from PyPDF2 import PdfFileWriter,PdfFileReader    def test1(file_path,folder_path,num,end_page,start_page=0):      """      :param file_path: pdf文件路径      :param folder_path: 存放路径      :param num: 拆分后的pdf存在几个原pdf页数      :param end_page: 拆分到的最后一页      :param start_page: 起始的页数，默认为0      :return:      """      # 打开PDF文件      pdf_file = PdfFileReader(open(file_path, 'rb'))      # 获取pdf的页数      pdf_file_num = pdf_file.getNumPages()      # 如果输入的end_page页数比pdf文件的页数大或者小于等于0，让停止的页数为pdf最大的页数      if end_page>pdf_file_num or end_page<=0:          end_page=pdf_file_num      # 从起始页到最后一页进行遍历      for i in range(start_page,end_page,num):          #创建一个PdfFileWriter的对象          out_put = PdfFileWriter()          # 给out_put这个对象传num数的页，项目中每个发票都只占了1页，所以num为1，如果发票占据2页，那么num为2          for k in range(num):              out_put.addPage(pdf_file.getPage(i))          # 设置保存的路径          out_file = folder_path + "\" + f"{i}.pdf"          # 把out_put里面的数据写入到文件中          out_put.write(open(out_file, 'wb'))

import fitz    def my_fitz(pdfPath, imagePath):      """      :param pdfPath: pdf的路径      :param imagePath: 图片文件夹的路径，不是图片路径      :return:      """      # 打开pdf文件      pdfDoc = fitz.open(pdfPath)      for pg in range(pdfDoc.pageCount):          page = pdfDoc[pg]          rotate = int(0)          # 每个尺寸的缩放系数为2，生成的图像的分辨率会提高，参数也可以自由设置，没有硬性要求          zoom_x = 2          zoom_y = 2          # 这个函数可以理解为，把zoom_x，zoom_y这两个参数保存起来          mat = fitz.Matrix(zoom_x, zoom_y).preRotate(rotate)          rect = page.rect  # 页面大小          # mp为截取矩形的左上角坐标          mp=rect.tr-(500/zoom_x,0)          # tem为截取矩形的右下角坐标          tem=rect.tr+(0,200/zoom_y)          # clip为截取的矩形          clip = fitz.Rect(mp, tem)          # 进行图片的截取          pix = page.getPixmap(matrix=mat, alpha=False,clip=clip)          if not os.path.exists(imagePath):  # 判断存放图片的文件夹是否存在              os.makedirs(imagePath)  # 若图片文件夹不存在就创建          new_img_path = imagePath + '/' + '0.png'          pix.writePNG(new_img_path)  # 将图片写入指定的文件夹内            return new_img_path

import cv2    def my_croping(imgpath):      # 读取图片的路径      img = cv2.imread(imgpath)      # 把该图片转换为灰度图      gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)      #设置固定级别的阈值应用于矩阵      ret, binary = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY)      # 寻找边缘，返回的contours为边缘数据的集合      _, contours, hierarchy = cv2.findContours(binary, cv2.RETR_TREE, cv2.CHAIN_APPROX_TC89_L1)      # 画出边缘,-1为画出所有的边缘，如果为任意自然数那么为contours的索引,(0,0,255)为颜色，最后的2是线条的粗细，数值越大，线条越粗      cv2.drawContours(img, contours, -1, (0, 0, 255), 2)      # 展示图片      cv2.imshow("pic", img)      # 等待，当参数为0时，为无限等待，直到有键盘指令      cv2.waitKey(0)

contours.sort(key=lambda x: len(x), reverse=True)  for i in range(len(contours)):          if len(contours[i]) > 10:              continue          else:              contours = contours[:i]              break

rect_list=[]  for i in range(len(contours)):          cont_ = contours[i]          # 找到boundingRect          rect = cv2.boundingRect(cont_)          print(rect)          rect_list.append(rect)          

def xyhw(li):      n=10      while n<30:          for i in range(len(li)):              tem_li=[li[i]]              for k in range(i+1,len(li)):                  if abs(li[i][1]-li[k][1])+abs(li[i][2]-li[k][2])+abs(li[i][3]-li[k][3])<n:                      tem_li.append(li[k])              if len(tem_li)>=8:                  return tem_li          n+=1

def filter_li(li):      if len(li)>8:          li = li[:9]      interval=li[0][0]-li[1][0]      test_interval=li[-2][0]-li[-1][0]      if test_interval/interval>1.5:          li=li[:-1]      return li

