seer_core/

diff.rs

//! Side-by-side comparison of two domains.
//!
//! [`DomainDiffer::diff`] runs a smart lookup (registration) and a status
//! check (DNS resolution + TLS certificate) for each domain concurrently and
//! lines the results up field by field in a [`DomainDiff`]. A leg that fails
//! leaves its fields empty (`None` / no records) and records why in
//! [`DomainDiff::errors`], so a failed check is never shown as a real answer
//! (a status error used to read as "does not resolve").

use chrono::{DateTime, Utc};
use serde::{Deserialize, Serialize};
use tracing::{debug, instrument};

use crate::error::{Result, SeerError};
use crate::lookup::{LookupResult, SmartLookup};
use crate::status::{StatusClient, StatusResponse};

/// Side-by-side comparison of two domains across registration, DNS, and SSL.
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct DomainDiff {
    pub domain_a: String,
    pub domain_b: String,
    pub registration: RegistrationDiff,
    pub dns: DnsDiff,
    pub ssl: SslDiff,
    /// Checks that failed (`"<domain>: registration lookup failed: …"`,
    /// sanitized); their fields above are left empty, not defaulted.
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub errors: Vec<String>,
}

/// Registration data comparison (registrar, organization, dates).
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct RegistrationDiff {
    pub registrar: (Option<String>, Option<String>),
    pub organization: (Option<String>, Option<String>),
    pub created: (Option<String>, Option<String>),
    pub expires: (Option<String>, Option<String>),
}

/// DNS resolution comparison (A records, nameservers, reachability).
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct DnsDiff {
    pub a_records: (Vec<String>, Vec<String>),
    pub nameservers: (Vec<String>, Vec<String>),
    /// Whether the domain resolves; `None` when its status check failed or
    /// did not check DNS.
    pub resolves: (Option<bool>, Option<bool>),
}

/// SSL certificate comparison (issuer, validity, remaining days).
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct SslDiff {
    pub issuer: (Option<String>, Option<String>),
    pub valid_until: (Option<String>, Option<String>),
    pub days_remaining: (Option<i64>, Option<i64>),
    pub is_valid: (Option<bool>, Option<bool>),
}

/// Compares two domains by running lookups and status checks concurrently.
#[derive(Default)]
pub struct DomainDiffer {
    lookup: SmartLookup,
    status_client: StatusClient,
}

impl DomainDiffer {
    pub fn new() -> Self {
        Self::default()
    }

    /// Builds a differ honoring `~/.seer/config.toml`: the lookup and status
    /// clients take their per-protocol timeouts from `config`.
    pub fn from_config(config: &crate::config::SeerConfig) -> Self {
        Self {
            lookup: SmartLookup::from_config(config),
            status_client: StatusClient::from_config(config),
        }
    }

    /// Compares two domains, returning their registration, DNS, and SSL differences.
    ///
    /// All four network calls (lookup + status for each domain) run concurrently.
    #[instrument(skip(self), fields(domain_a = %domain_a, domain_b = %domain_b))]
    pub async fn diff(&self, domain_a: &str, domain_b: &str) -> Result<DomainDiff> {
        let domain_a = crate::validation::normalize_domain(domain_a)?;
        let domain_b = crate::validation::normalize_domain(domain_b)?;

        // Box::pin each branch so the four heavyweight futures live on the heap
        // instead of inlining into the join! state. Each lookup/status future
        // itself joins many protocol futures; inlined, the combined `diff`
        // future is ~105 KB and a single poll overflows a tokio worker stack in
        // debug builds (TUI `:diff a b` crash). Boxing keeps the poll shallow.
        let (lookup_a, lookup_b, status_a, status_b) = tokio::join!(
            Box::pin(self.lookup.lookup(&domain_a)),
            Box::pin(self.lookup.lookup(&domain_b)),
            Box::pin(self.status_client.check(&domain_a)),
            Box::pin(self.status_client.check(&domain_b)),
        );

        let registration = build_registration_diff(lookup_a.as_ref().ok(), lookup_b.as_ref().ok());
        let dns = build_dns_diff(status_a.as_ref().ok(), status_b.as_ref().ok());
        let ssl = build_ssl_diff(status_a.as_ref().ok(), status_b.as_ref().ok());

        let errors = failures(&[
            (&domain_a, "registration lookup", lookup_a.as_ref().err()),
            (&domain_b, "registration lookup", lookup_b.as_ref().err()),
            (&domain_a, "status check", status_a.as_ref().err()),
            (&domain_b, "status check", status_b.as_ref().err()),
        ]);

        debug!("Domain diff complete");

        Ok(DomainDiff {
            domain_a,
            domain_b,
            registration,
            dns,
            ssl,
            errors,
        })
    }
}

