markdown-Heading — beginne direkt mit dem Text."
                     )},
                     {"role": "user", "content": transcript_text},
                 ],