【练习-071】批处理或VBS获取论坛首页指定数据

@Echo Off&SetLocal EnableDelayedExpansion
Mode Con Cols=130 Lines=20
Title 批处理获取论坛首页指定数据 『冷玉公子』
 
Set "Index=index.php"
Set "Url=http://www.bathome.net/index.php"
If Exist %Index% Del /f /q %Index%
 
Echo 正在更新首页信息，请稍后。
Curl -# -G -o %Index% %Url%
 
For /F "tokens=*" %%A In (index.php) Do (
	Set tStr="%%A"
	IF !e1Switch! == True Set e1Switch=False&Call :_TitleResovle !tStr! 最新主题&Pause>Nul
	IF !e2Switch! == True Set e2Switch=False&Call :_TitleResovle !tStr! 最新回复&Pause>Nul
	IF !e3Switch! == True Set e3Switch=False&Call :_TitleResovle !tStr! 热门主题&Pause>Nul
	Set e1Str=!tStr:homegrids_c_1=DandyMu!
	Set e2Str=!tStr:homegrids_c_2=DandyMu!
	Set e3Str=!tStr:homegrids_c_3=DandyMu!
	IF Not !e1Str! == !tStr! Set e1Switch=True
	IF Not !e2Str! == !tStr! Set e2Switch=True
	IF Not !e3Str! == !tStr! Set e3Switch=True
)
 
Goto :Eof
 
:_TitleResovle
Set trStr=%1
For /F "usebackq tokens=*" %%A In ('%trStr%') Do (
	@Rem Echo %%A
	Call :_LiResolve %%A %2
)
Goto :Eof
 
::Rem %1带处理的Li标签内容 %2标题
:_LiResolve
Cls
Echo [%2]
Echo=
Echo ---------------------------------------------------------------------------------------------------------------------------------
Set Str=%1
For /F "usebackq tokens=*" %%A In ('%Str%') Do (
	Set Html="%%A"&Set Code=!Html:^</li^>=^</li^>;!
	For /F "tokens=1-8 delims=;" %%1 In ("!Code!") Do (
		Call :_EchoInfo "%%1"&Call :_EchoInfo "%%2"&Call :_EchoInfo "%%3"&Call :_EchoInfo "%%4"
		Call :_EchoInfo "%%5"&Call :_EchoInfo "%%6"&Call :_EchoInfo "%%7"&Call :_EchoInfo "%%8"
	)
)
Goto :Eof
 
:_EchoInfo
For /F "usebackq tokens=1-20 delims=><=" %%A In ('%1') Do (
	For /F "tokens=1 delims='" %%1 In ("%%K") Do Set "HTURL=http://www.bathome.net/%%1" 
	Set "HTID=%%G"&Set "HTTITLE=%%M"
	IF "!HTTITLE:~1,5!" == "color" Set "HTTITLE=%%N"
	Echo 主题：!HTTITLE!	发帖ID：!HTID!	URL：!HTURL!
	Echo ---------------------------------------------------------------------------------------------------------------------------------
)
Goto :EofCOPY

@echo off&SetLocal EnableDelayEdexpansion&cd /d "%~dp0"
Mode Con Cols=130 Lines=30
del /f index.html>nul 2>nul
wget http://www.bathome.net 1>nul 2>nul
for /f "delims=" %%a in (index.html) do (
  set "STR=%%a"
  if defined flag (
    echo 【!cont!】
    call :tiqu "%%a"
    set "flag="
  )
  for %%b in (最新主题 最新回复 热门主题) do (
    if "!STR!" neq "!STR:%%b=!" set "flag=1"&set "cont=%%b"
  )
)
pause&exit
:tiqu
for %%b in ("%~1") do set "str=!str!%%b"
set "str=!str:'= !"
set "str=!str:<= !"
for %%b in (!str!) do (
  set "s=%%b"
  if "%%~xb" == ".html" (
    set "Url=http://www.bathome.net/%%b"
    set "flag1=1"
  )
  if "!s:~0,1!" equ ">" (
    set "t=!s:~1!"
    rem 发现下3句写成 "if not defined flag1 set "Id=!t!" else ("会出错，有时flag1为空时也会显示。
    if not defined flag1 (
      set "Id=!t!" 
    ) else (
      set "Title=!t!"
      set "flag1="
      echo 主题：!Title! 发帖ID：!Id! URL：!Url!
    )
  )
)        
goto :eofCOPY

