def __init__(self):
self.news = set()
self.updatedNews = set()
self.newsList = []
f = open(FILE_NAME)
for line in f:
self.news.add(line[0:-1])
f.close()
self.file = open(FILE_NAME, "wb")
for s in NEWSCATALOG["data"][1]["list"]:
newdb.update({'newCatalog': s, 'newsList': []})
def __init__(self):
self.news = set()
self.updatedNews = set()
self.newsList = []
f = open(FILE_NAME)
for line in f:
self.news.add(line[0:-1])
f.close()
self.file = open(FILE_NAME, "wb")
for s in NEWSCATALOG["data"][1]["list"]:
newdb.update({'newCatalog': s, 'newsList': []})
def process_item(self, item, spider):
self.updatedNews.add(json.dumps(dict(item), ensure_ascii=False, sort_keys=True).encode('utf-8'))
line = json.dumps(dict(item), ensure_ascii=False, sort_keys=True).encode('utf-8') + "\n"
if self.currentCatalog != item["newCatalog"] and self.currentCatalog != "":
newdb.update({'newCatalog': self.currentCatalog, 'newsList': self.currentCatalogList})
self.currentCatalogList = []
else :
self.currentCatalogList.append(dict(item))
self.currentCatalog = item["newCatalog"]
self.file.write(line)
return item