rec10/trunk/src/guess.py

   1 #!/usr/bin/python
   2 # coding: UTF-8
   3 # Rec10 TS Recording Tools
   4 # Copyright (C) 2009-2010 Yukikaze
   5
   6 import os
   7 import re
   8 import time
   9 import os
  10 import datetime
  11 import n_gram
  12 import recdblist
  13 import zenhan
  14 import auto_move
  15
  16 def detName(title,path):
  17     tt=detNameType(title, path)
  18     if tt['type']=="C"or tt['type']=="D":
  19         tt['num']=detNum(tt['title'],tt['folder'])
  20 def detNum(title,movepath):
  21     """
  22         #番組の話数を推測する。
  23     """
  24     files=os.listdir(movepath)
  25     ff=[]
  26     maxnum=0
  27     for file in files:
  28         recdblist.printutf8(file)
  29         file=os.path.join(movepath,file)
  30         if os.path.isfile(file):
  31             name=os.path.splitext(os.path.split(file)[1])
  32             name=name[0]
  33             name=zenhan.toHankaku(unicode(name))
  34             p1=detNameType(name,movepath)
  35             recdblist.printutf8(p1['title']+" "+str(p1['num']))
  36             time1=time.localtime(os.path.getmtime(file))
  37             time1=datetime.datetime.fromtimestamp(os.path.getmtime(file))
  38             if p1['num']!=-1:
  39                 ff.append([p1['num'],p1['title'],time1])
  40                 if maxnum<p1['num']+1:
  41                     maxnum=p1['num']+1
  42     detMultiDeltaDays(ff)
  43 def detMultiDeltaDays(num_with_title_with_dates):
  44     maxnum=0
  45     for ft in num_with_title_with_dates:
  46         if maxnum<ft[0]+1:
  47             maxnum=ft[0]+1
  48     ff=num_with_title_with_dates
  49     f3=[None]*maxnum
  50     fret={}
  51     for i in range(0, maxnum, 1):
  52         f3[i]=[]
  53     for f2 in ff:
  54         f3[f2[0]].append(f2)
  55     for i in range(maxnum):
  56         for j in range(i+1,maxnum,1):
  57             for ft3 in f3[i]:
  58                 for ft4 in f3[j]:
  59                     for gdd in detSingleDeltaDay(i, ft3[2], j, ft4[2]):
  60                         #print gdd
  61                         if fret.get(gdd[0])!=None:
  62                             #print fret
  63                             fret[gdd[0]]=fret[gdd[0]]+gdd[1]
  64                         else:
  65                             fret[gdd[0]]=gdd[1]
  66     #print fret
  67     maxk=0
  68     maxp=0
  69     for i in range(maxnum):
  70         if fret.get(i)!=None:
  71             if maxp<fret[i]:
  72                 maxk=i
  73                 maxp=fret[i]
  74     #print maxk
  75     #print maxp
  76     return maxk
  77 def detSingleDeltaDay(num1,date1,num2,date2):
  78     dd=date1-date2
  79     if date1<date2:
  80         dd=date2-date1
  81     dn=dd.days
  82     d=num2-num1
  83     if d<0:
  84         d=-1*d
  85     dp=dd+datetime.timedelta(hours=6)
  86     dp=dp.days
  87     dm=dd-datetime.timedelta(hours=6)
  88     dm=dm.days
  89     ret=[]
  90     if dn%d*2>d:
  91         dn=dn+d
  92     if dm%d*2>d:
  93         dm=dm+d
  94     if dp%d*2>d:
  95         dp=dp+d
  96     recdblist.printutf8(str(d)+":"+str(dn))
  97     if dp!=dn:
  98         ret.append([dn/d,60])
  99         ret.append([dp/d,40])
 100     elif dm!=dn:
 101         ret.append([dn/d,60])
 102         ret.append([dm/d,40])
 103     else:
 104         ret.append([dn/d,100])
 105     return ret
 106 def detNameType(title,path):
 107     """
 108     type A ---title#<number>
 109     type B ---title#<number>subtitle
 110     type C ---title subtitle
 111     type D ---title(without number)
 112     type Aj ---title第<number>話
 113     path --search reflexively
 114     """
 115     recdblist.printutf8(title)
 116     #rA=re.compile(".+(?P<title>)#\d(?P<num>)\s[0,10]\z")
 117     rA=re.compile("(.+)#(\d*)\s*\Z")
 118     tA=rA.match(title)
