2009-10-20 10:04:09 -06:00
|
|
|
// Copyright 2009 The Go Authors. All rights reserved.
|
|
|
|
// Use of this source code is governed by a BSD-style
|
|
|
|
// license that can be found in the LICENSE file.
|
|
|
|
|
2009-10-20 14:00:16 -06:00
|
|
|
// Package ascii85 implements the ascii85 data encoding
|
|
|
|
// as used in the btoa tool and Adobe's PostScript and PDF document formats.
|
|
|
|
package ascii85
|
2009-10-20 10:04:09 -06:00
|
|
|
|
|
|
|
import (
|
|
|
|
"io";
|
|
|
|
"os";
|
|
|
|
"strconv";
|
|
|
|
)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Encoder
|
|
|
|
*/
|
|
|
|
|
|
|
|
// Encode encodes src into at most MaxEncodedLen(len(src))
|
|
|
|
// bytes of dst, returning the actual number of bytes written.
|
|
|
|
//
|
|
|
|
// The encoding handles 4-byte chunks, using a special encoding
|
|
|
|
// for the last fragment, so Encode is not appropriate for use on
|
|
|
|
// individual blocks of a large data stream. Use NewEncoder() instead.
|
2009-10-20 14:00:16 -06:00
|
|
|
//
|
|
|
|
// Often, ascii85-encoded data is wrapped in <~ and ~> symbols.
|
|
|
|
// Encode does not add these.
|
2009-11-03 18:32:08 -07:00
|
|
|
func Encode(dst, src []byte) int {
|
2009-10-20 10:04:09 -06:00
|
|
|
if len(src) == 0 {
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
n := 0;
|
|
|
|
for len(src) > 0 {
|
|
|
|
dst[0] = 0;
|
|
|
|
dst[1] = 0;
|
|
|
|
dst[2] = 0;
|
|
|
|
dst[3] = 0;
|
|
|
|
dst[4] = 0;
|
|
|
|
|
|
|
|
// Unpack 4 bytes into uint32 to repack into base 85 5-byte.
|
|
|
|
var v uint32;
|
|
|
|
switch len(src) {
|
|
|
|
default:
|
|
|
|
v |= uint32(src[3]);
|
|
|
|
fallthrough;
|
|
|
|
case 3:
|
2009-11-09 22:13:17 -07:00
|
|
|
v |= uint32(src[2]) << 8;
|
2009-10-20 10:04:09 -06:00
|
|
|
fallthrough;
|
|
|
|
case 2:
|
2009-11-09 22:13:17 -07:00
|
|
|
v |= uint32(src[1]) << 16;
|
2009-10-20 10:04:09 -06:00
|
|
|
fallthrough;
|
|
|
|
case 1:
|
2009-11-09 22:13:17 -07:00
|
|
|
v |= uint32(src[0]) << 24
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
// Special case: zero (!!!!!) shortens to z.
|
|
|
|
if v == 0 && len(src) >= 4 {
|
|
|
|
dst[0] = 'z';
|
2009-11-20 12:45:05 -07:00
|
|
|
dst = dst[1:];
|
2009-10-20 10:04:09 -06:00
|
|
|
n++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Otherwise, 5 base 85 digits starting at !.
|
|
|
|
for i := 4; i >= 0; i-- {
|
2009-11-09 22:13:17 -07:00
|
|
|
dst[i] = '!' + byte(v%85);
|
2009-10-20 10:04:09 -06:00
|
|
|
v /= 85;
|
|
|
|
}
|
|
|
|
|
|
|
|
// If src was short, discard the low destination bytes.
|
|
|
|
m := 5;
|
|
|
|
if len(src) < 4 {
|
2009-11-09 22:13:17 -07:00
|
|
|
m -= 4 - len(src);
|
2009-10-20 10:04:09 -06:00
|
|
|
src = nil;
|
|
|
|
} else {
|
2009-11-20 12:45:05 -07:00
|
|
|
src = src[4:]
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
2009-11-20 12:45:05 -07:00
|
|
|
dst = dst[m:];
|
2009-10-20 10:04:09 -06:00
|
|
|
n += m;
|
|
|
|
}
|
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
|
|
|
// MaxEncodedLen returns the maximum length of an encoding of n source bytes.
