class BasicArticleCrawler(object):
def __init__(self):
self.mysql_client = MysqlClient()
self.mongo_client = MongoClient().tdb.tcoll
def insertSuccess(self, msg):
"""
success crawle the article msg, insert into the successed db, insert into mongodb
"""
try:
self.mysql_client.begin()
# print article
# print msg["url"]
article = self.mysql_client.getOne(
"select * from failed_url where url=%s", (msg["url"], ))
if article != False:
article = self.mysql_client.delete(
"delete from failed_url where url=%s", (msg["url"], ))
LOGGER.info("delete the article from failed_url: %s",
msg["url"])
article = self.mysql_client.getOne(
"select * from successed_url where url=%s", (msg["url"], ))
if article != False:
LOGGER.info("repeat crawler the article give up save: %s",
msg["url"])
return
self.mongo_client.save(msg)
LOGGER.debug("insert into mongo: %s@%s" %
(msg["title"], msg["url"]))
self.mysql_client.insertOne("insert into successed_url(url, tag, sub_tag, version, create_time) values(%s, %s, %s, %s, %s)", \
(msg["url"], msg["tag"], msg["sub_tag"], VERSION, msg["create_time"]))
LOGGER.debug("insert successed_url %s" % (msg["url"], ))
self.mysql_client.end("commit")
except Exception, e:
self.mysql_client.end("rollback")
self.mysql_client.begin()
self.insertFailed(msg)
LOGGER.error("insert into mongo/successed_url error: %s" %
(msg["url"]))
LOGGER.error(traceback.format_exc())
class BasicArticleCrawler(object):
def __init__(self):
self.mysql_client = MysqlClient()
self.mongo_client = MongoClient().tdb.tcoll
def insertSuccess(self, msg):
"""
success crawle the article msg, insert into the successed db, insert into mongodb
"""
try:
self.mysql_client.begin()
# print article
# print msg["url"]
article = self.mysql_client.getOne("select * from failed_url where url=%s", (msg["url"], ))
if article != False:
article = self.mysql_client.delete("delete from failed_url where url=%s", (msg["url"], ))
LOGGER.info("delete the article from failed_url: %s", msg["url"])
article = self.mysql_client.getOne("select * from successed_url where url=%s", (msg["url"], ))
if article != False:
LOGGER.info("repeat crawler the article give up save: %s", msg["url"])
return
self.mongo_client.save(msg)
LOGGER.debug("insert into mongo: %s@%s" %(msg["title"], msg["url"]))
self.mysql_client.insertOne("insert into successed_url(url, tag, sub_tag, version, create_time) values(%s, %s, %s, %s, %s)", \
(msg["url"], msg["tag"], msg["sub_tag"], VERSION, msg["create_time"]));
LOGGER.debug("insert successed_url %s" %(msg["url"], ))
self.mysql_client.end("commit")
except Exception, e:
self.mysql_client.end("rollback")
self.mysql_client.begin()
self.insertFailed(msg)
LOGGER.error("insert into mongo/successed_url error: %s" %(msg["url"]))
LOGGER.error(traceback.format_exc())
def __init__(self):
self.mysql_client = MysqlClient()
self.mongo_client = MongoClient().tdb.tcoll
def __init__(self):
self.mysql_client = MysqlClient()
self.mongo_client = MongoClient().tdb.tcoll
@author: lml
'''
import sys
sys.path.append("../")
sys.path.append("../../")
sys.path.append("/home/lml/webcrawler/webcrawler-nlp/crawler/")
from utils.dbmong import MongoClient
from utils.dbmysql import MysqlClient
from time import sleep
if __name__ == '__main__':
while True:
print "**************************************************"
mysql_client = MysqlClient()
mongo_client = MongoClient()
published_url_count = mysql_client.getOne(
"select count(*) as count from published_url")
print "published url count: %s" % published_url_count["count"]
successed_url_count = mysql_client.getOne(
"select count(*) as count from successed_url")
print "successed url count: %s" % successed_url_count["count"]
failed_url_count = mysql_client.getOne(
"select count(*) as count from failed_url")
print "failed url count: %s" % failed_url_count["count"]
count = mongo_client.tdb.tcoll.count()
print "mongo articles: %s" % count
sleep(10)