def process():
conn_old = mysql.connect('bsppr', '192.168.241.7')
mysql.insert(conn_old, 'set names utf8')
conn_new = mysql.connect('bsppr', '192.168.241.32')
mysql.insert(conn_new, 'set names utf8')
cinfos_old = get_cinfos(conn_old)
#print cinfos_old
cinfos_new = get_cinfos(conn_new)
mongo_conn = get_mongo_conn()
tablename = 'weixin'
tmpdatas = mongo.find(mongo_conn, tablename, {}, 50)
rawdatas = []
for raw in tmpdatas:
date = raw['pubtime']
now = datetime.datetime.now()
diff = now - date
print diff.days
rawdatas.append(raw)
if len(rawdatas) == 0:
time.sleep(10)
raw_old_qualified = filter(cinfos_old, rawdatas)
old_insert_num = feed_xpost.feed_data_to_xpost(conn_old, raw_old_qualified,
'old')
raw_new_qualified = filter(cinfos_new, rawdatas)
new_insert_num = feed_xpost.feed_data_to_xpost(conn_new, raw_new_qualified,
'new')
def process():
conn_old = mysql.connect('bsppr', '192.168.241.7')
mysql.insert(conn_old,'set names utf8')
conn_new = mysql.connect('bsppr', '192.168.241.32')
mysql.insert(conn_new,'set names utf8')
cinfos_old = get_cinfos_moa(conn_old)
cinfos_new = get_cinfos_moa(conn_new)
mongo_conn = get_mongo_conn()
tablename = 'weixin'
while True:
tmpdatas = mongo.find(mongo_conn, tablename, {},1000)
rawdatas = []
for raw in tmpdatas:
url = raw['url']
mongo.delete(mongo_conn, tablename, {'url':url})
date = raw['pubtime']
now = datetime.datetime.now()
diff = now - date
if diff.days>2:
continue
rawdatas.append(raw)
if len(rawdatas)==0:
print 'wait datas...'
time.sleep(300)
raw_old_qualified = filter(cinfos_old,rawdatas)
if raw_old_qualified:
old_insert_num = feed_xpost.feed_data_to_xpost(conn_old, raw_old_qualified,'old')
print 'old_insert_num: ',old_insert_num
raw_new_qualified = filter(cinfos_new,rawdatas)
if raw_new_qualified:
new_insert_num = feed_xpost.feed_data_to_xpost(conn_new, raw_new_qualified,'new')
print 'new_insert_num: ',new_insert_num
def __init__(self):
self.conn = mysql.connect('mx_kol')
self.flag = 1
timetag = datetime.datetime.now().strftime('%Y%m%d%H%M%S')
self.f = open(
'%s/kol_search_%s_%s.dat' %
(self.data_path, self.filetag, timetag), 'w')
self.total = 0
def connect(self):
self.conn = mysql.connect('app_crawler',host='192.168.241.17')
self.cursor = self.conn.cursor()
self.conn.set_character_set('utf8')
self.cursor.execute('set names utf8mb4')
self.cursor.execute('SET CHARACTER SET utf8;')
self.cursor.execute('SET character_set_connection=utf8;')
self.cursor.execute('set interactive_timeout=24*3600;')
self.cursor.execute('set wait_timeout=24*3600;')
def connect(self):
self.conn = mysql.connect('app_crawler', host='192.168.241.17')
self.cursor = self.conn.cursor()
self.conn.set_character_set('utf8')
self.cursor.execute('set names utf8mb4')
self.cursor.execute('SET CHARACTER SET utf8;')
self.cursor.execute('SET character_set_connection=utf8;')
self.cursor.execute('set interactive_timeout=24*3600;')
self.cursor.execute('set wait_timeout=24*3600;')
def process(terrace):
if terrace=='new':
conn = mysql.connect('bsppr', '192.168.241.32')
else:
conn = mysql.connect('bsppr', '192.168.241.7')
mysql.insert(conn,'set names utf8')
mongo_conn = get_mongo_conn()
tablename = 'comgeneral'
while True:
total = 0
before_total = 0
raw_total = 0
tmpdatas = mongo.find(mongo_conn, tablename, {'terrace':terrace},200)
print '...%s get data ...'%(terrace)
rawdatas = []
for raw in tmpdatas:
raw_total += 1
url = raw['url']
mongo.delete(mongo_conn, tablename, {'url':url})
datestr = raw['pubtime']
if not datestr:
continue
updatetime = raw.get('updatetime',time.time)