Set ie = CreateObject("InternetExplorer.Application")
ie.Navigate "http://www.bathome.net"
ie.Visible = False
Do While ie.Busy Or ie.ReadyState <> 4 :Wscript.Sleep 100 :Loop
Text = ie.Document.Body.InnerHtml :ie.Quit
Pattern1 = """>(.*?)</A>.*href=""(thread.*)"">(.*)</A>"
Pattern2 = "id=homegrids_t_[1-3]>(?:(最新主题|最新回复|热门主题))"
 
Set Re = New RegExp
Re.Pattern =  Pattern1 & "|" & Pattern2
Re.Global = True
Re.IgnoreCase = True
For Each a in Re.Execute(Text)
   Set FSO=CreateObject("Scripting.FilesystemObject")
   Str = "[" & a.SubMatches(3) & "]"
   if Str = "[]" Then
      Str = "主题：" & a.SubMatches(2) & " "
      Str = Str & "发贴ID：" & a.SubMatches(0) & " " & "URL："
      Str = Str & "http://www.bathome.net/" & a.SubMatches(1)
   End If
   FSO.OpenTextFile("Result.txt",8,True).WriteLine Str
   Set FSO = Nothing
Next
 
CreateObject("Wscript.Shell").Run "NotePad Result.txt"COPY

Dim objXML, Url
Url = "http://www.bathome.net"
Set objXML = CreateObject("MSXML2.XmlHttp")
objXML.open "GET", Url, False
objXML.send()
Do Until objXML.readyState = 4 : WScript.Sleep 200 : Loop
Dim objADODB
Set objADODB = CreateObject("Adodb.Stream")
objADODB.Type = 1
objADODB.Mode = 3
objADODB.Open()
objADODB.Write objXML.responseBody
Set objXML = Nothing
Dim objFSO, Temp
Set objFSO = CreateObject("Scripting.FileSystemObject")
Temp = objFSO.GetSpecialFolder(2) & "\"
objADODB.SaveToFile Temp & "bathome.html", 2
Set objADODB = Nothing
Dim objHTML, objIH
Set objHTML = GetObject(Temp & "bathome.html", "HtmlFile")
Do Until objHTML.ReadyState = "complete" : WScript.Sleep 200 : Loop
Dim A, B, objStr
A = Array("最新主题", "最新回复", "热门主题")
B = Array("发帖ID：", "最新回复ID：", "最新回复ID：")
For i = 1 To 3
  objIH = objHTML.GetElementByID("homegrids_c_" & i).InnerHtml
  objStr = objStr & "["& A(i - 1) & "]" & vbCrLf & GetInfo(objIH, B(i - 1)) & vbCrLf
Next
Set objHTML = Nothing
objFSO.DeleteFile Temp & "bathome.html"
objStr = Replace(objStr, "<UL class=textinfolist>" & vbCrLf, "")
objFSO.OpenTextFile("HomePage.txt", 2, True).Write Replace(objStr, vbLf, vbCrLf)
Set objFSO = Nothing
CreateObject("Wscript.Shell").Run "cmd /chomepage.txt", True, False
 
Function GetInfo(Str1, Str2)
  Dim objRE, Matches, Matche
  Set objRE = New RegExp
  objRE.Global = True
  objRE.IgnoreCase = True
  objRE.Pattern = ".*?"">(.*?)</a.*?(\d{1,}-\d{1,}-\d{1,}).*?"">(.*?)</a.*"
  GetInfo = objRE.Replace(Str1, "主题：$3 " & Str2 & "$1 URL：" & Url & "/thread-$2.html")
  Set objRE = Nothing 
End FunctionCOPY