糗百网页爬取

#!/bin/bash
page=1

function get_html(){

   curl -A "Mozilla/4.0 ??compatible?? MSIE 6.0?? Windows NT 5.0?©" http://www.qiushibaike.com/8hr/page/$page>fun

    echo "This is page$1"

    sed -n '//,/<\/div>/p' fun|sed 's/^<.*//g' |sed '/^ *$/d'>f
    read_by_line

    let page=$1+1

    have_fun $page

}



function have_fun()

{

for i in `seq 10`

do


    Input_user=n

    i=$page

    Input_user=0

    if [ $i -eq 1 ]

    then

        echo "have fun"

        get_html $i

    else

        echo "Continue? [y|N]"

        read Input_user

    fi



    if [[ $Input_user == N ]]

    then

        exit

    else

        get_html $i

    fi

done

}


function read_by_line()
{
    clear
    while read line
    do
        echo
        echo $line |sed 's//\n/g'
        echo
        echo "Continue [y|N]"

        read Input_user</dev/tty
        if [[ $Input_user == N ]]
        then
            exit
        fi
    done <f
}


have_fun

rm -rf f fun