Compare commits

..

No commits in common. "b8e6cffd317401d980600e339eb21b15b9bc64c1" and "c33886b7100f1b92c763435e59af262879817f76" have entirely different histories.

16 changed files with 139 additions and 102 deletions

View File

@ -22,21 +22,17 @@ import (
// UTF8BOM is the utf-8 byte-order marker
var UTF8BOM = []byte{'\xef', '\xbb', '\xbf'}
type ConvertOpts struct {
KeepBOM bool
}
// ToUTF8WithFallbackReader detects the encoding of content and converts to UTF-8 reader if possible
func ToUTF8WithFallbackReader(rd io.Reader, opts ConvertOpts) io.Reader {
func ToUTF8WithFallbackReader(rd io.Reader) io.Reader {
buf := make([]byte, 2048)
n, err := util.ReadAtMost(rd, buf)
if err != nil {
return io.MultiReader(bytes.NewReader(MaybeRemoveBOM(buf[:n], opts)), rd)
return io.MultiReader(bytes.NewReader(RemoveBOMIfPresent(buf[:n])), rd)
}
charsetLabel, err := DetectEncoding(buf[:n])
if err != nil || charsetLabel == "UTF-8" {
return io.MultiReader(bytes.NewReader(MaybeRemoveBOM(buf[:n], opts)), rd)
return io.MultiReader(bytes.NewReader(RemoveBOMIfPresent(buf[:n])), rd)
}
encoding, _ := charset.Lookup(charsetLabel)
@ -46,20 +42,20 @@ func ToUTF8WithFallbackReader(rd io.Reader, opts ConvertOpts) io.Reader {
return transform.NewReader(
io.MultiReader(
bytes.NewReader(MaybeRemoveBOM(buf[:n], opts)),
bytes.NewReader(RemoveBOMIfPresent(buf[:n])),
rd,
),
encoding.NewDecoder(),
)
}
// ToUTF8 converts content to UTF8 encoding
func ToUTF8(content []byte, opts ConvertOpts) (string, error) {
// ToUTF8WithErr converts content to UTF8 encoding
func ToUTF8WithErr(content []byte) (string, error) {
charsetLabel, err := DetectEncoding(content)
if err != nil {
return "", err
} else if charsetLabel == "UTF-8" {
return string(MaybeRemoveBOM(content, opts)), nil
return string(RemoveBOMIfPresent(content)), nil
}
encoding, _ := charset.Lookup(charsetLabel)
@ -74,22 +70,28 @@ func ToUTF8(content []byte, opts ConvertOpts) (string, error) {
result = append(result, content[n:]...)
}
result = MaybeRemoveBOM(result, opts)
result = RemoveBOMIfPresent(result)
return string(result), err
}
// ToUTF8WithFallback detects the encoding of content and converts to UTF-8 if possible
func ToUTF8WithFallback(content []byte, opts ConvertOpts) []byte {
bs, _ := io.ReadAll(ToUTF8WithFallbackReader(bytes.NewReader(content), opts))
func ToUTF8WithFallback(content []byte) []byte {
bs, _ := io.ReadAll(ToUTF8WithFallbackReader(bytes.NewReader(content)))
return bs
}
// ToUTF8 converts content to UTF8 encoding and ignore error
func ToUTF8(content string) string {
res, _ := ToUTF8WithErr([]byte(content))
return res
}
// ToUTF8DropErrors makes sure the return string is valid utf-8; attempts conversion if possible
func ToUTF8DropErrors(content []byte, opts ConvertOpts) []byte {
func ToUTF8DropErrors(content []byte) []byte {
charsetLabel, err := DetectEncoding(content)
if err != nil || charsetLabel == "UTF-8" {
return MaybeRemoveBOM(content, opts)
return RemoveBOMIfPresent(content)
}
encoding, _ := charset.Lookup(charsetLabel)
@ -115,14 +117,11 @@ func ToUTF8DropErrors(content []byte, opts ConvertOpts) []byte {
}
}
return MaybeRemoveBOM(decoded, opts)
return RemoveBOMIfPresent(decoded)
}
// MaybeRemoveBOM removes a UTF-8 BOM from a []byte when opts.KeepBOM is false
func MaybeRemoveBOM(content []byte, opts ConvertOpts) []byte {
if opts.KeepBOM {
return content
}
// RemoveBOMIfPresent removes a UTF-8 BOM from a []byte
func RemoveBOMIfPresent(content []byte) []byte {
if len(content) > 2 && bytes.Equal(content[0:3], UTF8BOM) {
return content[3:]
}

