#!/usr/bin/python#-*- coding: utf-8 -*-#filename: paxel.py
'''It is a multi-thread downloading tool
It was developed follow axel.
Author: volans
E-mail: volansw [at] gmail.com'''
importsysimportosimporttimeimporturllibfrom threading importThread
local_proxies= {'http': 'http://131.139.58.200:8080'}classAxelPython(Thread, urllib.FancyURLopener):'''Multi-thread downloading class.
run() is a vitural method of Thread.'''
def __init__(self, threadname, url, filename, ranges=0, proxies={}):
Thread.__init__(self, name=threadname)
urllib.FancyURLopener.__init__(self, proxies)
self.name=threadname
self.url=url
self.filename=filename
self.ranges=ranges
self.downloaded=0defrun(self):'''vertual function in Thread'''
try:
self.downloaded=os.path.getsize( self.filename )exceptOSError:#print 'never downloaded'
self.downloaded =0#rebuild start poind
self.startpoint = self.ranges[0] +self.downloaded#This part is completed
if self.startpoint >= self.ranges[1]:print 'Part %s has been downloaded over.' %self.filenamereturnself.oneTimeSize= 16384 #16kByte/time
print 'task %s will download from %d to %d' % (self.name, self.startpoint, self.ranges[1])
self.addheader("Range", "bytes=%d-%d" % (self.startpoint, self.ranges[1]))
self.urlhandle=self.open( self.url )
data=self.urlhandle.read( self.oneTimeSize )whiledata:
filehandle= open( self.filename, 'ab+')
filehandle.write( data )
filehandle.close()
self.downloaded+=len( data )#print "%s" % (self.name)
#progress = u'\r...'
data=self.urlhandle.read( self.oneTimeSize )def GetUrlFileSize(url, proxies={}):
urlHandler= urllib.urlopen( url, proxies=proxies )
headers=urlHandler.info().headers
length=0for header inheaders:if header.find('Length') != -1:
length= header.split(':')[-1].strip()
length=int(length)returnlengthdefSpliteBlocks(totalsize, blocknumber):
blocksize= totalsize/blocknumber
ranges=[]for i in range(0, blocknumber-1):
ranges.append((i*blocksize, i*blocksize +blocksize - 1))
ranges.append(( blocksize*(blocknumber-1), totalsize -1))returnrangesdefislive(tasks):for task intasks:iftask.isAlive():returnTruereturnFalsedef paxel(url, output, blocks=6, proxies=local_proxies):'''paxel'''size=GetUrlFileSize( url, proxies )
ranges=SpliteBlocks( size, blocks )
threadname= [ "thread_%d" % i for i inrange(0, blocks) ]
filename= [ "tmpfile_%d" % i for i inrange(0, blocks) ]
tasks=[]for i inrange(0,blocks):
task=AxelPython( threadname[i], url, filename[i], ranges[i] )
task.setDaemon( True )
task.start()
tasks.append( task )
time.sleep(2)whileislive(tasks):
downloaded= sum( [task.downloaded for task intasks] )
process= downloaded/float(size)*100show= u'\rFilesize:%d Downloaded:%d Completed:%.2f%%' %(size, downloaded, process)
sys.stdout.write(show)
sys.stdout.flush()
time.sleep(0.5)
filehandle= open( output, 'wb+')for i infilename:
f= open( i, 'rb')
filehandle.write( f.read() )
f.close()try:
os.remove(i)pass
except:passfilehandle.close()if __name__ == '__main__':
url= "http://www.pygtk.org/dist/pygtk2-tut.pdf"output= 'pygtk2.pdf'paxel( url, output, blocks=4, proxies={} )