利用cURL实现单个文件分多段同时下载，支持断点续传(修订版) [复制链接]

摘自 http://bbs.chinaunix.net/thread-917952-1-1.html

在ubuntu下测试通过, 适合在支持多线程下载的站点下载文件
可以配合flashgot在firefox中使用

用法：./mycurl url [referedUrl]
第一个参数url是要下载的文件的地址,第二个参数referedUrl是指需要参照的网址(一般不需要,有些网站,比如华军需要此参数)
例如:
./mycurl ftp://xx.xxx.xxx/xxx.rar
或者
./mycurl http://xx.xxx.xx/xxx.rar http://www.xxx.xxx/yy.htm

下面是代码：

#!/bin/bash
####################################################################
#
# Script for curl to support resumable multi-part download.
#
# Tested on Ubuntu
#

url=$1

# How many "parts" will the target file be divided into?
declare -i parts=5

read -ep "Please input the target directory: " targetdir
read -ep "Please input the outfile name: " outfile

[ -z "$targetdir" ] && targetdir="./"
cd $targetdir||exit 2

[ -z "$outfile" ] && outfile=`basename $1`

#Set the referer url
if [ -n "$2" ]; then
refurl="-L -e $2"
else refurl=""
fi

length=`curl $refurl -s -I $url|grep Content-Length|tail -n 1|sed s/[^0-9]//g`

if [ -z "$length" ]; then
echo "cann‘t get the length of the target file"
exit 1
fi
let "length = $length"

#lsession is used to record how many bytes of each subpart should be downloaded
declare -i lsession=$(($length/$parts))

finished="false"

#Assume the available maximum connections on server can reach "parts" at first
maxconn=$parts

while true;
do

for (( i=1; i<=parts ; i=i+1 ))
do

#Array offsetold is used to record how many bytes have been downloaded of each subpart

if [ -e $outfile$i ]; then
            offsetold[$i]=`ls -l $outfile$i|awk ‘{print $5}‘`
      else offsetold[$i]=0
      fi
      let "offsetold[$i] = ${offsetold[$i]}"

done

curr=0

for (( i=1; i<=parts && maxconn>0; i=i+1 ))
do

if [ $i -lt $parts ]; then
            if [ ${offsetold[$i]} -lt $lsession ]; then
                     curl $refurl -r $(($curr+${offsetold[$i]}))-$(($curr+$lsession-1)) $url >> $outfile$i &
         maxconn=$(($maxconn-1))
            fi
      else
            if [ ${offsetold[$i]} -lt $(($length-$(($lsession*$(($parts-1)))))) ]; then
                     curl $refurl -r $(($curr+${offsetold[$i]}))- $url >> $outfile$i &
         maxconn=$(($maxconn-1))
            fi
      fi

      curr=$(($curr+$lsession))

done

#To wait for all curl processes to terminate.

wait

finished="true"
maxconn=0
for (( i=1; i<=parts; i=i+1 ))
do

#Array offsetnew is used to record how many bytes have been downloaded of each subpart

if [ -e $outfile$i ]; then
            offsetnew[$i]=`ls -l $outfile$i|awk ‘{print $5}‘`
      else offsetnew[$i]=0
      fi
      let "offsetnew[$i] = ${offsetnew[$i]}"
      if [ $i -lt $parts ]; then
            if [ ${offsetnew[$i]} -lt $lsession ]; then
                     finished="false"
            fi
      else
            if [ ${offsetnew[$i]} -lt $(($length-$(($lsession*$(($parts-1)))))) ]; then
                     finished="false"
            fi
      fi

#Calculate the "real" available maximum connections supported by server

if [ ${offsetnew[$i]} -gt ${offsetold[$i]} ]; then
maxconn=$(($maxconn+1))
fi
done

if [ "$finished" == "true" ]; then
            break
elif [ $maxconn -eq 0 ]; then
      echo "Some errors may occur. retry 10 sec later..."
      sleep 10
      maxconn=parts
      fi
done

echo "All parts have been downloaded. Merging..."

mv --backup=t $outfile"1" $outfile
for (( i=2; i<=parts; i=i+1))
do
cat $outfile$i >> $outfile
rm $outfile$i
done

echo "Done."

[ 本帖最后由 ypxing 于 2007-4-4 21:45 编辑 ]