coremlit 0.1.0

use super::*;

/// Flatten spans to `(start, end)` pairs for terse geometry assertions.
fn offsets(plan: &WindowPlan, total: usize) -> Vec<(usize, usize)> {
  plan
    .spans(total)
    .unwrap()
    .iter()
    .map(|s| (s.start(), s.end()))
    .collect()
}

/// The naive soundevents `chunk_slices` geometry: starts at 0, H, 2H, … while
/// `start < total`, each chunk `[start, min(start + W, total))`. The reference
/// CED's multi-tail continuation must match for `total > WINDOW_SAMPLES` under
/// `TailPolicy::Pad` (spec §2: "matching soundevents' chunk_slices overlapped
/// semantics"). For `total <= WINDOW_SAMPLES` the short-clip guard deliberately
/// deviates (exactly one span regardless of hop — spec-settled), so this
/// reference is only consulted for long clips.
fn chunk_slices_reference(total: usize, window: usize, hop: usize) -> Vec<(usize, usize)> {
  let mut out = Vec::new();
  let mut start = 0usize;
  while start < total {
    out.push((start, total.min(start + window)));
    start += hop;
  }
  out
}

#[test]
fn window_samples_is_the_model_geometry() {
  assert_eq!(WINDOW_SAMPLES, 160_000); // 10 s × 16 kHz
  assert_eq!(DEFAULT_HOP_SAMPLES as usize, WINDOW_SAMPLES);
}

#[test]
fn default_plan_is_no_overlap_pad() {
  let plan = WindowPlan::default();
  assert_eq!(plan, WindowPlan::new());
  assert_eq!(plan.hop_samples(), DEFAULT_HOP_SAMPLES);
  assert_eq!(plan.tail_policy(), TailPolicy::Pad);
}

#[test]
fn empty_clip_plans_no_windows() {
  assert!(WindowPlan::new().spans(0).unwrap().is_empty());
}

#[test]
fn short_clip_is_one_window_regardless_of_hop() {
  // Clap contract 1: total <= window ⇒ exactly one span, even under a tiny hop
  // (where a literal soundevents chunk_slices would emit several sub-window
  // chunks — the recorded deviation, spec §2).
  let plan = WindowPlan::new().with_hop_samples(1_000);
  assert_eq!(offsets(&plan, 50_000), vec![(0, 50_000)]);
  assert_eq!(offsets(&plan, WINDOW_SAMPLES), vec![(0, WINDOW_SAMPLES)]);
}

#[test]
fn short_clip_survives_drop_below_min() {
  // A short clip's sole span is never dropped — there is nothing else to
  // represent it.
  let plan =
    WindowPlan::new().with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(100_000)));
  assert_eq!(offsets(&plan, 50_000), vec![(0, 50_000)]);
}

#[test]
fn short_clip_coverage_is_padding_aware() {
  let spans = WindowPlan::new().spans(40_000).unwrap();
  assert_eq!(spans.len(), 1);
  let coverage = spans[0].coverage();
  assert!(
    (coverage - 0.25).abs() < 1e-6,
    "40_000 / 160_000 = 0.25, got {coverage}"
  );
}

#[test]
fn no_overlap_tiling_with_padded_tail() {
  // 400_000 = 2 full windows + an 80_000-sample tail, kept under Pad.
  let plan = WindowPlan::new();
  assert_eq!(
    offsets(&plan, 400_000),
    vec![(0, 160_000), (160_000, 320_000), (320_000, 400_000)]
  );
}

#[test]
fn drop_below_min_drops_the_short_tail() {
  let plan =
    WindowPlan::new().with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(100_000)));
  // The 80_000-sample tail is below the 100_000 threshold.
  assert_eq!(
    offsets(&plan, 400_000),
    vec![(0, 160_000), (160_000, 320_000)]
  );
  // A tail AT the threshold is kept.
  assert_eq!(
    offsets(&plan, 420_000),
    vec![(0, 160_000), (160_000, 320_000), (320_000, 420_000)]
  );
}

