"""Prevent aggregation of observably incompatible measurement populations.

Legacy rows with uniformly absent metadata remain usable for diagnostics.

Missing and present metadata are distinct: absence never establishes equality

with a known environment. This check is not a public-provenance attestation.

"""

from __future__ import annotations

import json

from typing import Any

DIMENSIONS = (

    "lane", "benchmark_lane", "measurement_lane", "profile", "benchmark_track", "command_semantics_version",

    "fixture_version", "scenario_version", "oracle_version", "runner_class", "runner_id", "cache_state",

    "remote_transport", "remote_server", "netshape_profile", "env_isolation_version",

    "calibration_version", "platform_semantics_version", "binary_sha256", "load_tier",

    "instruction_level", "agent.name", "agent.model", "agent.adapter",

    "agent.provider", "agent.version", "agent.temperature", "agent.max_output_tokens",

    "agent.environment", "agent.instruction_level", "fixture.version", "fixture.fixture_version",

    "oracle.version", "task_prompt_id",

)

def population_conflicts(rows: list[dict[str, Any]]) -> list[str]:

    """Return differing identity fields within ONE subject/scenario/op cell."""

    fields = {name: set() for name in DIMENSIONS}

    fields.update({"subject_details.version": set(), "subject_details.binary_sha256": set()})

    for row in rows:

        for name in DIMENSIONS:

            value: Any = row

            for part in name.split("."):

                value = value.get(part) if isinstance(value, dict) else None

            fields[name].add(json.dumps(value, sort_keys=True))

        details = row.get("subject_details") or {}

        if isinstance(details, dict):

            details = details.get(str(row.get("subject")), details)

        for name in ("version", "binary_sha256"):

            value = details.get(name) if isinstance(details, dict) else None

            fields[f"subject_details.{name}"].add(json.dumps(value, sort_keys=True))

    return sorted(name for name, values in fields.items() if len(values) > 1)