 119     rB=re.compile("(.+)#(\d*)\s*(\D*)")
 120     tB=rB.match(title)
 121     rAj=re.compile("(.+)第(\d*)話\s*\Z")
 122     tAj=rAj.match(title)
 123     ret={'title':"",'type':"",'num':0,'subtitle':"",'folder':""}
 124     if tA:
 125         recdblist.printutf8("typeA")
 126         recdblist.printutf8("title="+tA.group(1))
 127         recdblist.printutf8("num="+tA.group(2))
 128         ret['type']="A"
 129         ret['title']=tA.group(1)
 130         ret['num']=int(tA.group(2))
 131         ret['folder']=searchFolder(tA.group(1),path)
 132     if tAj:
 133         recdblist.printutf8("typeA")
 134         recdblist.printutf8("title="+tAj.group(1))
 135         recdblist.printutf8("num="+tAj.group(2))
 136         ret['type']="Aj"
 137         ret['title']=tAj.group(1)
 138         ret['num']=int(tAj.group(2))
 139         ret['folder']=searchFolder(tAj.group(1),path)
 140     elif tB:
 141         recdblist.printutf8("typeB")
 142         recdblist.printutf8("title="+tB.group(1))
 143         recdblist.printutf8("num="+tB.group(2))
 144         recdblist.printutf8("subtitle="+tB.group(3))
 145         ret['type']="B"
 146         ret['title']=tB.group(1)
 147         ret['num']=int(tB.group(2))
 148         ret['folder']=searchFolder(tB.group(1),path)
 149         ret['subtitle']=tB.group(3)
 150     else:#type C or type D
 151         #fold=searchFolder(title, path)
 152         ts=title.split(" ")
 153         tt=""
 154         rt=["",0,""]
 155         for t in ts:
 156             tt=tt+" "+t
 157             ft1=searchFolder(tt,path)
 158             recdblist.printutf8(tt)
 159             #print ft1
 160             if ft1!="":
 161                 #recdblist.printutf8(rt)
 162                 #recdblist.printutf8(ft1[0]+" : "+str(ft1[1]))
 163                 if ft1[1]>rt[1]:
 164                     rt[0]=tt
 165                     rt[1]=ft1[1]
 166                     rt[2]=ft1[0]
 167                     #recdblist.printutf8(rt)
 168         #recdblist.printutf8("title="+rt[0][1:]+"/")
 169         #recdblist.printutf8("subtitle = "+title.replace(rt[0][1:],"")[1:])
 170         ret['title']=rt[0][1:]
 171         ret['num']=-1
 172         ret['folder']=rt[2]
 173         ret['subtitle']=title.replace(rt[0][1:],"")[1:]
 174         if ret['subtitle'].replace(" ","")=="":
 175             ret['type']="D"
 176         else:
 177             ret['type']="C"
 178     return ret
 179 def searchFolder(title,path,threshold=500):
 180     """
 181     titleからフォルダーを探す
 182     """
 183     folderpath=os.listdir(path)
 184     lfpath=[]
 185     ngram=[]
 186     for ft in folderpath:
 187         fullpath=os.path.join(path, ft)
 188         if os.path.isdir(fullpath):
 189             lfpath.append(fullpath)
 190             ftt=os.listdir(fullpath)
 191             if len(ftt)>0:
 192                 for ft2 in ftt:
 193                     folderpath.append(os.path.join(fullpath, ft2))
 194         else:
 195             lfpath.append(fullpath)
 196     for dirp in lfpath:
 197         cmpp=""
 198         appp=""
 199         if os.path.isdir(dirp):
 200             cmpp=os.path.dirname(dirp)
 201             appp=dirp
 202         else:
 203             cmpp=os.path.basename(dirp)
 204             appp=os.path.dirname(dirp)
 205         ntitle=auto_move.getTitle(title)
 206         ncmpp=auto_move.getTitle(cmpp)
 207         p=n_gram.trigram(ntitle,ncmpp)
 208         if p>0:
 209             ngram.append((p,appp))
 210     ngram=list(set(ngram))
 211     ngram.sort()
 212     ngram.reverse()
 213     if len(ngram)>0:
 214         #recdblist.printutf8(title + ngram[0][1] + " : "+str(ngram[0][0]))
 215         if ngram[0][0]>threshold:
 216             return ngram[0][1]
 217         else:
 218             return ""
 219     else:
 220         return ""