auto-downloader

自动下载网络学堂课件的脚本 (原创) only for THU

#!/bin/bash
# THU wlxt auto-downloader 
#
# Usage:
#   $chmod a+x auto-downloader
#   $./auto-downloader
#
# dinstein@FREE  [e-mail]    http://dinstein.cublog.cn
#
#

CHAR="gbk" # your local Characters Encoding
echo "Please input username:"
read  USER  #read username
echo "Please input password:"
read -s PASSWD #read password
#echo "Please input your local Characters Encoding  (gbk gbk or utf-8 etc.)"
#read CHAR


echo "Start"

if [ ! -d "gonggao" ]; then
	mkdir gonggao
fi
if [ ! -d "print" ]; then
	mkdir print
fi
if [  -f "new.txt" ]; then 
	cat new.txt >>history.txt
	rm new.txt
	unix2dos history.txt 2>/dev/null
fi

if [  -f "gonggao/new.html" ]; then 
	rm gonggao/new.html
	echo "<H1>What's New</H1>" >> gonggao/new.html
fi


for del in {MyCourse,MyCourse1,MyCourse2,CourseID,pw1,pw2,pw3,cookies,gonggao2.html,gonggao.html};do
if [  -f "$del" ]; then 
	rm $del
fi
done


echo "Get cookies..."
curl -s -c cookies "http://learn.tsinghua.edu.cn/use_go.jsp?userid=$USER&&userpass=$PASSWD" >> pw1  #get cookies
iconv -f gbk -t $CHAR pw1 >> pw2
echo "get Mycourse..."
cat pw2 |grep window.alert|cut -d'(' -f2|cut -d')' -f1 >> pw3
FLAG=`cat pw3|grep "" -c`
if [ $FLAG != 0 ];then
	cat pw3
	rm pw1 pw2 pw3 cookies
	exit
fi

curl -s -b cookies "http://learn.tsinghua.edu.cn/lesson/student/MyCourse.jsp?typepage=1" >>MyCourse1 #get MyCourse
iconv -f gbk -t $CHAR MyCourse1 >> MyCourse2
echo "get CourseID..."
grep course_id MyCourse2|cut -d'=' -f3-|cut -d'"' -f1 >>CourseID  #get CourseID
echo "get MyCourse Name..."
grep 200 MyCourse2|cut -d'(' -f1 |sed 's/[<> ~!@#$%^&*()]//g' >>MyCourse  #get MyCourse name


# gonggao.html begin
if [ -f "gonggao.html" ];then
	rm gonggao.html
fi
echo "
<TABLE height=483 cellSpacing=0 cellPadding=0 width="810%" border=0>
   <TBODY>
         <TR>
	   <TD width="90%" height="100%">
	     <IFRAME name=content_frame src=gonggao/new.html marginWidth=0 marginHeight=0 frameBorder=0 width=800 height="100%">
	     </IFRAME>
	   </TD>
                </TR>
   </TBODY>
</TABLE> 
<H2>Please click the link following</H2>" >>gonggao.html
echo "<A href="gonggao/new.html" target=content_frame>What's new</A> " >> gonggao.html
count1=1
for CourseName in `cat MyCourse`; do
	CourseID=`cat CourseID|head -$count1|tail -1`
	echo "<A href=\"gonggao/"$CourseID".html\" target=content_frame>"$CourseName"</A>" >> gonggao2.html
	count1=$[count1+1]
done
if [ -f gonggao2.html ];then
	iconv -f $CHAR -t gbk gonggao2.html >> gonggao.html
	rm gonggao2.html
fi

#gonggao.html end

count1=1
for CourseName in `cat MyCourse`; do
        echo "checking $CourseName..."
	CourseID=`cat CourseID|head -$count1|tail -1`
	if [ ! -d $CourseName ]; then
		mkdir $CourseName
	fi
	cd $CourseName
	if [ -f down ];then
		rm down*
	fi
	if [ -f cookies2 ];then
		rm cookies2
	fi
	curl -s -b ../cookies -c cookies2 "http://learn.tsinghua.edu.cn/lesson/student/course_locate.jsp?course_id=$CourseID" >>/dev/null #get Course cookies
	curl -s -b cookies2 "http://learn.tsinghua.edu.cn/lesson/student/download.jsp?course_id=$CourseID" >> down #get file informations
	
