agentshield/parser/
shell.rs

1use std::path::{Path, PathBuf};
2
3use once_cell::sync::Lazy;
4use regex::Regex;
5
6use super::{LanguageParser, ParsedFile};
7use crate::error::Result;
8use crate::ir::execution_surface::*;
9use crate::ir::{ArgumentSource, Language, SourceLocation};
10
11pub struct ShellParser;
12
13static CURL_WGET_RE: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?m)\b(curl|wget)\s+").unwrap());
14
15static EVAL_RE: Lazy<Regex> = Lazy::new(|| Regex::new(r"(?m)\beval\s+").unwrap());
16
17static INSTALL_RE: Lazy<Regex> = Lazy::new(|| {
18    Regex::new(r"(?m)\b(pip3?\s+install|npm\s+install|npm\s+i\b|yarn\s+add|pnpm\s+add)").unwrap()
19});
20
21static BACKTICK_RE: Lazy<Regex> = Lazy::new(|| Regex::new(r"`[^`]+`").unwrap());
22
23static SENSITIVE_VAR_RE: Lazy<Regex> =
24    Lazy::new(|| Regex::new(r"(?i)\$\{?(AWS_|SECRET|TOKEN|PASSWORD|API_KEY|PRIVATE_KEY)").unwrap());
25
26impl LanguageParser for ShellParser {
27    fn language(&self) -> Language {
28        Language::Shell
29    }
30
31    fn parse_file(&self, path: &Path, content: &str) -> Result<ParsedFile> {
32        let mut parsed = ParsedFile::default();
33        let file_path = PathBuf::from(path);
34
35        for (line_idx, line) in content.lines().enumerate() {
36            let line_num = line_idx + 1;
37            let trimmed = line.trim();
38
39            if trimmed.starts_with('#') || trimmed.is_empty() {
40                continue;
41            }
42
43            // curl/wget = network operations
44            if let Some(cap) = CURL_WGET_RE.find(trimmed) {
45                let func = cap.as_str().trim();
46                let arg_source = if trimmed.contains('$') {
47                    ArgumentSource::Interpolated
48                } else {
49                    ArgumentSource::Literal(trimmed.to_string())
50                };
51                parsed.network_operations.push(NetworkOperation {
52                    function: func.to_string(),
53                    url_arg: arg_source,
54                    method: None,
55                    sends_data: trimmed.contains("-d ") || trimmed.contains("--data"),
56                    location: loc(&file_path, line_num),
57                });
58            }
59
60            // eval
61            if EVAL_RE.is_match(trimmed) {
62                parsed.dynamic_exec.push(DynamicExec {
63                    function: "eval".into(),
64                    code_arg: ArgumentSource::Interpolated,
65                    location: loc(&file_path, line_num),
66                });
67            }
68
69            // backtick execution
70            if BACKTICK_RE.is_match(trimmed) {
71                parsed.commands.push(CommandInvocation {
72                    function: "backtick".into(),
73                    command_arg: ArgumentSource::Interpolated,
74                    location: loc(&file_path, line_num),
75                });
76            }
77
78            // pip/npm install
79            if INSTALL_RE.is_match(trimmed) {
80                parsed.commands.push(CommandInvocation {
81                    function: "package_install".into(),
82                    command_arg: ArgumentSource::Literal(trimmed.to_string()),
83                    location: loc(&file_path, line_num),
84                });
85            }
86
87            // Sensitive env var access
88            for cap in SENSITIVE_VAR_RE.captures_iter(trimmed) {
89                let var = cap.get(0).map(|m| m.as_str()).unwrap_or("");
90                parsed.env_accesses.push(EnvAccess {
91                    var_name: ArgumentSource::Literal(var.to_string()),
92                    is_sensitive: true,
93                    location: loc(&file_path, line_num),
94                });
95            }
96        }
97
98        Ok(parsed)
99    }
100}
101
102fn loc(file: &Path, line: usize) -> SourceLocation {
103    SourceLocation {
104        file: file.to_path_buf(),
105        line,
106        column: 0,
107        end_line: None,
108        end_column: None,
109    }
110}
111
112#[cfg(test)]
113mod tests {
114    use super::*;
115
116    #[test]
117    fn detects_curl() {
118        let code = "curl https://example.com/data\n";
119        let parsed = ShellParser.parse_file(Path::new("test.sh"), code).unwrap();
120        assert_eq!(parsed.network_operations.len(), 1);
121    }
122
123    #[test]
124    fn detects_eval() {
125        let code = "eval $USER_INPUT\n";
126        let parsed = ShellParser.parse_file(Path::new("test.sh"), code).unwrap();
127        assert_eq!(parsed.dynamic_exec.len(), 1);
128    }
129
130    #[test]
131    fn detects_pip_install() {
132        let code = "pip install requests\n";
133        let parsed = ShellParser.parse_file(Path::new("test.sh"), code).unwrap();
134        assert_eq!(parsed.commands.len(), 1);
135        assert!(parsed.commands[0].function.contains("package_install"));
136    }
137}
agentshield/parser/shell.rs

agentshield/parser/
shell.rs
