rec10/trunk/src/classify.py

   1 #!/usr/bin/python
   2 # coding: UTF-8
   3 # Rec10 TS Recording Tools
   4 # Copyright (C) 2009-2010 Yukikaze
   5
   6 import glob
   7 import n_gram
   8 import os
   9 import os.path
  10 import re
  11 import time
  12 import optparse
  13 import sys
  14 import shutil
  15
  16 import configreader
  17 import recdblist
  18 recordedpath=unicode(configreader.getpath("recorded"),'utf-8')
  19 recpath=unicode(configreader.getpath("recpath"),'utf-8')
  20 def search_b25(path):
  21     """
  22     録画フォルダを検索し、各QUEの実行状況を表示する
  23     """
  24     b25list = glob.glob(path + "/*.b25")
  25     for b25f in b25list:
  26         ##b25f is title.ts.b25  avi is title.avi
  27         dir = os.path.split(b25f)[0]
  28         title = os.path.split(b25f)[1]
  29         title = title.replace(".ts.b25", "")
  30         avipath = os.path.join(dir, title + ".avi")
  31         mp4path = os.path.join(dir, title + ".mp4")
  32         tspath = os.path.join(dir, title + ".ts")
  33         b25f = b25f.replace(".ts.b25", ".avi")
  34         mode = "ts"
  35         if os.path.isfile(tspath):##tsファイルが存在している
  36             dtime = time.time()-os.path.getmtime(tspath)
  37             dtime = int(dtime)
  38             if dtime > 120:
  39                 if os.path.getsize(tspath) > 1 * 1000 * 1000:##最終更新から22分以上経過かつ1MB以上
  40                     mode = "avi"
  41                 else:
  42                     mode = "tsmiss"
  43             else:
  44                 mode = "ts"
  45         if os.path.isfile(mp4path):##tsファイルが存在している
  46             dtime = time.time()-os.path.getmtime(mkvpath)
  47             dtime = int(dtime)
  48             if dtime > 120:
  49                 if os.path.getsize(avipath) > 1 * 1000 * 1000:##最終更新から22分以上経過かつ1MB以上
  50                     mode = "fin"
  51                 else:
  52                     mode = "avimiss"
  53             else:
  54                 mode = "avi"
  55         recdblist.printutf8(title + ":" + mode)
  56 def search_avi(temppath,recpath):
  57     """
  58     録画一時フォルダ内のaviファイルを検索
  59     """
  60     avilist = glob.glob(temppath + "/*.avi")
  61     ret=[]
  62     for avif in avilist:
  63         ##b25f is title.ts.b25  avi is title.avi
  64         dir = os.path.split(avif)[0]
  65         title = os.path.split(avif)[1]
  66         title = title.replace(".avi", "")
  67         avipath = os.path.join(dir, title + ".avi")
  68         dtime = time.time()-os.path.getmtime(avipath)
  69         dtime = int(dtime)
  70         if dtime > 300:
  71             if veryfySize(avipath):
  72                 ret.append(title)
  73     return ret
  74 def search_mkv(temppath,recpath):
  75     """
  76     録画一時フォルダ内mkvファイルを検索
  77     """
  78     avilist = glob.glob(temppath + "/*.mkv")
  79     ret=[]
  80     for avif in avilist:
  81         ##b25f is title.ts.b25  avi is title.avi
  82         dir = os.path.split(avif)[0]
  83         title = os.path.split(avif)[1]
  84         title = title.replace(".mkv", "")
  85         avipath = os.path.join(dir, title + ".mkv")
  86         dtime = time.time()-os.path.getmtime(avipath)
  87         dtime = int(dtime)
  88         if dtime > 300:
  89             if veryfySize(avipath):
  90                 ret.append(title)
  91     return ret
  92 def search_mp4(temppath,recpath):
  93     """
  94     録画一時フォルダ内mp4ファイルを検索
  95     """
  96     avilist = glob.glob(temppath + "/*.mp4")
  97     ret=[]
  98     for avif in avilist:
  99         ##b25f is title.ts.b25  avi is title.avi
 100         dir = os.path.split(avif)[0]
 101         title = os.path.split(avif)[1]
