feat(v3): T22-T23 harness 级工具扩展 + 工作区选择后端
- T22 工具内核：edit_file（old_string 唯一命中才替换，防误改）、search_files （跨文件内容搜索，跳过 .git/node_modules 与二进制大文件）、run_command （allow_shell 默认关；超时+输出截断+Windows CREATE_NO_WINDOW） - T23 工作区选择（参考 deepseek-harness 打开文件夹体验）：/agent/fs 磁盘目录浏览 （空 path 列 Windows 盘符）、/agent/workspaces 最近列表持久化、 POST /agent 接受 workspace（须存在目录），运行状态记录所用工作区 - 新增测试 12 项，全量 274 passed
This commit is contained in:
tzt
+198
-1
@@ -24,6 +24,15 @@ MAX_LIST_ENTRIES = 200
|
||||
MAX_RESULT_CHARS = 8000
|
||||
MAX_WRITE_CHARS = 200_000
|
||||
|
||||
# search_files 上限
|
||||
SEARCH_MAX_MATCHES = 30
|
||||
SEARCH_MAX_FILES = 400
|
||||
SEARCH_MAX_FILE_BYTES = 512 * 1024
|
||||
SEARCH_SKIP_DIRS = {".git", "node_modules", "__pycache__", ".venv", "venv", "dist", "build"}
|
||||
|
||||
# run_command 上限
|
||||
SHELL_OUTPUT_CHARS = 4000
|
||||
|
||||
# 默认循环上限
|
||||
DEFAULT_MAX_ROUNDS = 8
|
||||
|
||||
@@ -71,11 +80,89 @@ TOOLS_SPEC: List[Dict[str, Any]] = [
|
||||
},
|
||||
},
|
||||
},
|
||||
{
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": "edit_file",
|
||||
"description": "对工作区内已有文件做精确替换编辑：old_string 必须在文件中恰好出现一次，"
|
||||
"被替换为 new_string。适合小改动；大改用 write_file 整体重写。",
|
||||
"parameters": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"path": {"type": "string", "description": "工作区内的相对路径"},
|
||||
"old_string": {"type": "string", "description": "要替换的原文（须唯一匹配）"},
|
||||
"new_string": {"type": "string", "description": "替换后的新文"},
|
||||
},
|
||||
"required": ["path", "old_string", "new_string"],
|
||||
},
|
||||
},
|
||||
},
|
||||
{
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": "search_files",
|
||||
"description": "在工作区（或其子目录）内按关键词跨文件搜索文本内容，"
|
||||
"返回匹配的文件/行号/行内容（自动跳过 .git、node_modules 等目录与二进制大文件）。",
|
||||
"parameters": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"query": {"type": "string", "description": "搜索的关键词（大小写敏感）"},
|
||||
"path": {"type": "string", "description": "限定的子目录，默认整个工作区"},
|
||||
},
|
||||
"required": ["query"],
|
||||
},
|
||||
},
|
||||
},
|
||||
{
|
||||
"type": "function",
|
||||
"function": {
|
||||
"name": "run_command",
|
||||
"description": "在工作区根目录执行一条 shell 命令并返回退出码与输出（如运行测试、查看版本）。"
|
||||
"仅当系统开启 allow_shell 时可用。",
|
||||
"parameters": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"command": {"type": "string", "description": "要执行的命令行"},
|
||||
},
|
||||
"required": ["command"],
|
||||
},
|
||||
},
|
||||
},
|
||||
]
|
||||
|
||||
TOOL_NAMES = {t["function"]["name"] for t in TOOLS_SPEC}
|
||||
|
||||
|
||||
def browse_directories(path: str = "") -> Dict[str, Any]:
|
||||
"""目录选择器的本地文件系统浏览（只列目录，不读文件内容）。
|
||||
|
||||
path 为空时列出 Windows 盘符（POSIX 列根目录）。返回
|
||||
{"ok", "path", "parent", "dirs": [名称]}；用于智能体"选择工作区"。
|
||||
"""
|
||||
import os
|
||||
if not path or not path.strip():
|
||||
if os.name == "nt":
|
||||
drives = [f"{c}:\\" for c in "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
|
||||
if os.path.exists(f"{c}:\\")]
|
||||
return {"ok": True, "path": "", "parent": "", "dirs": drives}
|
||||
return {"ok": True, "path": "/", "parent": "",
|
||||
"dirs": sorted(os.listdir("/"))}
|
||||
p = Path(path).resolve()
|
||||
if not p.exists():
