# 奇巧淫技-二维码跨网络传输数据
# 概述
将文件压缩成7zip格式再按2950的长度分割，并给每一块打上序号分别生成二维码，大致是2K数据对应一张二维码。
使用方式:
使用qr_gen中的代码生成二维码
大约30000个中文汉字得到的压缩包为62K，最后生成了29张二维码。
每张二维码携带的数据大致如下：
{"`<count>`/`<total>`":"`<data>`"}python qr_gen.py test_file.txt使用相机或各种手段得到这些二维码中存储的数据并json.loads，最后放入一个大dict中-将这一部分逻辑补充到qr_decode.py的get_big_dict函数中
- 安卓系统可以去谷歌商店下载【二维码扫描仪】，去设置中打开批量扫描模式，然后扫完之后将扫描的历史记录导出成csv处理。
使用qr_decode中的代码恢复文件
python qr_decode.py temp.7z
# 代码
# qr_gen
import qrcode
import os
from py7zr import SevenZipFile
import base64
import json
import numpy as np
# import cv2
from PIL import Image
import sys
import shutil
import time
# def show(img, name='img'):
# maxHeight = 540
# maxWidth = 960
# scaleX = maxWidth / img.shape[1]
# scaleY = maxHeight / img.shape[0]
# scale = min(scaleX, scaleY)
# if scale < 1:
# img = cv2.resize(img, (0, 0), fx=scale, fy=scale)
# cv2.imshow(name, img)
# cv2.waitKey(0)
# cv2.destroyWindow(name)
def split_data(data, base=2954):
tmp = data
index = 0
res = []
x = {}
while len(tmp) > 0:
x[index] = tmp[0:min(len(tmp), base-len('{"' + str(index) + '": ""}') - 4)]
tmp = tmp[min(len(tmp), base-len('{"' + str(index) + '": ""}') - 4):]
index += 1
total = index - 1
for k, v in x.items():
y = {f'{k}/{total}': v}
res.append(json.dumps(y))
return res
def color_gen(size, color):
img = np.zeros((size[0], size[1], 3), np.uint8)
for i, v in enumerate(color):
img[:, :, i] = img[:, :, i] + np.ones((size[0], size[1]), np.uint8) * v
# show(img)
return img
def add_mark(img, f):
size = [30, 30]
size_2 = [30, 30]
m = color_gen(size, [255, 0, 255])
g = color_gen(size, [0, 255, 0])
b = color_gen(size, [255, 0, 0])
s = color_gen(size, [255, 255, 0])
c = color_gen(size_2, [0, 0, 0])
img[0:m.shape[0], 0:m.shape[1]] = m
img[0:g.shape[0], 0-g.shape[1]:] = g
img[0-b.shape[0]:, 0:b.shape[1]] = b
img[0-s.shape[0]:, 0-s.shape[1]:] = s
# if f:
# img[0 - c.shape[0]:, (img.shape[1] - c.shape[1])//2:(img.shape[1] - c.shape[1])//2+c.shape[1]] = c
# show(img)
return img
def add_mark_se(img_size):
img = np.ones((img_size[0], img_size[1], 3), np.uint8) * 255
size = [30, 30]
size_2 = [30, 30]
m = color_gen(size, [255, 0, 255])
g = color_gen(size, [0, 255, 0])
b = color_gen(size, [255, 0, 0])
s = color_gen(size, [255, 255, 0])
c = color_gen(size_2, [0, 255, 255])
img[0:m.shape[0], 0:m.shape[1]] = m
img[0:g.shape[0], 0-g.shape[1]:] = g
img[0-b.shape[0]:, 0:b.shape[1]] = b
img[0-s.shape[0]:, 0-s.shape[1]:] = s
img[0 - c.shape[0]:, (img.shape[1] - c.shape[1])//2:(img.shape[1] - c.shape[1])//2+c.shape[1]] = c
# show(img)
return img
# def vedio_gen(data_path, x, size=(1850, 1850)):
# # data_path = "2017-06-13-29/images/"
# fps = 8 # 视频帧率
# # size = (640, 480) # 需要转为视频的图片的尺寸
# res_pth = os.path.normpath(os.path.join(data_path, '../output.mp4'))
# video = cv2.VideoWriter(res_pth, cv2.VideoWriter_fourcc('a', 'v', 'c', '1'), fps, size)
#
# img_s = add_mark_se(size)
# video.write(img_s)
# for i in range(x):
# image_path = data_path + rf'\a{i}.png'
# # print(image_path)
# img = cv2.imread(image_path)
# video.write(img)
#
# video.write(img_s)
# video.write(img_s)
#
# video.release()
# cv2.destroyAllWindows()
def main(pth, base):
# print(pth)
if os.path.exists(os.path.join(pth, f'../a')):
shutil.rmtree(os.path.join(pth, f'../a'))
time.sleep(1)
os.makedirs(os.path.normpath(os.path.join(pth, f'../a')))
qr = qrcode.QRCode(
version=40,
error_correction=qrcode.constants.ERROR_CORRECT_L,
box_size=10,
border=4,
