From ebb49ce04ca24979c7bb7db437ddbeaf433f4ca6 Mon Sep 17 00:00:00 2001 From: caiwx86 Date: Thu, 8 Dec 2022 01:03:58 +0800 Subject: [PATCH] fixt --- utils/HtmlUtils.py | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) diff --git a/utils/HtmlUtils.py b/utils/HtmlUtils.py index cc93759..5d3b06b 100644 --- a/utils/HtmlUtils.py +++ b/utils/HtmlUtils.py @@ -11,7 +11,7 @@ class htmlUtils: url_data = {} @classmethod - def getHTML(cls, curl): + def getHTML(cls, curl,type=None): rstr = r"[\/\\\:\*\?\"\<\>\|\.]" #  '/ \ : * ? " < > |' #file_url = re.sub(rstr, "_", curl) keys = cls.url_data.keys() @@ -29,7 +29,10 @@ class htmlUtils: try: print("请求地址:",curl) res = s.get(curl,stream=True, headers=cls.headers, timeout=20) - url_text = html.fromstring(res.text) + if type == "json": + url_text = res.json() + else: + url_text = html.fromstring(res.text) except: print(f'Retry! 第{count}次') time.sleep(3) @@ -43,15 +46,9 @@ class htmlUtils: cls.url_data.update(data) return url_text - @classmethod - def getBytes(cls, url): - return requests.get(url, headers=cls.headers,stream=True,timeout=30) - @classmethod def getJSON(cls,curl): - res = requests.get(curl, headers=cls.headers,timeout=30) - data_json = res.json() - return data_json + return cls.getHTML(curl,type="json") @classmethod def xpathData(cls,c_xpath,url=None,num=None,not_eq=None):