lc_tools/
python_repl.rs

1// lc-tools/src/python_repl.rs
2//! Python 代码执行工具
3//!
4//! 调用系统 Python 解释器执行代码并返回结果。
5//! 需要系统中已安装 Python。
6//!
7//! # 安全警告
8//! 此工具默认**禁用**，必须调用 `with_dangerously_allow(true)` 显式启用。
9//! 启用后会执行任意 Python 代码，仅应在受控/沙箱环境中使用。
10//!
11//! 内置的"危险 import 黑名单"只是**噪音过滤，不是安全边界**——它挡不住
12//! `__import__`/`eval`/`exec`/字符串拼接等编码混淆，也会误伤字符串字面量。
13//! 真正的隔离必须走沙箱（[`crate::sandbox`]），黑名单只用于减少误入沙箱的噪音。
14
15use async_trait::async_trait;
16use regex::Regex;
17use schemars::JsonSchema;
18use serde::{Deserialize, Serialize};
19use tokio::process::Command;
20
21use lc_core::tools::{BaseTool, Tool, ToolError};
22
23/// Python REPL 工具输入
24#[derive(Debug, Deserialize, JsonSchema)]
25pub struct PythonREPLInput {
26    /// 要执行的 Python 代码
27    pub code: String,
28    /// 超时时间（秒，默认 30）
29    pub timeout_seconds: Option<u64>,
30}
31
32/// Python REPL 工具输出
33#[derive(Debug, Serialize)]
34pub struct PythonREPLOutput {
35    /// 执行的代码
36    pub code: String,
37    /// 标准输出
38    pub stdout: String,
39    /// 标准错误
40    pub stderr: String,
41    /// 退出码
42    pub exit_code: i32,
43}
44
45/// Dangerous Python modules that are blocked for security.
46const BLOCKED_IMPORTS: &[&str] = &[
47    "os",
48    "subprocess",
49    "sys",
50    "shutil",
51    "signal",
52    "ctypes",
53    "multiprocessing",
54    "socket",
55    "http.server",
56    "xmlrpc",
57    "pickle",
58    "shelve",
59    "importlib",
60    "code",
61    "codeop",
62    "compileall",
63    "pty",
64    "commands",
65    "pdb",
66    "webbrowser",
67    "antigravity",
68];
69
70/// 常见绕过 import 黑名单的危险内建调用（单词边界 + 函数调用形式）。
71const DANGEROUS_BUILTIN_CALLS: &[&str] = &[
72    "__import__",
73    "import_module",
74    "eval",
75    "exec",
76    "execfile",
77    "compile",
78];
79
80/// 匹配 `__import__(` / `import_module(` / `eval(` / `exec(` / `compile(` 等危险调用。
81///
82/// `\b` 保证不会误伤 `evaluate(` / `execute(` / `length(` 这类含子串的普通词；
83/// 但仍会误伤字符串字面量里的 `"eval(...)"` 字样——这是字符串级拦截的固有局限，
84/// 详见 [`contains_dangerous_code`] 的安全定位说明。
85static DANGEROUS_CALL_REGEX: std::sync::LazyLock<Regex> = std::sync::LazyLock::new(|| {
86    Regex::new(&format!(
87        r"\b(?:{})\s*\(",
88        DANGEROUS_BUILTIN_CALLS.join("|")
89    ))
90    .unwrap()
91});
92
93/// Check if Python code contains dangerous imports or builtin calls.