#[test]
fn overlapping_hop_produces_full_windows_then_tails() {
  // hop 80_000 over 400_000: full windows at 0/80k/160k/240k, then the
  // multi-tail continuation keeps striding past windit's first-tail stop
  // (clap contract 2), emitting the 80k tail at 320k.
  let plan = WindowPlan::new().with_hop_samples(80_000);
  assert_eq!(
    offsets(&plan, 400_000),
    vec![
      (0, 160_000),
      (80_000, 240_000),
      (160_000, 320_000),
      (240_000, 400_000),
      (320_000, 400_000),
    ]
  );
}

#[test]
fn spans_match_soundevents_chunk_slices_for_long_clips() {
  // The Pad-policy geometry for total > WINDOW_SAMPLES is exactly soundevents'
  // chunk_slices (start, end) sequence — including EVERY progressively shorter
  // tail an overlapped hop generates.
  for (total, hop) in [
    (400_000usize, 80_000u32),
    (500_000, 60_000),
    (160_001, 160_000),
    (1_000_000, 160_000),
    (330_000, 100_000),
  ] {
    let plan = WindowPlan::new().with_hop_samples(hop);
    assert_eq!(
      offsets(&plan, total),
      chunk_slices_reference(total, WINDOW_SAMPLES, hop as usize),
      "total={total} hop={hop}"
    );
  }
}

#[test]
fn window_just_over_boundary_keeps_a_one_sample_tail_under_pad() {
  assert_eq!(
    offsets(&WindowPlan::new(), WINDOW_SAMPLES + 1),
    vec![(0, WINDOW_SAMPLES), (WINDOW_SAMPLES, WINDOW_SAMPLES + 1)]
  );
}

#[test]
fn span_geometry_accessors() {
  let spans = WindowPlan::new().spans(400_000).unwrap();
  let tail = spans[2];
  assert_eq!(tail.start(), 320_000);
  assert_eq!(tail.len(), 80_000);
  assert_eq!(tail.end(), 400_000);
  assert!((tail.coverage() - 0.5).abs() < 1e-6);
}

#[test]
#[should_panic(expected = "hop_samples")]
fn zero_hop_setter_panics() {
  let _ = WindowPlan::new().with_hop_samples(0);
}

#[test]
#[should_panic(expected = "hop_samples")]
fn hop_past_window_setter_panics() {
  // hop > window would leave gaps of un-classified audio (the soundevents
  // sparse-skim mode is a recorded non-goal).
  let _ = WindowPlan::new().with_hop_samples(WINDOW_SAMPLES as u32 + 1);
}

#[test]
#[should_panic(expected = "min_samples")]
fn zero_drop_min_setter_panics() {
  let _ = WindowPlan::new().with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(0)));
}

#[test]
fn hop_one_twenty_second_clip_is_rejected_typed() {
  // The codex [high] regression: a serde-supplied hop of 1 over a 20 s clip
  // would plan 320 000 windows (~643 MiB retained + 320 000 CoreML inferences).
  // The O(1) cap refuses it typed BEFORE materializing anything — this test
  // completing at all (no OOM, no 320 000 pushes) is half the point; the exact
  // `got` pins the FULL-count semantics (not windit's abort-at-`max + 1`).
  let plan = WindowPlan::new().with_hop_samples(1);
  let total = 2 * WINDOW_SAMPLES; // 320_000 = 20 s at 16 kHz
  let err = plan.spans(total).unwrap_err();
  assert!(
    matches!(
      err,
      Error::Windowing(WinditError::TooManyWindows { got: 320_000, max })
        if max == DEFAULT_MAX_WINDOWS as usize
    ),
    "expected TooManyWindows {{ got: 320_000, max: {} }}, got {err:?}",
    DEFAULT_MAX_WINDOWS
  );
}

#[test]
fn cap_boundary_exact_count_passes_and_plus_one_fails() {
  // hop 80_000 over 400_000 is the pinned 5-span geometry
  // (`overlapping_hop_produces_full_windows_then_tails`). A cap of exactly the
  // planned count admits it unchanged; one below refuses with the full count.
  let expected = vec![
    (0, 160_000),
    (80_000, 240_000),
    (160_000, 320_000),
    (240_000, 400_000),
    (320_000, 400_000),
  ];
  let at_cap = WindowPlan::new()
    .with_hop_samples(80_000)
    .with_max_windows(5);
  assert_eq!(offsets(&at_cap, 400_000), expected);

  let under_cap = WindowPlan::new()
    .with_hop_samples(80_000)
    .with_max_windows(4);
  let err = under_cap.spans(400_000).unwrap_err();
  assert!(
    matches!(
      err,
      Error::Windowing(WinditError::TooManyWindows { got: 5, max: 4 })
    ),
    "got {err:?}"
  );
}

