lloeki/fix_latin
1
0
Fork 0
mirror of https://github.com/lloeki/fix_latin.git synced 2025-12-06 10:04:40 +01:00
Code Issues Projects Releases Packages Wiki Activity

iso-8859-15 support

Browse source
This commit is contained in:
Loic Nageleisen 2015-06-30 11:48:26 +02:00
parent 1afd423e30
commit 44c907a4bd
1 changed files with 33 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

37
fix_latin.go
View file

@ -11,13 +11,14 @@ type Encoding int
const ( const (
ISO_8859_1 Encoding = iota ISO_8859_1 Encoding = iota
// TODO: ISO_8859_15 ISO_8859_15
CP1252 CP1252
) )
type Fixer struct { type Fixer struct {
allowControl bool allowControl bool
handleCP1252 bool handleCP1252 bool
handleISO_8859_15 bool
} }
func AllowControl(f *Fixer) error { func AllowControl(f *Fixer) error {
@ -27,11 +28,28 @@ func AllowControl(f *Fixer) error {
func Assume(e Encoding) func(*Fixer) error { func Assume(e Encoding) func(*Fixer) error {
return func(f *Fixer) error { return func(f *Fixer) error {
f.handleCP1252 = e == CP1252 switch e {
case CP1252:
f.handleCP1252 = true
case ISO_8859_15:
f.handleISO_8859_15 = true
}
return nil return nil
} }
} }
// remainder is ISO-8859-1
var iso_8859_15 = map[byte][]byte{
0xA4: {0xE2, 0x82, 0xAC}, // EURO SIGN
0xA6: {0xC5, 0xA0}, // LATIN CAPITAL LETTER S WITH CARON
0xA8: {0xC5, 0xA1}, // LATIN SMALL LETTER S WITH CARON
0xB4: {0xC5, 0xBD}, // LATIN CAPITAL LETTER Z WITH CARON
0xB8: {0xC5, 0xBE}, // LATIN SMALL LETTER Z WITH CARON
0xBC: {0xC5, 0x92}, // LATIN CAPITAL LIGATURE OE
0xBD: {0xC5, 0x93}, // LATIN SMALL LIGATURE OE
0xBE: {0xC5, 0xB8}, // LATIN CAPITAL LETTER Y WITH DIAERESIS
}
// remainder is ISO-8859-1 // remainder is ISO-8859-1
// does not define 0x81, 0x8D, 0x8F, 0x90, 09D // does not define 0x81, 0x8D, 0x8F, 0x90, 09D
var cp1252 = map[byte][]byte{ var cp1252 = map[byte][]byte{
@ -147,6 +165,17 @@ func Fix(r io.Reader, w io.Writer, options ...func(*Fixer) error) {
} }
} }
// ISO-8859-15
if f.handleISO_8859_15 {
if bytes, ok := iso_8859_15[input[0]]; ok {
for _, b := range bytes {
output = append(output, b)
}
input = input[1:]
continue
}
}
// ISO-8859-1 high-order control chars // ISO-8859-1 high-order control chars
if !f.allowControl && input[0] >= 0x80 && input[0] <= 0x9F { if !f.allowControl && input[0] >= 0x80 && input[0] <= 0x9F {
panic("control char") panic("control char")