Skip to main content

systemprompt_analytics/services/
extractor.rs

1use http::{HeaderMap, Uri};
2use std::collections::HashMap;
3
4use axum::extract::Request;
5
6use super::bot_keywords::{matches_bot_ip_range, matches_bot_pattern};
7use super::detection;
8use super::user_agent::parse_user_agent;
9use crate::GeoIpReader;
10
11#[derive(Debug, Clone, Default)]
12pub struct SessionAnalytics {
13    pub ip_address: Option<String>,
14    pub user_agent: Option<String>,
15    pub device_type: Option<String>,
16    pub browser: Option<String>,
17    pub os: Option<String>,
18    pub fingerprint_hash: Option<String>,
19    pub preferred_locale: Option<String>,
20    pub country: Option<String>,
21    pub region: Option<String>,
22    pub city: Option<String>,
23    pub referrer_source: Option<String>,
24    pub referrer_url: Option<String>,
25    pub landing_page: Option<String>,
26    pub entry_url: Option<String>,
27    pub utm_source: Option<String>,
28    pub utm_medium: Option<String>,
29    pub utm_campaign: Option<String>,
30    pub utm_content: Option<String>,
31    pub utm_term: Option<String>,
32}
33
34impl SessionAnalytics {
35    pub fn from_headers(headers: &HeaderMap) -> Self {
36        Self::from_headers_with_geoip(headers, None)
37    }
38
39    pub fn from_headers_with_geoip(
40        headers: &HeaderMap,
41        geoip_reader: Option<&GeoIpReader>,
42    ) -> Self {
43        Self::from_headers_with_geoip_and_socket(headers, geoip_reader, None)
44    }
45
46    pub fn from_headers_with_geoip_and_socket(
47        headers: &HeaderMap,
48        geoip_reader: Option<&GeoIpReader>,
49        socket_addr: Option<std::net::SocketAddr>,
50    ) -> Self {
51        let user_agent = headers
52            .get("user-agent")
53            .and_then(|v| v.to_str().ok())
54            .map(ToString::to_string);
55
56        let ip_address = headers
57            .get("x-forwarded-for")
58            .and_then(|v| v.to_str().ok())
59            .and_then(|s| s.split(',').next())
60            .map(|s| s.trim().to_string())
61            .or_else(|| {
62                headers
63                    .get("x-real-ip")
64                    .and_then(|v| v.to_str().ok())
65                    .map(ToString::to_string)
66            })
67            .or_else(|| socket_addr.map(|addr| addr.ip().to_string()));
68
69        let fingerprint_hash = headers
70            .get("x-fingerprint")
71            .and_then(|v| v.to_str().ok())
72            .map(ToString::to_string);
73
74        let preferred_locale = headers
75            .get("accept-language")
76            .and_then(|v| v.to_str().ok())
77            .and_then(|s| s.split(',').next())
78            .map(|s| s.trim().split(';').next().unwrap_or(s).to_string());
79
80        let (device_type, browser, os) = user_agent
81            .as_ref()
82            .map_or((None, None, None), |ua| parse_user_agent(ua));
83
84        let (country, region, city) = ip_address
85            .as_ref()
86            .and_then(|ip_str| Self::lookup_geoip(ip_str, geoip_reader))
87            .unwrap_or((None, None, None));
88
89        let referrer_url = headers
90            .get("referer")
91            .and_then(|v| v.to_str().ok())
92            .map(ToString::to_string);
93
94        let referrer_source = referrer_url
95            .as_ref()
96            .and_then(|url| Self::parse_referrer_source(url));
97
98        Self {
99            ip_address,
100            user_agent,
101            device_type,
102            browser,
103            os,
104            fingerprint_hash,
105            preferred_locale,
106            country,
107            region,
108            city,
109            referrer_source,
110            referrer_url,
111            landing_page: None,
112            entry_url: None,
113            utm_source: None,
114            utm_medium: None,
115            utm_campaign: None,
116            utm_content: None,
117            utm_term: None,
118        }
119    }
120
121    pub fn from_headers_and_uri(
122        headers: &HeaderMap,
123        uri: Option<&Uri>,
124        geoip_reader: Option<&GeoIpReader>,
125        content_routing: Option<&dyn systemprompt_models::ContentRouting>,
126    ) -> Self {
127        let mut analytics = Self::from_headers_with_geoip(headers, geoip_reader);
128
129        if let Some(uri) = uri {
130            let query_params = Self::parse_query_params(uri);
131
132            analytics.utm_source = query_params.get("utm_source").cloned();
133            analytics.utm_medium = query_params.get("utm_medium").cloned();
134            analytics.utm_campaign = query_params.get("utm_campaign").cloned();
135            analytics.utm_content = query_params.get("utm_content").cloned();
136            analytics.utm_term = query_params.get("utm_term").cloned();
137
138            let is_html_page =
139                content_routing.is_some_and(|routing| routing.is_html_page(uri.path()));
140
141            if is_html_page {
