合并百度影音的离线数据( with python 2.3)


Posted in Python onAugust 04, 2015

四种格式的解析:

filelist

slicelist

download.cfg

third_party_download.cfg

还是2个文件。替换之前版本即可。

初步测试正常,但时间轴不对。

代码如下:

# -*- coding: UTF-8 -*-
import os
import io
import codecs
import string
import shutil
import moviefmt
import traceback
def coroutine(func):
 def start(*args,**kwargs):
  g = func(*args,**kwargs)
  g.next()
  return g
 return start
@coroutine
def enum_movie_file(target): 
 selectedFolder = (yield)
 for fileitem in os.listdir(selectedFolder):    
  targetFO = os.path.join(selectedFolder,fileitem)  
  if(os.path.isfile(targetFO) == True):
   continue;  
  target.send(targetFO)  
@coroutine
def read_movie_file(funcDisp,target):
 while(True):
  filmFolder = (yield)
  for fileitem in os.listdir(filmFolder):  
   targetfile = os.path.join(filmFolder,fileitem)
   if(os.path.isfile(targetfile) == False):
    continue;   
   filebasename , fext = os.path.splitext(fileitem);    
   if(filebasename[0] == "."):
    continue   
   if(funcDisp.has_key(string.lower(fext))):
    pfn = funcDisp[string.lower(fext)];
    (targetFmt,targetObjects,filmFolderEx,filebasenameEx) = pfn(targetfile)
    if(targetFmt is None):
     continue;
    outputs = filmFolder,filebasename, (targetFmt,targetObjects,filmFolderEx,filebasenameEx);
    # outputs = filmFolder,filebasename, pfn(targetfile);
    target.send(outputs)
    break
def getWindowsText(orgText): 
 try:
  return orgText.encode('gbk') 
 except Exception as exc: 
  print("%s" % exc); 
  return orgText
@coroutine
def create_report(outputfile):
 objTar = codecs.open(outputfile,"w+")  
 while(1):
  filmFolder,filebasename,(extname,piece_list,filmFolderEx,filebasenameEx) = (yield)
  if(extname == None) :
   break;
  objTar.write('cd \"%s\"\r\n' % (filmFolderEx if filmFolderEx else filmFolder))  
  objTar.write('attrib -h *.*\r\n ')  
  if(len(piece_list) >1):
   objTar.write('copy /b ')  
  else:
   objTar.write('copy ')  
  if(len(piece_list) == 1):
   objTar.write('\"%s \"' %(getWindowsText(piece_list[0])))
  else:   
   bFirst = True
   for line in piece_list:
    if(bFirst == False):     
     objTar.write('+\"%s\"' %(getWindowsText(line))) 
    else:     
     objTar.write('\"%s\"' %(getWindowsText(line)))
     bFirst = False
  relpath = os.path.dirname(filmFolder)  
  currentPath = os.path.join(getWindowsText(relpath),getWindowsText(filebasenameEx if filebasenameEx else filebasename)) 
  objTar.write(' %s.%s\r\n' % (currentPath,extname))    
  objTar.write('cd %s\r\n' %(relpath))
 objTar.close()    
if __name__ == '__main__': 
 funcDisp = dict()
 funcDisp['.bdv_0000']= moviefmt.read_bdv_file;
 funcDisp['.rmvb_0000']= moviefmt.read_rmvb_file;
 funcDisp['.mkv_0000']= moviefmt.read_mkv_file;
 funcDisp['.bdv']= moviefmt.read_bdv_index;   
 funcDisp['.mp4_0000']= moviefmt.read_mp4_file;   
 funcDisp['']= moviefmt.read_mkv_2_file; 
 funcDisp['.filelist'] = moviefmt.read_filelist_index;
 funcDisp['.slicelist'] = moviefmt.read_slicelist_index;
 funcDisp['.cfg'] = moviefmt.read_cfg_index;
 try:
  rmf = enum_movie_file(read_movie_file(funcDisp,create_report('film.bat')))
  rmf.send(os.getcwd())
  rmf.close()
 except StopIteration:
  pass
 except Exception as exc: 
  print("%s" % exc); 
  info = traceback.format_exc()
  print(info)      
 print("done");  
# -*- coding: UTF-8 -*-
import os
import io
import sys
import string
import shutil
import codecs
import glob
import re
def read_bdv_index_V1(objFile):
 piece_list= list() 
 for line in objFile:
  if(line[0:4] != 'file'):
   continue;
  strings = string.split(line,'/')
  tarfile = strings[len(strings)-1]  
  tempText = string.strip(tarfile)
  piece_list.append(tempText)
 return "avi",piece_list,None,None
def read_bdv_index_V2(objFile):
 piece_list= list() 
 for line in objFile:  
  nPos = line.count('bdv')
  if( nPos <=0):
   continue;  
  tarfile = line[0:len(line)-2] 
  piece_list.append(tarfile) 
 return "mpeg",piece_list,None,None
def read_bdv_index_V3(objFile):
 piece_list= list()   
 oneFile= os.path.basename(objFile) 
 piece_list.append(oneFile)
 return "avi",piece_list,None,None
def read_bdv_index_V4(objFile):
 piece_list= list() 
 #skip #EXT-X-MEDIA-SEQUENCE
 objFile.readline();
 for line in objFile:
  if(line.startswith('#')==True):
   continue;  
  piece_list.append(line.replace('\r\n',''))
 return "avi",piece_list,None,None
