SniperGold_ML/engine/verify/vinvariants.py

"""Independent invariant checks (spec 18.3).

All checks are implemented against the specification text; they never call
producer validation functions. Each returns (ok, details).
"""

from ..util import sha256_bytes


def check_monotonic_stream(records):
    """ts_ms must be non-decreasing across the merged stream."""
    prev = None
    bad = 0
    first_bad = None
    for rec in records:
        ts = rec[0]
        if prev is not None and ts < prev:
            bad += 1
            if first_bad is None:
                first_bad = (prev, ts)
        prev = ts
    return bad == 0, {"regressions": bad, "first": first_bad}


def check_ohlc_validity(bar_rows):
    """low <= open,close <= high; counts >= 1; avg = sum // count."""
    bad = 0
    for row in bar_rows:
        (bar_idx, _pid, _st, _en, o, h, l, c, n, _v, _smn, _smx, ss,
         avg, _final, _f, _l) = row
        if not (l <= o <= h and l <= c <= h and n >= 1):
            bad += 1
        if avg != (ss // n):
            bad += 1
    return bad == 0, {"invalid": bad}


def check_bar_uniqueness(bar_rows):
    """(TF period) uniqueness is implied per stream; bar_idx must be strictly
    increasing; period_id strictly increasing within one stream."""
    seen = {}
    bad = 0
    for row in bar_rows:
        bar_idx, period_id = row[0], row[1]
        if seen.get(period_id):
            bad += 1
        seen[period_id] = True
    return bad == 0, {"duplicates": bad}


def check_chunk_counts_sum(per_chunk_counts, total):
    return sum(per_chunk_counts) == total, {"sum": sum(per_chunk_counts),
                                            "total": total}


def check_carry_continuity(carry, first_tick):
    """Resume continuity: first ts must be >= carried last_ts; carried
    period must not be re-emitted."""
    if carry is None:
        return True, {}
    if first_tick[0] < carry["last_ts_ms"]:
        return False, {"first_ts": first_tick[0], "carried_last": carry["last_ts_ms"]}
    return True, {}


def independent_serialize_ticks(records):
    """Independent canonical tick serialization (spec 8.5).

    ``records`` is an iterable of (ts_ms, bid_u, ask_u, vol) tuples in
    rec_ord order."""
    from .vaggregate import v_tick_serialize
    return b"".join(v_tick_serialize(r[0], r[1], r[2], r[3]) for r in records)


def independent_ticks_content_id(records):
    return sha256_bytes(independent_serialize_ticks(records))


def independent_bars_content_id(rows):
    from .vaggregate import v_bar_serialize
    return sha256_bytes(b"".join(v_bar_serialize(r) for r in rows))


def independent_preservation_content_id(pairs):
    """Independent source-preservation content id (spec 8.6).

    ``pairs`` is an iterable of (src_line, last_u) integers in rec_ord
    order, extracted independently from the sidecar; the verifier re-derives
    the identity from its own row extraction with the fixed canonical
    serialization ``src_line|last_u\n`` (no floats)."""
    parts = []
    for src_line, last_u in pairs:
        parts.append(("%d|%d" % (src_line, last_u)).encode("ascii") + b"\n")
    return sha256_bytes(b"".join(parts))