def main():
for i in range(200):
ik = get_random_id()
u = image_api.format(id=ik)
r = requests.get(u, headers=img_api_headers)
url, filename = get_image_url_and_filename(r.text)
save_url_image_to_file(url, filename)
def send_query_until_true(num):
# 生成准考证号
new_id = myid.format(id=num)
# 获取验证码图片地址
img_api_url = image_api.format(id=new_id)
img_api_resp = requests.get(img_api_url, headers=img_api_headers)
img_url, filename = get_image_url_and_filename(img_api_resp.text)
print()
# 获取验证码图片并猜测
img_resp = requests.get(img_url)
code = get_validate_code_from_image(Image.open(BytesIO(img_resp.content)))
# 执行查询操作
data = {
"data": "CET4_181_DANGCI,{id},{name}".format(
id=new_id, name=name), #看您是四级还是六级考试的修改为CET4_181还是CET6_181
"v": code
}
log_info(data)
query_resp = requests.post(query_api, data=data, headers=query_api_headers)
query_text = query_resp.text
log_info(query_text)
if "验证码错误" in query_text:
query_text = send_query_until_true(num)
return query_text
def save_image_to_file():
myid = "340080181210211"
new_id = myid.format(id=myid)
img_api_url = image_api.format(id=new_id)
img_api_resp = requests.get(img_api_url,
headers=img_api_headers,
timeout=10)
img_url, filename = get_image_url_and_filename(img_api_resp.text)
r = requests.get(img_url)
with open("images/raw_picture/" + filename, "wb+") as f:
f.write(r.content)
'''
def send_query_until_true(num):
# 生成准考证号
global proxy
new_id = myid.format(id=num)
# 获取验证码图片地址
img_api_url = image_api.format(id=mid)
# img_api_url = 'http://cet.neea.edu.cn/imgs/28e6e9985b6e4d4880db307edfd66413.png'
while True:
try:
img_api_resp = requests.get(img_api_url,
headers=img_api_headers,
timeout=10)
img_url, filename = get_image_url_and_filename(img_api_resp.text)
print(img_url)
# 获取验证码图片并猜测
img_resp = requests.get(img_url, timeout=10)
if img_resp.status_code == 200:
images = Image.open(BytesIO(img_resp.content))
code = img_verify_code(images)
else:
code = "xxxx"
except Exception:
print("重新获取代理")
p = str(get_proxy())
proxy = {'http': 'http://' + p, 'https': 'http://' + p}
else:
break
# CET4成绩查询选项
# data = {"data": "CET4_181_DANGCI,{id},{name}".format(id=new_id, name=name),"v": code}
# CET6成绩查询选项
data = {
"data": "CET6_181_DANGCI,{id},{name}".format(id=new_id, name=name),
"v": code
}
query_resp = requests.post(query_api, data=data, headers=query_api_headers)
query_text = query_resp.text
log_info(query_text.split("'")[3], new_id)
if "验证码错误" in query_text:
query_text = send_query_until_true(num)
# elif "您查询的结果为空" in query_text:
# images.save("images/save_picture/" + code + ".png")
return query_text
def send_query_until_true(myid, name, type):
# 生成准考证号
new_id = myid
# 获取验证码图片地址
img_api_url = image_api.format(id=new_id)
img_api_resp = requests.get(img_api_url, headers=img_api_headers)
img_url, filename = get_image_url_and_filename(img_api_resp.text)
# 获取验证码图片并猜测
img_resp = requests.get(img_url)
code = get_validate_code_from_image(Image.open(BytesIO(img_resp.content)))
# 执行查询操作
data = {}
if type == 4:
data = {
"data": "CET4_172_DANGCI,{id},{name}".format(id=new_id, name=name),
"v": code
}
log_info(data)
query_resp = requests.post(query_api, data=data, headers=query_api_headers)
query_text = query_resp.text
log_info(query_text)
# if "验证码错误" in query_text:
# query_text = send_query_until_true(num, type)
if "验证码错误" in query_text:
return False
return query_text
else:
data = {
"data": "CET6_172_DANGCI,{id},{name}".format(id=new_id, name=name),
"v": code
}
log_info(data)
query_resp = requests.post(query_api, data=data, headers=query_api_headers)
query_text = query_resp.text
log_info(query_text)
if "验证码错误" in query_text:
return False
return query_text
def send_query_until_true(js, zw):
global cnt
# 生成准考证号
new_id = myid.format(js=js, zw=zw)
# 获取验证码图片地址
img_api_url = image_api.format(id=new_id)
img_api_resp = requests.get(img_api_url, headers=img_api_headers)
img_url, filename = get_image_url_and_filename(img_api_resp.text)
# 获取验证码图片并猜测
img_resp = requests.get(img_url, stream=True)
if img_resp.status_code == 200:
code = get_validate_code_from_image(Image.open(BytesIO(img_resp.content)))
else:
# 404后直接赋值一个错误的code, 以便接下来递归回来
code = "xxxx"
# 执行查询操作
cnt += 1
data = {
"data": "CET4_181_DANGCI,{id},{name}".format(id=new_id, name=name),
"v": code
}
#log_info(data)
print("Try: " + data['data'].split(',')[1] + "," +data['data'].split(',')[2])
# threadLock.acquire()
try:
query_resp = requests.post(query_api, data=data, headers=query_api_headers, timeout=5)
if query_resp.status_code == 200:
query_text = query_resp.text
except requests.exceptions.Timeout:
global NETWORK_STATUS
NETWORK_STATUS = False
if NETWORK_STATUS == False:
for i in range(1, 10):
print("第%s次超时,正在重新请求."%i)
query_resp = requests.post(query_api, data=data, headers=query_api_headers, timeout=5)
if query_resp.status_code == 200:
query_text = query_resp.text
break
# threadLock.release()
print("Result: " + query_text)
if "验证码错误" in query_text:
query_text = send_query_until_true(js, zw)
'''
if not os.path.exists("download_error_capture"):
os.mkdir('download_error_capture')
with open("download_error_capture/%s.png" % random.randint(10000, 99999), "wb") as f:
f.write(img_resp.content)
f.close()
'''
else:
if not os.path.exists("download_capture"):
os.mkdir('download_capture')
with open("download_capture/%s.png" %code, "wb") as f:
f.write(img_resp.content)
f.close()
return query_text