/// One sanitized line per failed `(domain, check, error)` leg; the full error
/// goes to the debug log.
fn failures(legs: &[(&str, &str, Option<&SeerError>)]) -> Vec<String> {
    legs.iter()
        .filter_map(|(domain, check, failure)| {
            let e = (*failure)?;
            debug!(%domain, error = %e, "{check} failed");
            Some(format!(
                "{domain}: {check} failed: {}",
                e.sanitized_message()
            ))
        })
        .collect()
}

/// `YYYY-MM-DD`, the date form every diff field uses.
fn ymd(date: DateTime<Utc>) -> String {
    date.format("%Y-%m-%d").to_string()
}

fn build_registration_diff(a: Option<&LookupResult>, b: Option<&LookupResult>) -> RegistrationDiff {
    // Both dates share LookupResult's RDAP-then-WHOIS fallback, so a diff is
    // never internally inconsistent about where its dates came from.
    let side = |r: Option<&LookupResult>| {
        (
            r.and_then(LookupResult::registrar),
            r.and_then(LookupResult::organization),
            r.and_then(LookupResult::creation_date).map(ymd),
            r.and_then(LookupResult::expiration_date).map(ymd),
        )
    };
    let ((registrar_a, org_a, created_a, expires_a), (registrar_b, org_b, created_b, expires_b)) =
        (side(a), side(b));

    RegistrationDiff {
        registrar: (registrar_a, registrar_b),
        organization: (org_a, org_b),
        created: (created_a, created_b),
        expires: (expires_a, expires_b),
    }
}

fn build_dns_diff(a: Option<&StatusResponse>, b: Option<&StatusResponse>) -> DnsDiff {
    let side = |s: Option<&StatusResponse>| match s.and_then(|s| s.dns_resolution.as_ref()) {
        Some(d) => (d.a_records.clone(), d.nameservers.clone(), Some(d.resolves)),
        None => (Vec::new(), Vec::new(), None),
    };
    let ((a_records_a, ns_a, resolves_a), (a_records_b, ns_b, resolves_b)) = (side(a), side(b));

    DnsDiff {
        a_records: (a_records_a, a_records_b),
        nameservers: (ns_a, ns_b),
        resolves: (resolves_a, resolves_b),
    }
}

fn build_ssl_diff(a: Option<&StatusResponse>, b: Option<&StatusResponse>) -> SslDiff {
    let side = |s: Option<&StatusResponse>| {
        let cert = s.and_then(|s| s.certificate.as_ref());
        (
            cert.map(|c| c.issuer.clone()),
            cert.map(|c| ymd(c.valid_until)),
            cert.map(|c| c.days_until_expiry),
            cert.map(|c| c.is_valid),
        )
    };
    let (
        (issuer_a, valid_until_a, days_a, is_valid_a),
        (issuer_b, valid_until_b, days_b, is_valid_b),
    ) = (side(a), side(b));

    SslDiff {
        issuer: (issuer_a, issuer_b),
        valid_until: (valid_until_a, valid_until_b),
        days_remaining: (days_a, days_b),
        is_valid: (is_valid_a, is_valid_b),
    }
}

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

    #[test]
    fn test_domain_diff_serialization() {
        let diff = DomainDiff {
            domain_a: "example.com".to_string(),
            domain_b: "google.com".to_string(),
            registration: RegistrationDiff {
                registrar: (Some("IANA".to_string()), Some("MarkMonitor".to_string())),
                organization: (None, Some("Google LLC".to_string())),
                created: (
                    Some("1995-08-14".to_string()),
                    Some("1997-09-15".to_string()),
                ),
                expires: (
                    Some("2026-08-13".to_string()),
                    Some("2028-09-14".to_string()),
                ),
            },
            dns: DnsDiff {
                a_records: (
                    vec!["93.184.216.34".to_string()],
                    vec!["142.250.185.46".to_string()],
                ),
                nameservers: (
                    vec!["a.iana-servers.net".to_string()],
                    vec!["ns1.google.com".to_string()],
                ),
                resolves: (Some(true), Some(true)),
            },
            ssl: SslDiff {
                issuer: (
                    Some("DigiCert Inc".to_string()),
                    Some("Google Trust Services".to_string()),
                ),
                valid_until: (
                    Some("2025-03-01".to_string()),
                    Some("2025-02-15".to_string()),
                ),
                days_remaining: (Some(89), Some(75)),
                is_valid: (Some(true), Some(true)),
            },
            errors: Vec::new(),
        };

        let json = serde_json::to_string(&diff).unwrap();
        assert!(json.contains("example.com"));
        assert!(json.contains("google.com"));
        assert!(json.contains("IANA"));
        assert!(json.contains("MarkMonitor"));
    }

