tidaldb/tidal/tests/visibility_export.rs

//! Integration tests for M7P4 Operational Visibility — RLHF signal export.
//!
//! - Task 08: RLHF signal export (with session context)
#![allow(clippy::unwrap_used)]

#[cfg(feature = "test-utils")]
use std::collections::HashMap;
use std::time::Duration;

use tidaldb::{
    ExportFormat, ExportRequest, ExportedSignal, TidalDb,
    schema::{DecaySpec, EntityId, EntityKind, Timestamp, Window},
};

// ── Shared test schema ───────────────────────────────────────────────────────

fn build_test_schema() -> tidaldb::schema::Schema {
    use tidaldb::{AgentPolicy, schema::SchemaBuilder};

    let mut builder = SchemaBuilder::new();
    let _ = builder
        .signal(
            "view",
            EntityKind::Item,
            DecaySpec::Exponential {
                half_life: Duration::from_secs(7 * 24 * 3600),
            },
        )
        .windows(&[Window::OneHour])
        .add();
    let _ = builder
        .signal(
            "like",
            EntityKind::Item,
            DecaySpec::Exponential {
                half_life: Duration::from_secs(30 * 24 * 3600),
            },
        )
        .windows(&[Window::OneHour])
        .add();
    builder.session_policy(
        "default",
        AgentPolicy {
            allowed_signals: vec!["view".to_string(), "like".to_string()],
            denied_signals: vec![],
            max_session_duration: Duration::from_secs(3600),
            max_signals_per_session: 1000,
        },
    );
    builder.build().unwrap()
}

// ── Task 08: RLHF Signal Export ──────────────────────────────────────────────

#[test]
#[cfg(feature = "test-utils")]
fn export_signals_ephemeral_returns_empty() {
    use tidaldb::{TidalDb, schema::SchemaBuilder};

    let mut builder = SchemaBuilder::new();
    let _ = builder
        .signal(
            "view",
            EntityKind::Item,
            DecaySpec::Exponential {
                half_life: Duration::from_secs(7 * 24 * 3600),
            },
        )
        .windows(&[Window::OneHour])
        .add();
    let schema = builder.build().unwrap();
    let db = TidalDb::builder()
        .ephemeral()
        .with_schema(schema)
        .open()
        .unwrap();

    // Ephemeral mode: no WAL on disk, always returns empty.
    let req = ExportRequest::time_range(0, u64::MAX);
    let signals = db.export_signals(&req).unwrap();
    assert!(signals.is_empty());
}

#[test]
#[cfg(feature = "test-utils")]
fn export_signals_unknown_type_returns_error() {
    use tidaldb::{TidalDb, schema::SchemaBuilder};

    let mut builder = SchemaBuilder::new();
    let _ = builder
        .signal(
            "view",
            EntityKind::Item,
            DecaySpec::Exponential {
                half_life: Duration::from_secs(7 * 24 * 3600),
            },
        )
        .windows(&[Window::OneHour])
        .add();
    let schema = builder.build().unwrap();
    let db = TidalDb::builder()
        .ephemeral()
        .with_schema(schema)
        .open()
        .unwrap();

    let req = ExportRequest::signals_in_range(vec!["nonexistent".into()], 0, u64::MAX);
    let result = db.export_signals(&req);
    assert!(
        result.is_err(),
        "requesting an unknown signal type must return an error"
    );
}

#[test]
fn export_format_json_lines_value() {
    assert_eq!(ExportFormat::JsonLines, ExportFormat::JsonLines);
}

#[test]
fn export_request_time_range_fields() {
    let req = ExportRequest::time_range(1000, 2000);
    assert_eq!(req.since, Some(1000));
    assert_eq!(req.until, Some(2000));
    assert!(req.signal_types.is_empty());
    assert_eq!(req.format, ExportFormat::JsonLines);
    assert!(req.limit.is_none());
    assert!(req.user_id.is_none());
}

