Merge pull request #2879 from yangziyu-hhh/master

feat: stream Bash progress and guard message actions during replies
This commit is contained in:
zhayujie
2026-06-12 18:19:13 +08:00
committed by GitHub
8 changed files with 456 additions and 31 deletions

32
.github/workflows/test-windows-bash.yml vendored Normal file
View File

@@ -0,0 +1,32 @@
name: Windows Bash Streaming Tests
on:
workflow_dispatch:
pull_request:
paths:
- "agent/tools/bash/bash.py"
- "tests/test_bash_streaming.py"
- ".github/workflows/test-windows-bash.yml"
jobs:
windows-bash-tests:
runs-on: windows-latest
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Set up Python
uses: actions/setup-python@v5
with:
python-version: "3.11"
cache: pip
- name: Install dependencies
run: |
python -m pip install --upgrade pip
python -m pip install pytest
python -m pip install -r requirements.txt
- name: Run Windows Bash streaming tests
run: python -m pytest tests/test_bash_streaming.py -v

View File

@@ -1174,10 +1174,21 @@ class AgentStreamExecutor:
# Set tool context
tool.model = self.model
tool.context = self.agent
tool.progress_callback = lambda message: self._emit_event(
"tool_execution_progress",
{
"tool_call_id": tool_id,
"tool_name": tool_name,
"message": message,
}
)
# Execute tool
start_time = time.time()
try:
result: ToolResult = tool.execute_tool(arguments)
finally:
tool.progress_callback = None
execution_time = time.time() - start_time
result_dict = {

View File

@@ -38,6 +38,16 @@ class BaseTool:
description: str = "Base tool"
params: dict = {} # Store JSON Schema
model: Optional[Any] = None # LLM model instance, type depends on bot implementation
progress_callback = None
def report_progress(self, message: str):
callback = getattr(self, "progress_callback", None)
if not callback:
return
try:
callback(str(message))
except Exception as e:
logger.debug(f"[{self.name}] progress callback failed: {e}")
@classmethod
def get_json_schema(cls) -> dict:

View File

@@ -4,9 +4,12 @@ Bash tool - Execute bash commands
import os
import re
import signal
import sys
import subprocess
import tempfile
import threading
import time
from typing import Dict, Any
from agent.tools.base_tool import BaseTool, ToolResult
@@ -19,6 +22,8 @@ class Bash(BaseTool):
"""Tool for executing bash commands"""
_IS_WIN = sys.platform == "win32"
_PROGRESS_MAX_BYTES = 4 * 1024
_PROGRESS_INTERVAL = 0.5
name: str = "bash"
description: str = f"""Execute a bash command in the current working directory. Returns stdout and stderr. Output is truncated to last {DEFAULT_MAX_LINES} lines or {DEFAULT_MAX_BYTES // 1024}KB (whichever is hit first). If truncated, full output is saved to a temp file.
@@ -113,17 +118,11 @@ SAFETY:
if command and not command.strip().lower().startswith("chcp"):
command = f"chcp 65001 >nul 2>&1 && {command}"
result = subprocess.run(
result = self._run_streaming(
command,
shell=True,
cwd=self.cwd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
text=True,
encoding="utf-8",
errors="replace",
timeout=timeout,
env=env,
timeout,
env,
dotenv_vars,
)
logger.debug(f"[Bash] Exit code: {result.returncode}")
@@ -236,6 +235,105 @@ SAFETY:
except Exception as e:
return ToolResult.fail(f"Error executing command: {str(e)}")
def _run_streaming(self, command: str, timeout: int, env: dict, dotenv_vars: dict):
process = subprocess.Popen(
command,
shell=True,
cwd=self.cwd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
env=env,
start_new_session=not self._IS_WIN,
)
stdout_chunks, stderr_chunks = [], []
recent = bytearray()
recent_lock = threading.Lock()
def drain(stream, chunks):
while True:
chunk = os.read(stream.fileno(), 4096)
if not chunk:
break
chunks.append(chunk)
with recent_lock:
recent.extend(chunk)
if len(recent) > self._PROGRESS_MAX_BYTES:
del recent[:-self._PROGRESS_MAX_BYTES]
readers = [
threading.Thread(target=drain, args=(process.stdout, stdout_chunks), daemon=True),
threading.Thread(target=drain, args=(process.stderr, stderr_chunks), daemon=True),
]
for reader in readers:
reader.start()
started = time.monotonic()
last_reported_at = started
last_snapshot = None
try:
while process.poll() is None:
now = time.monotonic()
elapsed = now - started
if elapsed >= timeout:
self._kill_process(process)
raise subprocess.TimeoutExpired(command, timeout)
if elapsed >= self._PROGRESS_INTERVAL and now - last_reported_at >= self._PROGRESS_INTERVAL:
with recent_lock:
snapshot = bytes(recent).decode("utf-8", errors="replace")
snapshot = self._redact_progress(snapshot, dotenv_vars)
if snapshot and snapshot != last_snapshot:
self.report_progress(snapshot)
last_snapshot = snapshot
last_reported_at = now
time.sleep(0.1)
finally:
if process.poll() is None:
self._kill_process(process)
process.wait()
join_deadline = time.monotonic() + 5
for reader in readers:
reader.join(timeout=max(0, join_deadline - time.monotonic()))
from types import SimpleNamespace
return SimpleNamespace(
returncode=process.returncode,
stdout=b"".join(stdout_chunks).decode("utf-8", errors="replace"),
stderr=b"".join(stderr_chunks).decode("utf-8", errors="replace"),
)
def _kill_process(self, process):
if self._IS_WIN:
try:
result = subprocess.run(
["taskkill", "/F", "/T", "/PID", str(process.pid)],
capture_output=True,
timeout=5,
)
if result.returncode != 0 and process.poll() is None:
process.kill()
except (OSError, subprocess.SubprocessError):
if process.poll() is None:
process.kill()
else:
try:
os.killpg(process.pid, signal.SIGKILL)
except (PermissionError, ProcessLookupError):
if process.poll() is None:
process.kill()
@staticmethod
def _redact_progress(text: str, dotenv_vars: dict) -> str:
text = re.sub(
r'(?i)\b(API_KEY|TOKEN|PASSWORD|AUTHORIZATION)\s*=\s*[^\s]+',
lambda match: f"{match.group(1)}=[REDACTED]",
text,
)
for value in dotenv_vars.values():
value = str(value or "")
if len(value) >= 6:
text = text.replace(value, "[REDACTED]")
return text
def _get_safety_warning(self, command: str) -> str:
"""
Get safety warning for absolutely catastrophic commands only.

View File

@@ -605,6 +605,18 @@
color: inherit;
}
.tool-error-text { color: #f87171; }
.tool-live-output:empty { display: none; }
.tool-streaming .tool-live-output:not(:empty)::after {
content: ' ';
display: inline-block;
width: 0.45em;
height: 1em;
margin-left: 0.15em;
vertical-align: -0.15em;
background: currentColor;
animation: tool-cursor-blink 1s steps(1) infinite;
}
@keyframes tool-cursor-blink { 50% { opacity: 0; } }
/* Log level highlighting */
.log-line { display: block; }
@@ -1579,3 +1591,9 @@
.delete-msg-btn:hover {
color: #ef4444 !important;
}
.edit-msg-btn:disabled,
.delete-msg-btn:disabled {
cursor: not-allowed !important;
opacity: 0.35 !important;
}