)
# filters = [{'id': FILTER_ZSTD, 'level': 3}]
z_pth = os.path.normpath(os.path.join(pth, '../a.7z'))
with SevenZipFile(z_pth, mode='w') as tar:
tar.write(pth, os.path.split(pth)[1])
with open(z_pth, 'rb') as f:
con = f.read()
# print(len(con))
con = base64.encodebytes(con).decode('utf-8')
con = con.replace('\n', '_')
# base = 2900
res = split_data(con, base)
flag = True
for i, v in enumerate(res):
qr.add_data(v)
qr.make(fit=False)
img = qr.make_image(fill_color="black", back_color="white")
# img.save(os.path.normpath(os.path.join(pth, f'../a/a{i}.png')))
qr.clear()
img = np.expand_dims(np.asarray(img, dtype='uint8'), axis=2)[:, :, [0, 0, 0]] * 255
img = add_mark(img, flag)
img = Image.fromarray(np.asarray(img)[:, :, [2,1,0]])
img.save(os.path.normpath(os.path.join(pth, f'../a/a{i}.png')))
flag = not flag
# vedio_gen(os.path.normpath(os.path.join(pth, f'../a')), i+1, size=img.shape[0:2])
print('Done!')
if __name__ == '__main__':
if len(sys.argv) == 2:
pth = sys.argv[1]
base = 2950
elif len(sys.argv) > 2:
pth = sys.argv[1]
base = sys.argv[2]
else:
pth = os.path.normpath(os.path.join(__file__, r'..\..\test_file\test.py'))
base = 2950
main(pth, base)
# qr_decode
import sys
import base64
def get_big_dict():
raise NotImplementedError
import pandas as pd
import json
fp = "/Users/lei/Documents/History_1691122776517.csv"
df = pd.read_csv(fp)
print(df.columns)
texts = list(df["text"])
res = {}
for text in texts:
res.update(json.loads(text))
return res
def check_data(res):
error = []
tar = ""
total = int(list(res.keys())[0].split("/")[1])
index = [int(i.split("/")[0]) for i in res.keys()]
index.sort()
for i in range(total + 1):
if i not in index:
error.append(i)
if error:
print(f"缺失数据：{error}")
return False
else:
print("数据完整")
for i in index:
tar += res[f"{i}/{total}"]
return tar
def main(big_dict, zip_file_name="temp.7z"):
big_dict = get_big_dict()
res = check_data(big_dict)
if res:
res = base64.decodebytes(res.encode("utf-8"))
with open("./temp.7z", "wb") as f:
f.write(res)
if __name__ == "__main__":
print(sys.argv)
if len(sys.argv) > 2:
print("Usage: python decode_qr.py [<zip_file_name>]")
elif len(sys.argv) == 2:
zip_file_name = sys.argv[1]
else:
zip_file_name = "temp.7z"
main(zip_file_name)
# decode_qr - 备用
# !-*- coding:utf-8 -*-
import os
import requests
from io import BytesIO
from pyzbar import pyzbar
from PIL import Image, ImageEnhance
# from py7zr import SevenZipFile
import re
import base64
import json
from manual_detect import manual_detect
def get_ewm(img_adds):
""" 读取二维码的内容： img_adds：二维码地址（可以是网址也可是本地地址 """
if os.path.isfile(img_adds):
# 从本地加载二维码图片
img = Image.open(img_adds)
else:
# 从网络下载并加载二维码图片
rq_img = requests.get(img_adds).content
img = Image.open(BytesIO(rq_img))
# img.show() # 显示图片，测试用
txt_list = pyzbar.decode(img)
barcodeData = b''
txt = txt_list[0]
barcodeData += txt.data
return barcodeData
# print(barcodeData)
def check_data(res):
error = []
tar = ''
total = int(list(res.keys())[0].split('/')[1])
index = [int(i.split('/')[0]) for i in res.keys()]
index.sort()
for i in range(total + 1):
if i not in index:
error.append(i)
if error:
print(f'缺失数据：{error}')
return False
else:
print('数据完整')
for i in index:
tar += res[f'{i}/{total}']
return tar
def main(pth):
tpl = re.compile(r'.*\.(png|jpg|JPG|PNG|jpeg)')
tar_list = []
for root, dirs, files in os.walk(pth):
for file in files:
if tpl.findall(file):
tar_list.append(os.path.join(root, file))
