处理图片替换

zhihu_login.py

@@ -239,6 +239,30 @@ class ZhihuAccount(object):
             print(item["content"])
             print(50*"*")
 
+
+    def _dispose_content_url(self,content,img_url_list,cur_image_index):
+        try:
+            for ori_img_url in img_url_list:
+                if ori_img_url.find(".jpg") >= 0 or ori_img_url.find(".png") >= 0:
+                    cur_image_index += 1
+
+                    local_img_url_path = "./image/img_" + str(cur_image_index) + ".png"
+                    print(ori_img_url, local_img_url_path)
+
+                    urlretrieve(ori_img_url, local_img_url_path)
+                    local_cv2_img = cv2.imread(local_img_url_path)
+                    height, weidth, channel = local_cv2_img.shape
+                    local_cropped_img = local_cv2_img[0:(height - 100), 0:weidth]
+                    local_cropped_img_url_path = "./image/cropped_image_" + str(cur_image_index) + ".png"
+                    cv2.imwrite(local_cropped_img_url_path, local_cropped_img)
+                    qiniu_url = upload_file(local_cropped_img_url_path)
+                    content = content.replace(ori_img_url, qiniu_url)
+
+            return content
+        except:
+            print(traceback.format_exc())
+            return content
+
     # 知乎搜索词搜索
     def zhihu_query_by_word(self,query_word,zhihu_spider_fd,cur_image_index):
 
@@ -265,21 +289,11 @@ class ZhihuAccount(object):
                             user_id = random.choice(majia_user_list)
                             question_id = ""
 
-                            img_url_list = re.findall('img src="(.*?)"', content)
-                            for ori_img_url in img_url_list:
-                                cur_image_index += 1
-
-                                local_img_url_path = "./image/img_" + str(cur_image_index) + ".png"
-                                print(ori_img_url,local_img_url_path)
+                            img_url_list = re.findall('src="(.*?)"', content)
+                            content = self._dispose_content_url(content=content,img_url_list=img_url_list,cur_image_index=cur_image_index)
 
-                                urlretrieve(ori_img_url, local_img_url_path)
-                                local_cv2_img = cv2.imread(local_img_url_path)
-                                height, weidth, channel = local_cv2_img.shape
-                                local_cropped_img = local_cv2_img[0:(height - 100), 0:weidth]
-                                local_cropped_img_url_path = "./image/cropped_image_" + str(cur_image_index) + ".png"
-                                cv2.imwrite(local_cropped_img_url_path,local_cropped_img)
-                                qiniu_url = upload_file(local_cropped_img_url_path)
-                                content = content.replace(ori_img_url,qiniu_url)
+                            img_url_list = re.findall('data-original="(.*?)"', content)
+                            content = self._dispose_content_url(content=content,img_url_list=img_url_list,cur_image_index=cur_image_index)