feat: handle foreign encodings

This commit is contained in:
James Houlahan
2020-08-05 17:00:06 +02:00
parent f4374a02da
commit ea01c155da
11 changed files with 194 additions and 78 deletions

View File

@ -2,8 +2,12 @@ package parser
import (
"errors"
"unicode/utf8"
pmmime "github.com/ProtonMail/proton-bridge/pkg/mime"
"github.com/emersion/go-message"
"golang.org/x/net/html/charset"
"golang.org/x/text/encoding"
)
type Parts []*Part
@ -30,35 +34,32 @@ func (p *Part) AddChild(child *Part) {
p.children = append(p.children, child)
}
func (p *Part) write(writer *message.Writer, w *Writer) (err error) {
if len(p.children) > 0 {
for _, child := range p.children {
if err = child.writeAsChild(writer, w); err != nil {
return
}
}
}
if _, err = writer.Write(p.Body); err != nil {
return
}
return
func (p *Part) isUTF8() bool {
return utf8.Valid(p.Body)
}
func (p *Part) writeAsChild(writer *message.Writer, w *Writer) (err error) {
if !w.shouldWrite(p) {
return
}
childWriter, err := writer.CreatePart(p.Header)
// TODO: Do we then need to set charset to utf-8? What if it's embedded in html?
func (p *Part) convertToUTF8() error {
t, params, err := p.Header.ContentType()
if err != nil {
return
return err
}
if err = p.write(childWriter, w); err != nil {
return
var decoder *encoding.Decoder
if knownCharset, ok := params["charset"]; !ok {
encoding, _, _ := charset.DetermineEncoding(p.Body, t)
decoder = encoding.NewDecoder()
} else if decoder, err = pmmime.SelectDecoder(knownCharset); err != nil {
return err
}
return childWriter.Close()
if p.Body, err = decoder.Bytes(p.Body); err != nil {
return err
}
params["charset"] = "utf-8"
p.Header.SetContentType(t, params)
return nil
}