compliance-scanner-agent/compliance-agent/src/pipeline/ics/mod.rs

//! Dynamic ICS (industrial control system) probing for PLC/SPS targets.
//!
//! Where the control-logic scanner is static (over ST / PLCopen XML), this probes
//! the *running* device over industrial protocols and reports exposed /
//! unauthenticated control interfaces. It is read-only: it never writes to a live
//! process. Modbus/TCP is implemented first; OPC UA / EtherNet-IP are follow-ons.

pub mod modbus;

use std::time::Duration;

use compliance_core::models::{Finding, ScanType, Severity};

use crate::pipeline::dedup;

/// Default Modbus/TCP port.
const MODBUS_PORT: u16 = 502;

/// Probe a PLC/SPS device's industrial-protocol surface and return findings.
/// `endpoint` is the target's live-URL / host reference.
pub async fn probe_target(endpoint: &str, repo_id: &str, budget: Duration) -> Vec<Finding> {
    let (host, port) = parse_endpoint(endpoint);
    let probe = modbus::probe(&host, port, budget).await;
    let mut findings = Vec::new();
    if !probe.speaks_modbus {
        // Not reachable, or the port does not speak Modbus — nothing to report.
        return findings;
    }
    let target = format!("{host}:{port}");

    // Reachable Modbus/TCP = unauthenticated, cleartext control access by design.
    let fp = dedup::compute_fingerprint(&[repo_id, "ics-modbus-exposed", &target]);
    let mut f = Finding::new(
        repo_id.to_string(),
        fp,
        "ics-probe".to_string(),
        ScanType::IcsProbe,
        "Modbus/TCP control interface exposed without authentication".to_string(),
        format!(
            "The device at {target} answers Modbus/TCP requests. Modbus/TCP has no \
             authentication or encryption in the protocol, so any host that can reach this \
             port can read and write process variables (coils/registers) and disrupt the \
             controlled process."
        ),
        Severity::Critical,
    );
    f.rule_id = Some("ics-modbus-exposed".to_string());
    f.cwe = Some("CWE-306".to_string());
    f.remediation = Some(
        "Restrict the Modbus/TCP port to a trusted control network (segmentation / \
         firewall / VPN), never expose it to IT or the internet, and prefer an authenticated \
         transport (e.g. Modbus/TLS) or a secure protocol gateway where available."
            .to_string(),
    );
    findings.push(f);

    if let Some(dev) = &probe.device {
        let details = [
            dev.vendor.as_deref(),
            dev.product.as_deref(),
            dev.revision.as_deref(),
        ]
        .into_iter()
        .flatten()
        .collect::<Vec<_>>()
        .join(" / ");
        let fp = dedup::compute_fingerprint(&[repo_id, "ics-device-disclosure", &target]);
        let mut f = Finding::new(
            repo_id.to_string(),
            fp,
            "ics-probe".to_string(),
            ScanType::IcsProbe,
            "PLC device identity disclosed over Modbus".to_string(),
            format!(
                "The device at {target} discloses its identity via Modbus Read Device \
                 Identification: {details}. This aids fingerprinting and targeting of \
                 known-vulnerable firmware/runtime versions."
            ),
            Severity::Low,
        );
        f.rule_id = Some("ics-device-disclosure".to_string());
        f.cwe = Some("CWE-200".to_string());
        f.remediation = Some(
            "Limit network reach to the device; Modbus device identification cannot be \
             disabled, so exposure is bounded by network segmentation."
                .to_string(),
        );
        findings.push(f);
    }
    findings
}

/// Extract `(host, port)` from a target reference. Modbus lives on its own port
/// (502 by default), independent of any HTTP/WebVisu URL, so unless the reference
/// explicitly carries `modbus://host:port` or a bare `host:port`, we probe 502.
fn parse_endpoint(endpoint: &str) -> (String, u16) {
    let s = endpoint.trim();
    let (scheme, rest) = match s.split_once("://") {
        Some((sch, r)) => (Some(sch.to_ascii_lowercase()), r),
        None => (None, s),
    };
    let hostport = rest.split(['/', '?']).next().unwrap_or(rest);
    let (host, port) = match hostport.rsplit_once(':') {
        Some((h, p)) => (h.to_string(), p.parse::<u16>().ok()),
        None => (hostport.to_string(), None),
    };
    let port = match (scheme.as_deref(), port) {
        // Explicit Modbus port, or a bare host:port the user chose.
        (Some("modbus"), Some(p)) | (None, Some(p)) => p,
        // An http(s)/WebVisu URL (or no port): Modbus is on its own port.
        _ => MODBUS_PORT,
    };
    (host, port)
}

#[cfg(test)]
mod tests {
    use super::parse_endpoint;

    #[test]
    fn endpoint_parsing_picks_the_modbus_port() {
        assert_eq!(parse_endpoint("10.0.0.5"), ("10.0.0.5".into(), 502));
        assert_eq!(parse_endpoint("10.0.0.5:1502"), ("10.0.0.5".into(), 1502));
        assert_eq!(
            parse_endpoint("modbus://plc.local:5020"),
            ("plc.local".into(), 5020)
        );
        // A WebVisu URL: the http port is ignored; Modbus is on 502.
        assert_eq!(
            parse_endpoint("http://plc.local:8080/webvisu"),
            ("plc.local".into(), 502)
        );
        assert_eq!(
            parse_endpoint("https://plc.local/"),
            ("plc.local".into(), 502)
        );
    }
}