rec10/branches/0.9.0/src/guess.py

   1 #!/usr/bin/python
   2 # coding: UTF-8
   3 # Rec10 TS Recording Tools
   4 # Copyright (C) 2009 Yukikaze
   5
   6 import os
   7 import re
   8 import time
   9 import os
  10 import datetime
  11 import n_gram
  12 import zenhan
  13
  14 def detName(title,path):
  15     tt=detNameType(title, path)
  16     if tt['type']=="C"or tt['type']=="D":
  17         tt['num']=detNum(tt['title'],tt['folder'])
  18 def detNum(title,movepath):
  19     """
  20         #
  21     """
  22     files=os.listdir(movepath)
  23     ff=[]
  24     maxnum=0
  25     for file in files:
  26         print file
  27         file=os.path.join(movepath,file)
  28         if os.path.isfile(file):
  29             name=os.path.splitext(os.path.split(file)[1])
  30             name=name[0]
  31             name=zenhan.toHankaku(unicode(name))
  32             p1=detNameType(name,movepath)
  33             print p1['title']+" "+str(p1['num'])
  34             time1=time.localtime(os.path.getmtime(file))
  35             time1=datetime.datetime.fromtimestamp(os.path.getmtime(file))
  36             if p1['num']!=-1:
  37                 ff.append([p1['num'],p1['title'],time1])
  38                 if maxnum<p1['num']+1:
  39                     maxnum=p1['num']+1
  40     detMultiDeltaDays(ff)
  41 def detMultiDeltaDays(num_with_title_with_dates):
  42     maxnum=0
  43     for ft in num_with_title_with_dates:
  44         if maxnum<ft[0]+1:
  45             maxnum=ft[0]+1
  46     ff=num_with_title_with_dates
  47     f3=[None]*maxnum
  48     fret={}
  49     for i in range(0, maxnum, 1):
  50         f3[i]=[]
  51     for f2 in ff:
  52         f3[f2[0]].append(f2)
  53     for i in range(maxnum):
  54         for j in range(i+1,maxnum,1):
  55             for ft3 in f3[i]:
  56                 for ft4 in f3[j]:
  57                     for gdd in detSingleDeltaDay(i, ft3[2], j, ft4[2]):
  58                         print gdd
  59                         if fret.get(gdd[0])!=None:
  60                             print fret
  61                             fret[gdd[0]]=fret[gdd[0]]+gdd[1]
  62                         else:
  63                             fret[gdd[0]]=gdd[1]
  64     print fret
  65     maxk=0
  66     maxp=0
  67     for i in range(maxnum):
  68         if fret.get(i)!=None:
  69             if maxp<fret[i]:
  70                 maxk=i
  71                 maxp=fret[i]
  72     print maxk
  73     print maxp
  74     return maxk
  75 def detSingleDeltaDay(num1,date1,num2,date2):
  76     dd=date1-date2
  77     if date1<date2:
  78         dd=date2-date1
  79     dn=dd.days
  80     d=num2-num1
  81     if d<0:
  82         d=-1*d
  83     dp=dd+datetime.timedelta(hours=6)
  84     dp=dp.days
  85     dm=dd-datetime.timedelta(hours=6)
  86     dm=dm.days
  87     ret=[]
  88     if dn%d*2>d:
  89         dn=dn+d
  90     if dm%d*2>d:
  91         dm=dm+d
  92     if dp%d*2>d:
  93         dp=dp+d
  94     print str(d)+":"+str(dn)
  95     if dp!=dn:
  96         ret.append([dn/d,60])
  97         ret.append([dp/d,40])
  98     elif dm!=dn:
  99         ret.append([dn/d,60])
 100         ret.append([dm/d,40])
 101     else:
 102         ret.append([dn/d,100])
 103     return ret
 104 def detNameType(title,path):
 105     """
 106     type A ---title#<number>
 107     type B ---title#<number>subtitle
 108     type C ---title subtitle
 109     type D ---title(without number)
 110     type Aj ---title第<number>話
 111     path --search reflexively
 112     """
 113     print title
 114     #rA=re.compile(".+(?P<title>)#\d(?P<num>)\s[0,10]\z")
