projectdiscovery/httpx
1
1
Fork 0
mirror of https://github.com/projectdiscovery/httpx.git synced 2024-09-11 17:46:08 +03:00
Code Issues Packages Projects Releases Wiki Activity

Improve GBK encode

Browse Source
This commit is contained in:
CaijiOrz 2020-11-29 14:05:52 +08:00
parent fd07f1cf15
commit 89351c1fdd
1 changed files with 21 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
common/httpx/title.go
View File

@ -28,6 +28,27 @@ func ExtractTitle(r *Response) (title string) {
return string(titleUtf8)
}
// Content-Type from head tag
re = regexp.MustCompile(`(?im)\s*charset="(.*?)"|charset=(.*?)"\s*`)
var match = re.FindSubmatch(r.Data)
var mcontentType = ""
if len(match) != 0 {
for i, v := range match {
if string(v) != "" && i != 0 {
mcontentType = string(v)
}
}
mcontentType = strings.ToLower(mcontentType)
}
if strings.Contains(mcontentType, "gb2312") || strings.Contains(mcontentType, "gbk") {
titleUtf8, err := Decodegbk([]byte(title))
if err != nil {
return
}
return string(titleUtf8)
}
}
return