142                analytics.entry_url = Some(uri.to_string());
143                analytics.landing_page = Some(uri.path().to_string());
144            }
145        }
146
147        analytics
148    }
149
150    pub fn from_request(
151        request: &Request,
152        geoip_reader: Option<&GeoIpReader>,
153        content_routing: Option<&dyn systemprompt_models::ContentRouting>,
154    ) -> Self {
155        Self::from_headers_and_uri(
156            request.headers(),
157            Some(request.uri()),
158            geoip_reader,
159            content_routing,
160        )
161    }
162
163    fn parse_query_params(uri: &Uri) -> HashMap<String, String> {
164        uri.query().map_or_else(HashMap::new, |q| {
165            q.split('&')
166                .filter_map(|param| {
167                    let mut parts = param.splitn(2, '=');
168                    Some((parts.next()?.to_string(), parts.next()?.to_string()))
169                })
170                .collect()
171        })
172    }
173
174    #[cfg(feature = "geolocation")]
175    fn lookup_geoip(
176        ip_str: &str,
177        geoip_reader: Option<&GeoIpReader>,
178    ) -> Option<(Option<String>, Option<String>, Option<String>)> {
179        let Some(reader) = geoip_reader else {
180            tracing::debug!(ip = %ip_str, "GeoIP lookup skipped: reader not configured");
181            return None;
182        };
183
184        let ip: std::net::IpAddr = match ip_str.parse() {
185            Ok(ip) => ip,
186            Err(e) => {
187                tracing::debug!(ip = %ip_str, error = %e, "GeoIP lookup failed: invalid IP address");
188                return None;
189            },
190        };
191
192        if ip.is_loopback() || ip.is_unspecified() {
193            tracing::debug!(ip = %ip_str, "GeoIP lookup skipped: loopback or unspecified address");
194            return None;
195        }
196
197        if let std::net::IpAddr::V4(ipv4) = ip {
198            if ipv4.is_private() || ipv4.is_link_local() {
199                tracing::debug!(ip = %ip_str, "GeoIP lookup skipped: private or link-local address");
200                return None;
201            }
202        }
203
204        let lookup_result = match reader.lookup(ip) {
205            Ok(result) => result,
206            Err(e) => {
207                tracing::debug!(ip = %ip_str, error = %e, "GeoIP lookup failed: database lookup error");
208                return None;
209            },
210        };
211
212        let city_data: maxminddb::geoip2::City = match lookup_result.decode() {
213            Ok(Some(data)) => data,
214            Ok(None) => {
215                tracing::debug!(ip = %ip_str, "GeoIP lookup returned empty result");
216                return None;
217            },
218            Err(e) => {
219                tracing::debug!(ip = %ip_str, error = %e, "GeoIP decode failed");
220                return None;
221            },
222        };
223
224        let country = city_data.country.iso_code.map(ToString::to_string);
225
226        let region = city_data
227            .subdivisions
228            .first()
229            .and_then(|s| s.iso_code)
230            .map(ToString::to_string);
231
232        let city_name = city_data.city.names.english.map(ToString::to_string);
233
234        Some((country, region, city_name))
235    }
236
237    #[cfg(not(feature = "geolocation"))]
238    const fn lookup_geoip(
239        _ip_str: &str,
240        _geoip_reader: Option<&GeoIpReader>,
241    ) -> Option<(Option<String>, Option<String>, Option<String>)> {
242        None
243    }
244
245    fn parse_referrer_source(url: &str) -> Option<String> {
246        url::Url::parse(url)
247            .ok()
248            .and_then(|parsed_url| parsed_url.host_str().map(ToString::to_string))
249            .filter(|host| host.parse::<std::net::IpAddr>().is_err())
250    }
251
252    pub fn is_bot(&self) -> bool {
253        self.user_agent
254            .as_ref()
255            .is_none_or(|ua| ua.is_empty() || matches_bot_pattern(ua))
256    }
257
258    pub fn is_bot_ip(&self) -> bool {
259        self.ip_address
260            .as_ref()
261            .is_some_and(|ip| matches_bot_ip_range(ip))
262    }
263
264    pub fn is_spam_referrer(&self) -> bool {
265        self.referrer_url
266            .as_ref()
267            .is_some_and(|url| detection::is_spam_referrer(url))
268    }
269
270    pub fn is_datacenter_ip(&self) -> bool {
271        self.ip_address
272            .as_ref()
273            .is_some_and(|ip| detection::is_datacenter_ip(ip))
274    }
275
276    pub fn is_high_risk_country(&self) -> bool {
277        self.country
278            .as_ref()
279            .is_some_and(|c| detection::is_high_risk_country(c))
280    }
281
282    pub fn should_skip_tracking(&self) -> bool {
283        self.is_bot()
284            || self.is_bot_ip()
285            || self.is_datacenter_ip()
286            || self.is_high_risk_country()
287            || self.is_spam_referrer()
288    }
289}