|
@@ -230,6 +230,21 @@ class BaseBusinessListSpider(Spider):
|
|
|
|
|
|
__business_type__ = "List"
|
|
|
|
|
|
+ def __auto_increment_page_number(self, request):
|
|
|
+ """翻页 - 页码自增"""
|
|
|
+ if request.page < int(request.item["crawl_page"]):
|
|
|
+ request.page += 1 # 采集页码自增
|
|
|
+ yield request
|
|
|
+
|
|
|
+ def infinite_pages(self, request, response):
|
|
|
+ """翻页"""
|
|
|
+ request_generator = self.__auto_increment_page_number(request)
|
|
|
+ try:
|
|
|
+ request = next(request_generator)
|
|
|
+ return request
|
|
|
+ except StopIteration:
|
|
|
+ pass
|
|
|
+
|
|
|
|
|
|
class MixBusinessSpider(BaseBusinessListSpider):
|
|
|
"""混采(列表页+详情页)采集基础爬虫"""
|