systemprompt_api/services/middleware/
bot_detector.rs

1use axum::extract::Request;
2use axum::middleware::Next;
3use axum::response::Response;
4use std::sync::Arc;
5use systemprompt_analytics::matches_bot_pattern;
6
7const DATACENTER_IP_PREFIXES: &[&str] = &[
8    "47.79.", "47.82.", "47.88.", "47.89.", "47.90.", "47.91.", "47.92.", "47.93.", "47.94.",
9    "47.95.", "47.96.", "47.97.", "47.98.", "47.99.", "47.100.", "47.101.", "47.102.", "47.103.",
10    "47.104.", "47.105.", "47.106.", "47.107.", "47.108.", "47.109.", "47.110.", "47.111.",
11    "47.112.", "47.113.", "47.114.", "47.115.", "47.116.", "47.117.", "47.118.", "47.119.",
12    "119.29.", "129.28.", "162.14.", "119.3.", "122.112.",
13];
14
15const CHROME_MIN_VERSION: i32 = 120;
16
17#[derive(Clone, Debug)]
18pub struct BotMarker {
19    pub is_bot: bool,
20    pub bot_type: BotType,
21    pub user_agent: String,
22    pub ip_address: Option<String>,
23}
24
25#[derive(Clone, Copy, Debug, PartialEq, Eq)]
26pub enum BotType {
27    KnownBot,
28    Scanner,
29    Suspicious,
30    Human,
31}
32
33pub async fn detect_bots_early(mut req: Request, next: Next) -> Response {
34    let user_agent = req
35        .headers()
36        .get("user-agent")
37        .and_then(|h| {
38            h.to_str()
39                .map_err(|e| {
40                    tracing::trace!(error = %e, "Invalid UTF-8 in user-agent header");
41                    e
42                })
43                .ok()
44        })
45        .unwrap_or("")
46        .to_string();
47
48    let ip_address = extract_ip_address(&req);
49    let uri_path = req.uri().path().to_string();
50
51    let marker = if is_known_bot(&user_agent) {
52        BotMarker {
53            is_bot: true,
54            bot_type: BotType::KnownBot,
55            user_agent: user_agent.clone(),
56            ip_address: ip_address.clone(),
57        }
58    } else if is_datacenter_ip(ip_address.as_deref()) || is_outdated_browser(&user_agent) {
59        BotMarker {
60            is_bot: true,
61            bot_type: BotType::Suspicious,
62            user_agent: user_agent.clone(),
63            ip_address: ip_address.clone(),
64        }
65    } else if is_scanner_request(&uri_path, &user_agent) {
66        BotMarker {
67            is_bot: false,
68            bot_type: BotType::Scanner,
69            user_agent: user_agent.clone(),
70            ip_address: ip_address.clone(),
71        }
72    } else {
73        BotMarker {
74            is_bot: false,
75            bot_type: BotType::Human,
76            user_agent: user_agent.clone(),
77            ip_address,
78        }
79    };
80
81    req.extensions_mut().insert(Arc::new(marker));
82    next.run(req).await
83}
84
85fn extract_ip_address(req: &Request) -> Option<String> {
86    req.headers()
87        .get("x-forwarded-for")
88        .and_then(|v| v.to_str().ok())
89        .and_then(|s| s.split(',').next())
90        .map(|s| s.trim().to_string())
91        .or_else(|| {
92            req.headers()
93                .get("x-real-ip")
94                .and_then(|v| v.to_str().ok())
95                .map(ToString::to_string)
96        })
97        .or_else(|| {
98            req.headers()
99                .get("cf-connecting-ip")
100                .and_then(|v| v.to_str().ok())
101                .map(ToString::to_string)
102        })
103}
104
105fn is_datacenter_ip(ip: Option<&str>) -> bool {
106    ip.is_some_and(|ip_addr| {
107        DATACENTER_IP_PREFIXES
108            .iter()
109            .any(|prefix| ip_addr.starts_with(prefix))
110    })
111}
112
113fn is_known_bot(user_agent: &str) -> bool {
114    matches_bot_pattern(user_agent)
115}
116
117fn is_outdated_browser(user_agent: &str) -> bool {
118    let ua_lower = user_agent.to_lowercase();
119
120    if let Some(pos) = ua_lower.find("chrome/") {
121        let version_str = &ua_lower[pos + 7..];
122        if let Some(dot_pos) = version_str.find('.') {
123            if let Ok(major) = version_str[..dot_pos].parse::<i32>() {
124                return major < CHROME_MIN_VERSION;
125            }
126        }
127    }
128
129    false
130}
131
132fn is_scanner_request(path: &str, user_agent: &str) -> bool {
133    let scanner_paths = [
134        ".env",
135        ".git",
136        ".php",
137        "admin",
138        "wp-admin",
139        "wp-login",
140        "administrator",
141        ".sql",
142        ".backup",
143        "config.php",
144        "web.config",
145        ".well-known",
146    ];
147
148    let scanner_agents = [
149        "masscan",
150        "nmap",
151        "nikto",
152        "sqlmap",
153        "metasploit",
154        "nessus",
155        "openvas",
156        "zap",
157        "burp",
158        "qualys",
159    ];
160
161    let path_lower = path.to_lowercase();
162    let ua_lower = user_agent.to_lowercase();
163
164    scanner_paths.iter().any(|p| path_lower.contains(p))
165        || scanner_agents.iter().any(|a| ua_lower.contains(a))
166}
systemprompt_api/services/middleware/bot_detector.rs

systemprompt_api/services/middleware/
bot_detector.rs