View File

@ -30,15 +30,15 @@ func resetDefaultCharsetsOrder() {
}
}
func TestMaybeRemoveBOM(t *testing.T) {
res := MaybeRemoveBOM([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
func TestRemoveBOMIfPresent(t *testing.T) {
res := RemoveBOMIfPresent([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, res)
res = MaybeRemoveBOM([]byte{0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
res = RemoveBOMIfPresent([]byte{0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, res)
}
func TestToUTF8(t *testing.T) {
func TestToUTF8WithErr(t *testing.T) {
resetDefaultCharsetsOrder()
var res string
var err error
@ -47,53 +47,53 @@ func TestToUTF8(t *testing.T) {
// locale, so some conversions might behave differently. For that reason, we don't
// depend on particular conversions but in expected behaviors.
res, err = ToUTF8([]byte{0x41, 0x42, 0x43}, ConvertOpts{})
res, err = ToUTF8WithErr([]byte{0x41, 0x42, 0x43})
assert.NoError(t, err)
assert.Equal(t, "ABC", res)
// "áéíóú"
res, err = ToUTF8([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
res, err = ToUTF8WithErr([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.NoError(t, err)
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, []byte(res))
// "áéíóú"
res, err = ToUTF8([]byte{
res, err = ToUTF8WithErr([]byte{
0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3,
0xc3, 0xba,
}, ConvertOpts{})
})
assert.NoError(t, err)
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, []byte(res))
res, err = ToUTF8([]byte{
res, err = ToUTF8WithErr([]byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63,
0xF3, 0x6D, 0x6F, 0x20, 0xF1, 0x6F, 0x73, 0x41, 0x41, 0x41, 0x2e,
}, ConvertOpts{})
})
assert.NoError(t, err)
stringMustStartWith(t, "Hola,", res)
stringMustEndWith(t, "AAA.", res)
res, err = ToUTF8([]byte{
res, err = ToUTF8WithErr([]byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63,
0xF3, 0x6D, 0x6F, 0x20, 0x07, 0xA4, 0x6F, 0x73, 0x41, 0x41, 0x41, 0x2e,
}, ConvertOpts{})
})
assert.NoError(t, err)
stringMustStartWith(t, "Hola,", res)
stringMustEndWith(t, "AAA.", res)
res, err = ToUTF8([]byte{
res, err = ToUTF8WithErr([]byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63,
0xF3, 0x6D, 0x6F, 0x20, 0x81, 0xA4, 0x6F, 0x73, 0x41, 0x41, 0x41, 0x2e,
}, ConvertOpts{})
})
assert.NoError(t, err)
stringMustStartWith(t, "Hola,", res)
stringMustEndWith(t, "AAA.", res)
// Japanese (Shift-JIS)
// 日属秘ぞしちゅ。
res, err = ToUTF8([]byte{
res, err = ToUTF8WithErr([]byte{
0x93, 0xFA, 0x91, 0xAE, 0x94, 0xE9, 0x82, 0xBC, 0x82, 0xB5, 0x82,
0xBF, 0x82, 0xE3, 0x81, 0x42,
}, ConvertOpts{})
})
assert.NoError(t, err)
assert.Equal(t, []byte{
0xE6, 0x97, 0xA5, 0xE5, 0xB1, 0x9E, 0xE7, 0xA7, 0x98, 0xE3,
@ -101,7 +101,7 @@ func TestToUTF8(t *testing.T) {
},
[]byte(res))
res, err = ToUTF8([]byte{0x00, 0x00, 0x00, 0x00}, ConvertOpts{})
res, err = ToUTF8WithErr([]byte{0x00, 0x00, 0x00, 0x00})
assert.NoError(t, err)
assert.Equal(t, []byte{0x00, 0x00, 0x00, 0x00}, []byte(res))
}
@ -109,22 +109,22 @@ func TestToUTF8(t *testing.T) {
func TestToUTF8WithFallback(t *testing.T) {
resetDefaultCharsetsOrder()
// "ABC"
res := ToUTF8WithFallback([]byte{0x41, 0x42, 0x43}, ConvertOpts{})
res := ToUTF8WithFallback([]byte{0x41, 0x42, 0x43})
assert.Equal(t, []byte{0x41, 0x42, 0x43}, res)
// "áéíóú"
res = ToUTF8WithFallback([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
res = ToUTF8WithFallback([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, res)
// UTF8 BOM + "áéíóú"
res = ToUTF8WithFallback([]byte{0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