#进行排序  rect_list.sort(key= lambda x:x[0],reverse=True)  #进行筛选  rect_list=filter_li(rect_list)  #x0,y0为矩阵的左上角，x1,y1为矩阵的右下角  y0=rect_list[0][1]  y1=rect_list[0][1]+rect_list[0][3]  x0=rect_list[-1][0]  x1=rect_list[0][0]+rect_list[0][2]  print(y0,y1,x0,x1)  #进行图片切割  cropImg = img2[y0:y1,x0:x1]  #写入图片  cv2.imwrite(img_path,cropImg)

import cv2  import numpy as np      def xyhw(li):      n=10      tem_li=[]      while n<30:          for i in range(len(li)):              tem_li=[li[i]]              for k in range(i+1,len(li)):                  if abs(li[i][1]-li[k][1])+abs(li[i][2]-li[k][2])+abs(li[i][3]-li[k][3])<n:                      tem_li.append(li[k])              if len(tem_li)>=8:                  return tem_li          n+=1      else:          return tem_li      # 将img的高度调整为28，先后对图像进行如下操作：直方图均衡化，形态学，阈值分割  def pre_treat(img):      height_ = 28      ratio_ = float(img.shape[1]) / float(img.shape[0])      gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)      gray = cv2.resize(gray, (int(ratio_ * height_), height_))      gray = cv2.equalizeHist(gray)      _, binary = cv2.threshold(gray, 190, 255, cv2.THRESH_BINARY)      img_ = 255 - binary  # 反转：文字置为白色，背景置为黑色      return img_      def get_roi(contours):      rect_list = []      for i in range(len(contours)):          rect = cv2.boundingRect(contours[i])          if rect[3] > 10:              rect_list.append(rect)      return rect_list      def get_rect(img):      _, contours, hierarchy = cv2.findContours(img,cv2.RETR_TREE, cv2.CHAIN_APPROX_TC89_L1)      rect_list = get_roi(contours)      rect_list.sort(key= lambda x:x[0],reverse=True)      rect_list=xyhw(rect_list)        return rect_list    def change_(img):      length = 28      h,w = img.shape      H = np.float32([[1,0,(length-w)/2],[0,1,(length-h)/2]])      img = cv2.warpAffine(img,H,(length,length))      M = cv2.getRotationMatrix2D((length/2,length/2),0,26/float(img.shape[0]))      return cv2.warpAffine(img,M,(length,length))    def fenge(img_path):      cont = 0      img = cv2.imread(img_path)      img = pre_treat(img)      contours = get_rect(img)      folder_path=r"C:Users86173Desktopjetbrains2019.2newtem"      file_list=[]      # img=cv2.drawContours(img,contours,2,(0, 0, 255),3)      print("*********************%s*************" %contours)      for i in range(len(contours)):          y0 = contours[i][1]          y1 = contours[i][1] + contours[i][3]          x0 = contours[i][0]          x1 = contours[i][0] + contours[i][2]          print(y0, y1, x0, x1)          cropImg = img[y0:y1, x0:x1]          cropImg = change_(cropImg)          fenge_img=rf"{folder_path}{cont}.png"          cv2.imwrite(fenge_img, cropImg)          cont += 1          file_list.append(fenge_img)      return file_list

