au3抓取不得姐网站

Posted 2020-10-07

au3抓取不得姐网站

网站地址：http://www.budejie.com/text/

用的正则有些别扭，见笑。

代码：

#include <IE.au3>
#include <File.au3>
#include <String.au3>
#include <Array.au3>
#include <Debug.au3>
#include <Date.au3>
;code try to collect budejie stories of www.budejie.com

Local $strUrl1 = "http://www.budejie.com/text/"
Local $filename1 = "budejie"
$filename1 = $filename1 & ‘_‘ & @MON
$filename1 = $filename1  & @MDAY
$filename1 = $filename1 & ‘.txt‘
Local $filesave = @TempDir & "\budejie.html"
Local $pageindex
Local $startindex = 1
Local $endindex = 5
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
  $ftemp = FileOpen($filesave, 0)
  $getsize=    FileGetSize ($filesave)
  $sHTML = FileRead($ftemp, $getsize)
  FileClose($ftemp)
  FileDelete($filesave)
  Local $aArray = StringRegExp($sHTML, ‘<div class="j-r-list-c-desc">[ \n\r]+<a href="/detail-\d+.html">[^<]+(?=</a>)‘, 3)
  ConsoleWrite("aArray size = " & UBound($aArray) & @CRLF)
    If UBound($aArray) <= 0 Then
        ContinueLoop
        EndIf
    $max = UBound($aArray)-1
  For $i = 0 To $max Step 1
     Local $item = $aArray[$i]
     If StringLen($item) > 0 Then
    $strnum = $storycount +1
    $strnum = $strnum & "." &@CRLF
    FileWrite($file, $strnum)
    $storycontent = StringRegExpReplace($item,‘<div class="j-r-list-c-desc">[ \n\r]+<a href="/detail-\d+.html">‘,"")
     $storycontent = $storycontent & @CRLF
     FileWrite($file, $storycontent)
     $storycount = $storycount + 1
     EndIf
     Next
   Next
FileClose($file)
MsgBox(0, "BUDEJIE", "Complete, story count = "&$storycount & ‘, story=‘ & $filename1)
Exit
Func MakeUpUrl($pagenum)
    If $pagenum == 1 Then
         $strUrl = ‘http://www.budejie.com/text/‘
    Else
        $strUrl = ‘http://www.budejie.com/text/‘ & $pagenum
   EndIf
   return $strUrl
   EndFunc