def __init__(self,fb,usr_id,queue,lockname):
threading.Thread.__init__(self)
self.fb = fb
self.usr_id = usr_id
self.fbsql = fb_mysql()
self.queue = queue
self.lock = lockname
def sharefan_num(page_name,limitnum,hot_fan_defind,percent=''):
import datetime
fbsql = fb_mysql()
msgdict = {}
onehoursdict = {}
hotfan = []
if page_name == 'all':
alllist = []
for i in pagelist:
dicttoid = {}
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' order by(created_time) limit %s'%(i[0],limitnum))
if i[1] in dicttoid:
dicttoid[i[1]] += 1
else:
dicttoid[i[1]] = 1
hotfan = dict(percentfan(hot_fan_defind,dicttoid,percent),**hotfan)
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' order by(created_time) limit %s'%(i[0],limitnum))
if i[1] in dicttoid:
dicttoid[i[1]] += 1
else:
dicttoid[i[1]] = 1
hotfan = percentfan(hot_fan_defind,dicttoid,percent)
drawhotfun_num(hotfan,onehoursdict,msgdict,page_name,limitnum,str(hot_fan_defind)+percent)
def seepercent(page_name):
import datetime
import numpy as np
from matplotlib.pyplot import *
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
#alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\')')
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
msgdict = {}
onehoursdict = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=11)
#temptime = datetime.datetime.strptime(msgdict[i[0]][1],"%Y-%m-%d %H:%M:%S")+datetime.timedelta(hours=1)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
dictxy = dict()
for i in msgdict:
if onehoursdict[i]:
dictxy[len(onehoursdict[i])] = fbsql.defind_by_self('select count(postid) from share_fb_msg_new where trend_mid=\'%s\' group by(trend_mid)'%i)[0]
for i in dictxy:
print str(i)+':'+str(dictxy[i])
xlabel('11_hours get_count')
ylabel('all_get')
title(page_name+' 11_hour:all_time')
#plot(dictavg.keys(),dictavg.values(),'ro')
plot(dictxy.keys(),dictxy.values(),'ro')
show()
def allshare(page_name,hours,justgetdata=None):
import datetime
import numpy as np
from draw_corrcoef import draw_corre
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
else:
alllist = fbsql.defind_by_self('select user_msg.msgid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
msgdict = {}
onehoursdict = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if not onehoursdict[i[0]]:
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours+3)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
dictx = []
dicty = []
allcount = 0
for i in msgdict:
if onehoursdict[i] and msgdict[i][0]:
allcount += 1
dictx.append(len(onehoursdict[i]))
dicty.append(msgdict[i][0])
if justgetdata:
return (onehoursdict,msgdict)
testdatax = []
testdatay = []
dicttrainx = []
dicttrainy = []
argrmdict = dict()
arragerdm = np.random.choice(len(dictx),int(len(dictx)*0.25),replace=False)
for i in arragerdm:
argrmdict[i] = None
for i in range(len(dictx)):
if i in argrmdict:
testdatax.append(dictx[i])
testdatay.append(dicty[i])
else:
dicttrainx.append(dictx[i])
dicttrainy.append(dicty[i])
print '总数据：'+str(allcount)
print '训练数据：'+str(len(dicttrainx))
print '测试数据：'+str(len(testdatax))
xlabel = str(hours)+'_hours get_count'
ylabel = 'share_count'
title = page_name+' '+str(hours)+'_hour:all_time_true_share_count'
return draw_corre(xdata=np.log10(dicttrainx),ydata=np.log10(dicttrainy),xlabelstr=xlabel, ylabelstr=ylabel, titlestr=title, testdatax=np.log10(testdatax),testdatay = np.log10(testdatay))
def __init__(self,cookies,dictt,uid,usr_name,dyn):
threading.Thread.__init__(self)
self.fbsql = fb_mysql()
self.fb = facebook_graph()
self.fb.uid = uid
self.fb.cookies = cookies
self.dyn = dyn
self.dictm = dictt
self.dictm['from_name'] = usr_name
self.usr_name = usr_name
def __init__(self,u_info,userid):
multiprocessing.Process.__init__(self)
self.fb = facebook_graph()
