def test_request_process(self):
vjkl5 = Vjkl5()
self.session.cookies["vjkl5"] = vjkl5
url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "关键词:合同",
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(vjkl5),
"number": Number(),
"guid": Guid(),
}
response = self.session.post(url, data=data)
json_data = json.loads(response.json())
print("列表数据:", json_data)
run_eval = json_data.pop(0)["RunEval"]
try:
key = parse_run_eval(run_eval)
except ValueError as e:
raise ValueError("返回脏数据") from e
else:
print("RunEval解析完成:", key, "\n")
key = key.encode()
for item in json_data:
cipher_text = item["文书ID"]
print("解密:", cipher_text)
plain_text = decrypt_doc_id(doc_id=cipher_text, key=key)
print("成功, 文书ID:", plain_text, "\n")
def test_list(self):
response = self.session.get("http://wenshu.court.gov.cn/list/list/")
text = response.content.decode()
if self.error_msg in text:
redirect_url1 = wzws_decrypt(text, url=response.url)
redirect_url2 = wzws_decrypt(text)
self.assertEqual(redirect_url1, redirect_url2)
_ = self.session.get(redirect_url1)
url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "案件类型:行政案件",
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(self.session.cookies["vjkl5"]),
"number": Number(),
"guid": Guid(),
}
response = self.session.post(url, data=data)
self.assertNotIn(self.error_msg, response.content.decode())
print(response.text)
def test_request_process(self):
vjkl5 = Vjkl5()
self.session.cookies["vjkl5"] = vjkl5
url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "案件类型:执行案件",
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(vjkl5),
"number": Number(),
"guid": Guid(),
}
response = self.session.post(url, data=data)
json_data = json.loads(response.json())
print("列表数据:", json_data)
run_eval = json_data.pop(0)["RunEval"]
try:
key = parse_run_eval(run_eval)
except ValueError as e:
raise ValueError("返回脏数据") from e
else:
print("RunEval解析完成:", key, "\n")
key = key.encode()
for item in json_data:
cipher_text = item["文书ID"]
print("解密:", cipher_text)
plain_text = decrypt_doc_id(doc_id=cipher_text, key=key)
print("成功, 文书ID:", plain_text, "\n")
# 请求详情页
detail_url = "http://wenshu.court.gov.cn/CreateContentJS/CreateContentJS.aspx"
params = {
"DocID": decrypt_doc_id(doc_id=json_data[0]["文书ID"], key=key),
}
response = self.session.get(detail_url, params=params)
text = response.content.decode()
if self.error_msg in text:
redirect_url1 = wzws_decrypt(text, url=response.url)
redirect_url2 = wzws_decrypt(text)
self.assertEqual(redirect_url1, redirect_url2)
response = self.session.get(redirect_url1)
self.assertNotIn(self.error_msg, response.content.decode())
group_dict = parse_detail(response.text)
pprint(group_dict)
def test_vl5x(self):
online_test_data = []
for data in local_test_data:
vjkl5 = data["vjkl5"]
vl5x = data["vl5x"]
py_vl5x = Vl5x(vjkl5)
self.assertEqual(py_vl5x, vl5x, "vl5x不一致")
online_test_data.append({
"vjkl5": vjkl5,
"vl5x": py_vl5x,
})
print("vl5x本地测试通过, 如需在线测试vl5x请复制以下js到文书网列表页console中测试:\n{}".format(
online_test_js % online_test_data))
def test_list_page(self):
url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "案件类型:执行案件",
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(self.session.cookies.setdefault("vjkl5", Vjkl5())),
"number": Number(),
"guid": Guid(),
}
response = self.session.post(url, data=data)
text = response.content.decode()
if self.error_msg in text:
retry = 3
for _ in range(retry):
redirect_url = decrypt_wzws(text)
response = self.session.post(redirect_url, data=data)
text = response.content.decode()
if self.error_msg not in text:
break
else:
self.fail("连续{}次获取数据失败".format(retry))
json_data = json.loads(response.json())
print("列表数据:", json_data)
run_eval = json_data.pop(0)["RunEval"]
try:
key = decrypt_runeval(run_eval)
except ValueError as e:
raise ValueError("返回脏数据") from e
else:
print("RunEval解析完成:", key, "\n")
key = key.encode()
for item in json_data:
cipher_text = item["文书ID"]
print("解密:", cipher_text)
plain_text = decrypt_doc_id(doc_id=cipher_text, key=key)
print("成功, 文书ID:", plain_text, "\n")
def request_list():
"""请求列表数据"""
session = requests.Session()
session.headers.update({
"User-Agent":
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/73.0.3683.103 Safari/537.36",
"X-Requested-With": "XMLHttpRequest",
})
vjkl5 = Vjkl5()
session.cookies["vjkl5"] = vjkl5
list_url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "案件类型:刑事案件", # TODO 太多人测试此检索条件的数据，如果报错或返回脏数据请更换别的检索条件进行测试
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(vjkl5),
"number": Number(),
"guid": Guid(),
}
response = session.post(list_url, data=data)
json_data = json.loads(response.json())
print("列表数据:", json_data)
run_eval = json_data.pop(0)["RunEval"]
try:
key = parse_run_eval(run_eval)
except ValueError as e:
raise ValueError("返回脏数据") from e
else:
print("RunEval解析完成:", key, "\n")
key = key.encode()
for item in json_data:
cipher_text = item["文书ID"]
print("解密:", cipher_text)
plain_text = decrypt_doc_id(doc_id=cipher_text, key=key)
print("成功, 文书ID:", plain_text, "\n")
def test_list(self):
response = self.session.get("http://wenshu.court.gov.cn/list/list/")
if "请开启JavaScript并刷新该页".encode() in response.content:
redirect_url = wzws_cid_decrypt(response.content)
_ = self.session.get(redirect_url)
url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "关键词:合同",
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(self.session.cookies["vjkl5"]),
"number": Number(),
"guid": Guid(),
}
response = self.session.post(url, data=data)
self.assertNotIn("请开启JavaScript并刷新该页", response.content.decode())
print(response.text)
def list_page(self):
"""文书列表页"""
url = "http://wenshu.court.gov.cn/List/ListContent"
data = {
"Param": "案件类型:刑事案件",
"Index": 1,
"Page": 10,
"Order": "法院层级",
"Direction": "asc",
"vl5x": Vl5x(self.session.cookies.setdefault("vjkl5", Vjkl5())),
"number": Number(),
"guid": Guid(),
}
response = self.session.post(url, data=data) # 请求1
text = response.content.decode()
if "请开启JavaScript并刷新该页" in text:
# 如果使用代理，确保请求1和请求2的ip为同一个，否则将继续返回"请开启JavaScript并刷新该页"
redirect_url = decrypt_wzws(text)
response = self.session.post(redirect_url, data=data) # 请求2
json_data = json.loads(response.json())
print("列表数据:", json_data)
run_eval = json_data.pop(0)["RunEval"]
try:
key = decrypt_runeval(run_eval)
except ValueError as e:
raise ValueError("返回脏数据") from e
else:
print("RunEval解析完成:", key, "\n")
key = key.encode()
for item in json_data:
cipher_text = item["文书ID"]
print("解密:", cipher_text)
plain_text = decrypt_doc_id(doc_id=cipher_text, key=key)
print("成功, 文书ID:", plain_text, "\n")