	iconv -f gbk -t $CHAR down >> down2
	cat down2 |grep uploadFile|cut -d'"' -f10-|cut -d'"' -f1 >>down3 #down3 save the filepath
	cat down2 |grep window.open|cut -d'&' -f2|cut -d'=' -f2|cut -d'_' -f1-|sed -e 's/ /_/g' >>down4 #down4 save the file name	
	#cat down2 |grep window.open|cut -d'&' -f2|cut -d'=' -f2|cut -d'_' -f1-|sed -e 's/ /_/g' -e 's/[0-9]\{7,10\}_//g' -e 's/_[0-9]\{7,10\}//g' >>down4 #down4 save the file name	
	count2=1;
	for FileName in `cat down4`; do
		 FilePath=`cat down3|head -$count2|tail -1`
		if [ ! -f $FileName ]; then
			echo "start download" $CourseName $FileName 
			curl -b cookies2 "http://learn.tsinghua.edu.cn$FilePath" >> $FileName #start download file
			echo $CourseName  $FileName >> ../new.txt
			if [ ! -f no.txt ];then
				cp $FileName ../print/
			fi
		fi
		 count2=$[count2+1]
	done

#gonggao begin	
	flag=0
	if [ -f Temp ];then
		for del in `ls Temp*`;do
			rm $del
		done
	fi
	if [  -f "../gonggao/$CourseID.html" ];then
		rm ../gonggao/$CourseID.html
	fi
	touch Temp
	echo "<H1>"$CourseName"</H1>" >> Temp.$CourseID.html.1
	URL1=`curl -s -b cookies2 "http://learn.tsinghua.edu.cn/public/bbs/getnoteid_student.jsp?course_id=$CourseID" |cut -d"\"" -f2-|cut -d"\"" -f1`
    curl -s -b cookies2 "$URL1" >> Temp.gg1
    cat Temp.gg1 |grep bbs_type|cut -d"'" -f2-|cut -d"'" -f1 >> Temp.URL2
    
    iconv -f gbk -t $CHAR Temp.gg1 >> Temp.gg1.1
   cat Temp.gg1.1 |grep bbs_type |sed -e 's/note_reply.*\&id=\([0-9]*\)/\1.html/g' -e 's/\&course_id=[0-9]*'//g >>Temp.$CourseID.html.1
    	iconv -f $CHAR -t gbk Temp.$CourseID.html.1 >> ../gonggao/$CourseID.html
	
	C=1
    for URL3 in `cat Temp.URL2`;do
       	NoteID=`echo $URL3 |cut -d'=' -f3-|cut -d'&' -f1`	
		if [ ! -f "../gonggao/$NoteID.html" ];then
			curl -s -b cookies2 "http://learn.tsinghua.edu.cn/public/bbs/$URL3" >>Temp.$NoteID
           	iconv -f gbk -t $CHAR Temp.$NoteID >> Temp.$NoteID.1
			cat Temp.$NoteID.1 | sed '1,72d'|grep -v sendmail | sed 's/onMouseOver=\".*\"//g'|sed 's/onMouseOut=\".*\"//g'  >> Temp.$NoteID.html.1
           	if [ $flag -eq 0 ];then
				echo "<H1>"$CourseName"</H1>" >> ../gonggao/new
				flag=1
			fi
			NEW=`cat Temp.$CourseID.html.1|grep $NoteID`
			echo $NEW >> ../gonggao/new
			echo " ">> ../gonggao/new
			C=$[C+1]
	    	iconv -f $CHAR -t gbk Temp.$NoteID.html.1 >> ../gonggao/$NoteID.html
		fi
        done
	rm Temp* 
#gonggao end
		
	rm down* cookies2
	cd ..
	count1=$[count1+1]
done
rm  cookies* MyCourse*  CourseID pw1 pw2 pw3
if [  -f "new.txt" ]; then 
	unix2dos new.txt 2>/dev/null
	echo ""
	echo "New Courseware:"
	cat new.txt
fi

if [  -f "gonggao/new" ]; then
	echo ""
	echo "New gonggao:"
	cat gonggao/new|sed s/\<[^\>]*\>//g
	echo "Please check gonggao.html"
	sed s/$/\<br\>/g gonggao/new >> gonggao/new.html.1
	iconv -f $CHAR -t gbk gonggao/new.html.1 >> gonggao/new.html
	rm gonggao/new gonggao/new.html.1
fi