self.fb.login(u_info['user_name'],u_info['password'])
self.fb.getaccess_token()
self.fbsql = fb_mysql()
self.act = self.fb.access_token
self.dyn = u_info['dyn']
self.uid = u_info['uid']
self.cookies = self.fb.cookies
self.fb.uid = u_info['uid']
self.fb.dyn = u_info['dyn']
self.usr_id = userid
def sharefan(page_name,hotfannum,hours,percent = '',justgetdata=None,hotnot=None):
import datetime
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
msgdict = {}
onehoursdict = {}
hotfan = []
for i in pagelist:
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
dicttoid = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time,fromname from user_msg where msgid=\'%s\''%i[0])[0]
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if not onehoursdict[i[0]]:
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours+3)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if i[1] in dicttoid:
dicttoid[i[1]] += 1
else:
dicttoid[i[1]] = 1
hotfan = dict(hotfan,**percentfan(hotfannum,dicttoid,percent))
#alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\')')
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
msgdict = {}
onehoursdict = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if not onehoursdict[i[0]]:
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours+3)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if i[1] in dicttoid:
dicttoid[i[1]] += 1
else:
dicttoid[i[1]] = 1
hotfan = percentfan(hotfannum,dicttoid,percent)
if justgetdata:
return (hotfan,onehoursdict,msgdict)
return drawhotfun(hotfan,onehoursdict,msgdict,page_name,hotfannum,hours,percent,hotnot)
def deep_fan(page_name,strchoose,strgetdata = None):
import numpy as np
import datetime
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
#alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\')')
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
timedict = dict()
for i in alllist:
if i[0] not in timedict:
tempdict = {}
tempdetail = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
tempdict['share_count'] = tempdetail[0]
tempdict['created_time'] = tempdetail[1]
tempdict['line_time']=tempdict['created_time']+datetime.timedelta(hours=11)
#tempdict['share'] = dict()
tempdict['like_num'] = 0
tempdict['like_max'] = 0
tempdict['all_num'] = 0
tempdict['like_sum'] = 0
if not tempdict['share_count']:
continue
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
#tempdict['share'][i[1]] = [i[2]]
tempdict['all_num'] = 1
if i[4] >0:
tempdict['like_num']+=1
tempdict['like_sum']+=i[4]
if i[4]>tempdict['like_max']:
tempdict['like_max'] = i[4]
timedict[i[0]] = tempdict
else:
tempdict = timedict[i[0]]
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] += 1
if i[4] >0:
tempdict['like_num']+=1
tempdict['like_sum']+=i[4]
if i[4]>tempdict['like_max']:
tempdict['like_max'] = i[4]
if strchoose == 'num':
drawlikenum(timedict,page_name,strgetdata)
elif strchoose == 'sum':
drawlikesum(timedict,page_name,strgetdata)
else:
drawlikedefind(timedict,page_name,strgetdata)
def share_time(page_name,hoursdefind):
import numpy as np
import datetime
import numpy as np
from matplotlib.pyplot import *
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.created_time from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
#alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.created_time from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\')')
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.created_time from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
timedict = dict()
time_count = {}
for i in range(hoursdefind):
time_count[i] = 0
for i in alllist:
if i[0] not in timedict:
tempdict = {}
tempdict['created_time'] = fbsql.defind_by_self('select created_time from user_msg where msgid=\'%s\''%i[0])[0][0]
tempdict['time_count'] = {}
minus = i[1] - tempdict['created_time']
