lc_tools/
python_repl.rs

1// lc-tools/src/python_repl.rs
2//! Python code execution tool
3//!
4//! Invokes the system Python interpreter to run code and return the result.
5//! Python must be installed on the system.
6//!
7//! # Security warning
8//! This tool is **disabled** by default; call `with_dangerously_allow(true)` to enable it explicitly.
9//! Once enabled it executes arbitrary Python code and should only be used in controlled/sandboxed environments.
10//!
11//! The built-in "dangerous import blacklist" is only **noise filtering, not a security boundary** —
12//! it cannot stop encoding obfuscation such as `__import__`/`eval`/`exec`/string concatenation,
13//! and it also has false positives on string literals.
14//! Real isolation must go through the sandbox ([`crate::sandbox`]); the blacklist only reduces noise reaching it.
15
16use async_trait::async_trait;
17use regex::Regex;
18use schemars::JsonSchema;
19use serde::{Deserialize, Serialize};
20use tokio::process::Command;
21
22use lc_core::tools::{BaseTool, Tool, ToolError};
23
24/// Python REPL tool input
25#[derive(Debug, Deserialize, JsonSchema)]
26pub struct PythonREPLInput {
27    /// The Python code to execute
28    pub code: String,
29    /// Timeout in seconds (default: 30)
30    pub timeout_seconds: Option<u64>,
31}
32
33/// Python REPL tool output
34#[derive(Debug, Serialize)]
35pub struct PythonREPLOutput {
36    /// The code that was executed
37    pub code: String,
38    /// Standard output
39    pub stdout: String,
40    /// Standard error
41    pub stderr: String,
42    /// Exit code
43    pub exit_code: i32,
44}
45
46/// Dangerous Python modules that are blocked for security.
47const BLOCKED_IMPORTS: &[&str] = &[
48    "os",
49    "subprocess",
50    "sys",
51    "shutil",
52    "signal",
53    "ctypes",
54    "multiprocessing",
55    "socket",
56    "http.server",
57    "xmlrpc",
58    "pickle",
59    "shelve",
60    "importlib",
61    "code",
62    "codeop",
63    "compileall",
64    "pty",
65    "commands",
66    "pdb",
67    "webbrowser",
68    "antigravity",
69];
70
71/// Common dangerous builtin calls that bypass the import blacklist (word boundary + function-call form).
72const DANGEROUS_BUILTIN_CALLS: &[&str] = &[
73    "__import__",
74    "import_module",
75    "eval",
76    "exec",
77    "execfile",
78    "compile",
79];
80
81/// Matches dangerous calls like `__import__(` / `import_module(` / `eval(` / `exec(` / `compile(`.
82///
83/// `\b` prevents false positives on ordinary words that contain these as substrings, such as
84/// `evaluate(` / `execute(` / `length(`; however it still matches the literal text `"eval(...)"`
85/// inside string literals — an inherent limitation of string-level interception,
86/// see the security-positioning note on [`contains_dangerous_code`].
87static DANGEROUS_CALL_REGEX: std::sync::LazyLock<Regex> = std::sync::LazyLock::new(|| {
88    Regex::new(&format!(
89        r"\b(?:{})\s*\(",
90        DANGEROUS_BUILTIN_CALLS.join("|")
91    ))
92    .unwrap()
93});
94
95/// Check if Python code contains dangerous imports or builtin calls.
96///
97/// **Security positioning**: this is a noise-filter layer, **not a security boundary**.
98/// Line-by-line substring/regex matching can always be bypassed by unicode obfuscation,
99/// `"o"+"s"` concatenation, `().__class__` reflection, etc., and it also has false
100/// positives on string literals. Untrusted code must go through the sandbox ([`crate::sandbox`]).
101fn contains_dangerous_code(code: &str) -> Option<String> {
102    for line in code.lines() {
103        let trimmed = line.trim();
104        if trimmed.starts_with('#') {
105            continue;
106        }
107        // Strip the inline comment (the part after `#`) so comment content does not cause false positives.
