建站常识

1、去除html标签正则<\/*[^<>]*>

Function LoseHtml(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<\/*[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

Set RegEx = Nothing

LoseHtml = ClsTempLoseStr

End function

2、去除网页中的class

Function LoseClassTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "(class=){1,}(""|\'){0,1}\S+(""|\'|>|\s){0,1}"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseClassTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

3、去除网页中的script

Function LoseScriptTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "(<script){1,}[^<>]*>[^\0]*(<\/script>){1,}"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseScriptTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

4、去除网页中的iframe

Function LoseIFrameTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "(<iframe){1,}[^<>]*>[^\0]*(<\/iframe>){1,}"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseIFrameTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

5、去除网页中的object

Function LoseObjectTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "(<object){1,}[^<>]*>[^\0]*(<\/object>){1,}"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseObjectTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

6、去除网页中的span

Function LoseSpanTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}span[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseSpanTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

7、去除网页中的font

Function LoseFontTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}font[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseFontTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

8、去除网页中的a

Function LoseATag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempL

oseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}a[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseATag = ClsTempLoseStr

Set RegEx = Nothing

End Function

9、除网页中的div

Function LoseDivTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}div[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseDivTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

11、去除网页中的style

Function LoseStyleTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "(<style){1,}[^<>]*>[^\0]*(<\/style>){1,}"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseStyleTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

12、去除网页中的注释

Function LoseNoteTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<!--\/*[^<>]*-->"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseNoteTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

13、去除网页中的table

Function LoseTableTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}table[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseTableTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

14、去除网页中的td

Function LoseTDTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}td[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseTDTag = ClsTempLoseStr

Set RegEx = Nothing

End Function

15、去除网页中的tr

Function LoseTRTag(ContentStr)

Dim ClsTempLoseStr,RegEx

ClsTempLoseStr = Cstr(ContentStr)

Set RegEx = New RegExp

RegEx.Pattern = "<(\/){0,1}tr[^<>]*>"

RegEx.IgnoreCase = True

RegEx.Global = True

ClsTempLoseStr = RegEx.Replace(ClsTempLoseStr,"")

LoseTRTag = ClsTempLoseStr

Set RegEx = Nothing

End Function