def step2(self, keyword: str, html_args: dict):
"""请求数据接口"""
url = "http://wsjs.saic.gov.cn/txnRead02.ajax"
path = parse.urlparse(url).path
page = 1
page_size = 50
request_args = {
"locale": "zh_CN", # 语言
"request:queryCom": "1",
"request:nc": "", # 国际分类
"request:sn": "", # 申请/注册号
"request:mn": keyword, # 商标名称
"request:hnc": "", # 申请人名称(中文)
"request:hne": "", # 申请人名称(英文)
"request:imf": "",
"request:maxHint": "",
"request:ncs": "",
"request:queryAuto": "",
"request:queryExp": f"mnoc = {keyword}*",
"request:queryMode": "",
"request:queryType": "",
"request:mi": html_args["request:mi"],
"request:tlong": html_args["request:tlong"],
"attribute-node:record_cache-flag": "false",
"attribute-node:record_page": page,
"attribute-node:record_page-row": page_size,
"attribute-node:record_sort-column": "RELEVANCE",
"attribute-node:record_start-row": (page - 1) * page_size + 1,
}
# 本地加密 TODO 已失效
# kwargs = self.local_encrypt(path=path, request_args=request_args)
# 在线加密
kwargs = online_encrypt(url=API, path=path, request_args=request_args)
response = self.session.post(url, **kwargs)
if response.status_code != 200:
raise Exception(response.status_code)
# 提取数据
for tag in html.fromstring(response.content).xpath("//record"):
print({
"tid": tag.xpath("tid/text()")[0],
"index": tag.xpath("index/text()")[0],
"申请/注册号": tag.xpath("tmid/text()")[0],
"国际分类": tag.xpath("nc/text()")[0],
"申请日期": tag.xpath("fd/text()")[0],
"商标名称": tag.xpath("mno/text()")[0],
"申请人名称": tag.xpath("hnc/text()")[0],
})
def _request(self, url: str, request_args: dict, *, method="POST", api: str = None) -> requests.Response:
path = parse.urlparse(url).path
if not api:
kwargs = local_encrypt(path=path, request_args=request_args) # 本地加密 TODO 已失效
else:
kwargs = online_encrypt(api=api, path=path, request_args=request_args) # 在线加密
response = self.session.request(method=method, url=url, **kwargs)
if response.status_code != 200:
raise Exception(response.status_code)
return response
def _request(self, url: str) -> requests.Response:
path = parse.urlparse(url).path
request_args = {
"request:tid": self.tid,
}
# 本地加密 TODO 已失效
# kwargs = self.local_encrypt(path=path, request_args=request_args)
# 在线加密
kwargs = online_encrypt(url=API, path=path, request_args=request_args)
response = self.session.post(url, **kwargs)
if response.status_code != 200:
raise Exception(response.status_code)
return response
def step1(self, keyword: str) -> dict:
"""访问html页面获取参数"""
url = "http://wsjs.saic.gov.cn/txnRead01.do"
path = parse.urlparse(url).path
request_args = {
"locale": "zh_CN", # 语言
"request:queryCom": "1", # 不明
"request:nc": "", # 国际分类
"request:sn": "", # 申请/注册号
"request:mn": keyword, # 商标名称
"request:hnc": "", # 申请人名称(中文)
"request:hne": "", # 申请人名称(英文)
"request:md5": None, # md5签名
}
request_args["request:md5"] = self.get_md5(request_args)
# 本地加密 TODO 已失效
# kwargs = self.local_encrypt(path=path, request_args=request_args)
# 在线加密
kwargs = online_encrypt(url=API, path=path, request_args=request_args)
response = self.session.post(url, **kwargs)
if response.status_code != 200:
raise Exception(response.status_code)
html_doc = html.fromstring(response.content)
if html_doc.xpath("//title/text()")[0] == "请继续":
raise Exception("出现验证码") # 换IP
# 提取后续请求所需参数
meta = html_doc.xpath("//meta")[3].get("content")
# 得到input隐藏域参数
input_tags = ctx.call("get_hidden_input_v2", meta)
html_args = {
tag.get("name"): tag.get("value")
for tag in html.fromstring(input_tags).xpath("//input")
}
print(html_args)
return html_args
def step1(self, keyword: str) -> dict:
"""访问html页面获取参数"""
url = "http://wsjs.saic.gov.cn/txnRead01.do"
path = parse.urlparse(url).path
request_args = {
"locale": "zh_CN", # 语言
"request:queryCom": "1", # 不明
"request:nc": "", # 国际分类
"request:sn": "", # 申请/注册号
"request:mn": keyword, # 商标名称
"request:hnc": "", # 申请人名称(中文)
"request:hne": "", # 申请人名称(英文)
"request:md5": None, # md5签名
}
request_args["request:md5"] = ListPageExample.get_md5(request_args)
# 本地加密 TODO 已失效
# kwargs = self.local_encrypt(path=path, request_args=request_args)
# 在线加密
kwargs = online_encrypt(url=API, path=path, request_args=request_args)
response = self.session.post(url, **kwargs)
if response.status_code != 200:
raise Exception(response.status_code)
# 解析meta标签(9DhefwqGPrzGxEp9hPaoag)
meta = html.fromstring(response.content).xpath(
"//*[@id='9DhefwqGPrzGxEp9hPaoag']")[0].get("content")
# 得到input隐藏域参数
input_tags = ctx.call("get_hidden_input", meta)
html_args = {
tag.get("name"): tag.get("value")
for tag in html.fromstring(input_tags).xpath("//input")
}
print(html_args)
return html_args