From 01760b73858a0eb4bf611928dfd25f7cd4eb9a48 Mon Sep 17 00:00:00 2001
From: PP1 <74917296+pengpengyi92@users.noreply.github.com>
Date: Tue, 11 Aug 2026 15:46:38 +0800
Subject: [PATCH] fix(channels): preserve indentation across message splits

Signed-off-by: PP1 <74917296+pengpengyi92@users.noreply.github.com>
---
 nanobot/channels/signal/runtime.py            | 17 ++++++-------
 .../signal/tests/test_signal_markdown.py      |  9 +++++++
 nanobot/utils/helpers.py                      | 25 +++++++++++++------
 tests/utils/test_helpers.py                   | 12 +++++++++
 4 files changed, 46 insertions(+), 17 deletions(-)

diff --git a/nanobot/channels/signal/runtime.py b/nanobot/channels/signal/runtime.py
index c3358e44e..18bebc6f3 100644
--- a/nanobot/channels/signal/runtime.py
+++ b/nanobot/channels/signal/runtime.py
@@ -269,19 +269,18 @@ def _partition_styles(
     if not text_styles:
         return [[] for _ in chunks]
 
-    # Locate each chunk's UTF-16 start in plain_text. split_message lstrips at
-    # boundaries (but not before the first chunk), so we skip whitespace
-    # between chunks to mirror that.
+    # Locate each chunk in the original text. This accounts for delimiters
+    # removed at split points while retaining indentation inside a chunk.
     chunk_ranges: list[tuple[int, int]] = []
     cursor = 0  # Python codepoint cursor in plain_text
-    for i, chunk in enumerate(chunks):
-        if i > 0:
-            while cursor < len(plain_text) and plain_text[cursor].isspace():
-                cursor += 1
-        utf16_start = _utf16_len(plain_text[:cursor])
+    for chunk in chunks:
+        chunk_start = plain_text.find(chunk, cursor)
+        if chunk_start < 0:
+            chunk_start = cursor
+        utf16_start = _utf16_len(plain_text[:chunk_start])
         utf16_end = utf16_start + _utf16_len(chunk)
         chunk_ranges.append((utf16_start, utf16_end))
-        cursor += len(chunk)
+        cursor = chunk_start + len(chunk)
 
     result: list[list[str]] = [[] for _ in chunks]
     for entry in text_styles:
diff --git a/nanobot/channels/signal/tests/test_signal_markdown.py b/nanobot/channels/signal/tests/test_signal_markdown.py
index 2ab8d4cce..33a4d7495 100644
--- a/nanobot/channels/signal/tests/test_signal_markdown.py
+++ b/nanobot/channels/signal/tests/test_signal_markdown.py
@@ -384,6 +384,15 @@ def test_partition_styles_drops_styles_outside_chunks():
     assert parts == [[], []]
 
 
+def test_partition_styles_keeps_offset_in_indented_chunk():
+    """Styles after preserved indentation remain relative to the chunk."""
+    plain = "header\n    code"
+    chunks = split_message(plain, 10)
+
+    assert chunks == ["header", "    code"]
+    assert _partition_styles(plain, chunks, ["11:4:BOLD"]) == [[], ["4:4:BOLD"]]
+
+
 def test_partition_styles_long_message_preserves_chunk_one_styles():
     """A bold span deep in the message must follow the message into chunk 1."""
     # Two ~30-char paragraphs separated by a blank line, then **tail**.
diff --git a/nanobot/utils/helpers.py b/nanobot/utils/helpers.py
index de6f89862..97f3df0b1 100644
--- a/nanobot/utils/helpers.py
+++ b/nanobot/utils/helpers.py
@@ -639,14 +639,23 @@ def split_message(content: str, max_len: int = 2000) -> list[str]:
             chunks.append(content)
             break
         cut = content[:max_len]
-        # Try to break at newline first, then space, then hard break
-        pos = cut.rfind("\n")
-        if pos <= 0:
-            pos = cut.rfind(" ")
-        if pos <= 0:
-            pos = max_len
-        chunks.append(content[:pos])
-        content = content[pos:].lstrip()
+        # Consume only the newline itself so indentation starts the next chunk.
+        newline_pos = cut.rfind("\n")
+        if newline_pos > 0:
+            chunks.append(content[:newline_pos])
+            content = content[newline_pos + 1 :]
+            continue
+
+        # Keep the existing word-boundary behavior, but avoid emitting a
+        # whitespace-only chunk when an indented line exceeds max_len.
+        space_pos = cut.rfind(" ")
+        if space_pos > 0 and cut[:space_pos].strip():
+            chunks.append(content[:space_pos])
+            content = content[space_pos:].lstrip()
+            continue
+
+        chunks.append(content[:max_len])
+        content = content[max_len:]
     return chunks
 
 
diff --git a/tests/utils/test_helpers.py b/tests/utils/test_helpers.py
index c43d760c4..f813a7170 100644
--- a/tests/utils/test_helpers.py
+++ b/tests/utils/test_helpers.py
@@ -17,6 +17,18 @@ def test_split_message_no_code_blocks_unchanged():
     assert split_message(content, max_len=12) == ["alpha beta", "gamma delta"]
 
 
+def test_split_message_preserves_indentation_after_newline():
+    content = "header\n    indented code"
+
+    assert split_message(content, max_len=18) == ["header", "    indented code"]
+
+
+def test_split_message_preserves_indentation_across_hard_break():
+    content = "head\n    abcdefghij"
+
+    assert split_message(content, max_len=8) == ["head", "    abcd", "efghij"]
+
+
 def test_split_message_nonpositive_maxlen_returns_unsplit():
     content = "alpha beta gamma delta"