#[test]
fn exported_signal_to_json_line_valid() {
    let sig = ExportedSignal {
        entity_id: 42,
        signal_type: "view".to_string(),
        weight: 1.0,
        timestamp_ns: 1_700_000_000_000_000_000,
        user_id: None,
        session_id: None,
        annotation: None,
    };
    let line = sig.to_json_line();
    assert!(line.starts_with('{'));
    assert!(line.ends_with('}'));
    assert!(line.contains("\"entity_id\":42"));
    assert!(line.contains("\"signal_type\":\"view\""));
    // Anonymous signals must not include optional keys.
    assert!(!line.contains("user_id"));
    assert!(!line.contains("session_id"));
    assert!(!line.contains("annotation"));
}

/// `export_signals` reads actual WAL events written to a persistent database.
///
/// Export reads live, uncompacted WAL segments — those fsynced to disk but
/// not yet removed by a clean shutdown. This test exercises the WAL scanning
/// code path (persistent mode) that the ephemeral early-return skips.
///
/// Note: a clean `close()` compacts WAL segments, so export is called on the
/// same open DB instance — not after reopening.
#[test]
#[cfg(feature = "test-utils")]
fn export_signals_reads_wal_events() {
    use tidaldb::TempTidalHome;

    let home = TempTidalHome::new().unwrap();
    let schema = build_test_schema();

    let db = TidalDb::builder()
        .with_data_dir(home.path())
        .with_schema(schema)
        .open()
        .unwrap();

    let since = Timestamp::now().as_nanos();
    db.signal("view", EntityId::new(1), 1.0, Timestamp::now())
        .unwrap();
    db.signal("view", EntityId::new(2), 1.5, Timestamp::now())
        .unwrap();
    db.signal("like", EntityId::new(3), 2.0, Timestamp::now())
        .unwrap();

    // Export BEFORE close — WAL segments are live on disk.
    // Clean close compacts (deletes) segments, so export must happen first.
    let req = ExportRequest::time_range(since, u64::MAX);
    let signals = db.export_signals(&req).unwrap();

    assert_eq!(signals.len(), 3, "all 3 WAL events must be exported");

    let view_count = signals.iter().filter(|s| s.signal_type == "view").count();
    assert_eq!(view_count, 2, "must export 2 view events");

    let like_count = signals.iter().filter(|s| s.signal_type == "like").count();
    assert_eq!(like_count, 1, "must export 1 like event");

    // Entity IDs and weights are preserved end-to-end.
    assert!(
        signals
            .iter()
            .any(|s| s.entity_id == 1 && (s.weight - 1.0).abs() < 0.001),
        "entity 1 view event must be present"
    );
    assert!(
        signals
            .iter()
            .any(|s| s.entity_id == 3 && s.signal_type == "like"),
        "entity 3 like event must be present"
    );

    // Anonymous batch WAL signals must have None for user/session/annotation.
    for sig in &signals {
        assert!(sig.user_id.is_none(), "batch WAL signals have no user_id");
        assert!(
            sig.session_id.is_none(),
            "batch WAL signals have no session_id"
        );
        assert!(
            sig.annotation.is_none(),
            "batch WAL signals have no annotation"
        );
    }

    db.close().unwrap();
}

/// `export_signals` respects the `limit` field.
///
/// Export is called on the same open DB instance (before close) because
/// clean shutdown compacts WAL segments, leaving nothing to scan.
#[test]
#[cfg(feature = "test-utils")]
fn export_signals_respects_limit() {
    use tidaldb::TempTidalHome;

    let home = TempTidalHome::new().unwrap();
    let schema = build_test_schema();

    let db = TidalDb::builder()
        .with_data_dir(home.path())
        .with_schema(schema)
        .open()
        .unwrap();

    for i in 1u64..=10 {
        db.signal("view", EntityId::new(i), 1.0, Timestamp::now())
            .unwrap();
    }

    let mut req = ExportRequest::time_range(0, u64::MAX);
    req.limit = Some(3);
    let signals = db.export_signals(&req).unwrap();
    assert_eq!(signals.len(), 3, "limit must cap the result count");