94///
95/// **安全定位**：这是噪音过滤层，**不是安全边界**。逐行子串/正则匹配永远可以被
96/// unicode 混淆、`"o"+"s"` 拼接、`().__class__` 反射等绕过，也会误伤字符串字面量。
97/// 不可信代码必须走沙箱（[`crate::sandbox`]）。
98fn contains_dangerous_code(code: &str) -> Option<String> {
99    for line in code.lines() {
100        let trimmed = line.trim();
101        if trimmed.starts_with('#') {
102            continue;
103        }
104        // 去掉行内注释（# 之后的部分），避免注释内容误伤。
105        let code_part = trimmed.split('#').next().unwrap_or(trimmed);
106
107        // 1) 危险 import 检查（BLOCKED_IMPORTS）
108        if code_part.contains("import") {
109            for blocked in BLOCKED_IMPORTS {
110                if code_part.contains(&format!("import {}", blocked))
111                    || code_part.contains(&format!("from {} ", blocked))
112                    || code_part.contains(&format!("from {}.", blocked))
113                    || code_part.contains(&format!("from {}import", blocked))
114                {
115                    return Some(blocked.to_string());
116                }
117            }
118        }
119
120        // 2) 危险内建调用检查（__import__ / import_module / eval / exec / compile 等常见绕过）
121        if let Some(call) = DANGEROUS_CALL_REGEX.find(code_part) {
122            return Some(call.as_str().to_string());
123        }
124    }
125    None
126}
127
128/// Python 代码执行工具
129///
130/// 在本地 Python 环境中执行代码并返回结果。
131/// 适用于数学计算、数据处理等需要 Python 生态的场景。
132///
133/// # 安全警告
134/// 此工具默认**禁用**。必须调用 [`PythonREPLTool::with_dangerously_allow`]
135/// 才能执行代码。在生产环境中使用时应确保在沙箱环境中运行。
136pub struct PythonREPLTool {
137    python_path: String,
138    /// 是否允许执行代码（默认 false，必须显式 opt-in）
139    dangerously_allow: bool,
140    /// 是否启用危险 import 检查（默认 true）
141    check_dangerous_imports: bool,
142}
143
144impl PythonREPLTool {
145    pub fn new() -> Self {
146        Self {
147            python_path: Self::find_python(),
148            dangerously_allow: false,
149            check_dangerous_imports: true,
150        }
151    }
152
153    /// 使用自定义 Python 路径
154    pub fn with_python_path(path: impl Into<String>) -> Self {
155        Self {
156            python_path: path.into(),
157            dangerously_allow: false,
158            check_dangerous_imports: true,
159        }
160    }
161
162    /// 显式启用代码执行（默认禁用）
163    pub fn with_dangerously_allow(mut self, allow: bool) -> Self {
164        self.dangerously_allow = allow;
165        self
166    }
167
168    /// Disable dangerous import checking (default: enabled).
169    pub fn with_skip_dangerous_imports_check(mut self, skip: bool) -> Self {
170        self.check_dangerous_imports = !skip;
171        self
172    }
173
174    /// 自动查找系统 Python
175    fn find_python() -> String {
176        for candidate in &["python3", "python"] {
177            if std::process::Command::new(candidate)
178                .arg("--version")
179                .output()
180                .is_ok()
181            {
182                return candidate.to_string();
183            }
184        }
185        "python3".to_string()
186    }
187}
188
189impl Default for PythonREPLTool {
190    fn default() -> Self {
191        Self::new()
192    }
193}
194
195#[async_trait]
196impl Tool for PythonREPLTool {
197    type Input = PythonREPLInput;
198    type Output = PythonREPLOutput;
199
200    async fn invoke(&self, input: Self::Input) -> Result<Self::Output, ToolError> {
201        if input.code.trim().is_empty() {
202            return Err(ToolError::InvalidInput(
203                "Python code must not be empty".to_string(),
204            ));
205        }
206
207        if !self.dangerously_allow {
208            return Err(ToolError::ExecutionFailed(
209                "PythonREPLTool is disabled by default for security. \
210                 Call .with_dangerously_allow(true) to enable execution."
211                    .to_string(),
212            ));
213        }
214
215        if self.check_dangerous_imports {
216            if let Some(blocked) = contains_dangerous_code(&input.code) {
217                return Err(ToolError::ExecutionFailed(format!(
218                    "Code contains dangerous import or builtin call: '{}'. \
219                     Blocked by the noise-filter blacklist (note: this is not a security boundary; \
220                     untrusted code must run in a sandbox). \
221                     Call .with_skip_dangerous_imports_check(true) to bypass (not recommended).",
222                    blocked
223                )));
224            }
225        }
226
227        let timeout_secs = input.timeout_seconds.unwrap_or(30);
228
229        let result = tokio::time::timeout(
230            std::time::Duration::from_secs(timeout_secs),
231            Command::new(&self.python_path)
232                .arg("-c")
233                .arg(&input.code)
234                .output(),
235        )
236        .await
237        .map_err(|_| {
238            ToolError::ExecutionFailed(format!(
239                "Python execution timed out after {} seconds",
240                timeout_secs
241            ))
242        })?