# print(tar_list)
res = {}
for i in tar_list:
tmp = get_ewm(i)
if tmp:
res.update(json.loads(tmp.decode('utf-8')))
res = check_data(res)
if res:
res = base64.decodebytes(res.encode('utf-8'))
with open(os.path.normpath(os.path.join(pth, '../tmp.7z')), 'wb') as f:
f.write(res)
if __name__ == '__main__':
# 解析本地二维码
# pth1 = os.path.normpath(os.path.join(__file__, r'..\..\test_file\res'))
pth = os.path.normpath(os.path.join(__file__, r'..\..\test_file\right'))
m = manual_detect()
pth1 = m.main(pth, force_manal_flag=False)
main(pth1)
# get_ewm(pth)
# manual_detect - 备用
import cv2
import os
import numpy as np
import shutil
import time
import numpy as np
def sigmoid(x):
s = 255/(1 + np.exp(127-x))
return s
class manual_detect:
def __init__(self):
self.draw_img = None
self.dots = []
self.m = ((135, 150, 150), (180, 255, 255))
self.g = ((38, 144, 147), (72, 255, 255))
self.b = ((96, 150, 150), (135, 255, 255))
self.s = ((74, 150, 150), (101, 255, 255))
self.manual_flag = False
@staticmethod
def show(img, name='img'):
# return
# return
maxHeight = 960
maxWidth = 960
scaleX = maxWidth / img.shape[1]
scaleY = maxHeight / img.shape[0]
scale = min(scaleX, scaleY)
if scale < 1:
img = cv2.resize(img, (0, 0), fx=scale, fy=scale)
cv2.imshow(name, img)
cv2.waitKey(0)
cv2.destroyWindow(name)
@staticmethod
def shrink_size(img):
maxHeight = 1080
maxWidth = 1080
scaleX = maxWidth / img.shape[1]
scaleY = maxHeight / img.shape[0]
scale = min(scaleX, scaleY)
if scale < 1:
img = cv2.resize(img, (0, 0), fx=scale, fy=scale, interpolation=cv2.INTER_AREA )
return img
@staticmethod
def get_trans(img, pts1, pts2):
M = cv2.getPerspectiveTransform(pts1, pts2)
dst = cv2.warpPerspective(img, M, (700, 700))
return dst
@staticmethod
def draw_dot(img, dots):
# img_tmp = cv2.copyTo(img)
img_tmp = img.copy()
for dot in dots:
cv2.circle(img_tmp, tuple(dot), 5, (0, 255, 0), -1)
return img_tmp
def show_confirm(self, img, name='img'):
return
maxHeight = 960
maxWidth = 960
scaleX = maxWidth / img.shape[1]
scaleY = maxHeight / img.shape[0]
scale = min(scaleX, scaleY)
if scale < 1:
img = cv2.resize(img, (0, 0), fx=scale, fy=scale)
while True:
cv2.imshow(name, img)
k = cv2.waitKey(5)
# print(k)
if k == ord('q') or k == 27:
break
elif k == ord('a'):
self.dots = []
self.manual_flag = True
# self.draw_img = img.copy()
break
cv2.destroyWindow(name)
def draw_circle(self, event, x, y, flags, param):
# global ix, iy, dots, draw_img
if event == cv2.EVENT_LBUTTONDBLCLK:
cv2.circle(self.draw_img, (x, y), 5, (255, 0, 0), -1)
self.dots.append([x, y])
def get_dots(self, img):
# global dots, draw_img
self.draw_img = img.copy()
self.dots = []
cv2.namedWindow('image')
cv2.setMouseCallback('image', self.draw_circle)
while True:
cv2.imshow('image', self.draw_img)
if len(self.dots) >= 4:
break
k = cv2.waitKey(5)
# print(k)
if k == 27:
break
elif k == ord('a'):
self.dots = []
self.draw_img = img.copy()
# break
cv2.destroyAllWindows()
# print(self.dots)
def main_sub(self, pth):
# global dots, drow_img
img = cv2.imread(pth)
img = self.shrink_size(img)
# self.show(img)
self.get_dots(img)
if len(self.dots) != 4:
print('标定点错误')
return
# pts1 = np.float32([[56, 65], [368, 52], [28, 387], [389, 390]])
pts2 = np.float32([[0, 0], [700, 0], [0, 700], [700, 700]])
pts1 = np.float32(self.dots)
img_dot = self.draw_dot(img, self.dots)
self.show(img_dot)
img_trs = self.get_trans(img, pts1, pts2)
img_trs = self.strach_color(img_trs)
