實例講解ASP實現抓取網上房產信息

2019-11-18 19:14:55

字體：大中小

來源：轉載

供稿：網友

<%@LANGUAGE="VBSCR

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "<html>
<head>
<title>Untitled Document</title>
<meta http-equiv="Content-Type" content="text/html; charset=gb2312">
<meta http-equiv="refresh" content="300;URL=steal_house.asp">
</head>

<body>
<%
on error resume next
'
Server.ScriptTimeout = 999999
'========================================================
'字符編碼函數
'====================================================
Function BytesToBstr(body,code)
 dim objstream
 set objstream = Server.CreateObject("adodb.stream")
 objstream.Type = 1
 objstream.Mode =3
 objstream.Open
 objstream.Write body
 objstream.Position = 0
 objstream.Type = 2
 objstream.Charset =code
 BytesToBstr = objstream.ReadText
 objstream.Close
 set objstream = nothing
End Function

'取行字符串在另一字符串中的出現位置
Function Newstring(wstr,strng)
Newstring=Instr(lcase(wstr),lcase(strng))
if Newstring<=0 then Newstring=Len(wstr)
End Function
'替換字符串函數
function ReplaceStr(ori,str1,str2)
ReplaceStr=replace(ori,str1,str2)
end function
'====================================================
function Read xml(url,code,start,ends)
set oSend=createobject("Microsoft.xmlhttp")
SourceCode = oSend.open ("GET",url,false)
oSend.send()
ReadXml=BytesToBstr(oSend.responseBody,code )
start=Instr(ReadXml,start)
ReadXml=mid(ReadXml,start)
ends=Instr(ReadXml,ends)
ReadXml=left(ReadXml,ends-1)
end function

function SubStr(body,start,ends)
start=Instr(body,start)
SubStr=mid(body,start+len(start)+1)
ends=Instr(SubStr,ends)
SubStr=left(SubStr,ends-1)
end function

dim getcont,NewsContent
dim url,title
url="http://www.***.com"'新聞網址VeVb.com
getcont=ReadXml(url,"gb2312","<table class=k2 border=""0""","</table>")
getcont=RegexHtml(getcont)
dim KeyId,NewsClass,City,Position,HouseType,Level,Area,PRice,Demostra

dim ContactMan,Contact
for i=2 to ubound(getcont)
response.Write(getcont(i)&"__ ")

tempLink=mid(getcont(i),instr(getcont(i),"href=""")+6,instr(getcont(i),""" onClick")-10)
tempLink=replace(tempLink,"../","")

response.Write(i&":"&tempLink&" ")
NewsContent=ReadXml(tempLink,"gb2312","<td valign=""bottom"" width=""400"">","<hr width=""760"" noshade size=""1"" color=""#808080""> ")
NewsContent=RemoveHtml(NewsContent)
NewsContent=replace(NewsContent,VbCrLf,"")
NewsContent=replace(NewsContent,vbNewLine,"")
NewsContent=replace(NewsContent," ","")
NewsContent=replace(NewsContent," ","")
NewsContent=replace(NewsContent," ","")
NewsContent=replace(NewsContent,"/n","")
NewsContent=replace(NewsContent,chr(10),"")
NewsContent=replace(NewsContent,chr(13),"")
'===============get Content=======================
response.Write(NewsContent)
KeyId=SubStr(NewsContent,"列號：","信息類別：")
NewsClass=SubStr(NewsContent,"類別：","所在城市：")
City=SubStr(NewsContent,"城市：","房屋具體位置：")
Position=SubStr(NewsContent,"位置：","房屋類型：")
HouseType=SubStr(NewsContent,"類型：","樓層：")
Level=SubStr(NewsContent,"樓層：","使用面積：")
Area=SubStr(NewsContent,"面積：","房價：")
Price=SubStr(NewsContent,"房價：","其他說明：")
Demostra=SubStr(NewsContent,"說明：","聯系人：")
ContactMan=SubStr(NewsContent,"聯系人：","聯系方式：")
Contact=SubStr(NewsContent,"聯系方式：","信息來源：")
response.Write("總序列號:"&KeyId&" ")
response.Write("信息類別:"&NewsClass&" ")
response.Write("所在城市:"&City&" ")
response.Write("房屋具體位置:"&Position&" ")
response.Write("房屋類型:"&HouseType&" ")
response.Write("樓層:"&Level&" ")
response.Write("使用面積:"&Area&" ")
response.Write("房價:"&Price&" ")
response.Write("其他說明:"&Demostra&" ")
response.Write("聯系人:"&ContactMan&" ")
response.Write("聯系方式:"&Contact&" ")
'title=RemoveHTML(aa(i))
'response.Write("title:"&title)
for n=0 to application.Contents.count
 if(application.Contents(n)=KeyId) then
 ifexit=true
 end if
next
if not ifexit then
 application(time&i)=KeyId
'添加到數據庫
'====================================================
set rs=server.CreateObject("adodb.recordset")
rs.open "select top 1 * from news order by id desc",conn,3,3
rs.addnew
rs("NewsClass")=NewsClass
rs("City")=City
rs("Position")=Position
rs("HouseType")=HouseType
rs("Level")=Level
rs("Area")=Area
rs("Price")=Price
rs("Demostra")=Demostra
rs("ContactMan")=ContactMan
rs("Contact")=Contact
rs.update
rs.close
set rs=nothing
end if
'==================================================

