yingtai_bot.py

#coding: utf-8
import tweepy, random, re, os, time, urllib, urllib2, cookielib, csv, MeCab
from xml.etree.ElementTree import *
from PIL import Image
from cv import *
from BeautifulSoup import BeautifulSoup

secrets=[i for i in csv.reader(open('Database\\password.csv'))]
consumer_key=secrets[2][1]; consumer_secret=secrets[3][1]
access_key=secrets[4][1]; access_secret=secrets[5][1]
auth = tweepy.OAuthHandler(consumer_key, consumer_secret)
auth.set_access_token(access_key, access_secret)
api = tweepy.API(auth_handler=auth)

class Maze():
    def __init__(self):
        self.ROAD=0; self.WALL=1; self.ROUTE=2
        self.STRROAD=u'□'; self.STRWALL=u'■'; self.STRROUTE=u'△'
        self.width=11; self.height=9
        self.sx=1; self.sy=0
        self.gx=self.width-2; self.gy=self.height-1

    def makeMazeData(self):
        bigdata=[[[] for x in range(self.width+2)] for y in range(self.height+2)]
        for x in range(self.width+2):
            for y in range(self.height+2):
                if x in set(i+1 for i in range(self.width)) and y in set(j+1 for j in range(self.height)):
                    bigdata[y][x]=self.WALL
                else: bigdata[y][x]=self.ROAD
        self.dig(self.sx+1, self.sy+2, bigdata)
        bigdata[self.sy+1][self.sx+1]=bigdata[self.gy+1][self.gx+1]=self.ROAD
        data=[L[1:self.width+1] for L in bigdata[1:self.height+1]]
        return data

    def dig(self,px,py,bigdata):
        for i in range(100):
            vector=[[0,1,0,-1],[1,0,-1,0]]; r=random.randint(0,3)
            dx=vector[0][r]; dy=vector[1][r]
            if bigdata[py+dy*2][px+dx*2]!=self.ROAD:
                bigdata [py+dy][px+dx]=bigdata[py+dy*2][px+dx*2]=self.ROAD
                self.dig(px+dx*2, py+dy*2, bigdata)

    def data2String(self,data):
        stringData=''
        for y in range(self.height):
            for x in range(self.width):
                if data[y][x]==self.ROAD: stringData+=self.STRROAD
                elif data[y][x]==self.WALL: stringData+=self.STRWALL
                elif data[y][x]==self.ROUTE: stringData+=self.STRROUTE
                else: print 'translate error'
            stringData+='\n'
        return stringData

    def string2Data(self,stringData):
        data=[[[] for x in range(self.width)] for y in range(self.height)]
        for x in range(width):
            for y in range(height):
                count=x+y*(self.width+1)
                if stringData[cnt]==self.STRWALL: data[y][x]=self.WALL
                elif stringData[cnt]==self.STRROAD: data[y][x]==self.ROAD
                elif stringData[cnt]==self.STRROUTE: data[y][x]==self.ROUTE
                else: print 'transtate error'
        return data

    def solve(self,data):
        INF=1000; vector=[[1,0,-1,0],[0,1,0,-1]]; d=[[INF for x in range(self.width)] for y in range(self.height)]
        d[self.sy][self.sx]=now=0
        que=[(self.sy,self.sx)]; goal=(self.gy,self.gx)
        while True:
            for count in que[:]:
                p=que.pop(0)
                for i in range(4):
                    newque=(p[0]+vector[0][i],p[1]+vector[1][i]);
                    if 0<=newque[0]<self.height and 0<=newque[1]<self.width and d[newque[0]][newque[1]]==INF and data[newque[0]][newque[1]]==self.ROAD:
                        d[newque[0]][newque[1]]=now+1; que.append(newque)
            if d[self.gy][self.gx]!=INF:
                data[self.gy][self.gx]=self.ROUTE; break
            now+=1
        while True:
            for i in range(4):
                x=map(lambda g,v:g+v,goal,(vector[0][i],vector[1][i]))
                if 0<=x[0]<self.height and 0<x[1]<self.width and d[x[0]][x[1]]==now:
                    goal=x; now-=1 ;data[x[0]][x[1]]=self.ROUTE; break
            if now==0: break
        data[self.sy][self.sx]=self.ROUTE
        return data
    def makeMain(self):
        return self.data2String(self.makeMazeData())
    def solveMain(self,text):
        text=re.split(u'(\W+)',text[3].lstrip(u'\n')+'\n')
        return self.data2String(self.solve(self.string2Data(tweet)))