def count_file_item(objFile,extText):
 cItem = 0;
 folderName = os.path.dirname(objFile)
 for fileitem in os.listdir(folderName): 
  filebasename , fext = os.path.splitext(fileitem); 
  if(fext == extText):
   cItem = cItem +1
 return cItem 
def read_bdv_index(filename):
 piece_list= list() 
 ext_type = None
 filmFolderEx = None
 filebasenameEx = None
 # total file count
 cItem = count_file_item(filename,".bdv")
 if(cItem == 1):
  ext_type,piece_list,filmFolderEx,filebasenameEx = read_bdv_index_V3(filename)
 else:
  objFile = codecs.open(filename,'r','utf-8') 
  topline = objFile.readline();
  bdv_ver = topline.replace("\r\n","")
  if(bdv_ver == '#EXTM3U'):
   v3Text = objFile.readline();
   if(v3Text.startswith('#EXT-X-TARGETDURATION') == False):
    ext_type,piece_list = read_bdv_index_V2(objFile)
   else:
    ext_type,piece_list = read_bdv_index_V4(objFile)
  else:
   ext_type,piece_list = read_bdv_index_V1(objFile)
  objFile.close()  
 return (ext_type ,piece_list,filmFolderEx,filebasenameEx)  
def read_bdv_file(filename):
 piece_list= list()
 piece_list.append('*.bdv_*')
 return ('avi',piece_list,None,None)
def read_rmvb_file(filename):
 piece_list= list()
 piece_list.append('*.rmvb_*')
 return ('rmvb',piece_list,None,None)
def read_mkv_file(filename):
 piece_list= list() 
 piece_list.append('*.mkv_*')
 return ('mkv',piece_list,None,None)
def read_mp4_file(filename):
 piece_list= list() 
 piece_list.append('*.mp4_*')
 return ('mp4',piece_list,None,None)
def read_mkv_2_file(filename):
 piece_list= list() 
 piece_list.append('video_*')
 return ('mkv',piece_list,None,None)
def read_filelist_index(filename):
 filmFolderEx = None
 filebasenameEx = None
 objFile = codecs.open(filename,'r','utf-8') 
 topline = objFile.readline();
 bdv_ver = topline.replace("\r\n","")
 if(bdv_ver == '#EXTM3U'):
  v3Text = objFile.readline();
  if(v3Text.startswith('#EXT-X-TARGETDURATION') == False):
   ext_type,piece_list,filmFolderEx,filebasenameEx = read_bdv_index_V2(objFile)
  else:
   ext_type,piece_list,filmFolderEx,filebasenameEx = read_bdv_index_V4(objFile)   
 else:
  ext_type,piece_list,filmFolderEx,filebasenameEx = read_bdv_index_V1(objFile)
 objFile.close()  
 return ("mp4" ,piece_list,filmFolderEx,filebasenameEx)    
def read_slicelist_index(filename):
 piece_list= list() 
 objFile = codecs.open(filename,'r','utf-8') 
 topline = objFile.readline();
 bdv_ver = topline.replace("\r\n","")
 if(bdv_ver == '#BD-SECTION'): 
  while(True):
   currentLine = objFile.readline();
   if(currentLine is None or len(currentLine) ==0):
    break
   piece_list.append(currentLine.replace("\r\n",""))
 objFile.close()  
 return ('mp4',piece_list,None,None)
def read_cfg_index(filename):
 piece_list= list() 
 simpleFileName = os.path.basename(filename)
 if(simpleFileName == 'download.cfg'):
  simpleFolderName = os.path.dirname(filename)
  chkLst =glob.glob(os.path.join(simpleFolderName, '*.filelist'));
  if(chkLst is not None and len(chkLst) >0):
   return (None,None,None,None)
  chkLst =glob.glob(os.path.join(simpleFolderName, '*.bdv'));
  piece_list.append(chkLst[0]);
  return ('avi',piece_list)
 if(simpleFileName != 'third_party_download.cfg'):
  raise Exception('unknown format file : %s' %(filename))
 reObj = re.compile('(?P<var_key>\S+):(?P<var_value>\S+)',re.IGNORECASE)
 lineDict = dict()
# [orgLines.append(orgLine.replace("\r\n","") ) for orgLine in codecs.open(filename,'r','utf-8')]
 for orgLine in codecs.open(filename,'r','utf-8'):
  orgLine = orgLine.replace("\r\n","")  
  results = reObj.findall(orgLine)
  if(len(results) > 0):
   lineDict[results[0][0]] =results[0][1]
 resID = lineDict["resource"] 
 vid = lineDict[ '%s_vid'%(resID)]
 targetName = lineDict["video_name"]
 simpleFolderName = os.path.dirname(filename)
 upLevelFolderName = os.path.dirname(simpleFolderName)
 targetFolder = os.path.join(upLevelFolderName,vid)
 targetFolder = os.path.join(targetFolder,vid)
 chkLst =glob.glob(os.path.join(targetFolder, '*.%s*' %(resID)));
 targetFile = chkLst[0] 
  #skip #EXT-X-MEDIA-SEQUENCE 
 for line in codecs.open(targetFile,'r','utf-8'):
  if(line.startswith('#')==True):
   continue;  
  line = os.path.basename(line.replace('\r\n',''))
  line = os.path.basename(line.replace('\r',''))
  line = os.path.basename(line.replace('\n',''))
  piece_list.append(line )
 return ('avi',piece_list,targetFolder,targetName)