next
function RemoveTag(body)

Set regEx = New RegExp
regEx.Pattern = "<[a].*?<//[a]>"
regEx.IgnoreCase = True
regEx.Global = True
Set Matches = regEx.Execute(body)
dim i,arr(15),ifexit
i=0
j=0
For Each Match in Matches
TempStr = Match.Value
TempStr=replace(TempStr,"<td>","")
TempStr=replace(TempStr,"</td>","")
TempStr=replace(TempStr,"<tr>","")
TempStr=replace(TempStr,"</tr>","")
arr(i)=TempStr
i=i+1
if(i>=15) then
exit for
end if
Next
Set regEx=nothing
Set Matches =nothing
RemoveTag=arr

end function
function RegexHtml(body)
dim r_arr(47),r_temp
Set regEx2 = New RegExp
regEx2.Pattern ="<a.*?<//a>"
regEx2.IgnoreCase = True
regEx2.Global = True
Set Matches2 = regEx2.Execute(body)
iii=0
For Each Match in Matches2

r_arr(iii)=Match.Value

iii=iii+1
Next
RegexHtml=r_arr
set regEx2=nothing
set Matches2=nothing
end function
'======================================================

conn.close
set conn=nothing
%>
</body>
</html>

　　function.asp

<%
'**************************************************
'函數名：gotTopic
'作用：截字符串，漢字一個算兩個字符，英文算一個字符
'參數：str ----原字符串
' strlen ----截取長度
'返回值：截取后的字符串
'**************************************************
function gotTopic(str,strlen)
if str="" then
gotTopic=""
exit function
end if
dim l,t,c, i
str=replace(replace(replace(replace(str," "," "),""",chr(34)),">",">"),"<","<")
str=replace(str,"?","")
l=len(str)
t=0
for i=1 to l
c=Abs(Asc(Mid(str,i,1)))
if c>255 then
 t=t+2
else
 t=t+1
end if
if t>=strlen then
 gotTopic=left(str,i) & "…"
 exit for
else
 gotTopic=str
end if
next
gotTopic=replace(replace(replace(replace(gotTopic," "," "),chr(34),"""),">",">"),"<","<")
end function
'=========================================================
'函數：RemoveHTML(strHTML)
'功能：去除HTML標記
'參數：strHTML --要去除HTML標記的字符串
'=========================================================
Function RemoveHTML(strHTML)
Dim objRegExp, Match, Matches
Set objRegExp = New Regexp

objRegExp.IgnoreCase = True
objRegExp.Global = True
'取閉合的<>
objRegExp.Pattern = "<.+?>"
'進行匹配
Set Matches = objRegExp.Execute(strHTML)

' 遍歷匹配集合，并替換掉匹配的項目
For Each Match in Matches
strHtml=Replace(strHTML,Match.Value,"")
Next
RemoveHTML=strHTML
Set objRegExp = Nothing
set Matches=nothing
End Function

　　conn.asp

<%
'on error resume next
set conn=server.CreateObject("adodb.connection")
con= "driver={Microsoft access Driver (*.mdb)};dbq=" & Server.MapPath("stest.mdb")
conn.open con

sub connclose
conn.close
set conn=nothing
end sub
%>

　　附：抓取信息的詳細頁面事例

總序列號：	479280
信息類別：	出租
所在城市：	濟南
房屋具體位置：	華龍路華信路交界口
房屋類型：	其他
樓層：	六層
使用面積：	24～240 平方米之間
房價：	0 　[租賃:元/月,買賣:萬元/套]
其他說明：	華信商務樓3至6層小空間對外出租（0.5元/平起），本樓屬純商務辦公投資使用，可用于辦公寫字間，周邊設施齊全、交通便利（37、80、K95在本樓前經過），全產權、市證，樓內設施包括水、電、暖、電梯設施齊全，有意者可電訊！
聯系人：	魯、王
聯系方式：	88017966、86812217
信息來源：	2005-8-4 8:28:55 來自：218.98.86.175
點擊次數：	19