fix: keep Telegram streamed code blocks balanced

Maintainer edit: split final streamed Telegram markdown before rendering to HTML so long fenced code blocks do not produce unbalanced <pre><code> chunks while still respecting Telegram's rendered HTML limit.

nanobot/channels/telegram.py

@@ -36,9 +36,9 @@ from nanobot.utils.helpers import split_message
 
 TELEGRAM_MAX_MESSAGE_LEN = 4000  # Telegram message character limit
 # Telegram's actual API limit is 4096; we split raw markdown at 4000 as a
-# safety margin for mid-stream edits (plain text).  For _stream_end, we
+# safety margin for mid-stream edits (plain text).  For _stream_end, we split
-# convert to HTML first and then split at the true 4096-char boundary so
+# raw markdown into chunks whose rendered HTML fits Telegram's true 4096-char
-# the final rendered message never overflows.
+# boundary so the final rendered message never overflows.
 TELEGRAM_HTML_MAX_LEN = 4096
 TELEGRAM_REPLY_CONTEXT_MAX_LEN = TELEGRAM_MAX_MESSAGE_LEN  # Max length for reply context in user message
 
@@ -285,6 +285,32 @@ def _markdown_to_telegram_html(text: str) -> str:
     return text
 
 
+def _split_telegram_markdown_html(content: str, max_html_len: int) -> list[str]:
+    """Split raw Telegram Markdown and return HTML chunks within Telegram's limit."""
+    chunks: list[str] = []
+    pending = _split_telegram_markdown(content, TELEGRAM_MAX_MESSAGE_LEN)
+    while pending:
+        chunk = pending.pop(0)
+        html = _markdown_to_telegram_html(chunk)
+        if len(html) <= max_html_len:
+            chunks.append(html)
+            continue
+
+        # Markdown can expand when rendered as HTML (tags/entities). Re-split
+        # the raw markdown with a smaller budget instead of slicing HTML tags.
+        next_limit = max(1, int(len(chunk) * max_html_len / len(html)) - 8)
+        next_limit = min(next_limit, len(chunk) - 1)
+        if next_limit <= 0:
+            chunks.extend(split_message(html, max_html_len))
+            continue
+        parts = _split_telegram_markdown(chunk, next_limit)
+        if len(parts) == 1 and parts[0] == chunk:
+            chunks.extend(split_message(html, max_html_len))
+            continue
+        pending = parts + pending
+    return chunks
+
+
 _SEND_MAX_RETRIES = 3
 _SEND_RETRY_BASE_DELAY = 0.5  # seconds, doubled each retry
 _STREAM_EDIT_INTERVAL_DEFAULT = 0.6  # min seconds between edit_message_text calls
@@ -800,14 +826,9 @@ class TelegramChannel(BaseChannel):
             if message_thread_id := meta.get("message_thread_id"):
                 thread_kwargs["message_thread_id"] = message_thread_id
             raw_text = buf.text
-            html = _markdown_to_telegram_html(raw_text)
+            html_chunks = _split_telegram_markdown_html(raw_text, TELEGRAM_HTML_MAX_LEN)
-            if len(html) <= TELEGRAM_HTML_MAX_LEN:
+            primary_html = html_chunks[0]
-                primary_html = html
+            extra_html_chunks = html_chunks[1:]
-                extra_html_chunks = []
-            else:
-                html_chunks = split_message(html, TELEGRAM_HTML_MAX_LEN)
-                primary_html = html_chunks[0]
-                extra_html_chunks = html_chunks[1:]
             try:
                 await self._call_with_retry(
                     self._app.bot.edit_message_text,

tests/channels/test_telegram_channel.py

@@ -719,6 +719,36 @@ async def test_send_delta_stream_end_html_expansion_does_not_overflow() -> None:
     assert "123" not in channel._stream_bufs
 
 
+@pytest.mark.asyncio
+async def test_send_delta_stream_end_splits_long_code_block_before_html_rendering() -> None:
+    """Final streamed replies must not split Telegram HTML inside <pre><code>."""
+    channel = TelegramChannel(
+        TelegramConfig(enabled=True, token="123:abc", allow_from=["*"]),
+        MessageBus(),
+    )
+    channel._app = _FakeApp(lambda: None)
+    channel._app.bot.edit_message_text = AsyncMock()
+    channel._app.bot.send_message = AsyncMock(return_value=SimpleNamespace(message_id=99))
+
+    raw_text = "```python\n" + ("print(\"line\")\n" * 450) + "```\nDone"
+    channel._stream_bufs["123"] = _StreamBuf(text=raw_text, message_id=7, last_edit=0.0)
+
+    await channel.send_delta("123", "", {"_stream_end": True})
+
+    html_chunks = [
+        channel._app.bot.edit_message_text.call_args.kwargs.get("text", ""),
+        *[
+            call.kwargs.get("text", "")
+            for call in channel._app.bot.send_message.call_args_list
+        ],
+    ]
+    assert len(html_chunks) > 1
+    for html in html_chunks:
+        assert len(html) <= 4096
+        assert html.count("<pre><code>") == html.count("</code></pre>")
+    assert "123" not in channel._stream_bufs
+
+
 @pytest.mark.asyncio
 async def test_send_delta_new_stream_id_replaces_stale_buffer() -> None:
     channel = TelegramChannel(