class Update_data():
def __init__(self,
db_name,
solr_name=SOLR_CORE_NAME,
mongodb_ip='127.0.0.1',
solr_ip='127.0.0.1',
port=27017):
self.db_name = db_name
self.db = MongoClient(mongodb_ip, port)[db_name]
self.solr_url = 'http://' + solr_ip + ':8999/solr'
self.solr_core = solr_name
self.solr = SOLR(self.solr_url)
def write_data2solr(self, collection_name):
query = 'scene_str:' + self.db_name + ' AND topic_str:' + collection_name
self.solr.delete_solr_by_query(self.solr_core, query)
for x in self.db[collection_name].find():
data_one = x.copy()
data_one['scene'] = self.db_name
data_one['topic'] = collection_name
data_one['_id'] = str(data_one['_id'])
if collection_name in ['refuse2chat', 'sentiment']:
self.solr.update_solr(data_one, self.solr_core)
continue
if 'super_intention' in data_one:
if data_one['super_intention'] == '':
data_one['super_intention'] = 'null'
data_one.pop('equal_questions')
for q in x['equal_questions']:
data_one['question'] = q
data_one['question_ik'] = q
data_one['question_cn'] = q
self.solr.update_solr(data_one, self.solr_core)
class BaseClass():
def __init__(self,
ip,
port,
db_name,
collection_name,
solr_name=SOLR_CORE_NAME):
self.dirpath = 'data/' + db_name
self.db_name = db_name
self.collection_name = collection_name
self.db = MongoClient(ip, port)[db_name]
self.collection = self.db[collection_name]
self.data = []
self.solr_url = 'http://' + ip + ':8999/solr'
self.solr_core = solr_name
self.solr = SOLR(self.solr_url)
def write_data2mongodb(self):
self.collection.drop()
self.collection.insert(self.data)
if self.collection_name in ['refuse2chat', 'sentiment']:
self.collection.create_index('question')
else:
self.collection.create_index('group')
self.collection.create_index('label')
def write_data2solr(self):
query = 'scene_str:'+self.db_name + ' AND topic_str:' +\
self.collection_name
self.solr.delete_solr_by_query(self.solr_core, query)
for x in self.collection.find():
data_one = x.copy()
data_one['scene'] = self.db_name
data_one['topic'] = self.collection_name
data_one['_id'] = str(data_one['_id'])
if self.collection_name in ['refuse2chat', 'sentiment']:
self.solr.update_solr(data_one, self.solr_core)
continue
if 'super_intention' in data_one:
if data_one['super_intention'] == '':
data_one['super_intention'] = 'null'
data_one.pop('equal_questions')
for q in x['equal_questions']:
data_one['question'] = q
data_one['question_ik'] = q
data_one['question_cn'] = q
self.solr.update_solr(data_one, self.solr_core)
class Mongodb():
def __init__(self, db_name, ip='127.0.0.1', port=27017):
self.db_name = db_name
self.db = MongoClient(ip, port)[db_name]
self.db_test = MongoClient(ip, port)[db_name + '_test']
self.solr_url = 'http://' + ip + ':8999/solr'
self.solr_core = SOLR_CORE_NAME
self.solr = SOLR(self.solr_url)
def write(self, collection, data):
try:
self.db[collection].drop()
self.db[collection].insert(data)
self.db_test[collection].drop()
self.db_test[collection].insert(data)
return 1
except:
traceback.print_exc()
return 0
def write_data2solr(self, collection):
query = 'scene_str:'+self.db_name+' AND topic_str:' +\
collection
self.solr.delete_solr_by_query(self.solr_core, query)
for x in self.db[collection].find():
data_one = x.copy()
data_one['scene'] = self.db_name
data_one['topic'] = collection
data_one['_id'] = str(data_one['_id'])
if collection in ['instruction']:
self.solr.update_solr(data_one, self.solr_core)
continue
if 'super_intention' in data_one:
if data_one['super_intention'] == '':
data_one['super_intention'] = 'null'
data_one.pop('questions')
for q in x['questions']:
data_one['question'] = q
data_one['question_ik'] = q
data_one['question_cn'] = q
self.solr.update_solr(data_one, self.solr_core)
class Update():
def __init__(self, ip, db_name):
self.db_name = db_name
self.db = MongoClient('127.0.0.1', 27017)[db_name]
self.core_name = SOLR_CORE_NAME
self.solr_url = 'http://127.0.0.1:8999/solr'
self.solr = SOLR(self.solr_url)
def check_solr_core(self):
if not self.solr.solr_core_exists(self.core_name):
self.solr.create_solr_core(self.core_name)
def update_data(self, collection):
def insert(data):
if not data:
return
data_one = data.copy()
data_one['_id'] = str(data_one['_id'])
data_one['scene'] = self.db_name
data_one['topic'] = collection
if 'super_intention' in data_one:
if data_one['super_intention'] == '':
data_one['super_intention'] = 'null'
if 'equal_questions' in data_one:
data_one.pop('equal_questions')
for q in data['equal_questions']:
data_one['question'] = q
data_one['question_ik'] = q
data_one['question_cn'] = q
self.solr.update_solr(data_one, self.core_name)
elif 'questions' in data_one:
data_one.pop('questions')
for q in data['questions']:
data_one['question'] = q
data_one['question_ik'] = q
data_one['question_cn'] = q
self.solr.update_solr(data_one, self.core_name)
else:
self.solr.update_solr(data_one, self.core_name)
self.solr.delete_solr_by_query(
self.core_name,
'scene_str:' + self.db_name + ' AND topic_str:' + collection)
data = [x for x in self.db[collection].find()]
for d in data:
insert(d)
def update(self):
try:
collections = self.db.collection_names()
if 'log' in collections:
collections.remove('log')
for collection in collections:
print('start ' + collection)
self.update_data(collection)
return 1
except Exception:
traceback.print_exc()
return 0
class Update():
def __init__(self, ip, db_name):
self.db_name = db_name
self.db = MongoClient('127.0.0.1', 27017)[db_name]
self.core_name = SOLR_CORE_NAME
self.solr_url = 'http://127.0.0.1:8999/solr'
self.solr = SOLR(self.solr_url)
def load_log(self, server_name):
#_id, collection, cmd, ids, comment, status, time
if server_name == 'develop':
query = {'status': '0'}
elif server_name == 'master':
query = {'status': '1'}
else:
query = {'status': '3'}
logs = [x for x in self.db.log.find(query).sort('time')]
return logs
def check_solr_core(self):
if not self.solr.solr_core_exists(self.core_name):
self.solr.create_solr_core(self.core_name)
def update_data(self, collection, cmd, _id):
def insert_automata(data, collection):
if collection in ['automata']:
questions = data['questions'].copy()
data.pop('questions')
for q in questions:
data['question'] = q
self.solr.update_solr(data, self.core_name)
elif collection in ['instruction']:
self.solr.update_solr(data, self.core_name)
else:
return None
def insert(collection, _id):
data = self.db[collection].find_one({'_id': ObjectId(_id)})
if not data:
return
data_one = data.copy()
data_one['_id'] = str(data_one['_id'])
data_one['scene'] = self.db_name
data_one['topic'] = collection
if self.db_name == 'automata':
return insert_automata(data_one)
if collection in ['refuse2chat', 'sentiment']:
self.solr.update_solr(data_one, self.core_name)
return None
if 'super_intention' in data_one:
if data_one['super_intention'] == '':
data_one['super_intention'] = 'null'
data_one.pop('equal_questions')
for q in data['equal_questions']:
data_one['question'] = q
data_one['question_ik'] = q
data_one['question_cn'] = q
self.solr.update_solr(data_one, self.core_name)
if cmd == 'create':
insert(collection, _id)
elif cmd == 'update':
self.solr.delete_solr_by_query(self.core_name, '_id_str:' + _id)
insert(collection, _id)
elif cmd == 'delete':
self.solr.delete_solr_by_query(self.core_name, '_id_str:' + _id)
else:
return 0
def update(self, server_name):
try:
logs = self.load_log(server_name)
if not logs:
print('no update!')
return 1
for log in logs:
if log['cmd'] == 'create':
self.check_solr_core()
for _id in log['ids']:
self.update_data(log['collection'], log['cmd'], _id)
if server_name == 'develop':
value = {'status': '1'}
elif server_name == 'master':
value = {'status': '2'}
else:
return 0
self.db.log.update_one({'_id': log['_id']}, {'$set': value})
return 1
except Exception:
traceback.print_exc()
return 0
class Data_backup():
def __init__(self, db_name):
self.db_name = db_name
self.client = MongoClient('127.0.0.1', 27017)
self.db = self.client[self.db_name]
self.solr_url = 'http://127.0.0.1:8999/solr'
self.solr = SOLR(self.solr_url)
self.core_name = SOLR_CORE_NAME
def data_dump(self, datapath, log_id):
if not os.path.exists(datapath):
os.mkdir(datapath)
dirpath = os.path.join(datapath, log_id)
if os.path.exists(dirpath):
shutil.rmtree(dirpath)
os.mkdir(dirpath)
cmd_dump = 'mongodump -d ' + self.db_name + ' -o ' + dirpath
try:
os.system(cmd_dump)
return 1
except:
traceback.print_exc()
return 0
def mongodb_restore(self, dirpath):
self.client.drop_database(self.db_name)
self.client.drop_database(self.db_name + '_test')
dbpath = os.path.join(dirpath, self.db_name)
cmd_restore1 = 'mongorestore -d ' + self.db_name + ' ' + dbpath
cmd_restore2 = 'mongorestore -d ' + self.db_name + '_test ' + dbpath
if os.system(cmd_restore1):
return 0
if os.system(cmd_restore2):
return 0
return 1
def solr_restore(self):
collections = self.db.collection_names()
if 'log' in collections:
collections.remove('log')
try:
for collection in collections:
query = '(scene_str:' + self.db_name + \
' AND topic_str:' + collection + ')'
self.solr.delete_solr_by_query(self.core_name, query)
for data in self.db[collection].find():
data_one = data.copy()
data_one['scene'] = self.db_name
data_one['topic'] = collection
data_one['_id'] = str(data_one['_id'])
if collection in ['refuse2chat', 'sentiment']:
self.solr.update_solr(data_one, self.core_name)
break
data_one.pop('equal_questions')
for q in data['equal_questions']:
data_one['question'] = q
self.solr.update_solr(data_one, self.core_name)
return 1
except:
traceback.print_exc()
return 0
def data_restore(self, dirpath, _id):
dirpath = os.path.join(dirpath, _id)
return self.mongodb_restore(dirpath) and self.solr_restore()