rec10/trunk/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
  13 def detName(title,path):
  14     tt=detNameType(title, path)
  15     if tt['type']=="C"or tt['type']=="D":
  16         tt['num']=detNum(tt['title'],tt['folder'])
  17 def detNum(title,movepath):
  18     """
  19         #
  20     """
  21     files=os.listdir(movepath)
  22     ff=[]
  23     maxnum=0
  24     for file in files:
  25         if os.path.isfile(file):
  26             name=os.path.splitext(os.path.split(file)[1])
  27             p1=detNameType(name,movepath)
  28             print p1['title']+" "+str(p1['num'])
  29             time1=time.localtime(os.path.getmtime(file))
  30             ff.append([p1['num'],p1['title'],time1])
  31             if maxnum<p1['num']+1:
  32                 maxnum=p1['num']+1
  33     f3=[maxnum]
  34     for i in range(0, maxnum, 1):
  35         f3[i]=[]
  36     for f2 in ff:
  37         f3[f2[0]].append(f2)
  38     for i in range(maxnum):
  39         for j in range(i+1,maxnum,1):
  40
  41 def guessDeltaDay(num1,date1,num2,date2):
  42     dd=date1-date2
  43     if date1<date2:
  44         dd=date2-date1
  45     d1=dd.days
  46     dd=dd+datetime.timedelta(hours=6)
  47     if dd!=d1:
  48
  49 def detNameType(title,path):
  50     """
  51     type A ---title#<number>
  52     type B ---title#<number>subtitle
  53     type C ---title subtitle
  54     type D ---title(without number)
  55     path --search reflexively
  56     """
  57     #print title
  58     #rA=re.compile(".+(?P<title>)#\d(?P<num>)\s[0,10]\z")
  59     rA=re.compile("(.+)#(\d*)\s*\Z")
  60     tA=rA.match(title)
  61     rB=re.compile("(.+)#(\d*)\s*(\D*)")
  62     tB=rB.match(title)
  63     ret={'title':"",'type':"",'num':0,'subtitle':"",'folder':""}
  64     if tA:
  65         print "typeA"
  66         print "title="+tA.group(1)
  67         print "num="+tA.group(2)
  68         ret['type']="A"
  69         ret['title']=tA.group(1)
  70         ret['num']=int(tA.group(2))
  71         ret['folder']=searchFolder(tA.group(1),path)
  72     elif tB:
  73         print "typeB"
  74         print "title="+tB.group(1)
  75         print "num="+tB.group(2)
  76         print "subtitle="+tB.group(3)
  77         ret['type']="B"
  78         ret['title']=tB.group(1)
  79         ret['num']=int(tB.group(2))
  80         ret['folder']=searchFolder(tB.group(1),path)
  81         ret['subtitle']=tB.group(3)
  82     else:#type C or type D
  83         #fold=searchFolder(title, path)
  84         ts=title.split(" ")
  85         tt=""
  86         rt=["",0,""]
  87         for t in ts:
  88             tt=tt+" "+t
  89             ft1=searchFolder(tt,path)
  90             print tt
  91             print ft1
  92             if ft1!="":
  93                 #print rt
  94                 #print ft1[0]+" : "+str(ft1[1])
  95                 if ft1[1]>rt[1]:
  96                     rt[0]=tt
  97                     rt[1]=ft1[1]
  98                     rt[2]=ft1[0]
  99                     #print rt
 100         #print "title="+rt[0][1:]+"/"
 101         #print "subtitle = "+title.replace(rt[0][1:],"")[1:]
 102         ret['title']=rt[0][1:]
 103         ret['num']=""
 104         ret['folder']=rt[2]
 105         ret['subtitle']=title.replace(rt[0][1:],"")[1:]
 106         if ret['subtitle'].replace(" ","")=="":
 107             ret['type']="D"
 108         else:
 109             ret['type']="C"
 110     return ret
 111 def searchFolder(title,path):
 112     """
 113     titleからフォルダーを探す
 114     """
 115     folderpath=os.listdir(path)
 116     lfpath=[]
 117     ngram=[]
 118     for ft in folderpath:
 119         fullpath=os.path.join(path, ft)
 120         if os.path.isdir(fullpath):
 121             lfpath.append(fullpath)
 122             ftt=os.listdir(fullpath)
 123             if len(ftt)>0:
 124                 for ft2 in ftt:
 125                     folderpath.append(os.path.join(fullpath, ft2))
 126         else:
 127             lfpath.append(fullpath)
 128     for dirp in lfpath:
 129         cmpp=""
 130         appp=""
 131         if os.path.isdir(dirp):
 132             cmpp=os.path.dirname(dirp)
 133             appp=dirp
 134         else:
 135             cmpp=os.path.basename(dirp)
 136             appp=os.path.dirname(dirp)
 137         p=n_gram.trigram(title.decode("utf-8"),cmpp.decode("utf-8"))
 138         if p>0:
 139             ngram.append((p,appp))
 140     ngram=list(set(ngram))
 141     ngram.sort()
 142     ngram.reverse()
 143     if len(ngram)>0:
 144         #print title + ngram[0][1] + " : "+str(ngram[0][0])
 145         if ngram[0][0]>300:
 146             return [ngram[0][1],ngram[0][0]]
 147         else:
 148             return ""
 149     else:
 150         return ""