class Weather():
    def main(self,text):
        text=unicode(text)
        regionlist=[{'jp':unicode(i[0]),'en':i[1]} for i in csv.reader(open('Database\\region.csv'))]
        conditionlist=[{'en':i[0],'jp':unicode(i[1])} for i in csv.reader(open('Database\\condition.csv'))]
        datelist=[{'reg':u'現在','date':u'現在','n':0},{'reg':u'(きょう|今日)','date':u'今日','n':1},{'reg':u'(あした|あす|明日)','date':u'明日','n':2},
                  {'reg':u'(!し|!明)(あさって|明後日)',u'date':'明後日','n':3},{'reg':u'[し明](あさって|明後日)','date':u'明後日','n':4}]
        week=['Sun','Mon','Tue','Wed','Thu','Fri','Sat']

        region={'jp':'神戸','en':'Kobe'}
        for i in regionlist:
            if re.search(i['jp'],text)!=None: region=i; break
        for i in datelist:
            if re.search(i['reg'],text)!=None: date=i;break
        elem=ElementTree(file=urllib2.urlopen('http://www.google.com/ig/api?weather=%s'%region['en']))
        current = elem.find('.//current_conditions')
        forecasts=elem.findall('.//forecast_conditions')

        if date['date']==u'現在':
            condition={'en':current.find('condition').get('data')}
            temp_c=current.find('temp_c').get('data')
            humidity={'en':current.find('humidity').get('data')}
            condition['jp']=condition['en']
            for i in conditionlist:
                if condition['en']==i['en']: condition['jp']=i['jp'];break
            humidity['jp']=humidity['en'].split(' ')[1]
            return u'現在の%sの天気は%sです、気温は%s℃、湿度は&sです'%(region['jp'],condition['jp'],temp_c,humidity['jp'])
        else:
            for forecast in forecasts:
                low=forecast.find('low').get('data')
                high=forecast.find('high').get('data')
                low=str((int(low)-32)*5/9); high=str((int(high)-32)*5/9)
                condition={'en':forecast.find('condition').get('data')}
                day_of_week=forecast.find('day_of_week').get('data')
                condition['jp']=condition['en']
                for i in conditionlist:
                    if condition['en']==i['en']: condition['jp']=i['jp'];break
                if week[(int(time.strftime('%w'))+date['n']-1)%7]==day_of_week:
                    return u'%sの%sの天気は%s、最高気温は%s℃、最低気温は%s℃でしょう'%(date['date'],region['jp'],condition['jp'],high,low)

class Sisoku():
    def __init__(self):
        self.comb=[[(i,j) for i in range(j) for j in range(j+1) if i<j] for j in range(6) if 0<j]

    def solve(self,opes,ans,n=6):
        print opes,ans
        length=len(opes)
        opes=map(float,opes); opes_copy=opes[:]; ans=float(ans)
        if not n in [1,2,3,4,5,6] or not length in [1,2,3,4,5,6]:return False
        funclist=[lambda x,y:x+y, lambda x,y:x*y, lambda x,y:x-y, lambda x,y:y-x, lambda x,y:x/y, lambda x,y:y/x]
        funcs=map(lambda i:i[:length],map(lambda list:map(lambda h:funclist[h],list),[(i,j,k,l,m) for i in range(n) for j in range(n) for k in range(n) for l in range(n) for m in range(n)]))
        turns=map(lambda i:i[:length-1],[(i,j,k,l,m) for i in range((length-1)*(length)/2) for j in range((length-2)*(length-1)/2 if 2<length else 1) for k in range((length-3)*(length-2)/2 if 3<length else 1) for l in range((length-4)*(length-3)/2 if 4<length else 1) for m in range((length-5)*(length-4)/2 if 5<length else 1)])
        for func in funcs:
            for turn in turns:
                opes=opes_copy[:]
                try:
                    for i in range(length-1):
                        opes[self.comb[length-(i+2)][turn[i]][0]] = func[i](opes[self.comb[length-(i+2)][turn[i]][0]],opes[self.comb[length-(i+2)][turn[i]][1]])
                        opes.pop(self.comb[length-(i+2)][turn[i]][1])
                    if -0.00001<opes[0]-ans<0.00001: return {'opes':opes_copy,'turn':turn,'func':func}
                except (ZeroDivisionError,OverflowError,ValueError):continue

