Sub 下载天津市规划局规划()
Dim strurl$, i%, n%, arr(), b() As Byte
For i = 1 To 110 '定义提取的页码
strurl = "http://gh.tj.gov.cn/newslist.aspx?id=CK0401"
With CreateObject("MSXML2.XMLHTTP")
'第一次GET,获取动态参数VIEWSTATE和EVENTVALIDATION
.Open "GET", strurl, False
.send
strText = .responseText
VIEWSTATE = encodeURI(CStr(Split(Split(strText, "__VIEWSTATE"" value=""")(1), """ />")(0)))
EVENTVALIDATION = encodeURI(CStr(Split(Split(strText, "__EVENTVALIDATION"" value=""")(1), """ />")(0)))
strText = .responseText
VIEWSTATE = encodeURI(CStr(Split(Split(strText, "__VIEWSTATE"" value=""")(1), """ />")(0)))
EVENTVALIDATION = encodeURI(CStr(Split(Split(strText, "__EVENTVALIDATION"" value=""")(1), """ />")(0)))
'这里的翻页动作是POST提交类型,将取得的动态参数写入需要send发送的参数中。
.Open "POST", strurl, False
.setRequestHeader "Content-Type", "application/x-www-form-urlencoded"
.send "&__EVENTARGUMENT=" & i _
& "&__EVENTTARGET=AspNetPager1" _
& "&__EVENTVALIDATION=" & EVENTVALIDATION _
& "&__VIEWSTATE=" & VIEWSTATE _
& "&__VIEWSTATEGENERATOR=14DD91A0" _
& "&AspNetPager1_input=" & i & "-1" _
& "&HiddenFieldPageFinished=1" _
& "&pkid=CK0401" _
& "&pkid2=3" _
& "&newskindid=CK0401" _
& "&Left1$ddl_cname=CK" _
& "&Left1$tb_search=" _
& "&Left1$rbl_site=title"
strText = .responseText
'正则获取单个规划的网址信息
Open ThisWorkbook.Path & "\图片\1.txt" For Output As #1
Print #1, strText
Close
Set reg = CreateObject("vbscript.regexp")
reg.Global = True
reg.IgnoreCase = True
reg.MultiLine = True
reg.Pattern = "<a href='(news.aspx\?id=\d+)'>(.*?)<\/a><\/td>\s*<td align=""right"" >(\d+-\d+-\d+)</td>"
n = 0
For Each mat In reg.Execute(strText)
n = n + 1
ReDim Preserve arr(1 To 3, 1 To n)
arr(1, n) = "http://gh.tj.gov.cn/" & mat.SubMatches(0) '正则取出的网址
arr(2, n) = mat.SubMatches(1) '正则取出的单项规划
arr(3, n) = mat.SubMatches(2) '正则取出的规划公示时间
Next mat
brr = Application.Transpose(arr)
rrow = ActiveSheet.Cells(Rows.Count, "a").End(3).Row + 1
ActiveSheet.Range("a" & rrow).Resize(UBound(brr), 3) = brr
'循环打开单个规划网址,保存图形文件
Set xml = CreateObject("MSXML2.XMLHTTP")
For r = 1 To UBound(brr)
xml.Open "GET", brr(r, 1), False
xml.send
Do While xml.ReadyState <> 4
DoEvents
Loop
strr = xml.responseText
reg.Pattern = "\/Files\/image\/\d+\.jpg"
If reg.Test(strr) Then '保存网页图片
k = 0
For Each mat In reg.Execute(strr)
Set xmlhttp = CreateObject("MSXML2.XMLHTTP")
k = k + 1
xmlhttp.Open "GET", "http://gh.tj.gov.cn" & mat, False
xmlhttp.send
Do While xmlhttp.ReadyState <> 4
DoEvents
Loop
b = xmlhttp.responseBody
On Error Resume Next '排除文件名过长的图片
Open "C:\图片\" & brr(r, 2) & k & ".jpg" For Binary As #1
Put #1, , b
Close
Next
Else
End If
Next
End With
Next
MsgBox "完成"
End Sub