|
2009-11-09 22:13:17 -07:00
|
|
|
func MaxEncodedLen(n int) int { return (n + 3) / 4 * 5 }
|
2009-10-20 10:04:09 -06:00
|
|
|
|
|
|
|
// NewEncoder returns a new ascii85 stream encoder. Data written to
|
|
|
|
// the returned writer will be encoded and then written to w.
|
|
|
|
// Ascii85 encodings operate in 32-bit blocks; when finished
|
|
|
|
// writing, the caller must Close the returned encoder to flush any
|
|
|
|
// trailing partial block.
|
2009-11-06 15:24:38 -07:00
|
|
|
func NewEncoder(w io.Writer) io.WriteCloser { return &encoder{w: w} }
|
2009-10-20 10:04:09 -06:00
|
|
|
|
|
|
|
type encoder struct {
|
|
|
|
err os.Error;
|
|
|
|
w io.Writer;
|
|
|
|
buf [4]byte; // buffered data waiting to be encoded
|
|
|
|
nbuf int; // number of bytes in buf
|
|
|
|
out [1024]byte; // output buffer
|
|
|
|
}
|
|
|
|
|
|
|
|
func (e *encoder) Write(p []byte) (n int, err os.Error) {
|
|
|
|
if e.err != nil {
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0, e.err
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
// Leading fringe.
|
|
|
|
if e.nbuf > 0 {
|
|
|
|
var i int;
|
|
|
|
for i = 0; i < len(p) && e.nbuf < 4; i++ {
|
|
|
|
e.buf[e.nbuf] = p[i];
|
|
|
|
e.nbuf++;
|
|
|
|
}
|
|
|
|
n += i;
|
2009-11-20 12:45:05 -07:00
|
|
|
p = p[i:];
|
2009-10-20 10:04:09 -06:00
|
|
|
if e.nbuf < 4 {
|
2009-11-09 13:07:39 -07:00
|
|
|
return
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
2009-11-03 18:32:08 -07:00
|
|
|
nout := Encode(&e.out, &e.buf);
|
2009-10-20 10:04:09 -06:00
|
|
|
if _, e.err = e.w.Write(e.out[0:nout]); e.err != nil {
|
2009-11-09 13:07:39 -07:00
|
|
|
return n, e.err
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
e.nbuf = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Large interior chunks.
|
|
|
|
for len(p) >= 4 {
|
2009-11-09 22:13:17 -07:00
|
|
|
nn := len(e.out) / 5 * 4;
|
2009-10-20 10:04:09 -06:00
|
|
|
if nn > len(p) {
|
2009-11-09 13:07:39 -07:00
|
|
|
nn = len(p)
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
2009-11-09 22:13:17 -07:00
|
|
|
nn -= nn % 4;
|
2009-10-20 10:04:09 -06:00
|
|
|
if nn > 0 {
|
2009-11-03 18:32:08 -07:00
|
|
|
nout := Encode(&e.out, p[0:nn]);
|
2009-10-20 10:04:09 -06:00
|
|
|
if _, e.err = e.w.Write(e.out[0:nout]); e.err != nil {
|
2009-11-09 13:07:39 -07:00
|
|
|
return n, e.err
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
n += nn;
|
2009-11-20 12:45:05 -07:00
|
|
|
p = p[nn:];
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
// Trailing fringe.
|
|
|
|
for i := 0; i < len(p); i++ {
|
2009-11-09 13:07:39 -07:00
|
|
|
e.buf[i] = p[i]
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
e.nbuf = len(p);
|
|
|
|
n += len(p);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Close flushes any pending output from the encoder.
|
|
|
|
// It is an error to call Write after calling Close.
|
|
|
|
func (e *encoder) Close() os.Error {
|
|
|
|
// If there's anything left in the buffer, flush it out
|
|
|
|
if e.err == nil && e.nbuf > 0 {
|
2009-11-09 22:13:17 -07:00
|
|
|
nout := Encode(&e.out, e.buf[0:e.nbuf]);
|
2009-10-20 10:04:09 -06:00
|
|
|
e.nbuf = 0;
|
|
|
|
_, e.err = e.w.Write(e.out[0:nout]);
|
|
|
|
}
|
|
|
|
return e.err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Decoder
|
|
|
|
*/
|
|
|
|
|
|
|
|
type CorruptInputError int64
|
|
|
|
|
|
|
|
func (e CorruptInputError) String() string {
|
2009-11-09 13:07:39 -07:00
|
|
|
return "illegal ascii85 data at input byte" + strconv.Itoa64(int64(e))
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
// Decode decodes src into dst, returning both the number
|
2009-11-03 18:32:08 -07:00
|
|
|
// of bytes written to dst and the number consumed from src.