108        let code_part = trimmed.split('#').next().unwrap_or(trimmed);
109
110        // 1) Dangerous import check (BLOCKED_IMPORTS)
111        if code_part.contains("import") {
112            for blocked in BLOCKED_IMPORTS {
113                if code_part.contains(&format!("import {}", blocked))
114                    || code_part.contains(&format!("from {} ", blocked))
115                    || code_part.contains(&format!("from {}.", blocked))
116                    || code_part.contains(&format!("from {}import", blocked))
117                {
118                    return Some(blocked.to_string());
119                }
120            }
121        }
122
123        // 2) Dangerous builtin-call check (common bypasses like __import__ / import_module / eval / exec / compile)
124        if let Some(call) = DANGEROUS_CALL_REGEX.find(code_part) {
125            return Some(call.as_str().to_string());
126        }
127    }
128    None
129}
130
131/// Python code execution tool
132///
133/// Executes code in a local Python environment and returns the result.
134/// Suitable for scenarios needing the Python ecosystem, such as math and data processing.
135///
136/// # Security warning
137/// This tool is **disabled** by default. Call [`PythonREPLTool::with_dangerously_allow`]
138/// to enable code execution. In production, ensure it runs inside a sandboxed environment.
139pub struct PythonREPLTool {
140    python_path: String,
141    /// Whether code execution is allowed (default false, must explicitly opt in)
142    dangerously_allow: bool,
143    /// Whether the dangerous-import check is enabled (default true)
144    check_dangerous_imports: bool,
145}
146
147impl PythonREPLTool {
148    /// Creates a Python code execution tool (execution disabled by default).
149    pub fn new() -> Self {
150        Self {
151            python_path: Self::find_python(),
152            dangerously_allow: false,
153            check_dangerous_imports: true,
154        }
155    }
156
157    /// Uses a custom Python path
158    pub fn with_python_path(path: impl Into<String>) -> Self {
159        Self {
160            python_path: path.into(),
161            dangerously_allow: false,
162            check_dangerous_imports: true,
163        }
164    }
165
166    /// Explicitly enables code execution (disabled by default)
167    pub fn with_dangerously_allow(mut self, allow: bool) -> Self {
168        self.dangerously_allow = allow;
169        self
170    }
171
172    /// Disable dangerous import checking (default: enabled).
173    pub fn with_skip_dangerous_imports_check(mut self, skip: bool) -> Self {
174        self.check_dangerous_imports = !skip;
175        self
176    }
177
178    /// Automatically finds the system Python
179    fn find_python() -> String {
180        for candidate in &["python3", "python"] {
181            if std::process::Command::new(candidate)
182                .arg("--version")
183                .output()
184                .is_ok()
185            {
186                return candidate.to_string();
187            }
188        }
189        "python3".to_string()
190    }
191}
192
193impl Default for PythonREPLTool {
194    fn default() -> Self {
195        Self::new()
196    }
197}
198
199#[async_trait]
200impl Tool for PythonREPLTool {
201    type Input = PythonREPLInput;
202    type Output = PythonREPLOutput;
203
204    async fn invoke(&self, input: Self::Input) -> Result<Self::Output, ToolError> {
205        if input.code.trim().is_empty() {
206            return Err(ToolError::InvalidInput(
207                "Python code must not be empty".to_string(),
208            ));
209        }
210
211        if !self.dangerously_allow {
212            return Err(ToolError::ExecutionFailed(
213                "PythonREPLTool is disabled by default for security. \
214                 Call .with_dangerously_allow(true) to enable execution."
215                    .to_string(),
216            ));
217        }
218
219        if self.check_dangerous_imports {
220            if let Some(blocked) = contains_dangerous_code(&input.code) {
221                return Err(ToolError::ExecutionFailed(format!(
222                    "Code contains dangerous import or builtin call: '{}'. \
223                     Blocked by the noise-filter blacklist (note: this is not a security boundary; \
224                     untrusted code must run in a sandbox). \
225                     Call .with_skip_dangerous_imports_check(true) to bypass (not recommended).",
226                    blocked
227                )));
228            }
229        }
230
231        let timeout_secs = input.timeout_seconds.unwrap_or(30);
232
233        // 0.22.0 H-A3: `kill_on_drop(true)` terminates the subprocess when the
234        // timeout fires and the `.output()` future is dropped. Without it the
235        // child kept running as an orphan after the parent gave up waiting,
236        // an infinite loop burned CPU in the background forever.
