PYTHON文件多线程下载

其实,在一般的文件编程中,这有两个概念要说明:

第一是,下载一个大文件,将这个大文件多为多线程。

第二是,下载N多小文件,将每个线程指定下载多个小文件。

现在实现的是多线程下载一个大文件。

今天完成了一个很有意思的下载图片的功能。想加入多线程功能。

#!/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
'''

import sys
import os
import time
import urllib
from threading import Thread

local_proxies = {'http': 'http://131.139.58.200:8080'}

class AxelPython(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 = 0

    def run(self):
        '''vertual function in Thread'''
        try:
            self.downloaded = os.path.getsize( self.filename )
        except OSError:
            #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.filename
            return
        
        self.oneTimeSize = 16384 #16kByte/time
        print 'task %s will download from %d to %d \n' % (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 )
        while data:
            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 = 0
    for header in headers:
        if header.find('Length') != -1:
            length = header.split(':')[-1].strip()
            length = int(length)
    return length

def SpliteBlocks(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 ))

    return ranges
def islive(tasks):
    for task in tasks:
        if task.isAlive():
            return True
    return False

def paxel(url, output, blocks=6, proxies=local_proxies):
    ''' paxel
    '''
    size = GetUrlFileSize( url, proxies )
    ranges = SpliteBlocks( size, blocks )

    threadname = [ "thread_%d" % i for i in range(0, blocks) ]
    filename = [ "tmpfile_%d" % i for i in range(0, blocks) ]
  
    tasks = []
    for i in range(0,blocks):
        task = AxelPython( threadname[i], url, filename[i], ranges[i] )
        task.setDaemon( True )
        task.start()
        tasks.append( task )
        
    time.sleep( 2 )
    while islive(tasks):
        downloaded = sum( [task.downloaded for task in tasks] )
        process = downloaded/float(size)*100
        show = u'\rFilesize:%d Downloaded:%d Completed:%.2f%%' % (size, downloaded, process)
        sys.stdout.write(show)
        sys.stdout.flush()
        time.sleep( 0.1 )
        print
            
    filehandle = open( output, 'wb+' )
    for i in filename:
        f = open( i, 'rb' )
        filehandle.write( f.read() )
        f.close()
        try:
            os.remove(i)
            pass
        except:
            pass

    filehandle.close()

if __name__ == '__main__':
    url = "http://6.duote.com.cn/geany_setup.zip"
    output = 'geany_setup.zip'
    paxel( url, output, blocks=5, proxies={} )

PYTHON文件多线程下载_第1张图片

你可能感兴趣的:(PYTHON文件多线程下载)