|
||||
return {"ok": False, "error": f"路径不存在: {path}"}
|
||||
if not p.is_dir():
|
||||
return {"ok": False, "error": f"不是目录: {path}"}
|
||||
dirs = []
|
||||
for child in sorted(p.iterdir(), key=lambda c: c.name.lower()):
|
||||
try:
|
||||
if child.is_dir():
|
||||
dirs.append(child.name)
|
||||
except OSError:
|
||||
continue # 无权限/符号链接坏点，跳过
|
||||
parent = str(p.parent) if p.parent != p else ""
|
||||
return {"ok": True, "path": str(p), "parent": parent, "dirs": dirs}
|
||||
|
||||
|
||||
class ToolError(Exception):
|
||||
"""工具执行失败（路径越界/不存在/参数非法）。"""
|
||||
|
||||
@@ -87,9 +174,12 @@ class WorkspaceTools:
|
||||
（根目录自身允许），否则抛 ToolError——防 ../ 越界与绝对路径逃逸。
|
||||
"""
|
||||
|
||||
def __init__(self, root: str | Path):
|
||||
def __init__(self, root: str | Path,
|
||||
allow_shell: bool = False, shell_timeout_s: int = 20):
|
||||
self.root = Path(root).resolve()
|
||||
self.root.mkdir(parents=True, exist_ok=True)
|
||||
self.allow_shell = bool(allow_shell)
|
||||
self.shell_timeout_s = max(1, int(shell_timeout_s))
|
||||
|
||||
# ---------- 路径关押 ----------
|
||||
def resolve(self, rel_path: str) -> Path:
|
||||
@@ -147,6 +237,103 @@ class WorkspaceTools:
|
||||
p.write_text(content, encoding="utf-8")
|
||||
return {"ok": True, "path": rel_path, "bytes_written": len(content.encode("utf-8"))}
|
||||
|
||||
def edit_file(self, rel_path: str, old_string: str, new_string: str) -> Dict[str, Any]:
|
||||
"""精确替换编辑：old_string 必须在文件中恰好出现一次（harness 式安全编辑）。"""
|
||||
if not old_string:
|
||||
return {"ok": False, "error": "old_string 不能为空"}
|
||||
if len(old_string) > MAX_READ_CHARS:
|
||||
return {"ok": False, "error": "old_string 过长（先 read_file 分段定位）"}
|
||||
p = self.resolve(rel_path)
|
||||
if not p.exists() or not p.is_file():
|
||||
return {"ok": False, "error": f"文件不存在: {rel_path}"}
|
||||
try:
|
||||
text = p.read_text(encoding="utf-8")
|
||||
except UnicodeDecodeError:
|
||||
return {"ok": False, "error": f"非文本文件: {rel_path}"}
|
||||
count = text.count(old_string)
|
||||
if count == 0:
|
||||
return {"ok": False, "error": "old_string 未在文件中找到（先 read_file 核对原文）"}
|
||||
if count > 1:
|
||||
return {"ok": False,
|
||||
"error": f"old_string 出现 {count} 次（要求唯一）；请扩大上下文使其唯一"}
|
||||
new_text = text.replace(old_string, new_string, 1)
|
||||
p.write_text(new_text, encoding="utf-8")
|
||||
return {
|
||||
"ok": True, "path": rel_path,
|
||||
"replaced": 1,
|
||||
"changed_chars": len(new_text) - len(text),
|
||||
}
|
||||
|
||||
def search_files(self, query: str, rel_path: str = "") -> Dict[str, Any]:
|
||||
"""跨文件文本搜索（跳过依赖/构建目录与二进制大文件，限量返回）。"""
|
||||
if not query:
|
||||
return {"ok": False, "error": "query 不能为空"}
|
||||
base = self.resolve(rel_path or "")
|
||||
if not base.exists() or not base.is_dir():
|
||||
return {"ok": False, "error": f"目录不存在: {rel_path}"}
|
||||
matches: List[Dict[str, Any]] = []
|
||||
scanned = 0
|
||||
truncated = False
|
||||
for p in sorted(base.rglob("*")):
|
||||
if len(matches) >= SEARCH_MAX_MATCHES:
|
||||
truncated = True
|
||||
break
|
||||
if not p.is_file():
|
||||
continue
|
||||
rel_parts = p.relative_to(base).parts
|
||||
