func DetermineEncoding(givenType *ParsedMimeType, head []byte) encoding.Encoding {
// Try to determine encoding ala whatwg "encoding sniffing algorithm"
enc, _, certain := charset.DetermineEncoding(head, givenType.String())
if certain {
return enc
}
// Before doing a byte-level encoding detection, check for xml
// declaration
s := strings.TrimSpace(string(head))
if strings.HasPrefix(s, "<?xml ") {
s = strings.TrimPrefix(s, "<?xml ")
end := strings.Index(s, "?>")
if end > 0 {
s = s[:end]
for _, match := range attrRe.FindAllStringSubmatch(s, -1) {
if match[1] == "encoding" {
if detectedEnc, _ := charset.Lookup(match[2]); detectedEnc != nil {
return detectedEnc
}
}
}
}
}
var detector *chardet.Detector
if givenType.IsHtml() || givenType.IsXml() {
detector = chardet.NewHtmlDetector()
} else {
detector = chardet.NewTextDetector()
}
detected, err := detector.DetectBest(head)
if err != nil {
fmt.Fprintln(os.Stderr, "Error detecting character set, using default.")
return enc
}
if detected.Charset == "GB-18030" {
detected.Charset = "GB18030"
}
if detectedEnc, _ := charset.Lookup(detected.Charset); detectedEnc != nil {
return detectedEnc
}
return enc
}
func encodingReader(body []byte, contentType string) (io.Reader, error) {
preview := make([]byte, 1024)
var r io.Reader = bytes.NewReader(body)
n, err := io.ReadFull(r, preview)
switch {
case err == io.ErrUnexpectedEOF:
preview = preview[:n]
r = bytes.NewReader(preview)
case err != nil:
return nil, err
default:
r = io.MultiReader(bytes.NewReader(preview), r)
}
e, _, certain := charset.DetermineEncoding(preview, contentType)
if !certain && e == charmap.Windows1252 && utf8.Valid(body) {
e = encoding.Nop
}
if e != encoding.Nop {
r = transform.NewReader(r, e.NewDecoder())
}
return r, nil
}