from PIL import Image  import numpy    def noise_remove_pil(image_name, k):      """      8邻域降噪      Args:          image_name: 图片文件命名          k: 判断阈值        Returns:        """        def calculate_noise_count(img_obj, w, h):          """          计算邻域非白色的个数          Args:              img_obj: img obj              w: width              h: height          Returns:              count (int)          """          count = 0          width, height = img_obj.size          for _w_ in [w - 1, w, w + 1]:              for _h_ in [h - 1, h, h + 1]:                  if _w_ > width - 1:                      continue                  if _h_ > height - 1:                      continue                  if _w_ == w and _h_ == h:                      continue                  if img_obj.getpixel((_w_, _h_)) < 190:  # 这里因为是灰度图像，设置小于230为非白色                      count += 1          return count        img = Image.open(image_name)      # 灰度      gray_img = img.convert('L')        w, h = gray_img.size      for _w in range(w):          for _h in range(h):              if _w == 0 or _h == 0:                  gray_img.putpixel((_w, _h), 255)                  continue              # 计算邻域非白色的个数              pixel = gray_img.getpixel((_w, _h))              if pixel == 255:                  continue                if calculate_noise_count(gray_img, _w, _h) < k:                  gray_img.putpixel((_w, _h), 255)      # gray_img = gray_img.resize((32, 32), Image.LANCZOS)      gray_img.save(image_name)      # gray_img.show()      im = numpy.array(gray_img)      for i in range(im.shape[0]):  # 转化为二值矩阵          for j in range(im.shape[1]):              if im[i, j] <190:                  im[i, j] = 1              else:                  im[i, j] = 0      return im          if __name__ == '__main__':      for i in range(0,10):          for k in range(0,100):              png_file_path=rf"C:Users86173Desktopjetbrains2019.2model_test{i}_{k}.png"              txt_file_path=rf"C:Users86173Desktopjetbrains2019.2model_testtxt_folder{i}_{k}.txt"              try:                  im = noise_remove_pil(png_file_path, 4)                  with open(txt_file_path,'at',encoding='utf-8')as f:                      for n in im:                          f.writelines(str(n).replace("[","").replace("]","").replace(" ","")+"n")              except Exception as e:                  continue

import numpy as np  from os import listdir  from sklearn.neighbors import KNeighborsClassifier as kNN    def np2vector(im):      returnVect = np.zeros((1, 784))      for i in range(28):          # 读一行数据          lineStr = im[i]          # 每一行的前28个元素依次添加到returnVect中          for j in range(28):              returnVect[0, 28 * i + j] = int(lineStr[j])      # 返回转换后的1x784向量      return returnVect  def img2vector(filename):      #创建1x784零向量      returnVect = np.zeros((1, 784))      #打开文件      fr = open(filename)      #按行读取      for i in range(28):          #读一行数据          lineStr = fr.readline()          #每一行的前28个元素依次添加到returnVect中          for j in range(28):                returnVect[0,28*i+j] = int(lineStr[j])      #返回转换后的1x784向量      return returnVect    def handwritingClassTest(im):      #测试集的Labels      hwLabels = []      #返回trainingDigits目录下的文件名      trainingFileList = listdir(r"C:Users86173Desktopjetbrains2019.2model_testtxt_folder")      #返回文件夹下文件的个数      m = len(trainingFileList)      #初始化训练的Mat矩阵,测试集      trainingMat = np.zeros((m, 784))      #从文件名中解析出训练集的类别      for i in range(m):          #获得文件的名字          fileNameStr = trainingFileList[i]          #获得分类的数字          classNumber = int(fileNameStr.split('_')[0])          #将获得的类别添加到hwLabels中          hwLabels.append(classNumber)          trainingMat[i,:] = img2vector(r'C:Users86173Desktopjetbrains2019.2model_testtxt_folder%s' % (fileNameStr))      #构建kNN分类器      neigh = kNN(n_neighbors = 4, algorithm = 'auto')      #拟合模型, trainingMat为测试矩阵,hwLabels为对应的标签      neigh.fit(trainingMat, hwLabels)        vectorUnderTest = np2vector(im)        classifierResult = neigh.predict(vectorUnderTest)      return classifierResult