237        let result = tokio::time::timeout(
238            std::time::Duration::from_secs(timeout_secs),
239            Command::new(&self.python_path)
240                .arg("-c")
241                .arg(&input.code)
242                .kill_on_drop(true)
243                .output(),
244        )
245        .await
246        .map_err(|_| {
247            ToolError::ExecutionFailed(format!(
248                "Python execution timed out after {} seconds",
249                timeout_secs
250            ))
251        })?
252        .map_err(|e| ToolError::ExecutionFailed(format!("Python execution failed: {}", e)))?;
253
254        let stdout = String::from_utf8_lossy(&result.stdout).to_string();
255        let stderr = String::from_utf8_lossy(&result.stderr).to_string();
256        let exit_code = result.status.code().unwrap_or(-1);
257
258        Ok(PythonREPLOutput {
259            code: input.code,
260            stdout,
261            stderr,
262            exit_code,
263        })
264    }
265}
266
267#[async_trait]
268impl BaseTool for PythonREPLTool {
269    fn name(&self) -> &str {
270        "python_repl"
271    }
272
273    fn description(&self) -> &str {
274        "Python code execution tool. Runs code in a local Python environment and returns results.
275
276Parameters:
277- code: Python code string to execute
278- timeout_seconds: Timeout in seconds (default: 30)
279
280Supports any Python syntax, including math, data processing, plotting, etc.
281
282SECURITY WARNING: Disabled by default. Must call .with_dangerously_allow(true) to enable.
283Only use in controlled/sandboxed environments.
284
285Examples:
286- Simple calc: {\"code\": \"print(1 + 2)\"}
287- List processing: {\"code\": \"print([x**2 for x in range(10)])\"}
288- Math: {\"code\": \"import math; print(math.pi)\"}"
289    }
290
291    async fn run(&self, input: String) -> Result<String, ToolError> {
292        let parsed: PythonREPLInput = serde_json::from_str(&input)
293            .map_err(|e| ToolError::InvalidInput(format!("JSON parse error: {}", e)))?;
294
295        let output = self.invoke(parsed).await?;
296
297        let mut result = String::new();
298        if !output.stdout.is_empty() {
299            result.push_str(&format!("stdout:\n{}\n", output.stdout));
300        }
301        if !output.stderr.is_empty() {
302            result.push_str(&format!("stderr:\n{}\n", output.stderr));
303        }
304        result.push_str(&format!("exit_code: {}", output.exit_code));
305
306        Ok(result)
307    }
308
309    fn args_schema(&self) -> Option<serde_json::Value> {
310        use schemars::schema_for;
311        serde_json::to_value(schema_for!(PythonREPLInput)).ok()
312    }
313}
314
315#[cfg(test)]
316mod tests {
317    use super::*;
318
319    #[test]
320    fn test_python_repl_tool_properties() {
321        let tool = PythonREPLTool::new();
322        assert_eq!(tool.name(), "python_repl");
323        assert!(tool.description().contains("Python"));
324        assert!(BaseTool::args_schema(&tool).is_some());
325    }
326
327    #[tokio::test]
328    async fn test_python_repl_empty_code() {
329        let tool = PythonREPLTool::new().with_dangerously_allow(true);
330        let result = tool.run(r#"{"code": ""}"#.to_string()).await;
331        assert!(result.is_err());
332    }
333
334    #[tokio::test]
335    async fn test_python_repl_disabled_by_default() {
336        let tool = PythonREPLTool::new();
337        let result = tool
338            .invoke(PythonREPLInput {
339                code: "print(1 + 2)".to_string(),
340                timeout_seconds: Some(10),
341            })
342            .await;
343        assert!(result.is_err());
344        let err_msg = result.unwrap_err().to_string();
345        assert!(
346            err_msg.contains("disabled by default"),
347            "Expected disabled error, got: {}",
348            err_msg
349        );
350    }
351
352    #[tokio::test]
353    async fn test_python_repl_basic_execution() {
354        let tool = PythonREPLTool::new().with_dangerously_allow(true);
355        let result = tool
356            .invoke(PythonREPLInput {
357                code: "print(1 + 2)".to_string(),
358                timeout_seconds: Some(10),
359            })
360            .await;
361
362        match result {
363            Ok(output) => {
364                if output.exit_code == 0 || !output.stdout.is_empty() {
365                    // Python available, verify functionality
366                } else {
367                    eprintln!(