res = ToUTF8WithFallback([]byte{0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, res)
// "Hola, así cómo ños"
res = ToUTF8WithFallback([]byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63,
0xF3, 0x6D, 0x6F, 0x20, 0xF1, 0x6F, 0x73,
}, ConvertOpts{})
})
assert.Equal(t, []byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xC3, 0xAD, 0x20, 0x63,
0xC3, 0xB3, 0x6D, 0x6F, 0x20, 0xC3, 0xB1, 0x6F, 0x73,
@ -133,65 +133,126 @@ func TestToUTF8WithFallback(t *testing.T) {
// "Hola, así cómo "
minmatch := []byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xC3, 0xAD, 0x20, 0x63, 0xC3, 0xB3, 0x6D, 0x6F, 0x20}
res = ToUTF8WithFallback([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x07, 0xA4, 0x6F, 0x73}, ConvertOpts{})
res = ToUTF8WithFallback([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x07, 0xA4, 0x6F, 0x73})
// Do not fail for differences in invalid cases, as the library might change the conversion criteria for those
assert.Equal(t, minmatch, res[0:len(minmatch)])
res = ToUTF8WithFallback([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x81, 0xA4, 0x6F, 0x73}, ConvertOpts{})
res = ToUTF8WithFallback([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x81, 0xA4, 0x6F, 0x73})
// Do not fail for differences in invalid cases, as the library might change the conversion criteria for those
assert.Equal(t, minmatch, res[0:len(minmatch)])
// Japanese (Shift-JIS)
// "日属秘ぞしちゅ。"
res = ToUTF8WithFallback([]byte{0x93, 0xFA, 0x91, 0xAE, 0x94, 0xE9, 0x82, 0xBC, 0x82, 0xB5, 0x82, 0xBF, 0x82, 0xE3, 0x81, 0x42}, ConvertOpts{})
res = ToUTF8WithFallback([]byte{0x93, 0xFA, 0x91, 0xAE, 0x94, 0xE9, 0x82, 0xBC, 0x82, 0xB5, 0x82, 0xBF, 0x82, 0xE3, 0x81, 0x42})
assert.Equal(t, []byte{
0xE6, 0x97, 0xA5, 0xE5, 0xB1, 0x9E, 0xE7, 0xA7, 0x98, 0xE3,
0x81, 0x9E, 0xE3, 0x81, 0x97, 0xE3, 0x81, 0xA1, 0xE3, 0x82, 0x85, 0xE3, 0x80, 0x82,
}, res)
res = ToUTF8WithFallback([]byte{0x00, 0x00, 0x00, 0x00}, ConvertOpts{})
res = ToUTF8WithFallback([]byte{0x00, 0x00, 0x00, 0x00})
assert.Equal(t, []byte{0x00, 0x00, 0x00, 0x00}, res)
}
func TestToUTF8(t *testing.T) {
resetDefaultCharsetsOrder()
// Note: golang compiler seems so behave differently depending on the current
// locale, so some conversions might behave differently. For that reason, we don't
// depend on particular conversions but in expected behaviors.
res := ToUTF8(string([]byte{0x41, 0x42, 0x43}))
assert.Equal(t, "ABC", res)
// "áéíóú"
res = ToUTF8(string([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}))
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, []byte(res))
// BOM + "áéíóú"
res = ToUTF8(string([]byte{
0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3,
0xc3, 0xba,
}))
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, []byte(res))
// Latin1
// Hola, así cómo ños
res = ToUTF8(string([]byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63,
0xF3, 0x6D, 0x6F, 0x20, 0xF1, 0x6F, 0x73,
}))
assert.Equal(t, []byte{
0x48, 0x6f, 0x6c, 0x61, 0x2c, 0x20, 0x61, 0x73, 0xc3, 0xad, 0x20, 0x63,
0xc3, 0xb3, 0x6d, 0x6f, 0x20, 0xc3, 0xb1, 0x6f, 0x73,
}, []byte(res))
// Latin1
// Hola, así cómo \x07ños
res = ToUTF8(string([]byte{
0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63,
0xF3, 0x6D, 0x6F, 0x20, 0x07, 0xA4, 0x6F, 0x73,
}))
// Hola,
bytesMustStartWith(t, []byte{0x48, 0x6F, 0x6C, 0x61, 0x2C}, []byte(res))
// This test FAILS
// res = ToUTF8("Hola, así cómo \x81ños")
// Do not fail for differences in invalid cases, as the library might change the conversion criteria for those
// assert.Regexp(t, "^Hola, así cómo", res)
// Japanese (Shift-JIS)
// 日属秘ぞしちゅ。
res = ToUTF8(string([]byte{
