Files
compliance-scanner-agent/compliance-agent/src/pipeline/patterns.rs
Sharang Parnerkar 3bb690e5bb
All checks were successful
CI / Format (push) Successful in 4s
CI / Clippy (push) Successful in 4m19s
CI / Security Audit (push) Successful in 1m44s
CI / Detect Changes (push) Successful in 5s
CI / Tests (push) Successful in 5m15s
CI / Deploy Agent (push) Successful in 2s
CI / Deploy Dashboard (push) Successful in 2s
CI / Deploy Docs (push) Has been skipped
CI / Deploy MCP (push) Successful in 2s
refactor: modularize codebase and add 404 unit tests (#13)
2026-03-13 08:03:45 +00:00

415 lines
15 KiB
Rust

use std::path::Path;
use compliance_core::models::{Finding, ScanType, Severity};
use compliance_core::traits::{ScanOutput, Scanner};
use compliance_core::CoreError;
use regex::Regex;
use crate::pipeline::dedup;
fn compile_regex(pattern: &str) -> Regex {
Regex::new(pattern).unwrap_or_else(|e| {
tracing::warn!("Invalid regex pattern '{pattern}': {e}, using empty fallback");
// SAFETY: "^$" is a known-valid regex that matches only empty strings
#[allow(clippy::unwrap_used)]
Regex::new("^$").unwrap()
})
}
pub struct GdprPatternScanner {
patterns: Vec<PatternRule>,
}
pub struct OAuthPatternScanner {
patterns: Vec<PatternRule>,
}
struct PatternRule {
id: String,
title: String,
description: String,
pattern: Regex,
severity: Severity,
file_extensions: Vec<String>,
}
impl GdprPatternScanner {
pub fn new() -> Self {
let patterns = vec![
PatternRule {
id: "gdpr-pii-logging".to_string(),
title: "PII data potentially logged".to_string(),
description: "Logging statements that may contain personally identifiable information (email, SSN, phone, IP address).".to_string(),
pattern: compile_regex(r#"(?i)(log|print|console\.|logger\.|tracing::)\s*[\.(].*\b(email|ssn|social.?security|phone.?number|ip.?addr|passport|date.?of.?birth|credit.?card)\b"#),
severity: Severity::High,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go", "rb"].into_iter().map(String::from).collect(),
},
PatternRule {
id: "gdpr-no-consent".to_string(),
title: "Data collection without apparent consent mechanism".to_string(),
description: "Data collection endpoint that doesn't reference consent or opt-in mechanisms.".to_string(),
pattern: compile_regex(r#"(?i)(collect|store|save|persist|record).*\b(personal|user.?data|pii|biometric)\b"#),
severity: Severity::Medium,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go"].into_iter().map(String::from).collect(),
},
PatternRule {
id: "gdpr-no-delete-endpoint".to_string(),
title: "Missing data deletion capability".to_string(),
description: "User data models or controllers without corresponding deletion endpoints (right to erasure).".to_string(),
pattern: compile_regex(r#"(?i)(class|struct|model)\s+User"#),
severity: Severity::Medium,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go", "rb"].into_iter().map(String::from).collect(),
},
PatternRule {
id: "gdpr-hardcoded-retention".to_string(),
title: "Hardcoded data retention period".to_string(),
description: "Data retention periods should be configurable for GDPR compliance.".to_string(),
pattern: compile_regex(r#"(?i)(retention|ttl|expire|keep.?for)\s*[=:]\s*\d+"#),
severity: Severity::Low,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go", "yaml", "yml", "toml", "json"].into_iter().map(String::from).collect(),
},
];
Self { patterns }
}
}
impl Scanner for GdprPatternScanner {
fn name(&self) -> &str {
"gdpr-patterns"
}
fn scan_type(&self) -> ScanType {
ScanType::Gdpr
}
#[tracing::instrument(skip_all)]
async fn scan(&self, repo_path: &Path, repo_id: &str) -> Result<ScanOutput, CoreError> {
let findings = scan_with_patterns(
repo_path,
repo_id,
&self.patterns,
ScanType::Gdpr,
"gdpr-patterns",
)?;
Ok(ScanOutput {
findings,
sbom_entries: Vec::new(),
})
}
}
impl OAuthPatternScanner {
pub fn new() -> Self {
let patterns = vec![
PatternRule {
id: "oauth-implicit-grant".to_string(),
title: "OAuth implicit grant flow detected".to_string(),
description: "Implicit grant flow is deprecated and insecure. Use authorization code flow with PKCE instead.".to_string(),
