check uri not start with /

2021-06-06 04:57:24 +08:00
parent 5abeb360af
commit e2cd1f09df
1 changed files with 4 additions and 1 deletions
--- a/WebCrawler/javbus.py
+++ b/WebCrawler/javbus.py
@@ -49,7 +49,10 @@ def getCover(htmlcode):  #获取封面链接
    image = doc('a.bigImage')
    uri = image.attr('href')
    if uri[0:4] != 'http':
-        return "https://www.javbus.com" + uri
+        if uri[0] == '/':
+            return "https://www.javbus.com" + uri
+        else:
+            return "https://www.javbus.com/" + uri
    return uri
 def getRelease(htmlcode): #获取出版日期
    html = etree.fromstring(htmlcode, etree.HTMLParser())