feat(llm): improve UTF-8 handling in response processing to prevent infinite loops
feat(shell): enhance output capturing by using threads for stdout and stderr feat(tui): update usage widget to display token counts and provider/model information refactor(tui): simplify status bar rendering by removing unnecessary token calculations
This commit is contained in:
@@ -386,13 +386,22 @@ impl LlmClient {
|
||||
break;
|
||||
}
|
||||
byte_buf.extend_from_slice(&chunk_buf[..n]);
|
||||
|
||||
// Drain any bytes that are not valid UTF-8 to prevent
|
||||
// infinite loop when a non-UTF-8 sequence is received.
|
||||
let valid_len = match std::str::from_utf8(&byte_buf) {
|
||||
Ok(s) => s.len(),
|
||||
Err(e) => e.valid_up_to(),
|
||||
Err(e) => {
|
||||
let n = e.valid_up_to();
|
||||
if n == 0 {
|
||||
// No valid UTF-8 prefix; skip the first byte (likely
|
||||
// a partial multi-byte sequence or stray byte).
|
||||
byte_buf.drain(..1);
|
||||
continue;
|
||||
}
|
||||
n
|
||||
}
|
||||
};
|
||||
if valid_len == 0 {
|
||||
continue;
|
||||
}
|
||||
let text =
|
||||
String::from_utf8_lossy(&byte_buf[..valid_len]).into_owned();
|
||||
byte_buf.drain(..valid_len);
|
||||
|
||||
@@ -89,30 +89,47 @@ impl Tool for Bash {
|
||||
let start = std::time::Instant::now();
|
||||
let timeout = Duration::from_millis(timeout_ms);
|
||||
|
||||
let mut child_stdout = child.stdout.take()
|
||||
.ok_or_else(|| anyhow::anyhow!("failed to capture stdout"))?;
|
||||
let mut child_stderr = child.stderr.take()
|
||||
.ok_or_else(|| anyhow::anyhow!("failed to capture stderr"))?;
|
||||
|
||||
let stdout_handle = std::thread::spawn(move || -> std::io::Result<Vec<u8>> {
|
||||
use std::io::Read;
|
||||
let mut buf = Vec::new();
|
||||
child_stdout.read_to_end(&mut buf)?;
|
||||
Ok(buf)
|
||||
});
|
||||
let stderr_handle = std::thread::spawn(move || -> std::io::Result<Vec<u8>> {
|
||||
use std::io::Read;
|
||||
let mut buf = Vec::new();
|
||||
child_stderr.read_to_end(&mut buf)?;
|
||||
Ok(buf)
|
||||
});
|
||||
|
||||
loop {
|
||||
match child.try_wait() {
|
||||
Ok(Some(status)) => {
|
||||
let stdout = stdout_handle.join().unwrap_or(Ok(Vec::new()))?;
|
||||
let stderr = stderr_handle.join().unwrap_or(Ok(Vec::new()))?;
|
||||
|
||||
let elapsed = start.elapsed().as_secs_f64();
|
||||
let output = child
|
||||
.wait_with_output()
|
||||
.map_err(|e| anyhow::anyhow!("failed to collect output: {e}"))?;
|
||||
let stdout = String::from_utf8_lossy(&output.stdout).to_string();
|
||||
let stderr = String::from_utf8_lossy(&output.stderr).to_string();
|
||||
let combined = if stderr.is_empty() {
|
||||
stdout
|
||||
let stdout_str = String::from_utf8_lossy(&stdout).to_string();
|
||||
let stderr_str = String::from_utf8_lossy(&stderr).to_string();
|
||||
let combined = if stderr_str.is_empty() {
|
||||
stdout_str
|
||||
} else {
|
||||
format!("{stdout}\n{stderr}")
|
||||
format!("{stdout_str}\n{stderr_str}")
|
||||
};
|
||||
let trimmed = combined.trim().to_string();
|
||||
|
||||
if status.success() {
|
||||
debug!(elapsed_secs = elapsed, "bash command completed successfully");
|
||||
return Ok(if trimmed.is_empty() {
|
||||
format!("Command completed in {elapsed:.2}s (exit code 0)")
|
||||
} else {
|
||||
format!("{trimmed}\n\nExit code: 0 ({elapsed:.2}s)")
|
||||
});
|
||||
}
|
||||
warn!(exit_code = status.code().unwrap_or(-1), elapsed_secs = elapsed, "bash command failed");
|
||||
return Ok(format!(
|
||||
"{}\n\nExit code: {} ({:.2}s)",
|
||||
trimmed,
|
||||
@@ -124,7 +141,6 @@ impl Tool for Bash {
|
||||
if start.elapsed() > timeout {
|
||||
let _ = child.kill();
|
||||
let _ = child.wait();