temphours = minus.days*24 + minus.seconds//3600
if temphours < hoursdefind and temphours > 0:
print i[1]
print tempdict['created_time']
print temphours
time_count[temphours] += 1
timedict[i[0]] = tempdict
else:
tempdict = timedict[i[0]]
minus = i[1] - tempdict['created_time']
temphours = minus.days*24 + minus.seconds//3600
if temphours < hoursdefind and temphours > 0:
time_count[temphours] += 1
dictxy={}
msgnum = len(timedict)
for i in time_count.keys():
print str(i)+':'+str(time_count[i])
y = time_count[i]/msgnum
dictxy[i] = y
xlabel('each_hours')
ylabel('avg_share')
title(page_name+'avg share num of each time')
plot(dictxy.keys(),dictxy.values(),'ro')
show()
def post_time(page_name,choose = 'page'):
import datetime
import numpy as np
from matplotlib.pyplot import *
fbsql = fb_mysql()
basictime = datetime.datetime.strptime('00:00:00','%H:%M:%S')
if choose == 'page':
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select created_time from user_msg where fromid = \'%s\''%i)
else:
alllist = fbsql.defind_by_self('select created_time from user_msg where fromname=\''+page_name+'\'')
else:
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select share_fb_msg_new.created_time from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromid = \'%s\''%i)
else:
alllist = fbsql.defind_by_self('select share_fb_msg_new.created_time from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
time_count = {}
for i in range(48):
time_count[i*0.5] = 0
for i in alllist:
if i[0]:
minus = i[0] - basictime
if choose == 'page':
time_count[(minus.seconds//1800)*0.5] += 1
else:
temp8 = ((minus.seconds//1800)+32)%48
time_count[temp8*0.5] += 1
xx = []
yy = []
for i in range(48):
xx.append(i*0.5)
yy.append(time_count[i*0.5])
bar(time_count.keys(),time_count.values(),width=0.5,align = 'edge',color='green')
plot(xx,yy, 'r--')
xlabel('each 0.5h')
if choose == 'page':
ylabel('page post count')
title(page_name+' page post count per 0.5h')
else:
ylabel('user post count')
title(page_name+' user post count per 0.5h')
show()
def deep_fan_num(page_name,limit_num):
import numpy as np
import datetime
fbsql = fb_mysql()
sharedict = dict()
if page_name=='all':
msglist = []
for i in pagelist:
msglist += fbsql.defind_by_self('select share_fb_msg_new.trend_mid, user_msg.share_count from user_msg,share_fb_msg_new where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\') group by(share_fb_msg_new.trend_mid)'%i)
#msglist = fbsql.defind_by_self('select share_fb_msg_new.trend_mid, user_msg.share_count from user_msg,share_fb_msg_new where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\') group by(share_fb_msg_new.trend_mid)')
else:
msglist = fbsql.defind_by_self('select share_fb_msg_new.trend_mid, user_msg.share_count from user_msg,share_fb_msg_new where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname =\'%s\' group by(share_fb_msg_new.trend_mid)'%(page_name))
for i in msglist:
templist = fbsql.defind_by_self('select like_count from share_fb_msg_new where trend_mid=\'%s\' order by(created_time) limit %s'%(i[0],limit_num))
sharedict[i[0]] = {}
sharedict[i[0]]['share_count'] = i[1]
sharedict[i[0]]['like_num'] = 0
for j in templist:
if j > 0:
sharedict[i[0]]['like_num'] += 1
drawlikenum_num(sharedict,page_name,limit_num)
def change_draw_fan(page_name,hotfannum,hours):
import datetime
import numpy as np
from matplotlib.pyplot import *
fbsql = fb_mysql()
begin_time = datetime.datetime(2015, 12, 15)
hotfan = list()
for k in range(40):
hotfan.append(0)
timenow_time = (begin_time + datetime.timedelta(days=k)).strftime("%Y-%m-%d %H:%M:%S")
print timenow_time
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\' and user_msg.created_time<\''+timenow_time+'\'')
dicttoid = {}
msgdict = {}
onehoursdict = {}
for i in alllist:
