add okmovieSpider/main.py

kangvcar · Dec 26, 2018 · 7147a4c · 7147a4c
1 parent 71bfb66
commit 7147a4c
Show file tree

Hide file tree

Showing 4 changed files with 99 additions and 63 deletions.
diff --git a/KoreaGirl/.idea/encodings.xml b/KoreaGirl/.idea/encodings.xml
diff --git a/KoreaGirl/.idea/vcs.xml b/KoreaGirl/.idea/vcs.xml
diff --git a/KoreaGirl/.idea/workspace.xml b/KoreaGirl/.idea/workspace.xml
diff --git a/okmovieSpider/main.py b/okmovieSpider/main.py
@@ -42,7 +42,7 @@ def getIndexPages(self, url, sumpage):
         '''获取需要爬取多少个列表页'''
         pageurls = []
         ul = "-".join(url.split('-')[:-1])
-        for page in range(6, sumpage+1):
+        for page in range(1, sumpage+1):
             pageurls.append(ul + "-" + str(page) + ".html")
         return pageurls
 
@@ -144,10 +144,10 @@ def getSourceCode(self, url):
 
 
 # ISO-8859-1
-ok = okSpider('http://www.okzy.co/?m=vod-index-pg-2.html', 10)
+ok = okSpider('http://www.okzy.co/?m=vod-index-pg-2.html', 2)
 # ok = okSpider('http://www.okzy.co/?m=vod-type-id-22-pg-1.html', 1)
 
 # url = 'http://www.okzy.co/?m=vod-type-id-1-pg-1.html'
 # url = 'http://www.okzy.co/?m=vod-detail-id-22396.html'
 # print(requests.get(url).encoding)
-# print("-".join(url.split('-')[:-1]))
+# print("-".join(url.split('-')[:-1]))