from os import chdir, listdir import numpy as np from time import gmtime def regulater(binary_array) : if not (len(binary_array.shape) == 2 and set([itm for row in binary_array for itm in row]) <= set([0,255]) ): print('binary_array is not two dimensional or not binary', len(binary_array.shape), set([itm for row in binary_array for itm in row])) raise TypeError newshape = tuple(list(binary_array.shape)+[5]) newarray = np.empty(newshape,dtype=np.uint8) height,width = binary_array.shape for indx in range(height) : for indx2 in range(width) : comparators = [int(binary_array[i,indx2]) for i in [max([0,indx-1]),indx,min([height-1,indx+1])] ] + [int(binary_array[indx,indx2]) for j in [max([0,indx2-1]),min([width-1,indx2+1])] ] if sum(comparators)/255 > 2 : newarray[indx,indx2] = 255 else : newarray[indx,indx2] = 0 return newarray def read_qrcodes(pdfdoc, inventory, pagenum=0) : '''note can improve accuracy by focusing on top right corner where the qrcode will be. this can be a larger or smaller sized box around where the nominal position of the qrcode. if larger it will be more robust against feed instability in the scanning process but less robust against obfuscating marks and clutter''' import csv, cv2, fitz, json from PIL import Image student_inventory = {} with open(inventory, 'r') as invent : csvreader = csv.reader(invent, delimiter = '\t') for line in csvreader : try : serial_number = line[4] filename = line[5] #student_inventory[ line[0] ] = line[ 1 ] except IndexError : if len(line) == 0 : continue else : print('index error csv', line) raise IndexError else : student_inventory[ line[ 4 ] ] = line[ 5 ] print('inventory', student_inventory) doc = fitz.open(pdfdoc) print('doc_length',doc.page_count) mat = fitz.Matrix(8.0,8.0) clip = fitz.Rect( (480,5), (630,120) ) #(650,120) qcd = cv2.QRCodeDetector() pageno = pagenum assignment = {} weirds = [] for indx in range(doc.page_count) : page = doc[indx] pix = page.get_pixmap(matrix = mat, clip = clip) im = np.frombuffer(pix.samples, dtype=np.uint8).reshape(pix.h,pix.w,pix.n) im = np.ascontiguousarray(im[...,[2,1,0]]) gray = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY) kernel = np.ones((3,3),dtype=np.uint8) ret,thresh = cv2.threshold(gray,200,255,cv2.THRESH_BINARY) thresh=cv2.bitwise_not(cv2.dilate(cv2.erode(cv2.bitwise_not(thresh),kernel,iterations=1),kernel,iterations=1)) data = [] for k in range(4) : dat = qcd.detectAndDecodeMulti(np.rot90(thresh,k)) if dat[0] == True : data += [dat[1]] if len(set(data)) == 1 and data[0]!=('',): print(data) assignment[pageno] = data[0] newdoc = fitz.open() newdoc.insert_pdf(doc,from_page=indx, to_page=indx, start_at=-1, rotate=-1, links=True, annots=True) newbytes = newdoc.tobytes(garbage=4,deflate=True,deflate_images=True, deflate_fonts=True) try : newfilename = student_inventory[data[1][0]] except KeyError : print('key error', pageno) pageno += 1 continue except IndexError : print('index error', pageno, data) else : '''it would be good to downsample this for smaller filesize''' with open('/path/to/directory/to/store/student/quizzes/'+newfilename, 'wb') as f: f.write(newbytes) else : weirds += [pageno] print("I can't read the qr code at page number " + str(pageno+1) +", could you enter the filename for me?") img = Image.fromarray(thresh.astype('uint8')) '''we want to keep an eye on any unreadable qr codes so we can keep track of what goes wrong''' img.save('/path/to/store/weird/qrs/'+str(gmtime().tm_yday)+'_'+str(pageno)+'.jpg') img.show() newfilename = input('page '+str(pageno+1)+'-->') newdoc = fitz.open() newdoc.insert_pdf(doc,from_page=indx, to_page=indx, start_at=-1, rotate=-1, links=True, annots=True) newbytes = newdoc.tobytes() with open('/path/to/directory/to/store/student/quizzes/' + newfilename, 'wb') as f: f.write(newbytes) pageno += 1 return (assignment, weirds, pageno) if __name__=='__main__' : inventory = [] outp = {} assignment,weirds,pageno = read_qrcodes('/path/to/scanned/quizzes/somefilename.pdf', '/path/to/inventory/file/q4_inventory_mat343.txt') print('pageno', pageno) for ky,val in assignment.items() : outp[ ky ] = val print(outp, weirds)