 115     rA=re.compile("(.+)#(\d*)\s*\Z")
 116     tA=rA.match(title)
 117     rB=re.compile("(.+)#(\d*)\s*(\D*)")
 118     tB=rB.match(title)
 119     rAj=re.compile("(.+)第(\d*)話\s*\Z")
 120     tAj=rAj.match(title)
 121     ret={'title':"",'type':"",'num':0,'subtitle':"",'folder':""}
 122     if tA:
 123         print "typeA"
 124         print "title="+tA.group(1)
 125         print "num="+tA.group(2)
 126         ret['type']="A"
 127         ret['title']=tA.group(1)
 128         ret['num']=int(tA.group(2))
 129         ret['folder']=searchFolder(tA.group(1),path)
 130     if tAj:
 131         print "typeA"
 132         print "title="+tAj.group(1)
 133         print "num="+tAj.group(2)
 134         ret['type']="Aj"
 135         ret['title']=tAj.group(1)
 136         ret['num']=int(tAj.group(2))
 137         ret['folder']=searchFolder(tAj.group(1),path)
 138     elif tB:
 139         print "typeB"
 140         print "title="+tB.group(1)
 141         print "num="+tB.group(2)
 142         print "subtitle="+tB.group(3)
 143         ret['type']="B"
 144         ret['title']=tB.group(1)
 145         ret['num']=int(tB.group(2))
 146         ret['folder']=searchFolder(tB.group(1),path)
 147         ret['subtitle']=tB.group(3)
 148     else:#type C or type D
 149         #fold=searchFolder(title, path)
 150         ts=title.split(" ")
 151         tt=""
 152         rt=["",0,""]
 153         for t in ts:
 154             tt=tt+" "+t
 155             ft1=searchFolder(tt,path)
 156             print tt
 157             print ft1
 158             if ft1!="":
 159                 #print rt
 160                 #print ft1[0]+" : "+str(ft1[1])
 161                 if ft1[1]>rt[1]:
 162                     rt[0]=tt
 163                     rt[1]=ft1[1]
 164                     rt[2]=ft1[0]
 165                     #print rt
 166         #print "title="+rt[0][1:]+"/"
 167         #print "subtitle = "+title.replace(rt[0][1:],"")[1:]
 168         ret['title']=rt[0][1:]
 169         ret['num']=-1
 170         ret['folder']=rt[2]
 171         ret['subtitle']=title.replace(rt[0][1:],"")[1:]
 172         if ret['subtitle'].replace(" ","")=="":
 173             ret['type']="D"
 174         else:
 175             ret['type']="C"
 176     return ret
 177 def searchFolder(title,path):
 178     """
 179     titleからフォルダーを探す
 180     """
 181     folderpath=os.listdir(path)
 182     lfpath=[]
 183     ngram=[]
 184     for ft in folderpath:
 185         fullpath=os.path.join(path, ft)
 186         if os.path.isdir(fullpath):
 187             lfpath.append(fullpath)
 188             ftt=os.listdir(fullpath)
 189             if len(ftt)>0:
 190                 for ft2 in ftt:
 191                     folderpath.append(os.path.join(fullpath, ft2))
 192         else:
 193             lfpath.append(fullpath)
 194     for dirp in lfpath:
 195         cmpp=""
 196         appp=""
 197         if os.path.isdir(dirp):
 198             cmpp=os.path.dirname(dirp)
 199             appp=dirp
 200         else:
 201             cmpp=os.path.basename(dirp)
 202             appp=os.path.dirname(dirp)
 203         p=n_gram.trigram(title.decode("utf-8"),cmpp.decode("utf-8"))
 204         if p>0:
 205             ngram.append((p,appp))
 206     ngram=list(set(ngram))
 207     ngram.sort()
 208     ngram.reverse()
 209     if len(ngram)>0:
 210         #print title + ngram[0][1] + " : "+str(ngram[0][0])
 211         if ngram[0][0]>300:
 212             return [ngram[0][1],ngram[0][0]]
 213         else:
 214             return ""
 215     else:
 216         return ""