以下内容为程序代码:
<HTML>
<HEAD>
<TITLE>网易分类广告</TITLE>
<META HTTP-EQUIV=Content-Type CONTENT="text/html; charset=gb2312">
<link rel=stylesheet href="http://classad.163.com/html/style.css">
</HEAD>
<BODY topmargin=0 marginheight=0 leftmargin=0 marginwidth=0 bgcolor=ffffff>
<table width=100% cellspacing=0 cellpadding=10 border=0>
<tr align=center>
<td class=p7><b>如何将html批量转换成xml</b><BR><font class=grey>发布时间：2003-11-04</font></td>
</tr>
<tr>
<td class=p4>
<table border=0 cellspacing=0 cellpadding=0 align=right>
<tr><td></td></tr>
</table>
现在很多人时髦写xml、xhtml的网页，但现在fp2002和dw4都不支持。那么我怎么画呢？<br />
那个xml的编辑工具xpy用起太不方便了。这个脚本是我自己修改的，可以做到95%了，自己再改一下就OK。<br />
另外建议你把要转的html代码用dw打开套用样式保存一次，再用这个转换。一般绝大多数可以转好。 <br />
<br />
<br />
html2xsl.asp<br />
<br />
<%<br />
strtag_source1="<html ,<body ,<table ,<td ,<tr ,<option ,<font ,<div ,<span ,<h1 ,<h2 ,<form "<br />
strtag1 = Split(strtag_source1, ",", -1, 1)<br />
<br />
strattri_source2="selected,checked,norwap,readonly,noshade" '单独属性<br />
strtag2 = Split(strattri_source2, ",", -1, 1)<br />
<br />
strtag_source3="<input ,<img ,<hr ,<br ,<meta" '单端标签<br />
strtag3 = Split(strtag_source3, ",", -1, 1)<br />
<br />
'-------------------以下是处理flag值的多个函数<br />
function func_flag1(tmp)'处理单个属性：(例：width=325)<br />
index=InStr(1,tmp,"=",1)<br />
z1=left(tmp,index)<br />
z2=""""<br />
z3=mid(tmp,index+1,len(tmp)-len(z1))<br />
func_flag1=z1+z2+z3+z2<br />
end function<br />
<br />
function func_flag2(tmp)'(例：selected)<br />
func_flag2=tmp+"="+""""+tmp+""""<br />
end function<br />
<br />
function func_flag3(tmp)'处理单端标签(例：<img…)<br />
func_flag3=replace(cstr(tmp),">","/>")<br />
end function<br />
<br />
<br />
function alone_tag(tag)<br />
dim tag1 '定义处理完以后的标签<br />
tag=LCase(tag) '所有字符串都改写成小写形式<br />
dim tmpattri '<br />
index=InStr(1,tag," ",1)<br />
tmpattri=right(tag,len(tag)-index) '除去左侧标签<br />
if len(tmpattri)>1 then<br />
tmpattri=trim(left(tmpattri,len(tmpattri)-1)) '除去右侧">"，并去除两端空格(如果标签长大于等于2个字符)<br />
end if<br />
<br />
tmpattri=replace(tmpattri,chr(13)," ") '对源码中，一个标签不在一行里的情况，尚等待另行考虑处理。<br />
tmpattri=replace(tmpattri,chr(10)," ")<br />
tmpattri=replace(tmpattri,chr(10)&chr(13)," ")<br />
tmpattri=replace(tmpattri," "," ") '【这两句是对付当属性串里有多个空格的时候，<br />
tmpattri=replace(tmpattri," "," ") '【替换成一个空格，不过只能处理不超过16个空格的情况。<br />
tmpattri=replace(tmpattri," "," ")<br />
tmpattri=replace(tmpattri," "," ")<br />
tmpattri=replace(tmpattri," "," ")<br />
tmpattri=replace(tmpattri," "," ")<br />
tmpattri=replace(tmpattri," "," ")<br />
attribute=Split(tmpattri, " ", -1, 1) <br />
<br />
For count=0 to UBound(attribute, 1)<br />
If InStr(1,attribute(count),"=",1)=0 Then<br />
flag=2 '单个属性串中没找到等于号。(例：selected)<br />
Else<br />
IF InStr(1,attribute(count),"""",1)=0 Then<br />
flag=1 '单个属性串中没找到等于号。(例：width=325)<br />
Else<br />
flag=0 '单个属性串找到了等于号。(例：width="325")<br />
IF InStr(1,attribute(count),"""",1)>0 Then<br />
<br />
End IF<br />
End If<br />
End If<br />
<br />
<br />
Select case flag<br />