View File

@@ -197,6 +197,8 @@ const I18N = {
delete_session_title: '删除会话',
delete_message_confirm: '确认删除这条消息?',
delete_message_title: '删除消息',
edit_disabled_reply_active: '正在生成回复,暂时无法编辑。',
delete_disabled_reply_active: '正在生成回复,暂时无法删除。',
untitled_session: '新对话',
context_cleared: '— 以上内容已从上下文中移除 —',
tip_new_chat: '新建对话',
@@ -410,6 +412,8 @@ const I18N = {
delete_session_title: 'Delete Session',
delete_message_confirm: 'Delete this message?',
delete_message_title: 'Delete Message',
edit_disabled_reply_active: 'Reply is being generated; editing is temporarily unavailable.',
delete_disabled_reply_active: 'Reply is being generated; deletion is temporarily unavailable.',
untitled_session: 'New Chat',
context_cleared: '— Context above has been cleared —',
tip_new_chat: 'New Chat',
@@ -907,6 +911,26 @@ let pollGeneration = 0; // incremented on each restart to cancel stale poll lo
let loadingContainers = {};
let activeStreams = {}; // request_id -> EventSource
let sessionActiveRequest = {}; // session_id -> request_id (in-flight stream per session)
function isCurrentSessionConversationActive() {
return !!sessionActiveRequest[sessionId];
}
function updateEditButtonsState() {
const active = isCurrentSessionConversationActive();
document.querySelectorAll('.edit-msg-btn, .delete-msg-btn').forEach(btn => {
btn.disabled = active;
if (btn.classList.contains('edit-msg-btn')) {
btn.title = active
? t('edit_disabled_reply_active')
: t('edit_message');
} else {
btn.title = active
? t('delete_disabled_reply_active')
: t('delete_message_title');
}
});
}
let streamBuffers = {}; // request_id -> { items: [event...], timestamp } for re-attach replay
let isComposing = false;
let appConfig = { use_agent: false, title: 'CowAgent', subtitle: '', providers: {}, api_bases: {} };
@@ -1195,6 +1219,7 @@ messagesDiv.addEventListener('click', (e) => {
const editBtn = e.target.closest('.edit-msg-btn');
if (editBtn) {
e.preventDefault();
if (isCurrentSessionConversationActive()) return;
const msgRoot = editBtn.closest('.user-message-group');
if (msgRoot) editUserMessage(msgRoot);
return;
@@ -1215,6 +1240,7 @@ messagesDiv.addEventListener('click', (e) => {
const deleteBtn = e.target.closest('.delete-msg-btn');
if (deleteBtn) {
e.preventDefault();
if (isCurrentSessionConversationActive()) return;
const userMsgEl = deleteBtn.closest('.user-message-group');
if (!userMsgEl) return;
@@ -2011,6 +2037,7 @@ function copyToClipboard(text) {
// Edit user message: extract content, remove this and subsequent messages, fill input
async function editUserMessage(msgEl) {
if (isCurrentSessionConversationActive()) return;
const rawContent = msgEl.dataset.rawContent;
if (!rawContent) return;
@@ -2061,9 +2088,9 @@ async function editUserMessage(msgEl) {
// Fill input with the original content
chatInput.value = rawContent;
chatInput.style.height = 'auto';
chatInput.style.height = chatInput.scrollHeight + 'px';
chatInput.dispatchEvent(new Event("input", { bubbles: true }));
chatInput.focus();
chatInput.selectionStart = chatInput.selectionEnd = chatInput.value.length;
scrollChatToBottom();
}
@@ -2268,7 +2295,7 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo, replayItems) {
let contentEl = null; // .answer-content (final streaming answer)
let mediaEl = null; // .media-content (images & file attachments)
let accumulatedText = '';
let currentToolEl = null;
const toolElements = new Map();
let currentReasoningEl = null; // live reasoning bubble
let reasoningText = '';
let reasoningStartTime = 0;
@@ -2283,12 +2310,14 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo, replayItems) {
const ownerSession = sessionId;
const isActive = () => ownerSession === sessionId;
sessionActiveRequest[ownerSession] = requestId;
updateEditButtonsState();
// Per-request event buffer used to rebuild the bubble on re-attach.
const buffer = streamBuffers[requestId] || { items: [], timestamp };
streamBuffers[requestId] = buffer;
const clearOwnerRequest = () => {
if (sessionActiveRequest[ownerSession] === requestId) {
delete sessionActiveRequest[ownerSession];
updateEditButtonsState();
}
delete streamBuffers[requestId];
};
@@ -2437,10 +2466,11 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo, replayItems) {
contentEl.innerHTML = '';
// Add tool execution indicator (collapsible)
currentToolEl = document.createElement('div');