if any(part in SEARCH_SKIP_DIRS for part in rel_parts):
|
||||
continue
|
||||
try:
|
||||
if p.stat().st_size > SEARCH_MAX_FILE_BYTES:
|
||||
continue
|
||||
text = p.read_text(encoding="utf-8")
|
||||
except (OSError, UnicodeDecodeError):
|
||||
continue # 二进制/不可读，跳过
|
||||
scanned += 1
|
||||
if scanned > SEARCH_MAX_FILES:
|
||||
truncated = True
|
||||
break
|
||||
for lineno, line in enumerate(text.splitlines(), 1):
|
||||
if query in line:
|
||||
rel = Path(*p.relative_to(self.root).parts).as_posix()
|
||||
matches.append({
|
||||
"file": rel, "line": lineno,
|
||||
"text": line.strip()[:300],
|
||||
})
|
||||
if len(matches) >= SEARCH_MAX_MATCHES:
|
||||
truncated = True
|
||||
break
|
||||
return {"ok": True, "query": query, "matches": matches,
|
||||
"scanned_files": scanned, "truncated": truncated}
|
||||
|
||||
def run_command(self, command: str) -> Dict[str, Any]:
|
||||
"""在工作区根目录执行 shell 命令（默认关闭，allow_shell 开启后可用）。"""
|
||||
if not self.allow_shell:
|
||||
return {"ok": False,
|
||||
"error": "run_command 未启用（系统设置 allow_shell 为关）。"
|
||||
"请让用户在智能体页打开「允许执行命令」后重试。"}
|
||||
command = (command or "").strip()
|
||||
if not command:
|
||||
return {"ok": False, "error": "command 不能为空"}
|
||||
import subprocess
|
||||
creationflags = 0x08000000 if __import__("os").name == "nt" else 0 # CREATE_NO_WINDOW
|
||||
try:
|
||||
proc = subprocess.run(
|
||||
command, shell=True, cwd=str(self.root), capture_output=True,
|
||||
timeout=self.shell_timeout_s, creationflags=creationflags,
|
||||
)
|
||||
out = (proc.stdout or b"").decode("utf-8", errors="replace")
|
||||
err = (proc.stderr or b"").decode("utf-8", errors="replace")
|
||||
combined = (out + ("\n[stderr]\n" + err if err.strip() else "")).strip()
|
||||
if len(combined) > SHELL_OUTPUT_CHARS:
|
||||
combined = combined[:SHELL_OUTPUT_CHARS] + "…(输出截断)"
|
||||
return {"ok": True, "exit_code": proc.returncode,
|
||||
"output": combined or "(无输出)", "command": command}
|
||||
except subprocess.TimeoutExpired:
|
||||
return {"ok": False, "error": f"命令超时（>{self.shell_timeout_s}s），已终止: {command}"}
|
||||
except OSError as e:
|
||||
return {"ok": False, "error": f"命令执行失败: {type(e).__name__}: {e}"}
|
||||
|
||||
# ---------- 统一执行入口 ----------
|
||||
def execute(self, name: str, arguments: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""按名字执行工具；任何异常折叠为 {"ok": False, "error": ...}。"""
|
||||
@@ -158,6 +345,16 @@ class WorkspaceTools:
|
||||
if name == "write_file":
|
||||
return self.write_file(
|
||||
str(arguments.get("path", "")), str(arguments.get("content", "")))
|
||||
if name == "edit_file":
|
||||
return self.edit_file(
|
||||
str(arguments.get("path", "")),
|
||||
str(arguments.get("old_string", "")),
|
||||
str(arguments.get("new_string", "")))
|
||||
if name == "search_files":
|
||||
return self.search_files(
|
||||
str(arguments.get("query", "")), str(arguments.get("path", "")))
|
||||
if name == "run_command":
|
||||
return self.run_command(str(arguments.get("command", "")))
|
||||
return {"ok": False, "error": f"未知工具: {name}"}
|
||||
except ToolError as e:
|
||||
return {"ok": False, "error": str(e)}
|
||||
|
||||
Reference in New Issue
Block a user