0x93, 0xFA, 0x91, 0xAE, 0x94, 0xE9, 0x82, 0xBC, 0x82, 0xB5, 0x82,
0xBF, 0x82, 0xE3, 0x81, 0x42,
}))
assert.Equal(t, []byte{
0xE6, 0x97, 0xA5, 0xE5, 0xB1, 0x9E, 0xE7, 0xA7, 0x98, 0xE3,
0x81, 0x9E, 0xE3, 0x81, 0x97, 0xE3, 0x81, 0xA1, 0xE3, 0x82, 0x85, 0xE3, 0x80, 0x82,
},
[]byte(res))
res = ToUTF8("\x00\x00\x00\x00")
assert.Equal(t, []byte{0x00, 0x00, 0x00, 0x00}, []byte(res))
}
func TestToUTF8DropErrors(t *testing.T) {
resetDefaultCharsetsOrder()
// "ABC"
res := ToUTF8DropErrors([]byte{0x41, 0x42, 0x43}, ConvertOpts{})
res := ToUTF8DropErrors([]byte{0x41, 0x42, 0x43})
assert.Equal(t, []byte{0x41, 0x42, 0x43}, res)
// "áéíóú"
res = ToUTF8DropErrors([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, res)
// UTF8 BOM + "áéíóú"
res = ToUTF8DropErrors([]byte{0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0xef, 0xbb, 0xbf, 0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba})
assert.Equal(t, []byte{0xc3, 0xa1, 0xc3, 0xa9, 0xc3, 0xad, 0xc3, 0xb3, 0xc3, 0xba}, res)
// "Hola, así cómo ños"
res = ToUTF8DropErrors([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0xF1, 0x6F, 0x73}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0xF1, 0x6F, 0x73})
assert.Equal(t, []byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73}, res[:8])
assert.Equal(t, []byte{0x73}, res[len(res)-1:])
// "Hola, así cómo "
minmatch := []byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xC3, 0xAD, 0x20, 0x63, 0xC3, 0xB3, 0x6D, 0x6F, 0x20}
res = ToUTF8DropErrors([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x07, 0xA4, 0x6F, 0x73}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x07, 0xA4, 0x6F, 0x73})
// Do not fail for differences in invalid cases, as the library might change the conversion criteria for those
assert.Equal(t, minmatch, res[0:len(minmatch)])
res = ToUTF8DropErrors([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x81, 0xA4, 0x6F, 0x73}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0x48, 0x6F, 0x6C, 0x61, 0x2C, 0x20, 0x61, 0x73, 0xED, 0x20, 0x63, 0xF3, 0x6D, 0x6F, 0x20, 0x81, 0xA4, 0x6F, 0x73})
// Do not fail for differences in invalid cases, as the library might change the conversion criteria for those
assert.Equal(t, minmatch, res[0:len(minmatch)])
// Japanese (Shift-JIS)
// "日属秘ぞしちゅ。"
res = ToUTF8DropErrors([]byte{0x93, 0xFA, 0x91, 0xAE, 0x94, 0xE9, 0x82, 0xBC, 0x82, 0xB5, 0x82, 0xBF, 0x82, 0xE3, 0x81, 0x42}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0x93, 0xFA, 0x91, 0xAE, 0x94, 0xE9, 0x82, 0xBC, 0x82, 0xB5, 0x82, 0xBF, 0x82, 0xE3, 0x81, 0x42})
assert.Equal(t, []byte{
0xE6, 0x97, 0xA5, 0xE5, 0xB1, 0x9E, 0xE7, 0xA7, 0x98, 0xE3,
0x81, 0x9E, 0xE3, 0x81, 0x97, 0xE3, 0x81, 0xA1, 0xE3, 0x82, 0x85, 0xE3, 0x80, 0x82,
}, res)
res = ToUTF8DropErrors([]byte{0x00, 0x00, 0x00, 0x00}, ConvertOpts{})
res = ToUTF8DropErrors([]byte{0x00, 0x00, 0x00, 0x00})
assert.Equal(t, []byte{0x00, 0x00, 0x00, 0x00}, res)
}
@ -241,6 +302,10 @@ func stringMustEndWith(t *testing.T, expected, value string) {
assert.Equal(t, expected, value[len(value)-len(expected):])
}
func bytesMustStartWith(t *testing.T, expected, value []byte) {
assert.Equal(t, expected, value[:len(expected)])
}
func TestToUTF8WithFallbackReader(t *testing.T) {
resetDefaultCharsetsOrder()
@ -252,7 +317,7 @@ func TestToUTF8WithFallbackReader(t *testing.T) {
}
input = input[:testLen]
input += "// Выключаем"
rd := ToUTF8WithFallbackReader(bytes.NewReader([]byte(input)), ConvertOpts{})
rd := ToUTF8WithFallbackReader(bytes.NewReader([]byte(input)))
r, _ := io.ReadAll(rd)
assert.EqualValuesf(t, input, string(r), "testing string len=%d", testLen)
}

