"""二维码检测与裁剪工具。

识别引擎优先用 zxing-cpp（对反色、旋转、小尺寸、艺术化二维码的容错明显更好），
拿不到时退回 OpenCV 自带检测器，保证不装 zxing-cpp 也能跑。
"""

from __future__ import annotations

from dataclasses import dataclass
from pathlib import Path

import cv2
import numpy as np

from app.core.logger import log

try:  # zxing-cpp 是可选依赖，缺失时自动退化为纯 OpenCV
    import zxingcpp

    _HAS_ZXING = True
except ImportError:  # pragma: no cover - 取决于部署环境
    zxingcpp = None
    _HAS_ZXING = False
    log.warning("未安装 zxing-cpp，二维码识别将退化为 OpenCV 检测器，识别率会明显下降")


# 要识别的二维码类型：标准 QR + 微型 QR + 矩形 QR
_QR_FORMATS = (
    [
        zxingcpp.BarcodeFormat.QRCode,
        zxingcpp.BarcodeFormat.MicroQRCode,
        zxingcpp.BarcodeFormat.RMQRCode,
    ]
    if _HAS_ZXING
    else []
)

# 放大重试的尺寸上限，避免长图被放大到内存爆掉
_MAX_UPSCALE_SIDE = 2600
# 裁剪区域重试时补的静默区宽度（像素）
_QUIET_ZONE = 16


@dataclass
class QrRegion:
    """单个二维码区域。"""

    points: tuple[tuple[int, int], ...]
    crop: np.ndarray | None = None
    text: str = ""


@dataclass
class QrDetectResult:
    """二维码检测结果。"""

    has_qr: bool
    items: list[QrRegion]


def _load_image(image: bytes | np.ndarray | str | Path) -> np.ndarray:
    """把输入转成 BGR 图像。"""
    if isinstance(image, np.ndarray):
        if image.ndim == 2:
            return cv2.cvtColor(image, cv2.COLOR_GRAY2BGR)
        return image.copy()

    if isinstance(image, (str, Path)):
        data = Path(image).read_bytes()
    else:
        data = image

    arr = np.frombuffer(data, dtype=np.uint8)
    img = cv2.imdecode(arr, cv2.IMREAD_COLOR)
    if img is None:
        raise ValueError("Failed to decode image data.")
    return img


def _order_points(points: np.ndarray) -> np.ndarray:
    """把四个角点整理成左上、右上、右下、左下。"""
    pts = np.asarray(points, dtype=np.float32).reshape(4, 2)
    rect = np.zeros((4, 2), dtype=np.float32)

    s = pts.sum(axis=1)
    diff = np.diff(pts, axis=1)

    rect[0] = pts[np.argmin(s)]
    rect[2] = pts[np.argmax(s)]
    rect[1] = pts[np.argmin(diff)]
    rect[3] = pts[np.argmax(diff)]
    return rect


def _warp_qr_image(img: np.ndarray, points: np.ndarray) -> np.ndarray:
    """按四边形点做透视矫正，截取二维码区域。"""
    rect = _order_points(points)
    (tl, tr, br, bl) = rect

    width_a = np.linalg.norm(br - bl)
    width_b = np.linalg.norm(tr - tl)
    height_a = np.linalg.norm(tr - br)
    height_b = np.linalg.norm(tl - bl)

    width = max(int(round(max(width_a, width_b))), 1)
    height = max(int(round(max(height_a, height_b))), 1)

    dst = np.array(
        [
            [0, 0],
            [width - 1, 0],
            [width - 1, height - 1],
            [0, height - 1],
        ],
        dtype=np.float32,
    )

    matrix = cv2.getPerspectiveTransform(rect, dst)
    return cv2.warpPerspective(img, matrix, (width, height))


def _normalize_points(points: np.ndarray | None) -> list[tuple[tuple[int, int], ...]]:
    if points is None:
        return []

    arr = np.asarray(points)
    if arr.ndim == 2 and arr.shape == (4, 2):
        arr = arr[None, ...]
    elif arr.ndim == 3 and arr.shape[-2:] == (4, 2):
        pass
    else:
        return []

    result: list[tuple[tuple[int, int], ...]] = []
    for item in arr:
        quad = tuple((int(round(x)), int(round(y))) for x, y in item)
        result.append(quad)
    return result


def _clip_points(
    quad: tuple[tuple[int, int], ...], shape: tuple[int, ...]
) -> tuple[tuple[int, int], ...]:
    """把角点裁进图像范围内，避免透视变换取到界外。"""
    height, width = shape[:2]
    return tuple(
        (min(max(x, 0), width - 1), min(max(y, 0), height - 1)) for x, y in quad
    )


def zxing_read(
    img: np.ndarray,
    *,
    binarizer: object | None = None,
    try_downscale: bool = True,
) -> list:
    """用 zxing-cpp 识别图中所有二维码，失败返回空列表。

    zxing-cpp 默认已开启 try_invert（反色）和 try_rotate（旋转），
    这是它比 OpenCV 检测器兼容性好的主要原因。
    """
    if not _HAS_ZXING:
        return []

    kwargs = {
        "formats": _QR_FORMATS,
        "try_rotate": True,
        "try_invert": True,
        "try_downscale": try_downscale,
    }
    if binarizer is not None:
        kwargs["binarizer"] = binarizer

    try:
        return list(zxingcpp.read_barcodes(img, **kwargs))
    except Exception as exc:  # zxing 内部异常不应该打断整个流程
        log.debug(f"zxing-cpp 识别异常: {exc}")
        return []


def zxing_variants(img: np.ndarray) -> list[tuple[str, np.ndarray]]:
    """构造 zxing 的重试图像变体：原图之外再补一版放大图。

