import cv2
from PyQt5.QtCore import pyqtSignal, QThread
from PyQt5.QtGui import *
from PyQt5.QtWidgets import *

# 子窗口布局
from sub_windows import ui_sub_window_8


class SubWindow(QMainWindow):
    def __init__(self):
        super().__init__(parent=None)
        self.ui = ui_sub_window_8.Ui_Form()
        self.ui.setupUi(self)
        self.FaceDetect_ = None
        self.ui_init()
        self.face_cascade = cv2.CascadeClassifier("./static/cascade.xml")

    def ui_init(self):
        self.ui.pushButton_video_captrue.clicked.connect(self.video_captrue)
        self.ui.pushButton_open_file.clicked.connect(self.open_file)
        self.FaceDetect_ = FaceDetect()
        self.FaceDetect_.DetectOneFrame.connect(self.update_frame_to_label)
        self.FaceDetect_.FaceNum.connect(self.update_face_num)

    def update_face_num(self, num):
        self.ui.label.setText(str(num))

    def open_file(self):
        file_path, file_type = QFileDialog.getOpenFileName(
            QFileDialog(), "选择图片", "", "图像文件(*.jpg *.bmp *.png)"
        )
        self.cv_srcImage = cv2.imread(file_path)
        if self.cv_srcImage is not None:
            gray = cv2.cvtColor(self.cv_srcImage, cv2.COLOR_BGR2GRAY)
            faces = self.face_cascade.detectMultiScale(
                gray, scaleFactor=1.15, minNeighbors=5, minSize=(5, 5)
            )
            for x, y, w, h in faces:
                cv2.rectangle(self.cv_srcImage, (x, y), (x + w, y + w), (0, 255, 0), 5)
            height, width = self.cv_srcImage.shape[0], self.cv_srcImage.shape[1]
            ui_image = QImage(
                cv2.cvtColor(self.cv_srcImage, cv2.COLOR_BGR2RGB),
                width,
                height,
                QImage.Format_RGB888,
            )
            if width > height:
                ui_image = ui_image.scaledToWidth(self.ui.label_image_1.width())
            else:
                ui_image = ui_image.scaledToHeight(self.ui.label_image_1.height())
            self.ui.label_image_1.setPixmap(QPixmap.fromImage(ui_image))
            self.ui.label.setText(str(len(faces)))

    def video_captrue(self):
        if not self.FaceDetect_.working:
            self.FaceDetect_.working = True
            self.FaceDetect_.start()
        else:
            self.FaceDetect_.working = None
            self.ui.label_image_1.setText("停止捕捉")

    def update_frame_to_label(self, frame):
        self.ui.label_image_1.setPixmap(QPixmap.fromImage(frame))


class FaceDetect(QThread):
    DetectOneFrame = pyqtSignal(QImage)
    FaceNum = pyqtSignal(int)

    def __init__(self):
        super().__init__()
        self.working = None

    def run(self):
        face_cascade = cv2.CascadeClassifier("./static/cascade.xml")
        capture = cv2.VideoCapture(0, cv2.CAP_DSHOW)
        codec = cv2.VideoWriter_fourcc("M", "J", "P", "G")
        capture.set(cv2.CAP_PROP_FOURCC, codec)
        while self.working:
            ret, frame_color = capture.read()
            (height, width, channels) = frame_color.shape
            frame_color = cv2.flip(frame_color, flipCode=1)  # 镜像
            gray = cv2.cvtColor(frame_color, cv2.COLOR_BGR2GRAY)
            faces = face_cascade.detectMultiScale(
                gray, scaleFactor=1.15, minNeighbors=5, minSize=(5, 5)
            )
            for x, y, w, h in faces:
                cv2.rectangle(frame_color, (x, y), (x + w, y + w), (255, 255, 0), 4)
            ui_image = QImage(
                cv2.cvtColor(frame_color, cv2.COLOR_BGR2RGB),
                width,
                height,
                QImage.Format_RGB888,
            )
            self.DetectOneFrame.emit(ui_image)
            self.FaceNum.emit(len(faces))
        capture.release()
        print("结束人脸检测")