currentToolEl.className = 'agent-step agent-tool-step';
const toolEl = document.createElement('div');
toolEl.className = 'agent-step agent-tool-step tool-streaming';
toolEl.dataset.progressReceived = 'false';
const argsStr = formatToolArgs(item.arguments || {});
currentToolEl.innerHTML = `
toolEl.innerHTML = `
<div class="tool-header" onclick="this.parentElement.classList.toggle('expanded')">
<i class="fas fa-cog fa-spin text-primary-400 flex-shrink-0 tool-icon"></i>
<span class="tool-name">${item.tool}</span>
@@ -2451,36 +2481,59 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo, replayItems) {
<div class="tool-detail-label">Input</div>
<pre class="tool-detail-content">${argsStr}</pre>
</div>
<div class="tool-detail-section tool-output-section"></div>
<div class="tool-detail-section tool-output-section">
<div class="tool-detail-label tool-output-label">Output</div>
<pre class="tool-detail-content tool-live-output"></pre>
</div>
</div>`;
stepsEl.appendChild(currentToolEl);
stepsEl.appendChild(toolEl);
toolElements.set(item.tool_call_id, toolEl);
scrollChatToBottom();
} else if (item.type === 'tool_progress') {
const toolEl = toolElements.get(item.tool_call_id);
if (toolEl) {
if (toolEl.dataset.progressReceived !== 'true') {
toolEl.classList.add('expanded');
toolEl.dataset.progressReceived = 'true';
}
toolEl.querySelector('.tool-live-output').textContent = String(item.content || '');
scrollChatToBottom();
}
} else if (item.type === 'tool_end') {
if (currentToolEl) {
const toolEl = toolElements.get(item.tool_call_id);
if (toolEl) {
const isError = item.status !== 'success';
const icon = currentToolEl.querySelector('.tool-icon');
const icon = toolEl.querySelector('.tool-icon');
icon.className = isError
? 'fas fa-times text-red-400 flex-shrink-0 tool-icon'
: 'fas fa-check text-primary-400 flex-shrink-0 tool-icon';
// Show execution time
const nameEl = currentToolEl.querySelector('.tool-name');
const nameEl = toolEl.querySelector('.tool-name');
if (item.execution_time !== undefined) {
nameEl.innerHTML += ` <span class="tool-time">${item.execution_time}s</span>`;
}
// Fill output section
const outputSection = currentToolEl.querySelector('.tool-output-section');
if (outputSection && item.result) {
outputSection.innerHTML = `
<div class="tool-detail-label">${isError ? 'Error' : 'Output'}</div>
<pre class="tool-detail-content ${isError ? 'tool-error-text' : ''}">${escapeHtml(String(item.result))}</pre>`;
const outputLabel = toolEl.querySelector('.tool-output-label');
const outputEl = toolEl.querySelector('.tool-live-output');
if (outputLabel) outputLabel.textContent = isError ? 'Error' : 'Output';
if (outputEl) {
outputEl.textContent = item.result ? String(item.result) : '';
outputEl.classList.toggle('tool-error-text', isError);
}
if (isError) currentToolEl.classList.add('tool-failed');
currentToolEl = null;
toolEl.classList.remove('tool-streaming');
toolEl.classList.remove('expanded');
if (!item.result) {
const outputSection = toolEl.querySelector('.tool-output-section');
if (outputSection) outputSection.remove();
}
if (isError) toolEl.classList.add('tool-failed');
toolElements.delete(item.tool_call_id);
}
} else if (item.type === 'image') {
@@ -2639,6 +2692,13 @@ function startSSE(requestId, loadingEl, timestamp, titleInfo, replayItems) {
// Record every event for re-attach replay (capped to avoid
// unbounded growth on very long streams).
if (item.type === 'tool_progress' && item.tool_call_id) {
const previousIndex = buffer.items.findIndex(
buffered => buffered.type === 'tool_progress'
&& buffered.tool_call_id === item.tool_call_id
);
if (previousIndex >= 0) buffer.items.splice(previousIndex, 1);
}
if (buffer.items.length < 5000) buffer.items.push(item);
// Background session: keep the stream alive so the reply finishes
@@ -3274,6 +3334,7 @@ function loadHistory(page) {
const sentinel = document.getElementById('history-load-more');
const insertBefore = sentinel ? sentinel.nextSibling : messagesDiv.firstChild;
messagesDiv.insertBefore(fragment, insertBefore);
updateEditButtonsState();
// Manage the "load more" sentinel at the very top
if (data.has_more) {
@@ -3729,6 +3790,7 @@ function switchSession(newSessionId) {
// Switching back re-attaches and resumes live streaming.
sessionId = newSessionId;
updateEditButtonsState();
localStorage.setItem(SESSION_ID_KEY, sessionId);
historyPage = 0;