try:
date = format_time(datestr,updatetime)
#open(self.terrace,'a+').write('%s\n'%str(date))
now = datetime.datetime.now()
diff = now - date
except:
open('date_error.dat','a+').write('%s\t%s\n'%(datestr,url))
continue
if diff.days>2:
continue
raw.update({'pubtime':date})
rawdatas.append(raw)
before_total += 1
if tmpdatas.count()==0:
break
insert_num = feed_xpost.feed_data_to_xpost(conn, rawdatas,terrace)
total += insert_num
open('total_%s.dat'%(terrace),'a+').write('%s\t%s\n'%(datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'),total))
open('raw_total_%s.dat'%(terrace),'a+').write('%s\t%s\n'%(datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'),raw_total))
open('before_total_%s.dat'%(terrace),'a+').write('%s\t%s\n'%(datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'),before_total))
def __init__(self):
self.conn = mysql.connect('coopinion',host='192.168.241.31')
self.cursor = self.conn.cursor()
self.conn.set_character_set('utf8')
self.cursor.execute('set names utf8mb4')
self.cursor.execute('SET CHARACTER SET utf8;')
self.cursor.execute('SET character_set_connection=utf8;')
self.cursor.execute('set interactive_timeout=24*3600;')
self.cursor.execute('set wait_timeout=24*3600;')
self.conn_weixin = ''
self.conn_local = mysql.connect('meta',host='localhost')
self.conn_local_cursor = self.conn_local.cursor()
self.conn_local.set_character_set('utf8')
self.conn_local_cursor.execute('set names utf8mb4')
self.conn_local_cursor.execute('SET CHARACTER SET utf8;')
self.conn_local_cursor.execute('SET character_set_connection=utf8;')
self.conn_local_cursor.execute('set interactive_timeout=24*3600;')
self.conn_local_cursor.execute('set wait_timeout=24*3600;')
self.conn_local_cursor.execute('set innodb_lock_wait_timeout=1000;')
# self.conn_local_cursor.execute('set global autocommit=1')
try:
self.meta_ip = get_meta_ip(network_card='enp7s0')
except:
self.meta_ip = get_meta_ip(network_card='eth0')
def process():
conn_old = mysql.connect('bsppr', '192.168.241.7')
mysql.insert(conn_old,'set names utf8')
conn_new = mysql.connect('bsppr', '192.168.241.32')
mysql.insert(conn_new,'set names utf8')
cinfos_old = get_cinfos(conn_old)
#print cinfos_old
cinfos_new = get_cinfos(conn_new)
mongo_conn = get_mongo_conn()
tablename = 'weixin'
tmpdatas = mongo.find(mongo_conn, tablename, {},50)
rawdatas = []
for raw in tmpdatas:
date = raw['pubtime']
now = datetime.datetime.now()
diff = now - date
print diff.days
rawdatas.append(raw)
if len(rawdatas)==0:
time.sleep(10)
raw_old_qualified = filter(cinfos_old,rawdatas)
old_insert_num = feed_xpost.feed_data_to_xpost(conn_old, raw_old_qualified,'old')
raw_new_qualified = filter(cinfos_new,rawdatas)
new_insert_num = feed_xpost.feed_data_to_xpost(conn_new, raw_new_qualified,'new')
def mx_kol(self,result):
if not self.conn_weixin:
self.conn_weixin = mysql.connect('mx_kol', host='192.168.241.29')
self.weixin_cousor = self.conn_weixin.cursor()
self.conn_weixin.set_character_set('utf8')
self.weixin_cousor.execute('set names utf8mb4')
self.weixin_cousor.execute('SET CHARACTER SET utf8;')
self.weixin_cousor.execute('SET character_set_connection=utf8;')
openid = process_openid(result['id'])
url = result['url']
author = result.get('author','')
tencent_id = get_tencent_id(url)
is_v = 0
sql = "insert into weixin_user_info(userid,openid,tencent_id,screen_name,is_v) values('%s','%s','%s','%s',%d) on duplicate key update screen_name='%s';"
sql = sql%(openid,openid,tencent_id,author,is_v,author)
self.weixin_cousor.execute(sql)
self.conn_weixin.commit()