#[test]
fn planned_windows_matches_materialized_len() {
  // The O(1) formula MUST equal the real materialized length for every
  // admissible geometry — otherwise the cap check would guard the wrong count.
  // The cap is lifted (`u32::MAX`) so only geometry, never the rail, is tested.
  let pad = |hop: u32| {
    WindowPlan::new()
      .with_hop_samples(hop)
      .with_max_windows(u32::MAX)
  };
  let drop_min = |hop: u32, min: u32| {
    WindowPlan::new()
      .with_hop_samples(hop)
      .with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(min)))
      .with_max_windows(u32::MAX)
  };
  let cases: [(WindowPlan, usize); 20] = [
    // Pad geometry (mirrors the chunk_slices grid) + boundary totals.
    (pad(80_000), 400_000),
    (pad(60_000), 500_000),
    (pad(160_000), 160_001),
    (pad(160_000), 1_000_000),
    (pad(100_000), 330_000),
    (pad(160_000), 0),
    (pad(160_000), 40_000),
    (pad(160_000), WINDOW_SAMPLES - 1),
    (pad(160_000), WINDOW_SAMPLES),
    (pad(160_000), WINDOW_SAMPLES + 1),
    (pad(1), WINDOW_SAMPLES),       // guard-1 immunity: 1 span
    (pad(1), WINDOW_SAMPLES + 100), // hop-1 multi-tail, materialized
    // DropBelowMin geometry (the pinned 400_000/420_000 @ 100_000 cases + edges).
    (drop_min(160_000, 100_000), 400_000),
    (drop_min(160_000, 100_000), 420_000),
    (drop_min(80_000, 100_000), 400_000),
    (drop_min(60_000, 40_000), 500_000),
    (drop_min(160_000, 160_000), 320_001),
    (drop_min(2, 100_000), 170_000), // small-hop multi-tail under a drop policy
    (drop_min(160_000, 100_000), 40_000),
    (drop_min(160_000, 100_000), 0),
  ];
  for (plan, total) in cases {
    assert_eq!(
      plan.planned_windows(total),
      plan.spans(total).unwrap().len(),
      "planned_windows != materialized len for hop={} tail={:?} total={total}",
      plan.hop_samples(),
      plan.tail_policy(),
    );
  }
}

#[test]
fn short_clip_never_trips_cap() {
  // Guard-1 immunity: a short clip is one span regardless of hop/cap, so even
  // the tightest cap admits it (planned == 1 <= any valid cap).
  let plan = WindowPlan::new().with_max_windows(1).with_hop_samples(1);
  let spans = plan.spans(WINDOW_SAMPLES).unwrap();
  assert_eq!(spans.len(), 1);
  assert_eq!((spans[0].start(), spans[0].end()), (0, WINDOW_SAMPLES));
}

#[test]
#[should_panic(expected = "max_windows")]
fn zero_max_windows_setter_panics() {
  let _ = WindowPlan::new().with_max_windows(0);
}

#[cfg(feature = "serde")]
mod serde_tests {
  use super::*;

  #[test]
  fn round_trips_through_json() {
    let plan = WindowPlan::new()
      .with_hop_samples(80_000)
      .with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(40_000)))
      .with_max_windows(50_000);
    let json = serde_json::to_string(&plan).unwrap();
    let back: WindowPlan = serde_json::from_str(&json).unwrap();
    assert_eq!(back, plan);
  }

  #[test]
  fn defaults_fill_for_a_partial_config() {
    let plan: WindowPlan = serde_json::from_str("{}").unwrap();
    assert_eq!(plan, WindowPlan::new());
    // The omitted cap fills the default — it is default-on for every config.
    assert_eq!(plan.max_windows(), DEFAULT_MAX_WINDOWS);
  }

