def __init__(self, cur_idepth, max_idepth, \
cur_xdepth, max_xdepth, content_group, \
pl_group, source, url):
self.cur_idepth = cur_idepth
self.max_idepth = max_idepth
self.cur_xdepth = cur_xdepth
self.max_xdepth = max_xdepth
self.content_group = content_group
self.pl_group = pl_group
self.source = source
self.url = url
self.uid = get_uid(url)
self.domain = get_domain(url)
self.host = urlparse(self.url).hostname
def put(self, url, name, cat, price, collection_name=None):
uid = get_uid(url)
domain = get_domain(url)
crawl_time = int(time.time())
item = self.get(url, collection_name)
if item:
if item.add_price(price, crawl_time):
self.dbclient.update_field(uid, collection_name, data=item.data, bottom_price=item.bottom_price)
else:
log.msg("duplicate price")
return
else:
item = {
"url": url,
"uid": uid,
"name": name,
"cat": cat,
"data": [(price, crawl_time)],
"bottom_price": (price, crawl_time),
"domain": domain,
}
self.dbclient.insert(item, uid, collection_name)
send_catch_log(signal=signals.item_saved, item=item)
def save_extract_info(self, url, item_num):
if item_num > 0:
domain = get_domain(url)
self.redis.hincrby(self.genkey(self.EXTRACT_ITEM_OK), domain, item_num)
else:
self.redis.sadd(self.genkey(self.EXTRACT_ITEM_FAIL), url)
