from __future__ import annotations

import json
import random
from collections import defaultdict
from dataclasses import dataclass
from pathlib import Path


@dataclass(frozen=True)
class PairRecord:
    path_s1: str
    path_s2: str
    identity_id: str
    finger_id: str
    sensor_s1: str
    sensor_s2: str
    dataset: str


def build_cross_sensor_pairs(
    samples: list[dict[str, str]],
    dataset: str,
    min_sensors: int = 2,
    max_pairs_per_finger: int = 5,
    seed: int = 42,
) -> list[PairRecord]:
    """Build cross-sensor pairs for the same identity/finger tuple."""

    rng = random.Random(seed)
    grouped: dict[tuple[str, str], list[dict[str, str]]] = defaultdict(list)

    for sample in samples:
        if sample.get("dataset") != dataset:
            continue
        key = (sample["identity_id"], sample["finger_id"])
        grouped[key].append(sample)

    all_pairs: list[PairRecord] = []
    for (identity_id, finger_id), items in grouped.items():
        by_sensor: dict[str, list[dict[str, str]]] = defaultdict(list)
        for item in items:
            by_sensor[item["sensor_id"]].append(item)

        if len(by_sensor) < min_sensors:
            continue

        sensor_names = sorted(by_sensor.keys())
        candidate_pairs: list[PairRecord] = []
        for i, sensor_a in enumerate(sensor_names):
            for sensor_b in sensor_names[i + 1 :]:
                for sa in by_sensor[sensor_a]:
                    for sb in by_sensor[sensor_b]:
                        candidate_pairs.append(
                            PairRecord(
                                path_s1=sa["image_path"],
                                path_s2=sb["image_path"],
                                identity_id=identity_id,
                                finger_id=finger_id,
                                sensor_s1=sensor_a,
                                sensor_s2=sensor_b,
                                dataset=dataset,
                            )
                        )

        if len(candidate_pairs) > max_pairs_per_finger:
            candidate_pairs = rng.sample(candidate_pairs, max_pairs_per_finger)
        all_pairs.extend(candidate_pairs)

    return all_pairs


def export_pairs_json(pairs: list[PairRecord], output_path: str) -> None:
    """Write pair records to disk as JSON list."""

    path = Path(output_path)
    path.parent.mkdir(parents=True, exist_ok=True)
    with path.open("w", encoding="utf-8") as f:
        json.dump([p.__dict__ for p in pairs], f, indent=2)