    #[test]
    fn extract_dates_rdap_falls_back_to_whois_creation_date() {
        use crate::rdap::RdapResponse;
        use crate::whois::WhoisResponse;

        // RDAP response with no `registration` event -> creation_date() is None.
        let rdap = RdapResponse::default();
        assert!(rdap.creation_date().is_none());

        // WHOIS fallback carries a creation date.
        let whois = WhoisResponse::parse_internal(
            "example.com",
            "whois.verisign-grs.com",
            "Domain Name: example.com\nCreation Date: 1995-08-14T04:00:00Z\n",
        );
        assert!(
            whois.creation_date.is_some(),
            "whois should parse creation date"
        );

        let result = LookupResult::Rdap {
            data: Box::new(rdap),
            whois_fallback: Some(whois),
        };
        // created must use the WHOIS fallback, matching the expiration ladder,
        // rather than being left None.
        let diff = build_registration_diff(Some(&result), None);
        assert_eq!(diff.created.0.as_deref(), Some("1995-08-14"));
    }

    #[test]
    fn test_build_registration_diff_both_none() {
        let diff = build_registration_diff(None, None);
        assert!(diff.registrar.0.is_none());
        assert!(diff.registrar.1.is_none());
        assert!(diff.organization.0.is_none());
        assert!(diff.organization.1.is_none());
        assert!(diff.created.0.is_none());
        assert!(diff.created.1.is_none());
        assert!(diff.expires.0.is_none());
        assert!(diff.expires.1.is_none());
    }

    #[test]
    fn test_build_dns_diff_both_none() {
        let diff = build_dns_diff(None, None);
        assert!(diff.a_records.0.is_empty());
        assert!(diff.a_records.1.is_empty());
        assert!(diff.nameservers.0.is_empty());
        assert!(diff.nameservers.1.is_empty());
        // A failed status check is unknown, not "does not resolve".
        assert_eq!(diff.resolves, (None, None));
    }

    #[test]
    fn failed_legs_are_reported_not_defaulted() {
        let err = SeerError::Timeout("status check timed out".to_string());
        let errors = failures(&[
            ("a.test", "registration lookup", None),
            ("b.test", "status check", Some(&err)),
        ]);
        assert_eq!(errors.len(), 1);
        assert!(
            errors[0].starts_with("b.test: status check failed: "),
            "{errors:?}"
        );
        // Serialized only when something failed.
        let json = |errors: Vec<String>| {
            serde_json::to_value(DomainDiff {
                domain_a: "a.test".into(),
                domain_b: "b.test".into(),
                registration: build_registration_diff(None, None),
                dns: build_dns_diff(None, None),
                ssl: build_ssl_diff(None, None),
                errors,
            })
            .unwrap()
        };
        assert!(json(Vec::new()).get("errors").is_none());
        assert_eq!(json(errors)["errors"].as_array().map(Vec::len), Some(1));
    }

    #[test]
    fn test_build_ssl_diff_both_none() {
        let diff = build_ssl_diff(None, None);
        assert!(diff.issuer.0.is_none());
        assert!(diff.issuer.1.is_none());
        assert!(diff.valid_until.0.is_none());
        assert!(diff.valid_until.1.is_none());
        assert!(diff.days_remaining.0.is_none());
        assert!(diff.days_remaining.1.is_none());
        assert!(diff.is_valid.0.is_none());
        assert!(diff.is_valid.1.is_none());
    }

    #[test]
    fn test_domain_differ_default() {
        let differ = DomainDiffer::default();
        // Just verify construction works
        let _ = differ;
    }

    /// Regression guard for a debug-build stack overflow (TUI `:diff a b`).
    ///
    /// `diff()` joins four heavyweight futures (lookup ×2, status ×2), each of
    /// which internally joins many protocol futures (RDAP/WHOIS/DNS/SSL/...).
    /// If those four are inlined into the `join!` state instead of boxed, the
    /// combined future balloons to tens of KB; a single inline poll then
    /// overflows a 2 MB tokio worker stack once nested under the TUI's extra
    /// `handle_action → data::fetch` wrapper frames. Boxing the joined futures
    /// keeps this future small and the poll shallow.
    #[test]
    fn diff_future_stays_small_enough_for_the_stack() {
        let differ = DomainDiffer::new();
        // Construct but never await — keeps the test hermetic (no network).
        let fut = differ.diff("example.com", "example.org");
        let size = size_of_val(&fut);
        assert!(
            size < 8_192,
            "diff() future is {size} bytes; the four joined lookup/status \
             futures must be boxed so they don't inline onto the poll stack"
        );
    }
}