def get_LDA_doc():
ms = MSSSQL(host="172.28.4.193", user="******", pwd="wy9756784750", db="pweb")
resList = ms.ExeQuery("select wsname,summary,categoryid,tags,Description from dbo.ws")
ws = {}
for (wsname,summary,categoryid,tags,Description) in resList:
s1 = wsname.strip().lower().replace(' ','-')
ll = summary+' '+categoryid+' '+tags+' '+Description
ll = ll.replace('\n',' ')
ws[s1] = ll
mp = {}
resList = ms.ExeQuery("select wsname,summary,categoryid,tags,Description from dbo.mp")
for (wsname,summary,categoryid,tags,Description) in resList:
s1 = wsname.strip().lower().replace(' ','-')
ll = summary + ' ' + categoryid + ' ' + tags + ' ' + Description
ll = ll.replace('\n', ' ')
ws[s1] = ll
input = open('webservice.txt')
output = open('LDA_train.txt','w')
cc = 0
for line in input:
s = line.strip('\r\n').split(' ')
cc+=1
if ws.has_key(s[0]):
output.write(ws[s[0]]+'\r\n')
elif mp.has_key(s[0]):
output.write(mp[s[0]]+'\r\n')
else:
output.write(s[0]+'\r\n')
print cc
input.close()
output.close()
def test():
t = write_api_mashupToTxt()
ms = MSSSQL(host="172.28.4.193", user="******", pwd="wy9756784750", db="pweb")
resList = ms.ExeQuery("select * from dbo.watchlist")
for (wsname, url, id, date) in resList:
str_array = url.split('/')
if str_array[-1] in t:
print wsname, url
def xiuzheng():
output = open('name_replace_1.txt', 'w')
t = write_api_mashupToTxt()
print len(t)
ms = MSSSQL(host="172.28.4.193", user="******", pwd="wy9756784750", db="pweb")
resList = ms.ExeQuery("select wsname,APIhome from dbo.mp")
for (wsname, APIhome) in resList:
s = APIhome.strip().split('/')
if s[-2] in t:
output.write(wsname.lower().strip().replace(' ', '-') + ' ' +
s[-2] + ' ' + '1' + '\r\n')
t.remove(s[-2])
print len(t)
print t
def get_user():
ms = MSSSQL(host="172.28.4.193", user="******", pwd="wy9756784750", db="pweb")
resList = ms.ExeQuery("select * from dbo.watchlist")
user_name = {}
for wsname, url, id, date in resList:
if user_name.has_key(wsname):
user_name[wsname] += 1
else:
user_name[wsname] = 1
s_name = []
count = 0
for key, value in user_name.items():
if value > 50:
##print key,value
s_name.append(key)
count += 1
##print count
##print len(s_name)
return s_name
def count_api_mashup():
ms = MSSSQL(host="172.28.4.193", user="******", pwd="wy9756784750", db="pweb")
resList = ms.ExeQuery("select wsname from dbo.ws")
apiAll = []
mashupAll = []
##output1 = open('api.txt','w')
##output2 = open('mashup.txt','w')
for (wsname, ) in resList:
s = wsname.strip().replace(' ', '-')
apiAll.append(s.upper())
##output1.write(s.upper()+'\r\n')
resList1 = ms.ExeQuery("select wsname from dbo.mp")
for (wsname, ) in resList1:
s = wsname.strip().replace(' ', '-')
mashupAll.append(s.upper())
##output2.write(s.upper()+'\r\n')
return apiAll, mashupAll
set_time = {}
c = 0
for line in input3:
s = line.strip('\r\n').split(' ')
t = d + datetime.timedelta(seconds=int(s[2]))
ut = (t-t_user[s[0]]).days
if ut<0:
ut = 0
'''
input5 = open('name_replace.txt')
rep = {}
for line in input5:
s = line.strip('\r\n').split(' ')
rep[s[1]] = s[0]
ms = MSSSQL(host="172.28.37.29", user="******", pwd="wy9756784750", db="pweb")
resList = ms.ExeQuery("select * from dbo.watchlist")
c = 0
for wsname,url,id,date in resList:
str_array = url.split('/')
url = str_array[-1]
url = url.replace('%2Fcomments', '')
if rep.has_key(url):
url = rep[url]
t = transfer_time(date)
if t_s.has_key(url):
if t<t_s[url]:
t_s[url] = t
if t_user.has_key(wsname):
if t<t_user[wsname]:
t_user[wsname] = t