pattern: compile_regex(r#"(?i)(response_type\s*[=:]\s*["']?token|grant_type\s*[=:]\s*["']?implicit)"#),
severity: Severity::High,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go", "yaml", "yml", "json"].into_iter().map(String::from).collect(),
},
PatternRule {
id: "oauth-missing-pkce".to_string(),
title: "OAuth flow without PKCE".to_string(),
description: "Authorization code flow should use PKCE (code_challenge/code_verifier) for public clients.".to_string(),
pattern: compile_regex(r#"(?i)authorization.?code"#),
severity: Severity::Medium,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go"].into_iter().map(String::from).collect(),
},
PatternRule {
id: "oauth-token-localstorage".to_string(),
title: "Token stored in localStorage".to_string(),
description: "Storing tokens in localStorage is vulnerable to XSS. Use httpOnly cookies or secure session storage.".to_string(),
pattern: compile_regex(r#"(?i)localStorage\.(set|get)Item\s*\(\s*["'].*token"#),
severity: Severity::High,
file_extensions: vec!["js", "ts", "jsx", "tsx"].into_iter().map(String::from).collect(),
},
PatternRule {
id: "oauth-token-url".to_string(),
title: "Token passed in URL parameters".to_string(),
description: "Tokens in URLs can leak via referrer headers, server logs, and browser history.".to_string(),
pattern: compile_regex(r#"(?i)(access_token|bearer)\s*[=]\s*.*\b(url|query|param|href)\b"#),
severity: Severity::High,
file_extensions: vec!["rs", "py", "js", "ts", "java", "go"].into_iter().map(String::from).collect(),
},
];
Self { patterns }
}
}
impl Scanner for OAuthPatternScanner {
fn name(&self) -> &str {
"oauth-patterns"
}
fn scan_type(&self) -> ScanType {
ScanType::OAuth
}
#[tracing::instrument(skip_all)]
async fn scan(&self, repo_path: &Path, repo_id: &str) -> Result<ScanOutput, CoreError> {
let findings = scan_with_patterns(
repo_path,
repo_id,
&self.patterns,
ScanType::OAuth,
"oauth-patterns",
)?;
Ok(ScanOutput {
findings,
sbom_entries: Vec::new(),
})
}
}
fn scan_with_patterns(
repo_path: &Path,
repo_id: &str,
patterns: &[PatternRule],
scan_type: ScanType,
scanner_name: &str,
) -> Result<Vec<Finding>, CoreError> {
let mut findings = Vec::new();
for entry in walkdir(repo_path)? {
let path = entry.path();
if !path.is_file() {
continue;
}
let ext = path
.extension()
.and_then(|e| e.to_str())
.unwrap_or("")
.to_string();
let content = match std::fs::read_to_string(path) {
Ok(c) => c,
Err(_) => continue, // skip binary files
};
let relative_path = path
.strip_prefix(repo_path)
.unwrap_or(path)
.to_string_lossy()
.to_string();
for pattern in patterns {
if !pattern.file_extensions.contains(&ext) {
continue;
}
for (line_num, line) in content.lines().enumerate() {
if pattern.pattern.is_match(line) {
let fingerprint = dedup::compute_fingerprint(&[
repo_id,
&pattern.id,
&relative_path,
&(line_num + 1).to_string(),
]);
let mut finding = Finding::new(
repo_id.to_string(),
fingerprint,
scanner_name.to_string(),
scan_type.clone(),
pattern.title.clone(),
pattern.description.clone(),
pattern.severity.clone(),
);
finding.rule_id = Some(pattern.id.clone());
finding.file_path = Some(relative_path.clone());
finding.line_number = Some((line_num + 1) as u32);
finding.code_snippet = Some(line.to_string());
findings.push(finding);
}
}
}
}
Ok(findings)
}
fn walkdir(path: &Path) -> Result<Vec<walkdir::DirEntry>, CoreError> {
// Simple recursive file walk, skipping hidden dirs and common non-source dirs
let skip_dirs = [
".git",
"node_modules",
"target",
"vendor",
".venv",
"__pycache__",
"dist",
"build",
];
let entries: Vec<_> = walkdir::WalkDir::new(path)
.into_iter()
.filter_entry(|e| {
let name = e.file_name().to_string_lossy();
!skip_dirs.contains(&name.as_ref())
})
.filter_map(|e| e.ok())
.collect();
Ok(entries)
}
#[cfg(test)]
mod tests {
use super::*;
// --- compile_regex tests ---
#[test]
fn compile_regex_valid_pattern() {
let re = compile_regex(r"\bfoo\b");
assert!(re.is_match("hello foo bar"));
assert!(!re.is_match("foobar"));
}
#[test]
fn compile_regex_invalid_pattern_returns_fallback() {
// An invalid regex should return the fallback "^$" that only matches empty strings
let re = compile_regex(r"[invalid");
assert!(re.is_match(""));
