From 5e647e99465d51dc900e64f1af75a5d059c38c24 Mon Sep 17 00:00:00 2001
From: Friende <35026241+pengyou200902@users.noreply.github.com>
Date: Thu, 7 May 2026 04:21:11 -0400
Subject: [PATCH] fix(memory): truncate at UTF-8 char boundary

Summary:
- Truncate oversized memory prompt content at the previous valid UTF-8 character boundary.
- Preserve the existing memory cap and truncation marker behavior.
- Add regression coverage for accented and emoji content crossing the byte cutoff.

Maintainer verification on current origin/main:
- cargo test -p deepseek-tui memory --locked
- cargo fmt --all -- --check
- git diff --check origin/main...HEAD
---
 crates/tui/src/memory.rs | 54 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 53 insertions(+), 1 deletion(-)

diff --git a/crates/tui/src/memory.rs b/crates/tui/src/memory.rs
index 8a61b6ef..fa7da78d 100644
--- a/crates/tui/src/memory.rs
+++ b/crates/tui/src/memory.rs
@@ -56,7 +56,8 @@ pub fn as_system_block(content: &str, source: &Path) -> Option<String> {
 
     let display = source.display();
     let payload = if content.len() > MAX_MEMORY_SIZE {
-        let mut head = content[..MAX_MEMORY_SIZE].to_string();
+        let cutoff = previous_char_boundary(content, MAX_MEMORY_SIZE);
+        let mut head = content[..cutoff].to_string();
         head.push_str("\n…(truncated, raise [memory].max_size or trim memory.md)");
         head
     } else {
@@ -68,6 +69,13 @@ pub fn as_system_block(content: &str, source: &Path) -> Option<String> {
     ))
 }
 
+fn previous_char_boundary(value: &str, mut index: usize) -> usize {
+    while !value.is_char_boundary(index) {
+        index -= 1;
+    }
+    index
+}
+
 /// Compose the `<user_memory>` block for the system prompt, honouring the
 /// opt-in toggle. Returns `None` when the feature is disabled or the file
 /// is missing / empty so the caller doesn't have to check both conditions.
@@ -161,6 +169,50 @@ mod tests {
         assert!(block.contains("(truncated"));
     }
 
+    #[test]
+    fn as_system_block_truncates_non_ascii_at_char_boundary() {
+        let mut content = "x".repeat(MAX_MEMORY_SIZE - 1);
+        content.push('é');
+        content.push_str("tail");
+
+        let block = as_system_block(&content, Path::new("/tmp/m.md")).unwrap();
+        let payload = block
+            .strip_prefix("<user_memory source=\"/tmp/m.md\">\n")
+            .unwrap()
+            .strip_suffix("\n</user_memory>")
+            .unwrap();
+        let (head, marker) = payload
+            .split_once("\n…(truncated, raise [memory].max_size or trim memory.md)")
+            .unwrap();
+
+        assert_eq!(head.len(), MAX_MEMORY_SIZE - 1);
+        assert!(head.bytes().all(|byte| byte == b'x'));
+        assert_eq!(marker, "");
+    }
+
+    #[test]
+    fn as_system_block_truncates_emoji_at_char_boundary() {
+        let mut content = "x".repeat(MAX_MEMORY_SIZE - 1);
+        content.push('😀');
+        content.push_str("tail");
+
+        let block = as_system_block(&content, Path::new("/tmp/m.md")).unwrap();
+        assert!(block.contains("…(truncated, raise [memory].max_size or trim memory.md)"));
+
+        let payload = block
+            .strip_prefix("<user_memory source=\"/tmp/m.md\">\n")
+            .unwrap()
+            .strip_suffix("\n</user_memory>")
+            .unwrap();
+        let head = payload
+            .strip_suffix("\n…(truncated, raise [memory].max_size or trim memory.md)")
+            .unwrap();
+
+        assert!(head.len() <= MAX_MEMORY_SIZE);
+        assert_eq!(head.len(), MAX_MEMORY_SIZE - 1);
+        assert!(head.bytes().all(|byte| byte == b'x'));
+    }
+
     #[test]
     fn append_entry_creates_file_and_writes_one_bullet() {
         let tmp = tempdir().unwrap();