View File

@@ -432,7 +432,15 @@ class WebChannel(ChatChannel):
elif event_type == "tool_execution_start":
tool_name = data.get("tool_name", "tool")
arguments = data.get("arguments", {})
q.put({"type": "tool_start", "tool": tool_name, "arguments": arguments})
q.put({"type": "tool_start", "tool_call_id": data.get("tool_call_id"), "tool": tool_name, "arguments": arguments})
elif event_type == "tool_execution_progress":
q.put({
"type": "tool_progress",
"tool_call_id": data.get("tool_call_id"),
"tool": data.get("tool_name", "tool"),
"content": str(data.get("message", ""))[-4 * 1024:],
})
elif event_type == "tool_execution_end":
tool_name = data.get("tool_name", "tool")
@@ -445,6 +453,7 @@ class WebChannel(ChatChannel):
result_str = result_str[:2000] + ""
q.put({
"type": "tool_end",
"tool_call_id": data.get("tool_call_id"),
"tool": tool_name,
"status": status,
"result": result_str,

View File

@@ -0,0 +1,185 @@
import subprocess
import sys
import time
from unittest.mock import patch
import pytest
from agent.tools.bash.bash import Bash
posix_only = pytest.mark.skipif(Bash._IS_WIN, reason="POSIX shell command")
windows_only = pytest.mark.skipif(not Bash._IS_WIN, reason="Windows integration test")
def _windows_pid_command(pid_file):
path = str(pid_file).replace("\\", "\\\\")
return (
f'"{sys.executable}" -c '
f'"import os,time;open(r\'{path}\',\'w\').write(str(os.getpid()));time.sleep(30)"'
)
def _windows_pid_is_running(pid):
result = subprocess.run(
["tasklist", "/FI", f"PID eq {pid}", "/FO", "CSV", "/NH"],
capture_output=True,
text=True,
errors="replace",
)
return f'"{pid}"' in result.stdout
def _windows_kill_pid_file(pid_file):
if pid_file.exists():
subprocess.run(
["taskkill", "/F", "/PID", pid_file.read_text()],
capture_output=True,
)
@posix_only
def test_fast_command_returns_output_without_progress(tmp_path):
tool = Bash({"cwd": str(tmp_path)})
progress = []
tool.progress_callback = progress.append
result = tool.execute({"command": "printf fast"})
assert result.status == "success"
assert result.result["output"] == "fast"
assert progress == []
@posix_only
def test_timeout_returns_promptly(tmp_path):
tool = Bash({"cwd": str(tmp_path)})
started = time.monotonic()
result = tool.execute({"command": "sleep 10", "timeout": 1})
assert result.status == "error"
assert "timed out after 1 seconds" in result.result
assert time.monotonic() - started < 3
@posix_only
def test_background_process_holding_pipe_does_not_hang(tmp_path):
tool = Bash({"cwd": str(tmp_path)})
started = time.monotonic()
result = tool.execute({"command": "sleep 10 & printf done", "timeout": 3})
assert result.status == "success"
assert result.result["output"] == "done"
assert time.monotonic() - started < 7
@posix_only
def test_output_without_trailing_newline_streams_progress(tmp_path):