View File

@ -174,7 +174,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
return batch.Index(id, &RepoIndexerData{
RepoID: repo.ID,
CommitID: commitSha,
Content: string(charset.ToUTF8DropErrors(fileContents, charset.ConvertOpts{})),
Content: string(charset.ToUTF8DropErrors(fileContents)),
Language: analyze.GetCodeLanguage(update.Filename, fileContents),
UpdatedAt: time.Now().UTC(),
})

View File

@ -135,7 +135,7 @@ func (b *Indexer) addUpdate(ctx context.Context, batchWriter git.WriteCloserErro
Id(id).
Doc(map[string]any{
"repo_id": repo.ID,
"content": string(charset.ToUTF8DropErrors(fileContents, charset.ConvertOpts{})),
"content": string(charset.ToUTF8DropErrors(fileContents)),
"commit_id": sha,
"language": analyze.GetCodeLanguage(update.Filename, fileContents),
"updated_at": timeutil.TimeStampNow(),

View File

@ -85,11 +85,9 @@ func (g *ASTTransformer) Transform(node *ast.Document, reader text.Reader, pc pa
// 2. If they're not wrapped with a link they need a link wrapper
// Check if the destination is a real link
if len(v.Destination) > 0 && !markup.IsLink(v.Destination) {
v.Destination = []byte(giteautil.URLJoin(
ctx.Links.ResolveMediaLink(ctx.IsWiki),
strings.TrimLeft(string(v.Destination), "/"),
))
link := v.Destination
if len(link) > 0 && !markup.IsLink(link) {
v.Destination = []byte(giteautil.URLJoin(ctx.Links.ResolveMediaLink(ctx.IsWiki), string(link)))
}
parent := n.Parent()
@ -105,7 +103,7 @@ func (g *ASTTransformer) Transform(node *ast.Document, reader text.Reader, pc pa
// Duplicate the current image node
image := ast.NewImage(ast.NewLink())
image.Destination = v.Destination
image.Destination = link
image.Title = v.Title
for _, attr := range v.Attributes() {
image.SetAttribute(attr.Name, attr.Value)