  /// The [`TailPolicy`] document, pinned byte-exactly in BOTH directions, plus
  /// the whole-plan document that carries it.
  ///
  /// The form is windit 0.4's: adjacently tagged, `kind` naming the variant and
  /// `value` carrying the payload. The externally tagged `"pad"` /
  /// `{"drop_below_min":{…}}` spellings this replaced are asserted to be
  /// REFUSED, so the wire break is a fact this suite states rather than a
  /// comment somebody has to trust.
  #[test]
  fn tail_policy_wire_spellings_are_pinned() {
    // Wildcard-free: a new variant fails to compile until its spelling is
    // pinned here (the ChunkAggregation golden pattern, `aggregate/tests.rs`).
    for kind in [
      TailPolicy::Pad,
      TailPolicy::DropBelowMin(DropBelowMin::new(40_000)),
    ] {
      let expected = match kind {
        TailPolicy::Pad => "{\"kind\":\"pad\"}".to_string(),
        TailPolicy::DropBelowMin(d) => {
          let min_samples = d.min_samples();
          format!("{{\"kind\":\"drop_below_min\",\"value\":{{\"min_samples\":{min_samples}}}}}")
        }
      };
      assert_eq!(serde_json::to_string(&kind).unwrap(), expected);
      let back: TailPolicy = serde_json::from_str(&expected).unwrap();
      assert_eq!(back, kind);
    }
    assert_eq!(
      serde_json::to_string(&WindowPlan::new()).unwrap(),
      "{\"hop_samples\":160000,\"tail\":{\"kind\":\"pad\"},\"max_windows\":100000}"
    );
    // A whole non-default plan, serialized and read back from the SAME literal
    // document — the payload variant reached as a struct field, which is how a
    // config file actually carries it.
    let doc = "{\"hop_samples\":80000,\"tail\":{\"kind\":\"drop_below_min\",\
                \"value\":{\"min_samples\":40000}},\"max_windows\":50000}";
    let plan = WindowPlan::new()
      .with_hop_samples(80_000)
      .with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(40_000)))
      .with_max_windows(50_000);
    assert_eq!(serde_json::to_string(&plan).unwrap(), doc);
    assert_eq!(serde_json::from_str::<WindowPlan>(doc).unwrap(), plan);
    // The retired externally tagged form is refused, both variants.
    assert!(serde_json::from_str::<TailPolicy>("\"pad\"").is_err());
    assert!(
      serde_json::from_str::<TailPolicy>("{\"drop_below_min\":{\"min_samples\":40000}}").is_err()
    );
  }

  /// Every [`TailPolicy`] variant and the plans that carry it survive a
  /// NON-self-describing format.
  ///
  /// The falsifier for the `is_human_readable` split. postcard carries no field
  /// names, so the adjacently tagged form cannot be read back at all: serde
  /// writes the tag as a struct FIELD and reads it through
  /// `deserialize_identifier`, which postcard refuses. Measured on the derived
  /// `tail_policy_serde::Document` mirror — the exact shape the branch replaced
  /// — EVERY variant serializes and then fails to deserialize, `Pad` (bytes
  /// `[0]`) and `DropBelowMin` (`[1, …]`) alike; the unit variant additionally
  /// has its missing content read through `deserialize_any`. Drop the
  /// `is_human_readable` branch and all four cases below red — both policies
  /// and both plans, the DEFAULT `WindowPlan` included — while every JSON pin
  /// above stays green.
  #[test]
  fn every_variant_round_trips_through_a_non_self_describing_format() {
    for policy in [
      TailPolicy::Pad,
      TailPolicy::DropBelowMin(DropBelowMin::new(40_000)),
    ] {
      let bytes = postcard::to_allocvec(&policy).unwrap();
      assert_eq!(
        postcard::from_bytes::<TailPolicy>(&bytes).unwrap(),
        policy,
        "postcard round-trip lost {policy:?} (bytes {bytes:?})"
      );
    }
    for plan in [
      WindowPlan::new(),
      WindowPlan::new()
        .with_hop_samples(80_000)
        .with_tail_policy(TailPolicy::DropBelowMin(DropBelowMin::new(40_000)))
        .with_max_windows(50_000),
    ] {
      let bytes = postcard::to_allocvec(&plan).unwrap();
      assert_eq!(
        postcard::from_bytes::<WindowPlan>(&bytes).unwrap(),
        plan,
        "postcard round-trip lost {plan:?} (bytes {bytes:?})"
      );
    }
  }

