parlov-elicit 0.3.0

//! `EmgAppVsServer404` -- application vs server 404 body differential.
//!
//! Compares an application-routed 404 (structured JSON with request metadata)
//! against the server's default 404 (plain text, no app headers). The baseline
//! hits the real route with a known-existing ID; the probe hits a garbled URL
//! that bypasses application routing entirely.

use http::Method;
use parlov_core::{NormativeStrength, OracleClass, ProbeDefinition, Technique, Vector};

use crate::context::ScanContext;
use crate::strategy::Strategy;
use crate::types::{ProbePair, ProbeSpec, RiskLevel, StrategyMetadata};
use crate::util::substitute_url;

fn metadata() -> StrategyMetadata {
    StrategyMetadata {
        strategy_id: "emg-app-vs-server-404",
        strategy_name: "EMG: App vs Server 404",
        risk: RiskLevel::Safe,
    }
}

fn technique() -> Technique {
    Technique {
        id: "emg-app-vs-server-404",
        name: "Application vs server 404 body differential",
        oracle_class: OracleClass::Existence,
        vector: Vector::ErrorMessageGranularity,
        strength: NormativeStrength::May,
    }
}

/// Replaces the path segment containing `{id}` with a nonsense segment that
/// bypasses application routing, producing the server's default 404 response.
fn garble_url(target: &str) -> String {
    // Find the scheme separator to avoid matching '/' inside "https://"
    let path_start = target
        .find("://")
        .map_or(0, |i| i + 3 + target[i + 3..].find('/').unwrap_or(0));

    if let Some(rel_pos) = target[path_start..].rfind('/') {
        let pos = path_start + rel_pos;
        let base = &target[..pos];
        if let Some(rel_pos2) = base[path_start..].rfind('/') {
            let pos2 = path_start + rel_pos2;
            return format!("{}/_parlov_no_route/0", &target[..pos2]);
        }
    }
    target.replace("{id}", "_parlov_no_route")
}

/// Elicits existence differentials by comparing app-routed vs server-default 404 bodies.
pub struct EmgAppVsServer404;

impl Strategy for EmgAppVsServer404 {
    fn id(&self) -> &'static str {
        "emg-app-vs-server-404"
    }

    fn name(&self) -> &'static str {
        "EMG: App vs Server 404"
    }

    fn risk(&self) -> RiskLevel {
        RiskLevel::Safe
    }

    fn methods(&self) -> &[Method] {
        &[Method::GET]
    }

    fn is_applicable(&self, _ctx: &ScanContext) -> bool {
        true
    }

    fn generate(&self, ctx: &ScanContext) -> Vec<ProbeSpec> {
        let baseline_url = substitute_url(&ctx.target, &ctx.baseline_id);
        let probe_url = garble_url(&ctx.target);

        let pair = ProbePair {
            baseline: ProbeDefinition {
                url: baseline_url,
                method: Method::GET,
                headers: ctx.headers.clone(),
                body: None,
            },
            probe: ProbeDefinition {
                url: probe_url,
                method: Method::GET,
                headers: ctx.headers.clone(),
                body: None,
            },
            metadata: metadata(),
            technique: technique(),
        };
        vec![ProbeSpec::Pair(pair)]
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use http::HeaderMap;

    fn make_ctx() -> ScanContext {
        ScanContext {
            target: "https://api.example.com/api/users/{id}".to_string(),
            baseline_id: "1001".to_string(),
            probe_id: "9999".to_string(),
            headers: HeaderMap::new(),
            max_risk: RiskLevel::Safe,
            known_duplicate: None,
            state_field: None,
            alt_credential: None,
            body_template: None,
        }
    }

    #[test]
    fn generates_correct_technique_vector() {
        let specs = EmgAppVsServer404.generate(&make_ctx());
        assert_eq!(specs[0].technique().vector, Vector::ErrorMessageGranularity);
    }

    #[test]
    fn generates_correct_technique_strength() {
        let specs = EmgAppVsServer404.generate(&make_ctx());
        assert_eq!(specs[0].technique().strength, NormativeStrength::May);
    }

    #[test]
    fn is_always_applicable() {
        assert!(EmgAppVsServer404.is_applicable(&make_ctx()));
    }

    #[test]
    fn risk_is_safe() {
        assert_eq!(EmgAppVsServer404.risk(), RiskLevel::Safe);
    }

    #[test]
    fn generates_get_method() {
        let specs = EmgAppVsServer404.generate(&make_ctx());
        let ProbeSpec::Pair(pair) = &specs[0] else {
            panic!("expected Pair variant")
        };
        assert_eq!(pair.baseline.method, Method::GET);
        assert_eq!(pair.probe.method, Method::GET);
    }

    #[test]
    fn baseline_url_has_baseline_id() {
        let specs = EmgAppVsServer404.generate(&make_ctx());
        let ProbeSpec::Pair(pair) = &specs[0] else {
            panic!("expected Pair variant")
        };
        assert!(
            pair.baseline.url.contains("1001"),
            "baseline URL must contain baseline_id: {}",
            pair.baseline.url
        );
    }

    #[test]
    fn probe_url_is_garbled() {
        let specs = EmgAppVsServer404.generate(&make_ctx());
        let ProbeSpec::Pair(pair) = &specs[0] else {
            panic!("expected Pair variant")
        };
        assert!(
            !pair.probe.url.contains("{id}"),
            "probe URL must not contain {{id}}: {}",
            pair.probe.url
        );
        assert!(
            pair.probe.url.contains("_parlov_no_route"),
            "probe URL must contain _parlov_no_route: {}",
            pair.probe.url
        );
    }

    #[test]
    fn garble_url_replaces_parent_segment() {
        let result = garble_url("https://api.example.com/api/users/{id}");
        assert_eq!(result, "https://api.example.com/api/_parlov_no_route/0");
    }

    #[test]
    fn garble_url_fallback_when_single_segment() {
        let result = garble_url("https://api.example.com/{id}");
        assert_eq!(result, "https://api.example.com/_parlov_no_route");
    }
}