From e7315e3ffa5d5c86b57c2e0afbbfa491e4949b00 Mon Sep 17 00:00:00 2001 From: Mathhew Date: Mon, 13 Jun 2022 10:01:03 +0800 Subject: [PATCH] fix getchu headers extrafanart --- scrapinglib/getchu.py | 21 +++++++++++++++++++++ scrapinglib/parser.py | 6 ++++++ 2 files changed, 27 insertions(+) diff --git a/scrapinglib/getchu.py b/scrapinglib/getchu.py index b29f4d8..f64b010 100644 --- a/scrapinglib/getchu.py +++ b/scrapinglib/getchu.py @@ -88,7 +88,17 @@ class wwwGetchu(Parser): extrafanart.append(i) return extrafanart + def extradict(self, dic: dict): + """ 额外新增的 headers + """ + dic['headers'] = {'referer': self.detailurl} + return dic + class dlGetchu(wwwGetchu): + """ 二者基本一致 + headers extrafanart 略有区别 + """ + imagecut = 4 allow_number_change = True @@ -127,3 +137,14 @@ class dlGetchu(wwwGetchu): def getCover(self, htmltree): return "https://dl.getchu.com" + super().getCover(htmltree) + + def extradict(self, dic: dict): + return dic + + def getExtrafanart(self, htmltree): + arts = self.getTreeAll(htmltree, self.expr_extrafanart) + extrafanart = [] + for i in arts: + i = "https://dl.getchu.com" + i + extrafanart.append(i) + return extrafanart diff --git a/scrapinglib/parser.py b/scrapinglib/parser.py index c052836..830c534 100644 --- a/scrapinglib/parser.py +++ b/scrapinglib/parser.py @@ -135,12 +135,18 @@ class Parser: 'userrating': self.getUserRating(htmltree), 'uservotes': self.getUserVotes(htmltree) } + dic = self.extradict(dic) except Exception as e: print(e) dic = {"title": ""} js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'), ) return js + def extradict(self, dic:dict): + """ 额外修改dict + """ + return dic + def getNum(self, htmltree): """ 增加 strip 过滤 """