 102         title = title.replace(".mp4", "")
 103         avipath = os.path.join(dir, title + ".mp4")
 104         dtime = time.time()-os.path.getmtime(avipath)
 105         dtime = int(dtime)
 106         if dtime > 300:
 107             if veryfySize(avipath):
 108                 ret.append(title)
 109     return ret
 110 def veryfySize(path):
 111     #vsize=[297,497,596,1196]#SD 30m 1h 1.5h 2h
 112     vsize=[245,275,295,591,830]
 113     vsize=vsize+[325,449,560,590,602,690,805,860,1014,1138,1237]
 114     vsize=vsize+[261,535,540,616,740]#HD 30m 1h
 115     #vsize=vsize+[381,895,447]
 116     ret = 0
 117     for size in vsize:
 118         if os.path.getsize(path)>(size-10)*1024*1024 and os.path.getsize(path)<(size+10)*1024*1024:
 119             ret=1
 120     if os.path.getsize(path)>270*1024*1024:
 121         ret=1
 122     return ret
 123 def searchFolder(title,path):
 124     """
 125     titleにマッチするフォルダを探し出す。
 126     """
 127     folderpath=os.listdir(path)
 128     lfpath=[]
 129     ngram=[]
 130     for ft in folderpath:
 131         fullpath=os.path.join(path.encode('utf-8'), ft)
 132         if os.path.isdir(fullpath):
 133             lfpath.append(fullpath)
 134             ftt=os.listdir(fullpath)
 135             if len(ftt)>0:
 136                 for ft2 in ftt:
 137                     try:
 138                         folderpath.append(os.path.join(fullpath, ft2))
 139                     except Exception, inst:
 140                         #print type(inst)
 141                         #print inst
 142                         ""
 143         else:
 144             lfpath.append(fullpath)
 145     for dirp in lfpath:
 146         cmpp=""
 147         appp=""
 148         ntitle=title
 149         if os.path.isdir(dirp):
 150             cmpp=os.path.dirname(dirp)
 151             appp=dirp
 152         else:
 153             cmpp=os.path.basename(dirp)
 154             appp=os.path.dirname(dirp)
 155         ntitle=getTitle(title)
 156         cmpp=getTitle(cmpp)
 157         p=n_gram.trigram(ntitle,cmpp)
 158         if p>0:
 159             ngram.append((p,appp))
 160     ngram=list(set(ngram))
 161     ngram.sort()
 162     ngram.reverse()
 163     if len(ngram)>0:
 164         #recdblist.printutf8(title + ngram[0][1] + " : "+str(ngram[0][0]))
 165         if ngram[0][0]>400:
 166             return ngram[0][1]
 167         else:
 168             return ""
 169     else:
 170         return ""
 171 def getTitle(title):
 172     rT=re.compile("(.+)_(\d*)\Z")
 173     tT=rT.match(title)
 174     rT2=re.compile("(.+)_(.+)_(\d*)\Z")#_(aichi)_2010-02-06T01:59:00.mkv
 175     tT2=rT2.match(title)
 176     rT3=re.compile("(.+)_(.+)_\d-\d-\dT\d:\d:\d\Z")#_(aichi)_2010-02-06T01:59:00.mkv
 177     tT3=rT3.match(title)
 178     rT4=re.compile("(.+)_(.+)_\d-\d-\dT\d-\d-\d\Z")#_(aichi)_2010-02-06T01-59-00.mkv
 179     tT4=rT4.match(title)
 180     ntitle=title
 181     if tT :
 182         ntitle=tT.group(0)
 183     elif tT2:
 184         ntitle=tT2.group(0)
 185     elif tT3:
 186         ntitle=tT3.group(0)
 187     elif tT4:
 188         ntitle=tT4.group(0)
 189     return ntitle
 190 def execMove(title,temppath,recpath):
 191     srcpath=os.path.join(temppath,title+".mkv")
 192     #desttitle=destName(title, temppath, recpath)
 193
 194     sf=searchFolder(title, recpath)
 195     if sf!="":
 196         destpath=os.path.join(sf,destNameMKV(title, temppath, sf)+".mkv")
 197         #os.path.join(sf,desttitle+".mkv")
 198         #recdblist.printutf8(srcpath)
 199         #recdblist.printutf8(destpath)
 200         recdblist.printutf8("moving now..")