    def data2String(self,data):
        if data:
            funcchr=['+','*','-','-','/','/']; func=[]
            opes=data['opes']; turn=data['turn']; length=len(opes)
            tests = map(lambda f:f(2.0,3.0),data['func'])
            for test in tests:
                if test==5: func.append(0)
                elif test==6: func.append(1)
                elif test==-1: func.append(2)
                elif test==1: func.append(3)
                elif test==2.0/3.0: func.append(4)
                else: func.append(5)
            opes=map(lambda i:str(int(i)),opes)
            for i in range(length-1):
                if func[i] in [3,5]:
                    opes[self.comb[length-(i+2)][turn[i]][0]]='('+opes[self.comb[length-(i+2)][turn[i]][1]]+funcchr[func[i]]+opes[self.comb[length-(i+2)][turn[i]][0]]+')'
                    opes.pop(self.comb[length-(i+2)][turn[i]][1])
                else:
                    opes[self.comb[length-(i+2)][turn[i]][0]]='('+opes[self.comb[length-(i+2)][turn[i]][0]]+funcchr[func[i]]+opes[self.comb[length-(i+2)][turn[i]][1]]+')'
                    opes.pop(self.comb[length-(i+2)][turn[i]][1])
            opes[0]=opes[0][1:len(opes[0])-1]
            return opes[0]
        else:
            return u'解がありません'

    def tweet2Data(self,text):
        textlist=re.split('[^0-9\.]',text)
        while True:
            if '' in textlist: textlist.remove('')
            else:break
        list=[float(i) for i in textlist]
        opes=list[:len(list)-1]
        ans=list[len(list)-1]
        return [opes,ans]


class Icon():
    def __init__(self):
        pixiv_id=secrets[0][1];pixiv_pw=secrets[1][1]
        self.connect =urllib2.build_opener(urllib2.HTTPCookieProcessor(cookielib.CookieJar()))
        req = urllib2.Request("http://www.pixiv.net/index.php")
        req.add_data( urllib.urlencode( {"mode":"login", "pixiv_id":pixiv_id, "pass":pixiv_pw} ))
        res = self.connect.open(req).read()

    def searchPageParser(self,html):
        info=[]
        illustList=BeautifulSoup(html).findAll('section',{'id':'search-result'})[0].findAll('li')
        for illust in illustList:
            a=illust.findAll('a')
            if len(a)==1:continue
            else:
                image_author_name=a[1].string; image_author_id=a[1]['href']
                illust_id=a[0]['href'].split('illust_id=')[1]
                if len(a)==2: bkm=0
                else: bkm=int(a[2]['data-tooltip'].split(u'件')[0].replace(',',''))
            info.append({'bkm':bkm, 'illust_id':illust_id, 'image_author_name':image_author_name, 'image_author_id':image_author_id})
        return info

    def illustPageParser(self,illust_id):
        illusturl='http://www.pixiv.net/member_illust.php?mode=big&illust_id=%s'%illust_id
        req=urllib2.Request(illusturl)
        req.add_header('Referer',illusturl.replace("big","medium"))
        html=self.connect.open(req).read().decode("utf8")
        imgurl=BeautifulSoup(html).findAll('img')[0]['src']
        return imgurl

    def download(self,illust_id):
        try:
            imgurl=self.illustPageParser(illust_id)
            req=urllib2.Request(imgurl)
            req.add_header('Referer','http://www.pixiv.net/member_illust.php?mode=big&illust_id=%s'%illust_id)
            image=self.connect.open(req).read()

            savedir=os.getcwd()+'\\Images'
            file=open(savedir+'\\'+os.path.basename(imgurl),'wb')
            for i in image:
                file.write(i)
            file.close()
        except Exception,e:print e

    def downloadMostFaved(self,query):
        imgdatas=[];maxbkm=0;illust_id=-1
        for i in range(200):
            url=u'http://www.pixiv.net/search.php?s_mode=s_tag&word=%s&order=date_d&p=%s'%(urllib.quote_plus(query.encode('utf-8')),str(i))
            html=self.connect.open(urllib2.Request(url)).read()
            if re.search(u'<div class="no-item">見つかりませんでした</div>',html)!=None:break
            imgdatas.extend(self.searchPageParser(html))
        imgdatas=sorted(imgdatas,key=lambda x:x['bkm'],reverse=True)[:(20 if len(imgdatas)>20 else len(imgdatas))]
        print imgdatas
        for i in imgdatas:
            self.download(i['illust_id'])

