@echo off&setlocal enabledelayedexpansion
title "期刊采集器(write by inittab 2010.3)"
rem 多次运行后，如出现“访问过于频繁，请稍后再试”请自行脚本中则修改适当的延时
if exist tmp1.txt del /Q tmp?.txt
echo;set oDOM = WScript.GetObject(WScript.Arguments(0))>htm2txt.vbs
echo;do until oDOM.readyState = ^"complete^">>htm2txt.vbs
echo;WScript.sleep 200>>htm2txt.vbs
echo;loop>>htm2txt.vbs
echo;WScript.Echo oDOM.Body.InnerText>>htm2txt.vbs
:start
set/p qs=请输入电脑爱好者期数:（1－24,按q退出):
if "%qs%"=="q" goto :eof
if %qs% gtr 24 echo\输入有误!!,按任意键重新输入！&pause>nul&goto start
if %qs% leq 0  echo\输入有误!!,按任意键重新输入！&pause>nul&goto start
if not exist 第!qs!期 md 第!qs!期
echo;正在下载，请稍等...
:::以下一行网址修改可采集http://lnlib.vip.qikan.com（要保留%qs%变量）其他杂志内容。
curl -s "http://lnlib.vip.qikan.com/Mag.aspx?issn=1005-0043&Year=2009&Issue=%qs%" | sed -n "/期刊目录/,/客服热线/p">tmp1.txt
:::以下一行为延时，改数字调整延时时间。
ping localhost -n 2 >nul
sed "s/article.aspx?/>/g;s/\" title=\"/@</g;1,2d;/客服/d" tmp1.txt>tmp2.txt
htox32c /ip tmp2.txt>tmp3.txt 2>nul
echo;建立目录结构并下载处理内容
set/a m=0
for /f "tokens=2,* delims=.@［］ " %%a in (tmp3.txt) do (
if "%%b"=="" (set var=00!m!&set dir1=!var:~-3!_%%a&set/a m+=1,n=0&md "第!qs!期\!dir1!" 2>nul) else (
set var=00!n!&set dir2=!var:~-3!_%%b&set/an+=1&md "第!qs!期\!dir1!\!dir2!" 2>nul
set "url=http://lnlib.vip.qikan.com/article.aspx?%%a"
cscript //NoLogo //e:vbscript htm2txt.vbs "!url!" | sed -n "/^首页/,/存入我的阅览室/p" | sed "s/首页 ->//g;s/字体.*//g">tmp1.txt
ping localhost -n 1 >nul
findstr /C:"[2]" tmp1.txt | sed "s/[^2-9 ]//g">tmp2.txt
set str=&set/p str=<tmp2.txt
for %%i in (!str!) do (
set/a subnum=%%i-1&cscript //NoLogo //e:vbscript htm2txt.vbs "!url!-!subnum!" | sed -n "/^首页/,/存入我的阅览室/p" | sed "/首页 ->/d">>tmp1.txt
ping localhost -n 1 >nul
)
sed "/存入我的阅览室/d" tmp1.txt>"第!qs!期\!dir1!\!dir2!\src.txt"
:::::以下两行为调试用，可删除
echo;&echo;======================第!qs!期\!dir1!\!dir2!===========================================
type "第!qs!期\!dir1!\!dir2!\src.txt"
)
)
:::::清理临时文件
del /Q tmp?.txt&echo;本期内容采集完毕！！！
 
复制代码