|
||||
warn!(timeout_ms, "bash command timed out");
|
||||
anyhow::bail!("command timed out after {timeout_ms}ms");
|
||||
}
|
||||
std::thread::sleep(Duration::from_millis(10));
|
||||
|
||||
@@ -96,19 +96,40 @@ fn draw_usage_widget(frame: &mut Frame, area: Rect, state: &crate::state::AppSta
|
||||
let lines: Vec<Line> = if let Some(ref rt) = state.session_runtime {
|
||||
let now_ms = chrono::Utc::now().timestamp_millis();
|
||||
let summary = compute_usage_summary(&rt.usage, rt.session_start, now_ms);
|
||||
vec![
|
||||
let max_tokens = crate::state::resolve_context_window(&state.app_config, &state.settings);
|
||||
let current_tokens = state.cached_token_count;
|
||||
|
||||
let mut items = vec![
|
||||
Line::from(Span::styled(
|
||||
format!(" {:>6}: {} tok", "total", summary.total_tokens),
|
||||
Style::default()
|
||||
.fg(Theme::TEXT)
|
||||
.add_modifier(Modifier::BOLD),
|
||||
)),
|
||||
Line::from(Span::styled(
|
||||
];
|
||||
|
||||
if summary.self_learning_tokens > 0 {
|
||||
items.push(Line::from(Span::styled(
|
||||
format!(" {:>6}: {} tok", "main", summary.main_tokens),
|
||||
Style::default().fg(Theme::TEXT_DIM),
|
||||
)));
|
||||
items.push(Line::from(Span::styled(
|
||||
format!(" {:>6}: {} tok", "learn", summary.self_learning_tokens),
|
||||
Style::default().fg(Theme::TEXT_DIM),
|
||||
)));
|
||||
}
|
||||
|
||||
items.extend(vec![
|
||||
Line::from(Span::styled(
|
||||
format!(" {:>6}: {}/{}", "ctx", current_tokens, max_tokens),
|
||||
Style::default().fg(Theme::TEXT_DIM),
|
||||
)),
|
||||
Line::from(Span::styled(
|
||||
format!(" {:>6}: {} tok", "learn", summary.self_learning_tokens),
|
||||
format!(" {:>6}: {}", "prov", state.settings.provider),
|
||||
Style::default().fg(Theme::TEXT_DIM),
|
||||
)),
|
||||
Line::from(Span::styled(
|
||||
format!(" {:>6}: {}", "model", state.settings.model),
|
||||
Style::default().fg(Theme::TEXT_DIM),
|
||||
)),
|
||||
Line::from(Span::styled(
|
||||
@@ -122,7 +143,9 @@ fn draw_usage_widget(frame: &mut Frame, area: Rect, state: &crate::state::AppSta
|
||||
),
|
||||
Style::default().fg(Theme::TEXT_DIM),
|
||||
)),
|
||||
]
|
||||
]);
|
||||
|
||||
items
|
||||
} else {
|
||||
vec![Line::from(Span::styled(
|
||||
" No active session.",
|
||||
|
||||
@@ -48,32 +48,7 @@ pub fn draw_status_bar(frame: &mut Frame, area: Rect, state: &crate::state::AppS
|
||||
status_badge,
|
||||
];
|
||||
|
||||
let max_tokens = crate::state::resolve_context_window(&state.app_config, &state.settings);
|
||||
|
||||
let right_str = if let Some(ref rt) = state.session_runtime {
|
||||
// Use cached token count — recomputed lazily only when new messages
|
||||
// arrive (token_count_dirty flag), not on every render frame.
|
||||
// This eliminates the expensive tiktoken_rs call from the hot path.
|
||||
let current_tokens = state.cached_token_count;
|
||||
|
||||
let mut parts = Vec::new();
|
||||
if rt.usage.last_tokens_in > 0 || rt.usage.last_tokens_out > 0 {
|
||||
parts.push(format!(
|
||||
"↑{} ↓{}",
|
||||
rt.usage.last_tokens_in, rt.usage.last_tokens_out
|
||||
));
|
||||
}
|
||||
parts.push(format!("{current_tokens}/{max_tokens}"));
|
||||
parts.push(state.settings.provider.clone());
|
||||
parts.push(state.settings.model.clone());
|
||||
|
||||
format!(" {} ", parts.join(" · "))
|
||||
} else {
|
||||
format!(
|
||||
" 0/{max_tokens} · {} · {} ",
|
||||
state.settings.provider, state.settings.model
|
||||
)
|
||||
};
|
||||
let right_str = String::new();
|
||||
|
||||
let left_line = Line::from(left_spans);
|
||||
let right_line = Line::from(Span::styled(
|
||||
|
||||
Reference in New Issue
Block a user