rpki/src/tools/sequence_triage_ccr_cir/analysis.rs

use std::collections::BTreeSet;

use super::args::Args;
use super::io::object_hash_key;
use super::model::{
    AnalysisResult, DiffEvent, EventOccurrence, SampleRecord, SequenceSample, Side,
};

pub(super) fn analyze_set<F>(
    result: &mut AnalysisResult,
    event_type: &'static str,
    left: &[SequenceSample],
    right: &[SequenceSample],
    extract: F,
    resolved_class: &'static str,
    persistent_class: &'static str,
    args: &Args,
) where
    F: for<'a> Fn(&'a SequenceSample) -> &'a BTreeSet<String>,
{
    analyze_direction(
        result,
        event_type,
        Side::Left,
        left,
        right,
        &extract,
        resolved_class,
        persistent_class,
        args,
    );
    analyze_direction(
        result,
        event_type,
        Side::Right,
        right,
        left,
        &extract,
        resolved_class,
        persistent_class,
        args,
    );
}

fn analyze_direction<F>(
    result: &mut AnalysisResult,
    event_type: &'static str,
    source_side: Side,
    source: &[SequenceSample],
    peer: &[SequenceSample],
    extract: &F,
    resolved_class: &'static str,
    persistent_class: &'static str,
    args: &Args,
) where
    F: for<'a> Fn(&'a SequenceSample) -> &'a BTreeSet<String>,
{
    for sample in source {
        let source_set = extract(sample);
        for key in source_set {
            if peer_sample_at_seq(peer, sample.raw.seq)
                .is_some_and(|peer_sample| extract(peer_sample).contains(key))
            {
                continue;
            }
            if let Some(matched) = find_future_match(peer, sample, key, extract, args) {
                result.add(
                    resolved_class,
                    SampleRecord {
                        classification: resolved_class,
                        event_type,
                        key: key.clone(),
                        source_side,
                        source_seq: sample.raw.seq,
                        source_run_id: sample.raw.run_id.clone(),
                        matched_seq: Some(matched.seq),
                        matched_run_id: Some(matched.run_id),
                        note: format!(
                            "matched in {} sequence within alignment window",
                            matched.side.as_str()
                        ),
                    },
                    args.sample_limit,
                );
            } else {
                result.add(
                    persistent_class,
                    SampleRecord {
                        classification: persistent_class,
                        event_type,
                        key: key.clone(),
                        source_side,
                        source_seq: sample.raw.seq,
                        source_run_id: sample.raw.run_id.clone(),
                        matched_seq: None,
                        matched_run_id: None,
                        note: "no matching event in peer sequence alignment window".to_string(),
                    },
                    args.sample_limit,
                );
            }
        }
    }
}

pub(super) fn analyze_hash_rollover(
    result: &mut AnalysisResult,
    left: &[SequenceSample],
    right: &[SequenceSample],
    args: &Args,
) {
    for (source_side, source, peer) in [(Side::Left, left, right), (Side::Right, right, left)] {
        for sample in source {
            for (uri, hash) in &sample.objects {
                if peer_sample_at_seq(peer, sample.raw.seq)
                    .and_then(|peer_sample| peer_sample.objects.get(uri))
                    .is_some_and(|peer_hash| peer_hash == hash)
                {
                    continue;
                }
                if let Some(peer_sample) = peer_sample_at_seq(peer, sample.raw.seq)
                    && peer_sample.objects.contains_key(uri)
                    && find_future_hash_match(peer, sample, uri, hash, args).is_some()
                {
                    let matched =
                        find_future_hash_match(peer, sample, uri, hash, args).expect("match");
                    result.add(
                        "CONTENT_ROLLOVER_RESOLVED",
                        SampleRecord {
                            classification: "CONTENT_ROLLOVER_RESOLVED",
                            event_type: "object_content_rollover",
                            key: object_hash_key(uri, hash),
                            source_side,
                            source_seq: sample.raw.seq,
                            source_run_id: sample.raw.run_id.clone(),
                            matched_seq: Some(matched.seq),
                            matched_run_id: Some(matched.run_id),
                            note: "same URI hash appeared in peer sequence later".to_string(),
                        },
                        args.sample_limit,
                    );
                }
            }
        }
    }
}

