XMLHTTP抓取远程数据的后期处理
程序员文章站
2022-04-14 18:50:11
<% hehe = hello("http://mmsg.qq.com/cgi-bin/gddylist?type=13&sort=1...
<%
hehe = hello("http://mmsg.qq.com/cgi-bin/gddylist?type=13&sort=1&page=3", "<html>", "</html>", ".*(<td width=""35%"" bgcolor=""#[\dabcde]{6}"">(.*)</td>)[.\n]*", "<font style=""font-size:9pt;"" color=blue>$2</font><br>")
response.write hehe
function hello(strurl, strstart, strend, patrn, replstr)
str = getbody(strurl)
str = mymid(str, strstart, strend)
str = replacetest(patrn, replstr, str)
hello = str
end function
function mymid(str, strstart, strend)
if strstart = "" then
i = 0
else
i = instr(str, strstart)
end if
if strend = "" then
j = len(str)
else
j = instr(i, str, strend)
end if
mymid = mid(str, i, j - i + 1)
end function
function replacetest(patrn, replstr, str1)
dim regex, match, matches
set regex = new regexp
regex.pattern = patrn
regex.ignorecase = true
regex.global = true
set matches = regex.execute(str1)
for each match in matches
replacetest = replacetest®ex.replace(match.value, replstr)
next
end function
function getbody(url)
set objxml = createobject("microsoft.xmlhttp")
with objxml
.open "get", url, false, "", ""
.send
getbody = .responsebody
end with
getbody = bytestobstr(getbody, "gb2312")
set objxml = nothing
end function
function bytestobstr(strbody, codebase)
set objstream = server.createobject("adodb.stream")
with objstream
.type = 1
.mode = 3
.open
.write strbody
.position = 0
.type = 2
.charset = codebase
bytestobstr = .readtext
.close
end with
set objstream = nothing
end function
%>
其他调用示例：
hehe = hello("http://list.mp3.baidu.com/song/a.htm", "<table width=""90%"" border=""0"" align=""center"" cellpadding=""3"" cellspacing=""0"" bgcolor=""#f5f5f5"" >", "<div align=center>", ".*(<td width=""20%""><a href="".*\.htm"" target=_blank>)(.*)(</a></td>)[.\n]*", "<font style=""font-size:9pt;"" color=blue>$2</font><br>")
hehe = hello("http://mmsg.qq.com/cgi-bin/gddylist?type=13&sort=1&page=3", "<html>", "</html>", ".*(<td width=""35%"" bgcolor=""#[\dabcde]{6}"">(.*)</td>)[.\n]*", "<font style=""font-size:9pt;"" color=blue>$2</font><br>")
response.write hehe
function hello(strurl, strstart, strend, patrn, replstr)
str = getbody(strurl)
str = mymid(str, strstart, strend)
str = replacetest(patrn, replstr, str)
hello = str
end function
function mymid(str, strstart, strend)
if strstart = "" then
i = 0
else
i = instr(str, strstart)
end if
if strend = "" then
j = len(str)
else
j = instr(i, str, strend)
end if
mymid = mid(str, i, j - i + 1)
end function
function replacetest(patrn, replstr, str1)
dim regex, match, matches
set regex = new regexp
regex.pattern = patrn
regex.ignorecase = true
regex.global = true
set matches = regex.execute(str1)
for each match in matches
replacetest = replacetest®ex.replace(match.value, replstr)
next
end function
function getbody(url)
set objxml = createobject("microsoft.xmlhttp")
with objxml
.open "get", url, false, "", ""
.send
getbody = .responsebody
end with
getbody = bytestobstr(getbody, "gb2312")
set objxml = nothing
end function
function bytestobstr(strbody, codebase)
set objstream = server.createobject("adodb.stream")
with objstream
.type = 1
.mode = 3
.open
.write strbody
.position = 0
.type = 2
.charset = codebase
bytestobstr = .readtext
.close
end with
set objstream = nothing
end function
%>
其他调用示例：
hehe = hello("http://list.mp3.baidu.com/song/a.htm", "<table width=""90%"" border=""0"" align=""center"" cellpadding=""3"" cellspacing=""0"" bgcolor=""#f5f5f5"" >", "<div align=center>", ".*(<td width=""20%""><a href="".*\.htm"" target=_blank>)(.*)(</a></td>)[.\n]*", "<font style=""font-size:9pt;"" color=blue>$2</font><br>")
上一篇： ASP中一个字符串处理类