243        .map_err(|e| ToolError::ExecutionFailed(format!("Python execution failed: {}", e)))?;
244
245        let stdout = String::from_utf8_lossy(&result.stdout).to_string();
246        let stderr = String::from_utf8_lossy(&result.stderr).to_string();
247        let exit_code = result.status.code().unwrap_or(-1);
248
249        Ok(PythonREPLOutput {
250            code: input.code,
251            stdout,
252            stderr,
253            exit_code,
254        })
255    }
256}
257
258#[async_trait]
259impl BaseTool for PythonREPLTool {
260    fn name(&self) -> &str {
261        "python_repl"
262    }
263
264    fn description(&self) -> &str {
265        "Python code execution tool. Runs code in a local Python environment and returns results.
266
267Parameters:
268- code: Python code string to execute
269- timeout_seconds: Timeout in seconds (default: 30)
270
271Supports any Python syntax, including math, data processing, plotting, etc.
272
273SECURITY WARNING: Disabled by default. Must call .with_dangerously_allow(true) to enable.
274Only use in controlled/sandboxed environments.
275
276Examples:
277- Simple calc: {\"code\": \"print(1 + 2)\"}
278- List processing: {\"code\": \"print([x**2 for x in range(10)])\"}
279- Math: {\"code\": \"import math; print(math.pi)\"}"
280    }
281
282    async fn run(&self, input: String) -> Result<String, ToolError> {
283        let parsed: PythonREPLInput = serde_json::from_str(&input)
284            .map_err(|e| ToolError::InvalidInput(format!("JSON parse error: {}", e)))?;
285
286        let output = self.invoke(parsed).await?;
287
288        let mut result = String::new();
289        if !output.stdout.is_empty() {
290            result.push_str(&format!("stdout:\n{}\n", output.stdout));
291        }
292        if !output.stderr.is_empty() {
293            result.push_str(&format!("stderr:\n{}\n", output.stderr));
294        }
295        result.push_str(&format!("exit_code: {}", output.exit_code));
296
297        Ok(result)
298    }
299
300    fn args_schema(&self) -> Option<serde_json::Value> {
301        use schemars::schema_for;
302        serde_json::to_value(schema_for!(PythonREPLInput)).ok()
303    }
304}
305
306#[cfg(test)]
307mod tests {
308    use super::*;
309
310    #[test]
311    fn test_python_repl_tool_properties() {
312        let tool = PythonREPLTool::new();
313        assert_eq!(tool.name(), "python_repl");
314        assert!(tool.description().contains("Python"));
315        assert!(BaseTool::args_schema(&tool).is_some());
316    }
317
318    #[tokio::test]
319    async fn test_python_repl_empty_code() {
320        let tool = PythonREPLTool::new().with_dangerously_allow(true);
321        let result = tool.run(r#"{"code": ""}"#.to_string()).await;
322        assert!(result.is_err());
323    }
324
325    #[tokio::test]
326    async fn test_python_repl_disabled_by_default() {
327        let tool = PythonREPLTool::new();
328        let result = tool
329            .invoke(PythonREPLInput {
330                code: "print(1 + 2)".to_string(),
331                timeout_seconds: Some(10),
332            })
333            .await;
334        assert!(result.is_err());
335        let err_msg = result.unwrap_err().to_string();
336        assert!(
337            err_msg.contains("disabled by default"),
338            "Expected disabled error, got: {}",
339            err_msg
340        );
341    }
342
343    #[tokio::test]
344    async fn test_python_repl_basic_execution() {
345        let tool = PythonREPLTool::new().with_dangerously_allow(true);
346        let result = tool
347            .invoke(PythonREPLInput {
348                code: "print(1 + 2)".to_string(),
349                timeout_seconds: Some(10),
350            })
351            .await;
352
353        match result {
354            Ok(output) => {
355                if output.exit_code == 0 || !output.stdout.is_empty() {
356                    // Python available, verify functionality
357                } else {
358                    eprintln!(
359                        "Python may not be installed (exit_code={})",
360                        output.exit_code
361                    );
362                }
363            }
364            Err(e) => {