  #[test]
  fn invalid_hop_fails_to_deserialize() {
    // The validated repr makes the checked setters unbypassable via serde:
    // hop 0 would loop forever, hop > window would skip audio.
    assert!(serde_json::from_str::<WindowPlan>("{\"hop_samples\":0}").is_err());
    assert!(serde_json::from_str::<WindowPlan>("{\"hop_samples\":160001}").is_err());
  }

  #[test]
  fn invalid_tail_min_fails_to_deserialize() {
    let json = "{\"tail\":{\"kind\":\"drop_below_min\",\"value\":{\"min_samples\":0}}}";
    assert!(serde_json::from_str::<WindowPlan>(json).is_err());
    let json = "{\"tail\":{\"kind\":\"drop_below_min\",\"value\":{\"min_samples\":160001}}}";
    assert!(serde_json::from_str::<WindowPlan>(json).is_err());
  }

  /// The rejection MESSAGE, pinned byte-exactly. `TailPolicy::DropBelowMin`
  /// carries a payload struct that shares its variant's name, so interpolating
  /// the PAYLOAD renders `DropBelowMin { min_samples: 0 }` — what the
  /// struct-shaped variant rendered before it was newtyped. Interpolating the
  /// whole policy instead would double the name, and nothing else asserts this.
  #[test]
  fn invalid_tail_min_rejection_message_names_the_payload_once() {
    let err = WindowPlan::try_from(WindowPlanRepr {
      hop_samples: DEFAULT_HOP_SAMPLES,
      tail: TailPolicy::DropBelowMin(DropBelowMin::new(0)),
      max_windows: DEFAULT_MAX_WINDOWS,
    })
    .unwrap_err();
    assert_eq!(
      err,
      "tail DropBelowMin.min_samples must be > 0 and <= WINDOW_SAMPLES \
       (160000), got DropBelowMin { min_samples: 0 }"
    );
  }

  #[test]
  fn zero_max_windows_fails_to_deserialize() {
    // A zero cap can never score any clip; the validated repr rejects it, just
    // as the setter panics on it.
    assert!(serde_json::from_str::<WindowPlan>("{\"max_windows\":0}").is_err());
  }

  /// A MISSPELLED key is REFUSED, not silently discarded.
  ///
  /// Every field defaults, so without `deny_unknown_fields` `{"max_window":1}`
  /// — the plural dropped — deserializes with the typo thrown away and
  /// `max_windows` filled from [`DEFAULT_MAX_WINDOWS`]: an operator capping the
  /// door at ONE window gets 100 000 instead, so a misspelled RESOURCE LIMIT
  /// silently becomes up to 100 000 CoreML predictions. A misspelled `hop` or
  /// `tail` key changes the scored geometry the same silent way. Each rejection
  /// is paired with the spelling that must still parse, so this cannot pass by
  /// the whole struct having stopped deserializing.
  #[test]
  fn a_misspelled_key_is_refused_rather_than_silently_defaulted() {
    for (doc, key) in [
      (r#"{"max_window":1}"#, "max_window"),
      (r#"{"hop":80000}"#, "hop"),
      (r#"{"tail_policy":{"kind":"pad"}}"#, "tail_policy"),
      // Beside a well-spelled key, where a permissive impl is likeliest to let
      // it through.
      (r#"{"hop_samples":80000,"max_window":1}"#, "max_window"),
    ] {
      let err = match serde_json::from_str::<WindowPlan>(doc) {
        Ok(plan) => panic!(
          "{doc} must be refused; it deserialized to {plan:?} (max_windows {})",
          plan.max_windows()
        ),
        Err(e) => e.to_string(),
      };
      assert!(
        err.contains(key),
        "the refusal must name {key}, got {err:?}"
      );
    }
    // Positive control: the correct spellings still parse and land.
    let ok: WindowPlan = serde_json::from_str(r#"{"hop_samples":80000,"max_windows":1}"#).unwrap();
    assert_eq!(ok.hop_samples(), 80_000);
    assert_eq!(ok.max_windows(), 1);
  }
}