View File

@ -580,8 +580,6 @@ https://example.com/file.bin
[[local link|file.bin]]
[[remote link|https://example.com]]
![local image](image.jpg)
![local image](path/file)
![local image](/path/file)
![remote image](https://example.com/image.jpg)
[[local image|image.jpg]]
[[remote link|https://example.com/image.jpg]]
@ -611,8 +609,6 @@ mail@domain.com
<a href="/src/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/image.jpg" target="_blank" rel="nofollow noopener"><img src="/image.jpg" alt="local image"/></a><br/>
<a href="/path/file" target="_blank" rel="nofollow noopener"><img src="/path/file" alt="local image"/></a><br/>
<a href="/path/file" target="_blank" rel="nofollow noopener"><img src="/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/image.jpg" rel="nofollow"><img src="/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -638,8 +634,6 @@ space</p>
<a href="/wiki/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/wiki/raw/image.jpg" target="_blank" rel="nofollow noopener"><img src="/wiki/raw/image.jpg" alt="local image"/></a><br/>
<a href="/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/wiki/raw/image.jpg" rel="nofollow"><img src="/wiki/raw/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -667,8 +661,6 @@ space</p>
<a href="https://gitea.io/src/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="https://gitea.io/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://gitea.io/image.jpg" alt="local image"/></a><br/>
<a href="https://gitea.io/path/file" target="_blank" rel="nofollow noopener"><img src="https://gitea.io/path/file" alt="local image"/></a><br/>
<a href="https://gitea.io/path/file" target="_blank" rel="nofollow noopener"><img src="https://gitea.io/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="https://gitea.io/image.jpg" rel="nofollow"><img src="https://gitea.io/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -696,8 +688,6 @@ space</p>
<a href="https://gitea.io/wiki/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="https://gitea.io/wiki/raw/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://gitea.io/wiki/raw/image.jpg" alt="local image"/></a><br/>
<a href="https://gitea.io/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="https://gitea.io/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://gitea.io/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="https://gitea.io/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="https://gitea.io/wiki/raw/image.jpg" rel="nofollow"><img src="https://gitea.io/wiki/raw/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -725,8 +715,6 @@ space</p>
<a href="/relative/path/src/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/relative/path/image.jpg" target="_blank" rel="nofollow noopener"><img src="/relative/path/image.jpg" alt="local image"/></a><br/>
<a href="/relative/path/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/path/file" alt="local image"/></a><br/>
<a href="/relative/path/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/relative/path/image.jpg" rel="nofollow"><img src="/relative/path/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -754,8 +742,6 @@ space</p>
<a href="/relative/path/wiki/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/relative/path/wiki/raw/image.jpg" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/image.jpg" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/relative/path/wiki/raw/image.jpg" rel="nofollow"><img src="/relative/path/wiki/raw/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -784,8 +770,6 @@ space</p>
<a href="/user/repo/src/branch/main/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/user/repo/media/branch/main/image.jpg" target="_blank" rel="nofollow noopener"><img src="/user/repo/media/branch/main/image.jpg" alt="local image"/></a><br/>
<a href="/user/repo/media/branch/main/path/file" target="_blank" rel="nofollow noopener"><img src="/user/repo/media/branch/main/path/file" alt="local image"/></a><br/>