case 0 attribute(count)=attribute(count)<br />
case 1 attribute(count)=func_flag1(attribute(count)) '调用函数func_flag1处理。(例：width=325)<br />
case 2 attribute(count)=func_flag2(attribute(count)) '调用函数func_flag2处理。(例：selected)<br />
case 3 attribute(count)=func_flag3(attribute(count)) '调用函数func_flag3处理单端标签(例：<img…)<br />
case 4 attribute(count)=(attribute(count)) '另行处理属性串之间包含分号、空格的情况<br />
End Select<br />
Next<br />
<br />
count=0<br />
for count=0 to UBound(attribute, 1)<br />
attribute_tmp=attribute_tmp+" "+attribute(count) '属性值之间要有空格<br />
next<br />
index=InStr(1,tag," ",1)<br />
if InStr(1,tag," ",1)=0 and len(tag)<>"" then <br />
tag1=Replace(tag,">"," >") <br />
else<br />
tag_self=left(tag,index-1)<br />
tag1=tag_self+attribute_tmp+">"<br />
end if<br />
for count=0 to UBound(strtag3,1)<br />
if InStr(1,tag1,strtag3(count),1)<>0 then '这里利用到前面已切分好的属性标签<br />
tag1=func_flag3(tag1) '对付单端标签——flag=3(例：<img…)<br />
end if<br />
next<br />
alone_tag=tag1<br />
end function<br />
<br />
<br />
<br />
<br />
Function transform(txt)<br />
dim alltmp '定义此字符串变量，只保留未处理的字符串部分。<br />
alltmp=txt<br />
alltmp=replace(alltmp," "," ") '是替换掉文本中的字符实体&#nbsp;<br />
'处理htm源代码<br />
do while trim(alltmp)<>""<br />
'msgbox alltmp<br />
index=0<br />
index=InStr(1,alltmp,"<",1)<br />
if index=1 then<br />
index_right=InStr(1,alltmp,">",1)<br />
tag=left(alltmp,index_right) <br />
if mid(tag,2,1)<>"/" then<br />
tag1=alone_tag(tag)<br />
txt1=txt1+tag1<br />
del_tag=len(tag)<br />
else <br />
txt1=txt1+LCase(tag)<br />
del_tag=len(tag)<br />
end if<br />
else<br />
if index>1 then<br />
str_tmp=left(alltmp,index-1)<br />
txt1=txt1+str_tmp<br />
del_tag=len(left(alltmp,index-1))<br />
end if<br />
if index=0 then <br />
txt1=txt1+alltmp<br />
del_tag=len(alltmp)<br />
end if<br />
end if<br />
<br />
'response.write "alltmp="+alltmp<br />
'减掉处理完的<br />
alltmp=right(alltmp,len(alltmp)-del_tag) '(如果标签长大于等于2个字符)这里有问题！12.14，下次再作！！<br />
<br />
loop<br />
<br />
txt1=replace(txt1," ="""" "," ") ' =""<br />
txt1=replace(txt1," >",">") ' ><br />
txt1=replace(txt1,"<tbody>","") '<tbody><br />
transform=replace(txt1,"</tbody>","") '</tbody><br />
<br />
End Function<br />
<br />
<br />
dim txt '源文本<br />
dim txt1 '目标文本<br />
dim tmpreadline '=thisfile.readline<br />
<br />
txt="":txt1="":tmpReadAll=""<br />
txt=request("intxt")<br />
<br />
if txt="" then<br />
%><meta http-equiv="Content-Type" content="text/html; charset=gb2312"><br />
<form method="POST" action><br />
<p><textarea rows="10" name="intxt" cols="47"></textarea><br><br />
<input type="submit" value="提交" name="B1"></p><br />
</form><br />
<br />
<%<br />
end if<br />
<br />
txt1=transform(cstr(txt))<br />
<br />
txt=replace(server.htmlencode(txt),chr(13)&chr(10),"<br>")+""<br />
<br />
txt1=replace(server.htmlencode(txt1),chr(13)&chr(10),"<br>")+""<br />
<br />
response.write txt1<br />
<br />
%> <br />
<br />
<br /> </td>
</tr>
</table>
</BODY>
</HTML>
|