    def detectFace(self):
        colors=[(0,0,255), (0,128,255), (0,255,255), (0,255,0), (255,128,0), (255,255,0), (255,0,0), (255,0,255)]
        cascade=Load('Database\\cascade.xml')
        for i in  os.listdir(os.getcwd()+'\\Images'):
            path=os.getcwd()+'\\Images\\'+i
            img=LoadImage(path, CV_LOAD_IMAGE_COLOR)
            faces = HaarDetectObjects(img, cascade, CreateMemStorage(0),1.11, 4, 0,(40, 40))
            if faces:
                pilimg=Image.open(path)
                x=faces[0][0][0];y=faces[0][0][1];width=faces[0][0][2];height=faces[0][0][3]
                x-=width*0.2; y-=width*0.2; width*=1.4; height*=1.4
                croped=pilimg.crop((int(x),int(y),int(x+width),int(y+height)))
                croped.save(os.getcwd()+'\\Images\\icon.gif')
                try:
                    api.update_profile_image(os.getcwd()+'\\Images\\icon.gif')
                    for j in  os.listdir(os.getcwd()+'\\Images'):
                        os.remove(os.getcwd()+'\\Images\\'+j)
                    return 0
                except:continue

    def main(self,query):
        query=query.split(u'を')[1].split('に')[0]
        if re.search(u'(ちん|たん|きゅん)?の?(画像|絵)',query)!=None:
            query=re.split(u'(ちん|たん|きゅん)?の?(画像|絵)?',query)[0]
        self.downloadMostFaved(query)
        self.detectFace()
        return u'アイコンを%sの画像に変更しました'%query

class Twitter():
    def returnNewStatuslist(self,method):
        statuses=method()
        try:
            statuses_data_path='Database\\mention_statuses.csv' if all([0<=i.text.find('@yingtai_bot') for i in statuses]) else 'Database\\timeline_statuses.csv'
            old_statuses=[i for i in csv.reader(open(statuses_data_path))]
            last_status_id=int(old_statuses[0][0].replace('\xef\xbb\xbf',''))

            newstatuslist=[]
            for status in statuses:
                if status.id >last_status_id: newstatuslist.append(status)
            new_status_writer=csv.writer(file(statuses_data_path,'w'))
            for newstatus in newstatuslist:
                new_status_writer.writerow([newstatus.id,newstatus.text,newstatus.author.id,newstatus.author.screen_name])
            new_status_writer.writerows(old_statuses)
            return newstatuslist
        except tweepy.error.TweepError,e:
            print e
            if not str(e).find('Rate limit exceeded.')>-1:
                self.returnNewStatuslist(method)

    def tweetResponser(self,statuses):
        followers_ids=api.followers_ids()
        #if statuses:
        if False:
            for status in statuses:
                if status.author.id in followers_ids:
                    def reply(text):
                        try: api.update_status(u'@%s %s'%(status.author.screen_name,text),status.id)
                        except:
                            try:api.update_status(u'@%s %s　'%(status.author.screen_name,text),status.id)
                            except:pass
                    if status.text.find('@yingtai_bot')>-1:
                        if status.text.find(u'天気')>-1:
                            api.update_status
                        elif re.search(u'迷路.*(昨|書|出)',status.text)!=None:
                            reply(u'\n%s'%m.makeMain())
                        elif re.search(u'解[いきくけこ]',status.text)!=None:
                            try:
                                reply(u'\n%s'%m.solveMain(api.get_status(status.in_reply_to_status_id).text))
                            except:
                                if status.id!=None: reply(u'迷路以外のツイートにリプライ飛ばしてませんか')
                                else: api.update_status(u'@%s えっ'%status.author.screen_name,status.id)
                        elif re.search(u'@yingtai_bot.*@[\w]+に[0-9]個の?(fav|ふぁぼ)',status.text):
                            self.favbomb(status)
                        elif re.search(u'(アイコン|icon|プロフィール画像|profile_image)を.+に')!=None:
                            reply(icon.main(status.text))

                        else: reply('はい')
                    else:
                        if status.text.find('@')<0:
                            if re.search(u'((!ちゃん)|[ね寝](る|ます))|[nNｎＮ][eｅEＥ](1|[rｒRＲlｌLＬ])(!vｖ)|おやす',status.text)!=None:
                                reply([u'おやすみ',u'おやすみー',u'おやすみなさい'][random.randint(0,2)])
                            elif re.search(u'(お[はそ][およ]|ohayo|[gGｇＧ](ood|ｏｏｄ|OOD|ＯＯＤ).*[mｍMＭ](orning,ｏｒｎｉｎｇ,ORNING,ＯＲＮＩＮＧ)|ごおｄもｒにんｇ|むくり|mkr|[起お]きた|すらま|スラマ|早上好)',status.text)!=None:
                                reply([u'おはよう',u'おはよー',u'おはようございます'][random.randint(0,2)])
                            elif re.search(u'離脱|りだ[つる]|([pｐPＰ][cｃCＣ]|パソ).*([閉と]じる|[け消]す|[き切]る｜[お落]とす)|[で出][か掛]ける|を[で出]る|[い行]って[くき来](る|ます)|出発',status.text)!=None:
                                reply('いってらっしゃい')
                            elif re.search(u'(!聞|!吐|!嘆)きたく|帰宅|ただいま|kitaku|KITAKU',status.text):
                                reply([u'おかえり',u'おかえりなさい'][random.randint(0,1)])
                    if re.search(u'([0-9.]+,)+[0-9\.] *(→|から|⇒|[-=]>) *[0-9\.]+',status.text)!=None:
                        data=sisoku.tweet2Data(status.text)
                        reply(sisoku.data2String(sisoku.solve(data[0],data[1])))



