"""
change VOL格式(images,xmls:未归一化的左上角右下角) 
to YOLO格式:images and txts(per txt:some lines(class,归一化的cx、cy、w、h))
"""
from xml.dom.minidom import parse
import glob


# 统计有哪些类别
if 0:
    classes = []
    for xml_file in glob.glob(r"D:\dataset\NEU-DET\train\ANNOTATIONS\*.xml"):
        DOMTree = parse(xml_file)
        objects = DOMTree.documentElement.getElementsByTagName("object")
        for item in objects:
            name = item.getElementsByTagName("name")[0]
            name = name.childNodes[0].data
            if name not in classes:
                classes.append(name)
    print(classes)      # ['crazing', 'patches', 'inclusion', 'pitted_surface', 'rolled-in_scale', 'scratches']

if 1:
    classes = ['crazing', 'patches', 'inclusion', 'pitted_surface', 'rolled-in_scale', 'scratches']
    i = 0
    for xml_file in glob.glob(r"D:\dataset\NEU-DET\train\ANNOTATIONS\*.xml"):
        i += 1
        DOMTree = parse(xml_file)

        # 拿到这张图片的wh
        size = DOMTree.documentElement.getElementsByTagName("size")[0]
        img_w = size.getElementsByTagName("width")[0]
        img_w = img_w.childNodes[0].data
        img_w = float(img_w)
        img_h = size.getElementsByTagName("height")[0]
        img_h = img_h.childNodes[0].data
        img_h = float(img_h)

        objects = DOMTree.documentElement.getElementsByTagName("object")
        msg = ""
        for item in objects:
            name = item.getElementsByTagName("name")[0]
            name = name.childNodes[0].data
            # 根据name获取class
            name = classes.index(str(name))
            msg = msg + str(name) + " "

            # 转换bbox
            bbox = item.getElementsByTagName("bndbox")[0]
            xmin = bbox.getElementsByTagName("xmin")[0]
            xmin = xmin.childNodes[0].data

            ymin = bbox.getElementsByTagName("ymin")[0]
            ymin = ymin.childNodes[0].data

            xmax = bbox.getElementsByTagName("xmax")[0]
            xmax = xmax.childNodes[0].data

            ymax = bbox.getElementsByTagName("ymax")[0]
            ymax = ymax.childNodes[0].data

            w = int(xmax) - int(xmin)
            h = int(ymax) - int(ymin)
            x_center = int(xmin) + int(w) / 2.0
            y_center = int(ymin) + int(h) / 2.0

            # 坐标归一化
            w /= img_w
            h /= img_h
            x_center /= img_w
            y_center /= img_h

            msg = msg + str(x_center) + " "
            msg = msg + str(y_center) + " "
            msg = msg + str(w) + " "
            msg = msg + str(h) + "\n"

        # save
        txt_name = xml_file.split("\\")[-1].replace(".xml", ".txt")
        with open("D:\\dataset\\NEU-DET\\train\\labels\\" + txt_name, "w") as f:
            f.write(msg)
        f.close()
    print(i)