if i[0] not in msgdict:
msgdict[i[0]] = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if not onehoursdict[i[0]]:
temptime = msgdict[i[0]][1]+datetime.timedelta(hours=hours+3)
strtime = temptime.strftime("%Y-%m-%d %H:%M:%S")
onehoursdict[i[0]] = fbsql.defind_by_self('select toid from share_fb_msg_new where trend_mid=\'%s\' and created_time<\'%s\''%(i[0],strtime))
if i[1] in dicttoid:
dicttoid[i[1]] += 1
else:
dicttoid[i[1]] = 1
for j in dicttoid.keys():
if dicttoid[j]>=hotfannum:
hotfan[k] += 1
hotfan[k] = hotfan[k]/len(dicttoid)
print hotfan[k]
xlabel('begin at 2015-12-15 each 1 day')
ylabel('hot_fan_num')
title('Tsai Ing-wen hotfan percent')
#plot(dictavg.keys(),dictavg.values(),'ro')
plot(range(40), hotfan)
show()
def fandefind(page_name):
fbsql = fb_mysql()
if page_name=='all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
#alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\')')
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,toid from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
for i in alllist:
if i[1] in dicttoid:
dicttoid[i[1]] += 1
else:
dicttoid[i[1]] = 1
dictnum = {}
peoplenum = len(dicttoid)
for i in dicttoid.keys():
if dicttoid[i] in dictnum:
dictnum[dicttoid[i]] += 1
else:
dictnum[dicttoid[i]] = 1
drawline(dictnum,page_name,peoplenum)
def __init__(self, fb, fid, act):
threading.Thread.__init__(self)
self.fid = fid
self.fbsql = fb_mysql()
self.fb = fb
def __init__(self,queue,fb,lockname):
threading.Thread.__init__(self)
self.fbsql = fb_mysql()
self.fb = fb
self.lock = lockname
self.queue = queue
from trend_topic import get_share_post
from trend_topic import getmsgsend
import datetime
import time
from save_post import fb_post_comment
from save_post import fb_post_sharelike
from _fb_commensql import fb_mysql
import threading
from fb_friend_thread import fb_friend
from trend_topic import get_user_msg
while(1):
uid = '100008105866583'
fb = facebook_graph()
fb.login("*****@*****.**","blueapple")
fb.getaccess_token()
fbsql = fb_mysql()
act = fb.access_token
cookies = fb.cookies
fb.uid = uid
usr_id = '326683984410'
#(dictmsg,usr_name) = get_user_msg(fb,fbsql,usr_id)
gettime = datetime.datetime.now()+datetime.timedelta(days=-15)
msglist = fbsql.get_line_msg(gettime.strftime("%y-%m-%d %H:%M:%S"))
threads = []
for i in msglist:
print i
msgtemp = dict()
msgtemp['msgid'] = i[0]
msgtemp['fromid'] = i[1]
msgtemp['from_name'] = i[2]
msgtemp['name'] = i[3]
def likesum_nothotfan(page_name,hotnum,timeline,percent = '',justgetdata=None):
import numpy as np
import datetime
from draw_corrcoef import draw_corre
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
hotfan = []
for i in pagelist:
eachlist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
alllist += eachlist
dicttoid = {}
for j in eachlist:
if j[2] in dicttoid:
dicttoid[j[2]] += 1
else:
dicttoid[j[2]] = 1
hotfan = dict(hotfan,**percentfan(hotnum,dicttoid,percent))
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
dicttoid = {}
for i in alllist:
if i[2] in dicttoid:
dicttoid[i[2]] += 1
else:
dicttoid[i[2]] = 1
hotfan = percentfan(hotnum,dicttoid,percent)
timedict = dict()
for i in alllist:
if i[0] not in timedict:
tempdict = {}
tempdetail = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
tempdict['share_count'] = tempdetail[0]
tempdict['created_time'] = tempdetail[1]
tempdict['line_time']=tempdict['created_time']+datetime.timedelta(hours=timeline)
tempdict['all_num'] = 0
tempdict['like_sum'] = 0
tempdict['all_like_sum'] = 0
if not tempdict['share_count']:
continue
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] = 1
if i[4] >0:
if i[2] not in hotfan:
tempdict['like_sum']+=i[4]
tempdict['all_like_sum']+=i[4]
timedict[i[0]] = tempdict
else:
tempdict = timedict[i[0]]
