# 这是一个在树莓派上运行人脸识别的案例
# 本案例会在命令行控制面板上输出识别出的人脸数量和身份结果。

# 你需要一个2代以上的树莓派，并在树莓派上安装face_recognition，并连接上picamera摄像头
# 并确保picamera这个模块已经安装（树莓派一般会内置安装）
# 你可以参考这个教程配制你的树莓派：
# https://gist.github.com/ageitgey/1ac8dbe8572f3f533df6269dab35df65

import time
import face_recognition
import picamera
import numpy as np
import cv2
from img_base_api import *

out_video_name = 'video'

# 你需要在sudo raspi-config中把camera功能打开
camera = picamera.PiCamera()
camera.resolution = (320, 240)  # 设置图像的width和height
camera.framerate = 35
# camera.saturation = 80  # 设置图像视频的饱和度
camera.brightness = 45  # 设置图像的亮度(50表示白平衡的状态)
camera.shutter_speed = 6000000  # 相机快门速度
# camera.iso = 800        # ISO标准实际上就是来自胶片工业的标准称谓，ISO是衡量胶片对光线敏感程度的标准。如50 ISO, 64 ISO, 100 ISO表示在曝光感应速度上要比高数值的来得慢，高数值ISO是指超过200以上的标准，如200 ISO, 400 ISO
# camera.sharpness = 0    #设置图像的锐度值，默认是0，取值范围是-100~100之间
# camera.hflip = True   # 是否进行水平翻转50
camera.vflip = True  # 是否进行垂直翻转
# camera.rotation = 0   # 是否对图像进行旋转
# camera.led = True      # 值为False那么led为关灯的状态，True为开灯的状态
output = np.empty((240, 320, 3), dtype=np.uint8)

# 载入样本图片（奥巴马和拜登）
print("Loading known face image(s)")
obama_image = face_recognition.load_image_file("obama_small.jpg")
obama_face_encoding = face_recognition.face_encodings(obama_image)[0]

# 初始化变量
face_locations = []
face_encodings = []
face_names = []
process_this_frame = True

# first face flag
first_face_flag = False
no_face_flag = False
# face blur value
face_blur_val = 0
# before face encode
before_face_encoding = None
last_face_exist_time = 0
same_face_interval_time = 5

blur_var_init = 50
# video property
height = 0
width = 0

while True:
    print("Capturing image.")
    # 以numpy array的数据结构从picamera摄像头中获取一帧图片
    camera.capture(output, format="bgr")
    height, width, _ = output.shape
    # 获得所有人脸的位置以及它们的编码
    # face_locations = face_recognition.face_locations(output)
    # print("Found {} faces in image.".format(len(face_locations)))
    # face_encodings = face_recognition.face_encodings(output, face_locations)
    if process_this_frame:
        # Find all the faces and face encodings in the current frame of video
        face_locations = face_recognition.face_locations(output)
        face_encodings = face_recognition.face_encodings(output, face_locations)
    no_face_flag = True if len(face_locations) == 0 else False
    process_this_frame = not process_this_frame

    for index, ((top, right, bottom, left), face_encoding) in enumerate(zip(face_locations, face_encodings)):
        # Scale back up face locations since the frame we detected in was scaled to 1/4 size
        left -= int(width * 0.1)
        top -= int(height * 0.168)
        right += int(width * 0.1)
        bottom += int(height * 0.1)
        if left < 0 or top < 0 or right < 0 or right > width or bottom < 0 or bottom > height:
            continue
        # blur var calc
        gray = cv2.cvtColor(output, cv2.COLOR_BGR2GRAY)
        blur_var = cv2.Laplacian(gray, cv2.CV_64F).var()
        if blur_var < blur_var_init:
            print(f'face blur val: {blur_var}')
            continue
        print(left, top, right, bottom)
        top *= 4
        right *= 4
        bottom *= 4
        left *= 4

        if before_face_encoding is None:
            matches = [False]
        else:
            matches = face_recognition.compare_faces([before_face_encoding], face_encoding)
        if all(matches):
            cv2.rectangle(output, (left, top), (right, bottom), color=[0, 0, 255], thickness=2)
            continue

        # -------------------  face compare && face quality evaluate end  -----------------------------
        print(f'top: {top} bottom: {bottom} left: {left} right: {right} ')
        image = cv2.rectangle(output, (left, top), (right, bottom), (0, 0, 255), 2)
        try:
            cv2.imwrite('.face_img.jpg', image)
            img_base64_str = cv2_base64(image)
            stat, res = send_img_to_server(img_base64_str)
            print("send img to server")
            # if not stat:
            #     continue
            # search_img_to_server(img_base64_str)
            last_face_exist_time = time.time()
            first_face_flag = True
            before_face_encoding = face_encoding

            # break
        except:
            pass

    no_face_time = time.time()
    if no_face_flag and last_face_exist_time != 0 and no_face_time - last_face_exist_time > same_face_interval_time:
        print('no face')
        before_face_encoding = None
        first_face_flag = False
    # Display the resulting image
    cv2.imshow(out_video_name, output)

    # Hit 'q' on the keyboard to quit!
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# Release handle to the webcam
camera.release()
cv2.destroyAllWindows()