以上就是本文的全部内容,希望大家喜欢。

Python 相关文章推荐
python 获取et和excel的版本号
Apr 09 Python
python编写简单爬虫资料汇总
Mar 22 Python
Python编程产生非均匀随机数的几种方法代码分享
Dec 13 Python
python的staticmethod与classmethod实现实例代码
Feb 11 Python
使用python 3实现发送邮件功能
Jun 15 Python
Python使用numpy产生正态分布随机数的向量或矩阵操作示例
Aug 22 Python
python3中property使用方法详解
Apr 23 Python
使用Python的Turtle库绘制森林的实例
Dec 18 Python
python3实现在二叉树中找出和为某一值的所有路径(推荐)
Dec 26 Python
Django+Celery实现动态配置定时任务的方法示例
May 26 Python
利用PyQt5+Matplotlib 绘制静态/动态图的实现代码
Jul 13 Python
Python LMDB库的使用示例
Feb 14 Python
python利用datetime模块计算时间差
Aug 04 #Python
使用python加密自己的密码
Aug 04 #Python
Python实现字典的key和values的交换
Aug 04 #Python
python利用装饰器进行运算的实例分析
Aug 04 #Python
浅谈python多线程和队列管理shell程序
Aug 04 #Python
python中使用序列的方法
Aug 03 #Python
python实现备份目录的方法
Aug 03 #Python
You might like
php获取文件内容最后一行示例
2014/01/09 PHP
smarty高级特性之对象的使用方法
2015/12/25 PHP
PHP内核学习教程之php opcode内核实现
2016/01/27 PHP
JSP中使用JavaScript动态插入删除输入框实现代码
2014/06/13 Javascript
JavaScript动态改变表格单元格内容的方法
2015/03/30 Javascript
文本框只能输入数字的js代码(含小数点)
2016/07/10 Javascript
文件上传,iframe跨域数据提交的实现
2016/11/18 Javascript
JS中实现函数return多个返回值的实例
2017/02/21 Javascript
jQuery实现三级联动效果
2017/03/02 Javascript
vue2.0 axios跨域并渲染的问题解决方法
2018/03/08 Javascript
浅谈React之状态(State)
2018/09/19 Javascript
element-ui table行点击获取行索引(index)并利用索引更换行顺序
2020/02/27 Javascript
[02:47]2018年度DOTA2最佳辅助位选手4号位-完美盛典
2018/12/17 DOTA
在Django中进行用户注册和邮箱验证的方法
2016/05/09 Python
python魔法方法-属性访问控制详解
2016/07/25 Python
Python 遍历列表里面序号和值的方法(三种)
2017/02/17 Python
答题辅助python代码实现
2018/01/16 Python
使用Python实现企业微信的自动打卡功能
2019/04/30 Python
Python实现爬取亚马逊数据并打印出Excel文件操作示例
2019/05/16 Python
Python使用scrapy爬取阳光热线问政平台过程解析
2019/08/14 Python
Python的控制结构之For、While、If循环问题
2020/06/30 Python
flask项目集成swagger的方法
2020/12/09 Python
真正了解CSS3背景下的@font face规则
2017/05/04 HTML / CSS
KIKO MILANO荷兰网上商店:意大利专业化妆品品牌
2017/05/12 全球购物
流行文化收藏品:Sideshow(DC漫画,星球大战,漫威)
2019/03/17 全球购物
世界经理人咨询有限公司面试
2014/09/23 面试题
初一生物教学反思
2014/01/18 职场文书
优秀公益广告词大全
2014/03/19 职场文书
办公室主任岗位承诺书
2014/05/29 职场文书
爬山的活动方案
2014/08/16 职场文书
十佳标兵事迹材料
2014/08/18 职场文书
收款委托书范本
2014/09/11 职场文书
员工离职证明范本
2015/06/12 职场文书
怎样写观后感
2015/06/19 职场文书
干部培训简讯
2015/07/20 职场文书
MySQL 8.0 驱动与阿里druid版本兼容问题解决
2021/07/01 MySQL