<a href="/user/repo/media/branch/main/path/file" target="_blank" rel="nofollow noopener"><img src="/user/repo/media/branch/main/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/user/repo/media/branch/main/image.jpg" rel="nofollow"><img src="/user/repo/media/branch/main/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -814,8 +798,6 @@ space</p>
<a href="/relative/path/wiki/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/relative/path/wiki/raw/image.jpg" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/image.jpg" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/relative/path/wiki/raw/image.jpg" rel="nofollow"><img src="/relative/path/wiki/raw/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -844,8 +826,6 @@ space</p>
<a href="/user/repo/src/sub/folder/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/user/repo/image.jpg" target="_blank" rel="nofollow noopener"><img src="/user/repo/image.jpg" alt="local image"/></a><br/>
<a href="/user/repo/path/file" target="_blank" rel="nofollow noopener"><img src="/user/repo/path/file" alt="local image"/></a><br/>
<a href="/user/repo/path/file" target="_blank" rel="nofollow noopener"><img src="/user/repo/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/user/repo/image.jpg" rel="nofollow"><img src="/user/repo/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -874,8 +854,6 @@ space</p>
<a href="/relative/path/wiki/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/relative/path/wiki/raw/image.jpg" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/image.jpg" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/relative/path/wiki/raw/image.jpg" rel="nofollow"><img src="/relative/path/wiki/raw/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -905,8 +883,6 @@ space</p>
<a href="/user/repo/src/branch/main/sub/folder/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/user/repo/media/branch/main/sub/folder/image.jpg" target="_blank" rel="nofollow noopener"><img src="/user/repo/media/branch/main/sub/folder/image.jpg" alt="local image"/></a><br/>
<a href="/user/repo/media/branch/main/sub/folder/path/file" target="_blank" rel="nofollow noopener"><img src="/user/repo/media/branch/main/sub/folder/path/file" alt="local image"/></a><br/>
<a href="/user/repo/media/branch/main/sub/folder/path/file" target="_blank" rel="nofollow noopener"><img src="/user/repo/media/branch/main/sub/folder/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/user/repo/media/branch/main/sub/folder/image.jpg" rel="nofollow"><img src="/user/repo/media/branch/main/sub/folder/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>
@ -936,8 +912,6 @@ space</p>
<a href="/relative/path/wiki/file.bin" rel="nofollow">local link</a><br/>
<a href="https://example.com" rel="nofollow">remote link</a><br/>
<a href="/relative/path/wiki/raw/image.jpg" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/image.jpg" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="/relative/path/wiki/raw/path/file" target="_blank" rel="nofollow noopener"><img src="/relative/path/wiki/raw/path/file" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" target="_blank" rel="nofollow noopener"><img src="https://example.com/image.jpg" alt="remote image"/></a><br/>
<a href="/relative/path/wiki/raw/image.jpg" rel="nofollow"><img src="/relative/path/wiki/raw/image.jpg" title="local image" alt="local image"/></a><br/>
<a href="https://example.com/image.jpg" rel="nofollow"><img src="https://example.com/image.jpg" title="remote link" alt="remote link"/></a><br/>

View File

@ -384,7 +384,7 @@ func Diff(ctx *context.Context) {
Metas: ctx.Repo.Repository.ComposeMetas(),
GitRepo: ctx.Repo.GitRepo,
Ctx: ctx,
}, template.HTMLEscapeString(string(charset.ToUTF8WithFallback(note.Message, charset.ConvertOpts{}))))
}, template.HTMLEscapeString(string(charset.ToUTF8WithFallback(note.Message))))
if err != nil {
ctx.ServerError("RenderCommitMessage", err)
return