img_trs = self.strach_color(img_trs)
img_trs = self.strach_color(img_trs)
# img_trs = cv2.cvtColor(img_trs, cv2.COLOR_BGR2GRAY)
# _, img_trs = cv2.threshold(img_trs,100,255,cv2.THRESH_BINARY)
# self.show(img_trs)
cv2.imwrite(os.path.normpath(os.path.join(pth, r'..\..\res', os.path.split(pth)[-1])), img_trs)
def get_p(self, thr):
contours, hierarchy = cv2.findContours(thr, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
if contours:
cnt = contours[0]
x, y, w, h = cv2.boundingRect(cnt)
img = cv2.cvtColor(thr, cv2.COLOR_GRAY2BGR)
cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
# print(M)
# cv2.circle(thr, )
# self.show(img)
self.dots.append([x + w // 2, y + h // 2])
def find_mark(self, img, color):
frame_HSV = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
frame_threshold = cv2.inRange(frame_HSV, color[0], color[1])
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (3, 3))
opening = cv2.morphologyEx(frame_threshold, cv2.MORPH_OPEN, kernel)
self.get_p(opening)
# self.show(frame_threshold)
# print(111)
def get_dots_auto(self, img):
self.dots = []
self.find_mark(img, self.m)
self.find_mark(img, self.g)
self.find_mark(img, self.b)
self.find_mark(img, self.s)
# self.show(img)
def strach_color(self, img):
img_t = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
h,s,v = cv2.split(img_t)
v_max = np.max(v)
v_min = np.min(v)
v1 = np.uint8(np.clip(cv2.add(255 / (v_max - v_min) * 1.2 * v, (0 - v_min) / (v_max - v_min)), 0, 255))
img1 = np.uint8(cv2.merge((h,s,v1)))
return cv2.cvtColor(img1, cv2.COLOR_HSV2BGR)
def strach_color2(self, img):
img_t = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
h,s,v = cv2.split(img_t)
v_max = np.max(v)
v_min = np.min(v)
v1 = np.clip(cv2.add(255 / (v_max - v_min) * 1.2 * v, (0 - v_min) / (v_max - v_min)), 0, 255)
v1 = np.uint8(sigmoid(v1))
img1 = np.uint8(cv2.merge((h,s,v1)))
return cv2.cvtColor(img1, cv2.COLOR_HSV2BGR)
def main_sub_fur(self, pth):
img = cv2.imread(pth)
img = self.shrink_size(img)
img2 = self.strach_color2(img)
# self.show(img2)
self.get_dots_auto(img2)
# pts1 = np.float32([[56, 65], [368, 52], [28, 387], [389, 390]])
pts2 = np.float32([[0, 0], [700, 0], [0, 700], [700, 700]])
pts1 = np.float32(self.dots)
img_dot = self.draw_dot(img, self.dots)
self.show_confirm(img_dot)
if len(self.dots) != 4:
print('标定点错误，需要人工标定')
self.manual_flag = True
if self.manual_flag:
return
img_trs = self.get_trans(img, pts1, pts2)
img_trs = self.strach_color(img_trs)
img_trs = self.strach_color(img_trs)
img_trs = self.strach_color(img_trs)
# img_trs = cv2.cvtColor(img_trs, cv2.COLOR_BGR2GRAY)
# _, img_trs = cv2.threshold(img_trs,100,255,cv2.THRESH_BINARY)
# self.show(img_trs)
cv2.imwrite(os.path.normpath(os.path.join(pth, r'..\..\res', os.path.split(pth)[-1])), img_trs)
def main(self, pth, force_manal_flag=False):
res_pth = os.path.normpath(os.path.join(pth, r'..\res'))
if os.path.exists(res_pth):
shutil.rmtree(res_pth)
time.sleep(2)
os.mkdir(res_pth)
for root, dirs, files in os.walk(pth):
for file in files:
self.manual_flag = False
if force_manal_flag:
self.manual_flag = True
else:
self.main_sub_fur(os.path.join(root, file))
if self.manual_flag:
self.main_sub(os.path.join(root, file))
return os.path.normpath(os.path.join(pth, r'..\res'))
if __name__ == '__main__':
pth = os.path.normpath(os.path.join(__file__, r'..\..\test_file\a1'))
m = manual_detect()
m.main(pth)
← qrcode Sandman2简介及示例 →