tool = Bash({"cwd": str(tmp_path)})
progress = []
tool.progress_callback = progress.append
result = tool.execute({
"command": "for i in 1 2 3 4 5; do printf .; sleep 0.3; done",
"timeout": 5,
})
assert result.status == "success"
assert result.result["output"] == "....."
assert progress
assert progress[-1].endswith(".")
def test_windows_kill_uses_taskkill_for_process_tree():
tool = Bash()
process = type("Process", (), {"pid": 1234, "poll": lambda self: None, "kill": lambda self: None})()
with patch.object(tool, "_IS_WIN", True), patch.object(subprocess, "run") as run:
run.return_value.returncode = 0
tool._kill_process(process)
run.assert_called_once_with(
["taskkill", "/F", "/T", "/PID", "1234"],
capture_output=True,
timeout=5,
)
def test_windows_kill_falls_back_when_taskkill_fails():
tool = Bash()
process = type("Process", (), {"pid": 1234, "poll": lambda self: None, "kill": lambda self: None})()
with patch.object(tool, "_IS_WIN", True), \
patch.object(subprocess, "run", side_effect=OSError), \
patch.object(process, "kill") as kill:
tool._kill_process(process)
kill.assert_called_once()
@windows_only
def test_windows_timeout_kills_process_tree(tmp_path):
pid_file = tmp_path / "timeout-child.pid"
child_command = _windows_pid_command(pid_file)
tool = Bash({"cwd": str(tmp_path)})
started = time.monotonic()
try:
result = tool.execute({
"command": f'start "" /b {child_command} & ping -n 30 127.0.0.1 >nul',
"timeout": 1,
})
assert result.status == "error"
assert "timed out after 1 seconds" in result.result
assert time.monotonic() - started < 8
assert pid_file.exists()
time.sleep(0.5)
assert not _windows_pid_is_running(pid_file.read_text())
finally:
_windows_kill_pid_file(pid_file)
@windows_only
def test_windows_long_running_command_streams_progress(tmp_path):
tool = Bash({"cwd": str(tmp_path)})
progress = []
tool.progress_callback = progress.append
result = tool.execute({
"command": (
f'"{sys.executable}" -u -c '
'"import sys,time; '
"[(sys.stdout.write('.'),sys.stdout.flush(),time.sleep(0.3)) "
'for _ in range(5)]"'
),
"timeout": 5,
})
assert result.status == "success"
assert result.result["output"] == "....."
assert progress
assert progress[-1].endswith(".")
@windows_only
def test_windows_background_process_holding_pipe_does_not_hang(tmp_path):
pid_file = tmp_path / "background-child.pid"
child_command = _windows_pid_command(pid_file)
tool = Bash({"cwd": str(tmp_path)})
started = time.monotonic()
try:
result = tool.execute({
"command": f'start "" /b {child_command} & echo done',
"timeout": 3,
})
assert result.status == "success"
assert result.result["output"].strip() == "done"
assert time.monotonic() - started < 8
assert pid_file.exists()
finally:
_windows_kill_pid_file(pid_file)