pub(super) fn collect_persistent_events(
    left: &[SequenceSample],
    right: &[SequenceSample],
    args: &Args,
) -> Vec<DiffEvent> {
    let mut events = Vec::new();
    collect_persistent_set(
        &mut events,
        "object_uri",
        "PERSISTENT_OBJECT_SET_DIVERGENCE",
        left,
        right,
        |sample| &sample.object_uris,
        args,
    );
    collect_persistent_set(
        &mut events,
        "object_hash",
        "PERSISTENT_CONTENT_DIVERGENCE",
        left,
        right,
        |sample| &sample.object_hashes,
        args,
    );
    collect_persistent_set(
        &mut events,
        "reject_uri",
        "PERSISTENT_REJECT_DIVERGENCE",
        left,
        right,
        |sample| &sample.rejects,
        args,
    );
    collect_persistent_set(
        &mut events,
        "trust_anchor",
        "PERSISTENT_TA_DIFFERENCE",
        left,
        right,
        |sample| &sample.trust_anchors,
        args,
    );
    collect_persistent_set(
        &mut events,
        "vrp_output",
        "PERSISTENT_OUTPUT_DIVERGENCE",
        left,
        right,
        |sample| &sample.vrps,
        args,
    );
    collect_persistent_set(
        &mut events,
        "vap_output",
        "PERSISTENT_OUTPUT_DIVERGENCE",
        left,
        right,
        |sample| &sample.vaps,
        args,
    );
    events
}

fn collect_persistent_set<F>(
    events: &mut Vec<DiffEvent>,
    event_type: &'static str,
    raw_class: &'static str,
    left: &[SequenceSample],
    right: &[SequenceSample],
    extract: F,
    args: &Args,
) where
    F: for<'a> Fn(&'a SequenceSample) -> &'a BTreeSet<String>,
{
    collect_persistent_direction(
        events,
        event_type,
        raw_class,
        Side::Left,
        left,
        right,
        &extract,
        args,
    );
    collect_persistent_direction(
        events,
        event_type,
        raw_class,
        Side::Right,
        right,
        left,
        &extract,
        args,
    );
}

fn collect_persistent_direction<F>(
    events: &mut Vec<DiffEvent>,
    event_type: &'static str,
    raw_class: &'static str,
    source_side: Side,
    source: &[SequenceSample],
    peer: &[SequenceSample],
    extract: &F,
    args: &Args,
) where
    F: for<'a> Fn(&'a SequenceSample) -> &'a BTreeSet<String>,
{
    for sample in source {
        for key in extract(sample) {
            if peer_sample_at_seq(peer, sample.raw.seq)
                .is_some_and(|peer_sample| extract(peer_sample).contains(key))
            {
                continue;
            }
            if find_future_match(peer, sample, key, extract, args).is_some() {
                continue;
            }
            events.push(DiffEvent {
                event_type,
                raw_class,
                key: key.clone(),
                source_side,
                source_seq: sample.raw.seq,
                source_run_id: sample.raw.run_id.clone(),
            });
        }
    }
}

impl AnalysisResult {
    fn add(&mut self, class: &'static str, record: SampleRecord, sample_limit: usize) {
        let stats = self.stats.entry(class).or_default();
        stats.total += 1;
        if stats.samples.len() < sample_limit {
            stats.samples.push(record);
        }
    }
}

pub(super) fn peer_sample_at_seq(peer: &[SequenceSample], seq: u32) -> Option<&SequenceSample> {
    peer.iter().find(|sample| sample.raw.seq == seq)
}

fn find_future_match<F>(
    peer: &[SequenceSample],
    source: &SequenceSample,
    key: &str,
    extract: &F,
    args: &Args,
) -> Option<EventOccurrence>
where
    F: for<'a> Fn(&'a SequenceSample) -> &'a BTreeSet<String>,
{
    peer.iter()
        .filter(|candidate| is_in_alignment_window(source, candidate, args))
        .find(|candidate| extract(candidate).contains(key))
        .map(|candidate| {
            occurrence(
                candidate,
                if candidate.raw.side.as_deref() == Some("left") {
                    Side::Left
                } else {
                    Side::Right
                },
            )
        })
}

fn find_future_hash_match(
    peer: &[SequenceSample],
    source: &SequenceSample,
    uri: &str,
    hash: &str,
    args: &Args,
) -> Option<EventOccurrence> {
    peer.iter()
        .filter(|candidate| is_in_alignment_window(source, candidate, args))
        .find(|candidate| {
            candidate
                .objects
                .get(uri)
                .is_some_and(|peer_hash| peer_hash == hash)
        })
        .map(|candidate| {
            occurrence(
                candidate,
                if candidate.raw.side.as_deref() == Some("left") {
                    Side::Left
                } else {
                    Side::Right
                },
            )
        })
}

fn is_in_alignment_window(
    source: &SequenceSample,
    candidate: &SequenceSample,
    args: &Args,
) -> bool {
    if candidate.raw.seq < source.raw.seq {
        return false;
    }
    let run_delta = candidate.raw.seq.saturating_sub(source.raw.seq);
    let time_delta = candidate.validation_time - source.validation_time;
    let secs = time_delta.whole_seconds().abs();
    run_delta <= args.align_window_runs || secs <= args.align_window_secs
}

fn occurrence(sample: &SequenceSample, side: Side) -> EventOccurrence {
    EventOccurrence {
        side,
        seq: sample.raw.seq,
        run_id: sample.raw.run_id.clone(),
    }
}