368                        "Python may not be installed (exit_code={})",
369                        output.exit_code
370                    );
371                }
372            }
373            Err(e) => {
374                eprintln!("Python not available (may be expected): {}", e);
375            }
376        }
377    }
378
379    #[test]
380    fn test_dangerous_import_detection() {
381        assert!(contains_dangerous_code("import os").is_some());
382        assert!(contains_dangerous_code("import subprocess").is_some());
383        assert!(contains_dangerous_code("from sys import path").is_some());
384        assert!(contains_dangerous_code("from os.path import join").is_some());
385        // Safe imports should pass
386        assert!(contains_dangerous_code("import math").is_none());
387        assert!(contains_dangerous_code("import json").is_none());
388        assert!(contains_dangerous_code("from datetime import datetime").is_none());
389        // Comments should be ignored
390        assert!(contains_dangerous_code("# import os").is_none());
391    }
392
393    #[test]
394    fn test_dangerous_builtin_calls_detected() {
395        // Common bypass: call builtin/imported functions directly without an import statement
396        assert!(contains_dangerous_code("__import__('os').system('ls')").is_some());
397        assert!(contains_dangerous_code("importlib.import_module('os')").is_some());
398        assert!(contains_dangerous_code("eval('os')").is_some());
399        assert!(contains_dangerous_code("exec('import os')").is_some());
400        assert!(contains_dangerous_code("compile('import os', '<x>', 'exec')").is_some());
401        assert!(contains_dangerous_code("execfile('/tmp/x.py')").is_some());
402    }
403
404    #[test]
405    fn test_dangerous_builtin_calls_no_false_positive_on_words() {
406        // Word boundary: no false positives on ordinary words containing these as substrings, like evaluate / execute / length
407        assert!(contains_dangerous_code("print('evaluate the result')").is_none());
408        assert!(contains_dangerous_code("result = execute_query()").is_none());
409        assert!(contains_dangerous_code("print(len([1, 2, 3]))").is_none());
410        assert!(contains_dangerous_code("x = len('hello')").is_none());
411    }
412
413    #[tokio::test]
414    async fn test_python_repl_blocks_dangerous_import() {
415        let tool = PythonREPLTool::new().with_dangerously_allow(true);
416        let result = tool
417            .invoke(PythonREPLInput {
418                code: "import os; print(os.getcwd())".to_string(),
419                timeout_seconds: Some(10),
420            })
421            .await;
422        assert!(result.is_err());
423        let err_msg = result.unwrap_err().to_string();
424        assert!(
425            err_msg.contains("dangerous import"),
426            "Expected dangerous import error, got: {}",
427            err_msg
428        );
429    }
430
431    #[tokio::test]
432    async fn test_python_repl_allows_safe_import() {
433        let tool = PythonREPLTool::new().with_dangerously_allow(true);
434        let result = tool
435            .invoke(PythonREPLInput {
436                code: "import math; print(math.pi)".to_string(),
437                timeout_seconds: Some(10),
438            })
439            .await;
440        match result {
441            Ok(output) => {
442                if output.exit_code == 0 {
443                    assert!(output.stdout.contains("3.14"));
444                }
445            }
446            Err(e) => {
447                assert!(
448                    !e.to_string().contains("dangerous import"),
449                    "math should not be blocked: {}",
450                    e
451                );
452            }
453        }
454    }
455
456    #[tokio::test]
457    async fn test_python_repl_with_error() {
458        let tool = PythonREPLTool::new().with_dangerously_allow(true);
459        let result = tool
460            .invoke(PythonREPLInput {
461                code: "print(undefined_var)".to_string(),
462                timeout_seconds: Some(10),
463            })
464            .await;
465
466        match result {
467            Ok(output) => {
468                if output.exit_code == 0 {
469                    // occasionally Python available but no error reported
470                }
471            }
472            Err(_) => {
473                // No Python available, skip
474            }
475        }
476    }
477}
lc_tools/python_repl.rs

lc_tools/
python_repl.rs