View File

@ -142,7 +142,7 @@ func setCsvCompareContext(ctx *context.Context) {
return nil, nil, err
}
csvReader, err := csv_module.CreateReaderAndDetermineDelimiter(ctx, charset.ToUTF8WithFallbackReader(reader, charset.ConvertOpts{}))
csvReader, err := csv_module.CreateReaderAndDetermineDelimiter(ctx, charset.ToUTF8WithFallbackReader(reader))
return csvReader, reader, err
}

View File

@ -166,8 +166,8 @@ func editFile(ctx *context.Context, isNewFile bool) {
}
buf = append(buf, d...)
if content, err := charset.ToUTF8(buf, charset.ConvertOpts{KeepBOM: true}); err != nil {
log.Error("ToUTF8: %v", err)
if content, err := charset.ToUTF8WithErr(buf); err != nil {
log.Error("ToUTF8WithErr: %v", err)
ctx.Data["FileContent"] = string(buf)
} else {
ctx.Data["FileContent"] = content

View File

@ -43,7 +43,7 @@ func RenderFile(ctx *context.Context) {
st := typesniffer.DetectContentType(buf)
isTextFile := st.IsText()
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc), charset.ConvertOpts{})
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc))
if markupType := markup.Type(blob.Name()); markupType == "" {
if isTextFile {

View File

@ -303,7 +303,7 @@ func LFSFileGet(ctx *context.Context) {
break
}
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc), charset.ConvertOpts{})
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc))
// Building code view blocks with line number on server side.
escapedContent := &bytes.Buffer{}

View File

@ -303,7 +303,7 @@ func renderReadmeFile(ctx *context.Context, subfolder string, readmeFile *git.Tr
return
}
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc), charset.ConvertOpts{})
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc))
if markupType := markup.Type(readmeFile.Name()); markupType != "" {
ctx.Data["IsMarkup"] = true
@ -449,7 +449,7 @@ func renderFile(ctx *context.Context, entry *git.TreeEntry) {
break
}
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc), charset.ConvertOpts{})
rd := charset.ToUTF8WithFallbackReader(io.MultiReader(bytes.NewReader(buf), dataRc))
shouldRenderSource := ctx.FormString("display") == "source"
readmeExist := util.IsReadmeFileName(blob.Name())

View File

@ -38,7 +38,8 @@
data-url="{{.Repository.Link}}/markup"
data-context="{{.RepoLink}}"
data-previewable-extensions="{{.PreviewableExtensions}}"
data-line-wrap-extensions="{{.LineWrapExtensions}}">{{.FileContent}}</textarea>
data-line-wrap-extensions="{{.LineWrapExtensions}}">
{{.FileContent}}</textarea>
<div class="editor-loading is-loading"></div>
</div>
<div class="ui bottom attached tab segment markup" data-tab="preview">

View File

@ -141,7 +141,7 @@ func readSQLFromFile(version string) (string, error) {
if err != nil {
return "", err
}
return string(charset.MaybeRemoveBOM(bytes, charset.ConvertOpts{})), nil
return string(charset.RemoveBOMIfPresent(bytes)), nil
}
func restoreOldDB(t *testing.T, version string) bool {

View File

@ -22,7 +22,7 @@
height: min(4em, 66.6%);
aspect-ratio: 1;
transform: translate(-50%, -50%);
animation: isloadingspin 1000ms infinite linear;
animation: isloadingspin 500ms infinite linear;
border-width: 4px;
border-style: solid;
border-color: var(--color-secondary) var(--color-secondary) var(--color-secondary-dark-8) var(--color-secondary-dark-8);

View File

@ -114,7 +114,7 @@ export async function createMonaco(textarea, filename, editorOpts) {
const model = editor.getModel();
model.onDidChangeContent(() => {
textarea.value = editor.getValue({preserveBOM: true});
textarea.value = editor.getValue();
textarea.dispatchEvent(new Event('change')); // seems to be needed for jquery-are-you-sure
});