 201         recdblist.printutf8(srcpath+" : "+destpath)
 202         print srcpath
 203         shutil.move(srcpath, destpath)
 204         #shutil.copy(srcpath, destpath)
 205         delpath=[os.path.join(temppath,title+".ts")]
 206         delpath.append(os.path.join(temppath,title+".avi"))
 207         delpath.append(os.path.join(temppath,title+".120.avi"))
 208         delpath.append(os.path.join(temppath,title+".timecode.txt"))
 209         delpath.append(os.path.join(temppath,title+".aac"))
 210         delpath.append(os.path.join(temppath,title+".ts.b25"))
 211         delpath.append(os.path.join(temppath,title+".ts.tsmix"))
 212         delpath.append(os.path.join(temppath,title+".ts.log"))
 213         delpath.append(os.path.join(temppath,title+".sa.avi"))
 214         delpath.append(os.path.join(temppath,title+".sa.avi.log"))
 215         delpath.append(os.path.join(temppath,title+".log"))
 216         for dp in delpath:
 217             try:
 218                 os.remove(dp)
 219                 ""
 220             except:
 221                 ""
 222 def destName(title,temppath,recpath):
 223     """
 224     titleから移動先に作るべきファイル名を作り出す。
 225     """
 226     dstpath=os.path.join(recpath,title+".avi")
 227     srcpath=os.path.join(temppath,title+".avi")
 228     if os.path.exists(dstpath):
 229         gmtime=time.gmtime(os.path.getmtime(srcpath))
 230         times=time.strftime("%y%m%d",gmtime)
 231         title=title+"_"+times
 232     return title
 233 def destNameMKV(title,temppath,recpath):
 234     """
 235     titleから移動先に作るべきファイル名を作り出す。
 236     """
 237     dstpath=os.path.join(recpath,title+".mkv")
 238     srcpath=os.path.join(temppath,title+".mkv")
 239     if os.path.exists(dstpath):
 240         gmtime=time.gmtime(os.path.getmtime(srcpath))
 241         times=time.strftime("%y%m%d",gmtime)
 242         title=title+"_"+times
 243     return title
 244 def destNameMP4(title,temppath,recpath):
 245     """
 246     titleから移動先に作るべきファイル名を作り出す。
 247     """
 248     dstpath=os.path.join(recpath,title+".mp4")
 249     srcpath=os.path.join(temppath,title+".mp4")
 250     if os.path.exists(dstpath):
 251         gmtime=time.gmtime(os.path.getmtime(srcpath))
 252         times=time.strftime("%y%m%d",gmtime)
 253         title=title+"_"+times
 254     return title
 255 def detName(path,title):
 256     """
 257     type A ---title#<number>
 258     type B ---title#<number>subtitle
 259     type C ---title subtitle
 260     type D ---title(without number)
 261     """
 262     #if re.match("#\d\s[0,10]\z|#[0-9]\s[0,10]\z", title)
 263     #    recdblist.printutf8("typeA")
 264     #elif re.match("#\d\s[0,10].|#[0-9]\s[0,10].", title)
 265     #    recdblist.printutf8("typeB")
 266
 267 def detNameType(title):
 268     """
 269     type A ---title#<number>
 270     type B ---title#<number>subtitle
 271     type C ---title subtitle
 272     type D ---title(without number)
 273     """
 274     recdblist.printutf8(title)
 275     #rA=re.compile(".+(?P<title>)#\d(?P<num>)\s[0,10]\z")
 276     rA=re.compile("(.+)#(\d*)\s*\Z")
 277     tA=rA.match(title)
 278     rB=re.compile("(.+)#(\d*)\s*(\D*)")
 279     tB=rB.match(title)
 280     if tA:
 281         recdblist.printutf8("typeA")
 282         recdblist.printutf8("title="+tA.group(1))
 283         recdblist.printutf8("num="+tA.group(2))
 284     elif tB:
 285         recdblist.printutf8("typeB")
 286         recdblist.printutf8("title="+tB.group(1))
 287         recdblist.printutf8("num="+tB.group(2))
 288         recdblist.printutf8("subtitle="+tB.group(3))
 289 if __name__ == "__main__":
 290     usage="usage: %prog read -h"
 291     version="%prog 0.7.0beta1"
 292     parser=optparse.OptionParser(usage=usage,version=version)
 293     parser.add_option("-s","--search",action="store",type="string",dest="ltitle",default="",metavar="TITLE",help="test to search where the title should be contained(test for -e)")
 294     parser.add_option("-A","--Auto",action="store_true",dest="auto",default=False,help="auto classifying mode(not a test)")
 295     parser.add_option("-e","--exec",action="store",type="string",dest="etitle",default="",metavar="TITLE",help="exec move(not a test)")
 296     parser.add_option("-l","--list",action="store_true",dest="list",default=False,help="File listing mode(test for -a)")
 297     (opts,args)=parser.parse_args(sys.argv)
 298     if opts.ltitle!="":
 299         #print opts.ltitle
 300         ltitle=unicode(opts.ltitle,'utf-8')
 301         sf=searchFolder(ltitle, recordedpath)
 302         recdblist.printutf8(os.path.join(sf, destNameMKV(ltitle,recpath,sf)+".mkv"))
 303         #print os.path.join(sf, destNameMP4(ltitle,recpath,sf)+".mp4")
 304     elif opts.auto:
 305         sa=search_mkv(recpath, recordedpath)
 306         for t in sa:
 307             recdblist.printutf8(u"自動推測実行中")
 308             sf=searchFolder(t,recordedpath)
 309             if sf!="":
 310                 recdblist.printutf8(u"移動先")
 311                 recdblist.printutf8(t+" : "+os.path.join(sf, destNameMKV(t, recpath,sf)+".mkv"))
 312                 recdblist.printutf8(u"実行中")
 313                 execMove(t, recpath, recordedpath)
 314             else:
 315                 recdblist.printutf8(t+" can't find matching folder.")
 316     elif opts.list:
 317         sa=search_mkv(recpath, recordedpath)
 318         for t in sa:
 319             sf=searchFolder(t,recordedpath)
 320             if sf!="":
 321                 recdblist.printutf8(t+u" : "+os.path.join(sf, destNameMKV(t, recpath,sf)+".mkv"))
 322     elif opts.etitle!="":
 323         etitle=unicode(opts.etitle,'utf-8')
 324         execMove(etitle,recpath, recordedpath)
 325