365                eprintln!("Python not available (may be expected): {}", e);
366            }
367        }
368    }
369
370    #[test]
371    fn test_dangerous_import_detection() {
372        assert!(contains_dangerous_code("import os").is_some());
373        assert!(contains_dangerous_code("import subprocess").is_some());
374        assert!(contains_dangerous_code("from sys import path").is_some());
375        assert!(contains_dangerous_code("from os.path import join").is_some());
376        // Safe imports should pass
377        assert!(contains_dangerous_code("import math").is_none());
378        assert!(contains_dangerous_code("import json").is_none());
379        assert!(contains_dangerous_code("from datetime import datetime").is_none());
380        // Comments should be ignored
381        assert!(contains_dangerous_code("# import os").is_none());
382    }
383
384    #[test]
385    fn test_dangerous_builtin_calls_detected() {
386        // 常见绕过：不经过 import 语句，直接调用内建/导入函数
387        assert!(contains_dangerous_code("__import__('os').system('ls')").is_some());
388        assert!(contains_dangerous_code("importlib.import_module('os')").is_some());
389        assert!(contains_dangerous_code("eval('os')").is_some());
390        assert!(contains_dangerous_code("exec('import os')").is_some());
391        assert!(contains_dangerous_code("compile('import os', '<x>', 'exec')").is_some());
392        assert!(contains_dangerous_code("execfile('/tmp/x.py')").is_some());
393    }
394
395    #[test]
396    fn test_dangerous_builtin_calls_no_false_positive_on_words() {
397        // 单词边界：不误伤 evaluate / execute / length 等含子串的普通词
398        assert!(contains_dangerous_code("print('evaluate the result')").is_none());
399        assert!(contains_dangerous_code("result = execute_query()").is_none());
400        assert!(contains_dangerous_code("print(len([1, 2, 3]))").is_none());
401        assert!(contains_dangerous_code("x = len('hello')").is_none());
402    }
403
404    #[tokio::test]
405    async fn test_python_repl_blocks_dangerous_import() {
406        let tool = PythonREPLTool::new().with_dangerously_allow(true);
407        let result = tool
408            .invoke(PythonREPLInput {
409                code: "import os; print(os.getcwd())".to_string(),
410                timeout_seconds: Some(10),
411            })
412            .await;
413        assert!(result.is_err());
414        let err_msg = result.unwrap_err().to_string();
415        assert!(
416            err_msg.contains("dangerous import"),
417            "Expected dangerous import error, got: {}",
418            err_msg
419        );
420    }
421
422    #[tokio::test]
423    async fn test_python_repl_allows_safe_import() {
424        let tool = PythonREPLTool::new().with_dangerously_allow(true);
425        let result = tool
426            .invoke(PythonREPLInput {
427                code: "import math; print(math.pi)".to_string(),
428                timeout_seconds: Some(10),
429            })
430            .await;
431        match result {
432            Ok(output) => {
433                if output.exit_code == 0 {
434                    assert!(output.stdout.contains("3.14"));
435                }
436            }
437            Err(e) => {
438                assert!(
439                    !e.to_string().contains("dangerous import"),
440                    "math should not be blocked: {}",
441                    e
442                );
443            }
444        }
445    }
446
447    #[tokio::test]
448    async fn test_python_repl_with_error() {
449        let tool = PythonREPLTool::new().with_dangerously_allow(true);
450        let result = tool
451            .invoke(PythonREPLInput {
452                code: "print(undefined_var)".to_string(),
453                timeout_seconds: Some(10),
454            })
455            .await;
456
457        match result {
458            Ok(output) => {
459                if output.exit_code == 0 {
460                    // occasionally Python available but no error reported
461                }
462            }
463            Err(_) => {
464                // No Python available, skip
465            }
466        }
467    }
468}
lc_tools/python_repl.rs

lc_tools/
python_repl.rs