def fetch_query_results(self, query_string, num=200):
index_path = "/disk1/kol_search_index/index"
query_index = QueryEnvironment()
query_index.addIndex(index_path)
# 根据query_string查询结果
# print query_string
docs = query_index.runQuery(query_string, num)
# 解析查询的结果
results = get_query_results(query_index, docs)
datas = {}
flag = 0
conn = ""
now = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
flag = 0
for result in results:
if flag >= 200:
break
data = {}
userid = result["userid"]
site = result["site"]
relevance = process_relevance(result["relevance"])
data.update({"userid": userid})
data.update({"site": site})
data.update({"relevance": relevance})
datas.update({flag: data})
flag += 1
if datas:
conn = mysql.connect("kol_search")
results = json.dumps(datas)
results = conn.escape_string(results)
query = str(self.keyword) + "#$#" + str(self.site)
sql = "insert into search_result_cache(query,result,update_time) values('%s','%s','%s');" % (
query,
results,
now,
)
# print sql
mysql.insert(conn, sql)
mysql.commit(conn)
mysql.close(conn)
query_index.close()
return datas
def fetch_query_results(self, query_string, num=200):
index_path = '/disk1/kol_search_index/index'
query_index = QueryEnvironment()
query_index.addIndex(index_path)
#根据query_string查询结果
#print query_string
docs = query_index.runQuery(query_string, num)
#解析查询的结果
results = get_query_results(query_index, docs)
datas = {}
flag = 0
conn = ''
now = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
flag = 0
for result in results:
if flag >= 200:
break
data = {}
userid = result['userid']
site = result['site']
relevance = process_relevance(result['relevance'])
data.update({'userid': userid})
data.update({'site': site})
data.update({'relevance': relevance})
datas.update({flag: data})
flag += 1
if datas:
conn = mysql.connect('kol_search')
results = json.dumps(datas)
results = conn.escape_string(results)
query = str(self.keyword) + '#$#' + str(self.site)
sql = "insert into search_result_cache(query,result,update_time) values('%s','%s','%s');" % (
query, results, now)
#print sql
mysql.insert(conn, sql)
mysql.commit(conn)
mysql.close(conn)
query_index.close()
return datas
def get_conn():
conn_new = mysql.connect('bsppr', '192.168.241.32')
conn_old = mysql.connect('bsppr', '192.168.241.7')
return conn_new, conn_old
def get_conn():
conn_new = mysql.connect('bsppr', '192.168.241.32')
conn_old = mysql.connect('bsppr', '192.168.241.7')
return conn_new,conn_old
import mysql_api as mysql
conn_local = mysql.connect('meta',host='localhost')
conn_local_cursor = conn_local.cursor()
conn_local.set_character_set('utf8')
conn_local_cursor.execute('set names utf8mb4')
conn_local_cursor.execute('SET CHARACTER SET utf8;')
conn_local_cursor.execute('SET character_set_connection=utf8;')
conn_local_cursor.execute('set interactive_timeout=24*3600;')
conn_local_cursor.execute('set wait_timeout=24*3600;')
conn_local_cursor.execute('set innodb_lock_wait_timeout=1000;')
sql = 'select * from meta_result limit 1;'
conn_local_cursor.execute(sql)
print conn_local_cursor.fetchall()
conn_local.close()
def __init__(self):
self.conn = mysql.connect('mx_kol')
self.flag = 1
timetag = datetime.datetime.now().strftime('%Y%m%d%H%M%S')
self.f = open('%s/kol_search_%s_%s.dat'%(self.data_path,self.filetag,timetag),'w')
self.total = 0
def get_appnameid(appname):
sql = 'select id from appname_maps where e_appname="%s";'%(appname)
conn = mysql.connect('app_crawler', '192.168.241.17')
id = mysql.query_one(conn, sql)[0]
mysql.close(conn)
return id
