au3抓取糗事百科网站

au3抓取糗事百科网站

网址：'http://www.qiushibaike.com/8hr/page/' & $pagenum & '?s=4512150'

#include <IE.au3>
#include <File.au3>
#include <String.au3>
#include <Array.au3>
#include <Debug.au3>
#include <Date.au3>

;code try to collect Qiushibaike stories in qiushibaike.com
Local $strUrl1 = "http://www.qiushibaike.com/8hr/page/2?s=4512150"
Local $filename1 = "qiushibaike"
$filename1 = $filename1 & '_' & @MON
$filename1 = $filename1  & @MDAY
$filename1 = $filename1 & '.txt'
Local $filesave = @TempDir & "\qb.html"
Local $pageindex
Local $startindex = 2
Local $endindex = 10
Local $sHTML
Local $storycount = 0
_FileCreate($filename1)
Local $file = FileOpen($filename1, 1)
If $file = -1 Then
    MsgBox(0, "Error", "Unable to open file.")
    Exit
 EndIf
For $pageindex = $startindex To $endindex Step 1
   $strUrl1 = MakeUpUrl($pageindex)
   Local $hDownload = InetGet($strUrl1, $filesave, 1, 1)
   Do
       Sleep(250)
   Until InetGetInfo($hDownload, 2)
   Local $nBytes = InetGetInfo($hDownload, 0)
   InetClose($hDownload)
  ConsoleWrite ($pageindex & '/' &  $endindex &" --- down bytes = " &$nBytes & @LF)
  $fsize = $nBytes
  ;ConsoleWrite($pageindex & '- filesize = '& $fsize & @LF)
  $ftemp = FileOpen($filesave, 0)
  $getsize=    FileGetSize ($filesave)
  $sHTML = FileRead($ftemp, $getsize)
  FileClose($ftemp)
  FileDelete($filesave)
  Local $aArray = StringRegExp($sHTML, '(?<=<span>)\n+[^/]+\n+(?=</span>)', 3)
  ConsoleWrite("    array size = " & UBound($aArray) & @CRLF)
  For $i = 0 To (UBound($aArray) - 1) Step 1
     Local $item = $aArray[$i]
     If StringLen($item) > 0 Then
    $strnum = $storycount +1
    $strnum = $strnum & "." &@CRLF
    FileWrite($file, $strnum)
    $storycontent = StringReplace($item, @LF, '')
     $storycontent = $storycontent & @CRLF
     FileWrite($file, $storycontent)
     $storycount = $storycount + 1
     EndIf
     Next
   Next
FileClose($file)
MsgBox(0, "QSBK", "Complete, story count = "&$storycount & ', story=' & $filename1)
Exit
Func MakeUpUrl($pagenum)
   $strUrl = 'http://www.qiushibaike.com/8hr/page/' & $pagenum & '?s=4512150'
   return $strUrl
   EndFunc