assert!(!re.is_match("anything"));
}
// --- GDPR pattern tests ---
#[test]
fn gdpr_pii_logging_matches() {
let scanner = GdprPatternScanner::new();
let pattern = &scanner.patterns[0]; // gdpr-pii-logging
// Regex: (log|print|console\.|logger\.|tracing::)\s*[\.(].*\b(pii_keyword)\b
assert!(pattern.pattern.is_match("console.log(email)"));
assert!(pattern.pattern.is_match("console.log(user.ssn)"));
assert!(pattern.pattern.is_match("print(phone_number)"));
assert!(pattern.pattern.is_match("tracing::(ip_addr)"));
assert!(pattern.pattern.is_match("log.debug(credit_card)"));
}
#[test]
fn gdpr_pii_logging_no_false_positive() {
let scanner = GdprPatternScanner::new();
let pattern = &scanner.patterns[0];
// Regular logging without PII fields should not match
assert!(!pattern
.pattern
.is_match("logger.info(\"request completed\")"));
assert!(!pattern.pattern.is_match("let email = user.email;"));
}
#[test]
fn gdpr_no_consent_matches() {
let scanner = GdprPatternScanner::new();
let pattern = &scanner.patterns[1]; // gdpr-no-consent
assert!(pattern.pattern.is_match("collect personal data"));
assert!(pattern.pattern.is_match("store user_data in db"));
assert!(pattern.pattern.is_match("save pii to disk"));
}
#[test]
fn gdpr_user_model_matches() {
let scanner = GdprPatternScanner::new();
let pattern = &scanner.patterns[2]; // gdpr-no-delete-endpoint
assert!(pattern.pattern.is_match("struct User {"));
assert!(pattern.pattern.is_match("class User(Model):"));
}
#[test]
fn gdpr_hardcoded_retention_matches() {
let scanner = GdprPatternScanner::new();
let pattern = &scanner.patterns[3]; // gdpr-hardcoded-retention
assert!(pattern.pattern.is_match("retention = 30"));
assert!(pattern.pattern.is_match("ttl: 3600"));
assert!(pattern.pattern.is_match("expire = 86400"));
}
// --- OAuth pattern tests ---
#[test]
fn oauth_implicit_grant_matches() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[0]; // oauth-implicit-grant
assert!(pattern.pattern.is_match("response_type = \"token\""));
assert!(pattern.pattern.is_match("grant_type: implicit"));
assert!(pattern.pattern.is_match("response_type='token'"));
}
#[test]
fn oauth_implicit_grant_no_false_positive() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[0];
assert!(!pattern.pattern.is_match("response_type = \"code\""));
assert!(!pattern.pattern.is_match("grant_type: authorization_code"));
}
#[test]
fn oauth_authorization_code_matches() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[1]; // oauth-missing-pkce
assert!(pattern.pattern.is_match("uses authorization_code flow"));
assert!(pattern.pattern.is_match("authorization code grant"));
}
#[test]
fn oauth_token_localstorage_matches() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[2]; // oauth-token-localstorage
assert!(pattern
.pattern
.is_match("localStorage.setItem('access_token', tok)"));
assert!(pattern
.pattern
.is_match("localStorage.getItem(\"refresh_token\")"));
}
#[test]
fn oauth_token_localstorage_no_false_positive() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[2];
assert!(!pattern
.pattern
.is_match("localStorage.setItem('theme', 'dark')"));
assert!(!pattern
.pattern
.is_match("sessionStorage.setItem('token', t)"));
}
#[test]
fn oauth_token_url_matches() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[3]; // oauth-token-url
assert!(pattern.pattern.is_match("access_token = build_url(query)"));
assert!(pattern.pattern.is_match("bearer = url.param"));
}
// --- Pattern rule file extension filtering ---
#[test]
fn gdpr_patterns_cover_common_languages() {
let scanner = GdprPatternScanner::new();
for pattern in &scanner.patterns {
assert!(
pattern.file_extensions.contains(&"rs".to_string()),
"Pattern {} should cover .rs files",
pattern.id
);
}
}
#[test]
fn oauth_localstorage_only_js_ts() {
let scanner = OAuthPatternScanner::new();
let pattern = &scanner.patterns[2]; // oauth-token-localstorage
assert!(pattern.file_extensions.contains(&"js".to_string()));
assert!(pattern.file_extensions.contains(&"ts".to_string()));
assert!(!pattern.file_extensions.contains(&"rs".to_string()));
assert!(!pattern.file_extensions.contains(&"py".to_string()));
}
}