    db.close().unwrap();
}

/// Anonymous batch WAL signals are excluded when a `user_id` filter is set.
/// In ephemeral mode there is no WAL or session journal on disk, so the
/// result is empty regardless. This test verifies that anonymous signals
/// written via `db.signal()` (no session) do NOT appear in user-filtered
/// exports.
#[test]
fn export_signals_user_id_filter_excludes_anonymous() {
    let schema = build_test_schema();
    let db = TidalDb::builder()
        .ephemeral()
        .with_schema(schema)
        .open()
        .unwrap();

    db.signal("view", EntityId::new(1), 1.0, Timestamp::now())
        .unwrap();

    let mut req = ExportRequest::time_range(0, u64::MAX);
    req.user_id = Some(42);
    let signals = db.export_signals(&req).unwrap();
    assert!(
        signals.is_empty(),
        "user_id filter must exclude anonymous batch WAL signals"
    );
}

/// Session signals carry `user_id`, `session_id`, and annotation through to export.
///
/// This test exercises the full session-journal export path:
/// 1. Persistent DB with schema and session policy
/// 2. Session signals written with annotations via `session_signal()`
/// 3. Export with `user_id` filter returns only that user's session signals
/// 4. Export with a different `user_id` returns empty
#[test]
#[cfg(feature = "test-utils")]
fn export_signals_with_session_context() {
    use tidaldb::TempTidalHome;

    let home = TempTidalHome::new().unwrap();
    let schema = build_test_schema();

    let db = TidalDb::builder()
        .with_data_dir(home.path())
        .with_schema(schema)
        .open()
        .unwrap();

    let user_id = 42u64;
    let meta = HashMap::new();

    // Start a session for user 42.
    let handle = db
        .start_session(user_id, "test-agent", "default", meta)
        .unwrap();

    let since = Timestamp::now().as_nanos();

    // Write session signals — one with annotation, one without.
    db.session_signal(
        &handle,
        "view",
        EntityId::new(100),
        1.0,
        Timestamp::now(),
        Some("good content".to_string()),
    )
    .unwrap();
    db.session_signal(
        &handle,
        "like",
        EntityId::new(101),
        2.0,
        Timestamp::now(),
        None,
    )
    .unwrap();

    // Close the session so it is fully flushed to the journal.
    db.close_session(handle).unwrap();

    // Session journal writes are fire-and-forget through the WAL writer thread.
    // Allow time for the writer to drain and fsync the session commands.
    std::thread::sleep(Duration::from_millis(100));

    // Export with user_id filter = Some(42).
    let mut req = ExportRequest::time_range(since, u64::MAX);
    req.user_id = Some(user_id);
    let signals = db.export_signals(&req).unwrap();

    assert!(
        !signals.is_empty(),
        "export with user_id filter must return session signals"
    );
    assert_eq!(signals.len(), 2, "expected 2 session signals for user 42");

    // All results must have user_id == Some(42).
    for sig in &signals {
        assert_eq!(
            sig.user_id,
            Some(user_id),
            "all exported signals must carry the user_id"
        );
        assert!(
            sig.session_id.is_some(),
            "all exported signals must carry a session_id"
        );
    }

    // At least one result has the annotation set.
    assert!(
        signals.iter().any(|s| s.annotation.is_some()),
        "at least one signal must have an annotation"
    );
    let annotated = signals.iter().find(|s| s.annotation.is_some()).unwrap();
    assert_eq!(
        annotated.annotation.as_deref(),
        Some("good content"),
        "annotation text must round-trip through the journal"
    );

    // Export with a different user_id returns empty.
    let mut req2 = ExportRequest::time_range(since, u64::MAX);
    req2.user_id = Some(999);
    let signals2 = db.export_signals(&req2).unwrap();
    assert!(
        signals2.is_empty(),
        "export for non-existent user must return empty"
    );

    db.close().unwrap();
}