def _process_item(self, item, spider):
cols,vals,key = self.item_key(item, spider)
print cols
print vals
print key
mutations = [Mutation(column=col, value=val) for col,val in zip(cols,vals)]
self.client.mutateRow(self.tableName,confUtil.getMd5(key),mutations,None)
return item
def hbase_tables(self):
tables = self.client.getTableNames()
print tables
cols =['detail:publish_time', 'detail:site_source', 'detail:site_type', 'detail:site_url', 'detail:task_id',
'detail:author',
'detail:catch_date'
]
vals = ['2015-03-10 02:39', 'news.sina.com.cn', 'news', 'http://news.sina.com.cn/c/2015-03-10/023931587440.shtml', '-1',
u'\u4eac\u534e\u65f6\u62a5'.encode("utf-8"),
'2015-03-27'
]
key = "http://news.sina.com.cn/c/2015-03-10/023931587440.shtml"
print confUtil.getMd5(key)
mutations = [Mutation(column=col, value=val) for col,val in zip(cols,vals)]
self.client.mutateRow(self.tableName,confUtil.getMd5(key),mutations,None)
def _process_item(self, item, spider):
cols, vals, key = self.item_key(item, spider)
print cols
print vals
print key
mutations = [
Mutation(column=col, value=val) for col, val in zip(cols, vals)
]
self.client.mutateRow(self.tableName, confUtil.getMd5(key), mutations,
None)
return item
def hbase_tables(self):
tables = self.client.getTableNames()
print tables
cols = [
'detail:publish_time', 'detail:site_source', 'detail:site_type',
'detail:site_url', 'detail:task_id', 'detail:author',
'detail:catch_date'
]
vals = [
'2015-03-10 02:39', 'news.sina.com.cn', 'news',
'http://news.sina.com.cn/c/2015-03-10/023931587440.shtml', '-1',
u'\u4eac\u534e\u65f6\u62a5'.encode("utf-8"), '2015-03-27'
]
key = "http://news.sina.com.cn/c/2015-03-10/023931587440.shtml"
print confUtil.getMd5(key)
mutations = [
Mutation(column=col, value=val) for col, val in zip(cols, vals)
]
self.client.mutateRow(self.tableName, confUtil.getMd5(key), mutations,
None)