    def favbomb(self,status):
        splited=re.split(u'(\W+)',status.text)
        userid=splited[4];number=int(splited[6])
        print userid,"に",number,"個"
        for p in tweepy.Cursor(api.user_timeline,id=userid).items(number):
            try: api.create_favorite(p.id)
            except: continue
        api.update_status(u'@%sの代行として%s個のfavを贈りました'%(userid,number),status.id)

    def follow_remove(self):
        global hour
        try:
            fol=set(api.followers_ids())
            fri=set(api.friends_ids())
            #fol+=set('rh_ruson','alsuce10','ConstantineJP','e__ki','isennite','akatikukitiku')
            for i in list(fol-fri):
                api.create_friendship(i)
            #for i in list(fri-fol):
            #    api.destroy_friendship(i)
        except: hour-=1

    def main(self):
        global hour
        if time.localtime()[3]!=hour:
            self.follow_remove()
            api.update_status(analyzer.markov(analyzer.getwordlist()))
        self.tweetResponser(self.returnNewStatuslist(api.mentions))
        self.tweetResponser(self.returnNewStatuslist(api.friends_timeline))
        hour=time.localtime()[3]

class Analyzer():
    def __init__(self):
        self.wkt=lambda text:MeCab.Tagger("-Owakati").parse(text)
        self.cha=lambda text:MeCab.Tagger('-Ochasen').parse(text)

    def markov(self,texts):
        wordlist=[]
        for text in texts:
            wordlist.extend(self.split(text))
        markov={}; word1=word2=''
        for word in wordlist:
            if word1 and word2:
                if (word1,word2) not in markov: markov[(word1,word2)]=[]
                markov[(word1,word2)].append(word)
            word1,word2=word2,word
        sentence=''
        word1,word2=random.choice(markov.keys())
        while True:
            while True:
                if any([self.cha(sentence).split('\n')[0].find(pos)>-1 for pos in [u'フィラー',u'動詞',u'助詞',u'助動詞',u'記号']]):
                    sentence=''
                try: tmp=random.choice(markov[(word1,word2)]);break
                except:
                    word1,word2=random.choice(markov.keys())
            sentence+=tmp
            word1,word2=word2,tmp
            sentence=re.sub('(@(\w+)|#(\w+)|\w*_\w*|h?ttp://[\w\./]+)','',sentence)
            if re.search('終助詞|助動詞|。',self.cha(sentence))!=None:
                if 50<len(sentence)<140:
                    f=lambda list:list[len(list)-3]
                    #for i in self.cha(sentence).split('\n'):
                    #    print '"%s"'%i
                    if not any(f(self.cha(sentence).split('\n')).find(pos)>-1 for pos in [u'連用',u'連体',u'未然',u'接頭',u'接続']):
                        if not sentence.endswith('。'):sentence+='。'
                        break
                else:sentence=''
        print self.cha(sentence)
        #print '>',f(self.cha(sentence).split('\n'))
        #print any(f(self.cha(sentence).split('\n')).find(pos)>-1 for pos in [u'連用',u'連体','未然','接頭'])
        return sentence

    def split(self,text):
        m = self.wkt(text)
        result = m.rstrip("\n").split(" ")
        return result

    def getwordlist(self):
        return [i[1] for i in csv.reader(open('Database\\timeline_statuses.csv'))]

maze=Maze()
weather=Weather()
sisoku=Sisoku()
icon=Icon()
twitter=Twitter()
analyzer=Analyzer()

#print icon.detectFace()

if __name__ == '__main__':
    hour=time.localtime()[3]
    while True:
        time.sleep(0.1)
        #print analyzer.markov(analyzer.getwordlist())
        #"""
        print u'残りAPI数: %s リセット: %s'%(api.rate_limit_status()['remaining_hits'],api.rate_limit_status()['reset_time'])
        time.sleep(50)
        print 'start'
        twitter.main()
        print 'end'
        #"""