    放大主要救「长图里的小二维码」和「低分辨率二维码」两类。
    """
    variants: list[tuple[str, np.ndarray]] = [("原图", img)]

    longest = max(img.shape[:2])
    if longest and longest * 2 <= _MAX_UPSCALE_SIDE:
        variants.append(
            ("放大2倍", cv2.resize(img, None, fx=2, fy=2, interpolation=cv2.INTER_CUBIC))
        )
    return variants


def pad_quiet_zone(img: np.ndarray, border: int = _QUIET_ZONE) -> np.ndarray:
    """给裁剪出来的二维码补静默区。

    有些海报把二维码贴边放，裁出来没有留白，补一圈能提高解码成功率。
    边框颜色取图像四角的中位数，反色码补深色、正常码补浅色，避免破坏极性。
    """
    corners = np.array(
        [img[0, 0], img[0, -1], img[-1, 0], img[-1, -1]], dtype=np.float32
    )
    color = np.median(corners, axis=0)
    value = tuple(int(round(c)) for c in np.atleast_1d(color))
    if len(value) == 1:
        value = value * 3
    return cv2.copyMakeBorder(
        img, border, border, border, border, cv2.BORDER_CONSTANT, value=value
    )


def _points_from_zxing(barcode: object) -> tuple[tuple[int, int], ...] | None:
    """把 zxing 的 position 转成四角点。"""
    position = getattr(barcode, "position", None)
    if position is None:
        return None

    quad: list[tuple[int, int]] = []
    for name in ("top_left", "top_right", "bottom_right", "bottom_left"):
        point = getattr(position, name, None)
        if point is None:
            return None
        quad.append((int(round(point.x)), int(round(point.y))))
    return tuple(quad)


def _detect_by_opencv(img: np.ndarray) -> list[tuple[tuple[int, int], ...]]:
    """OpenCV 检测器兜底，额外补一次反色重试。"""
    for candidate in (img, cv2.bitwise_not(img)):
        for detector in (cv2.QRCodeDetector(), cv2.QRCodeDetectorAruco()):
            try:
                if hasattr(detector, "detectMulti"):
                    ok, points = detector.detectMulti(candidate)
                    if ok:
                        normalized = _normalize_points(points)
                        if normalized:
                            return normalized

                ok, points = detector.detect(candidate)
                if ok:
                    normalized = _normalize_points(points)
                    if normalized:
                        return normalized
            except cv2.error:
                continue
    return []


def _detect_regions(img: np.ndarray) -> list[QrRegion]:
    """检测二维码区域：zxing 优先（顺带拿到内容），OpenCV 兜底。"""
    for name, variant in zxing_variants(img):
        barcodes = zxing_read(variant)
        if not barcodes:
            continue

        scale = variant.shape[1] / img.shape[1] if img.shape[1] else 1
        regions: list[QrRegion] = []
        for barcode in barcodes:
            quad = _points_from_zxing(barcode)
            if quad is None:
                continue
            if scale != 1:
                quad = tuple(
                    (int(round(x / scale)), int(round(y / scale))) for x, y in quad
                )
            quad = _clip_points(quad, img.shape)
            regions.append(
                QrRegion(
                    points=quad,
                    crop=_warp_qr_image(img, np.array(quad, dtype=np.float32)),
                    text=getattr(barcode, "text", "") or "",
                )
            )
        if regions:
            if name != "原图":
                log.debug(f"二维码检测命中变体: {name}")
            return regions

    return [
        QrRegion(
            points=quad,
            crop=_warp_qr_image(img, np.array(quad, dtype=np.float32)),
        )
        for quad in _detect_by_opencv(img)
    ]


def scan_qr(image: bytes | np.ndarray | str | Path) -> QrDetectResult:
    """扫描图片中是否存在二维码，并返回二维码区域。"""
    img = _load_image(image)
    items = _detect_regions(img)
    return QrDetectResult(has_qr=bool(items), items=items)


def has_qr(image: bytes | np.ndarray | str | Path) -> bool:
    """判断图片里有没有二维码。"""
    return scan_qr(image).has_qr


def crop_qr(image: bytes | np.ndarray | str | Path) -> list[np.ndarray]:
    """裁剪出图片中的二维码区域。"""
    return [item.crop for item in scan_qr(image).items if item.crop is not None]