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] += 1
if i[4] >0:
if i[2] not in hotfan:
tempdict['like_sum']+=i[4]
tempdict['all_like_sum']+=i[4]
dictx = []
dicty = []
if justgetdata:
return timedict
for i in timedict.keys():
print i+':'+str(timedict[i]['like_sum'])
if timedict[i]['all_num'] > 0 and timedict[i]['like_sum']>0:
x = timedict[i]['like_sum']
dictx.append(x)
dicty.append(timedict[i]['share_count'])
testdatax = []
testdatay = []
dicttrainx = []
dicttrainy = []
argrmdict = dict()
arragerdm = np.random.choice(len(dictx),int(len(dictx)*0.25),replace=False)
for i in arragerdm:
argrmdict[i] = None
for i in range(len(dictx)):
if i in argrmdict:
testdatax.append(dictx[i])
testdatay.append(dicty[i])
else:
dicttrainx.append(dictx[i])
dicttrainy.append(dicty[i])
print '训练数据：'+str(len(dicttrainx))
print '测试数据：'+str(len(testdatax))
xlabel='like_sum_notfan/all_num '+str(timeline)+'h'
ylabel='share_count'
title=page_name+' like_sum_notfan:share_count time:'+str(timeline-8)+'h fan_defind:'+str(hotnum)+percent+'up'
return draw_corre(xdata=np.log10(dicttrainx), ydata=np.log10(dicttrainy), xlabelstr=xlabel, ylabelstr=ylabel, titlestr=title, testdatax=np.log10(testdatax), testdatay=np.log10(testdatay))
def twolikesum_nothotfan(page_name,hotnum,hotnum2,timeline):
import numpy as np
import datetime
from matplotlib.pyplot import *
fbsql = fb_mysql()
if page_name == 'all':
alllist = []
for i in pagelist:
alllist += fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromid = \'%s\')'%i)
#alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and (user_msg.fromname = \'history\' or user_msg.fromname = \'Will Smith\' or user_msg.fromname = \'nba\')')
else:
alllist = fbsql.defind_by_self('select user_msg.msgid,share_fb_msg_new.fromid,toid,share_fb_msg_new.created_time,share_fb_msg_new.like_count from share_fb_msg_new,user_msg where share_fb_msg_new.trend_mid = user_msg.msgid and user_msg.fromname=\''+page_name+'\'')
timedict = dict()
dicttoid = {}
for i in alllist:
if i[2] in dicttoid:
dicttoid[i[2]] += 1
else:
dicttoid[i[2]] = 1
hotfan = {}
for i in dicttoid.keys():
if dicttoid[i] >= hotnum:
hotfan[i] = None
hotfan2 = {}
for i in dicttoid.keys():
if dicttoid[i] >= hotnum2:
hotfan2[i] = None
print 'hotfan > 80:'+str(hotfan2)
for i in alllist:
if i[0] not in timedict:
tempdict = {}
tempdetail = fbsql.defind_by_self('select share_count,created_time from user_msg where msgid=\'%s\''%i[0])[0]
tempdict['share_count'] = tempdetail[0]
tempdict['created_time'] = tempdetail[1]
tempdict['line_time']=tempdict['created_time']+datetime.timedelta(hours=timeline)
tempdict['all_num'] = 0
tempdict['like_sum'] = 0
tempdict['like_sum2'] = 0
if not tempdict['share_count']:
continue
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] = 1
if i[4] >0 and (i[2] not in hotfan):
tempdict['like_sum']+=i[4]
if i[4] >0 and (i[2] in hotfan2):
tempdict['like_sum2']+=i[4]
timedict[i[0]] = tempdict
else:
tempdict = timedict[i[0]]
if i[3] < tempdict['line_time'] and i[1]!=i[2]:
tempdict['all_num'] += 1
if i[4] >0 and (i[2] not in hotfan):
tempdict['like_sum']+=i[4]
if i[4] >0 and (i[2] in hotfan2):
tempdict['like_sum2']+=i[4]
dictxy={}
dictxy2={}
for i in timedict.keys():
print i+':'+str(timedict[i]['like_sum'])
if timedict[i]['all_num'] > 0:
#x = timedict[i]['like_sum']/timedict[i]['all_num']
x = timedict[i]['like_sum']
dictxy[x] = timedict[i]['share_count']
x = timedict[i]['like_sum2']
dictxy2[x] = timedict[i]['share_count']
xlabel('like_sum_notfan/all_num '+str(timeline)+'h')
ylabel('share_count')
print dictxy2
title(page_name+' like_sum_notfan:share_count time:'+str(timeline)+'h fan_defind:'+str(hotnum)+'-'+str(hotnum2)+'up')
#plot(np.log10(dictxy.keys()),np.log10(dictxy.values()),'ro')
plot(np.log10(dictxy2.keys()),np.log10(dictxy2.values()),'*')
show()