|
2009-10-20 10:04:09 -06:00
|
|
|
// If src contains invalid ascii85 data, Decode will return the
|
|
|
|
// number of bytes successfully written and a CorruptInputError.
|
|
|
|
// Decode ignores space and control characters in src.
|
2009-10-20 14:00:16 -06:00
|
|
|
// Often, ascii85-encoded data is wrapped in <~ and ~> symbols.
|
|
|
|
// Decode expects these to have been stripped by the caller.
|
2009-10-20 10:04:09 -06:00
|
|
|
//
|
|
|
|
// If flush is true, Decode assumes that src represents the
|
|
|
|
// end of the input stream and processes it completely rather
|
|
|
|
// than wait for the completion of another 32-bit block.
|
|
|
|
//
|
|
|
|
// NewDecoder wraps an io.Reader interface around Decode.
|
|
|
|
//
|
2009-11-03 18:32:08 -07:00
|
|
|
func Decode(dst, src []byte, flush bool) (ndst, nsrc int, err os.Error) {
|
2009-10-20 10:04:09 -06:00
|
|
|
var v uint32;
|
|
|
|
var nb int;
|
|
|
|
for i, b := range src {
|
2009-11-05 09:55:18 -07:00
|
|
|
if len(dst)-ndst < 4 {
|
2009-11-09 13:07:39 -07:00
|
|
|
return
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
switch {
|
|
|
|
case b <= ' ':
|
2009-11-09 13:07:39 -07:00
|
|
|
continue
|
2009-10-20 10:04:09 -06:00
|
|
|
case b == 'z' && nb == 0:
|
|
|
|
nb = 5;
|
|
|
|
v = 0;
|
|
|
|
case '!' <= b && b <= 'u':
|
2009-11-05 09:55:18 -07:00
|
|
|
v = v*85 + uint32(b-'!');
|
2009-10-20 10:04:09 -06:00
|
|
|
nb++;
|
|
|
|
default:
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0, 0, CorruptInputError(i)
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
if nb == 5 {
|
2009-11-09 22:13:17 -07:00
|
|
|
nsrc = i + 1;
|
|
|
|
dst[ndst] = byte(v >> 24);
|
|
|
|
dst[ndst+1] = byte(v >> 16);
|
|
|
|
dst[ndst+2] = byte(v >> 8);
|
2009-10-20 10:04:09 -06:00
|
|
|
dst[ndst+3] = byte(v);
|
|
|
|
ndst += 4;
|
|
|
|
nb = 0;
|
|
|
|
v = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if flush {
|
|
|
|
nsrc = len(src);
|
|
|
|
if nb > 0 {
|
|
|
|
// The number of output bytes in the last fragment
|
|
|
|
// is the number of leftover input bytes - 1:
|
|
|
|
// the extra byte provides enough bits to cover
|
|
|
|
// the inefficiency of the encoding for the block.
|
|
|
|
if nb == 1 {
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0, 0, CorruptInputError(len(src))
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
for i := nb; i < 5; i++ {
|
|
|
|
// The short encoding truncated the output value.
|
|
|
|
// We have to assume the worst case values (digit 84)
|
|
|
|
// in order to ensure that the top bits are correct.
|
2009-11-09 13:07:39 -07:00
|
|
|
v = v*85 + 84
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
for i := 0; i < nb-1; i++ {
|
2009-11-09 22:13:17 -07:00
|
|
|
dst[ndst] = byte(v >> 24);
|
2009-10-20 10:04:09 -06:00
|
|
|
v <<= 8;
|
|
|
|
ndst++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
// NewDecoder constructs a new ascii85 stream decoder.
|
2009-11-06 15:24:38 -07:00
|
|
|
func NewDecoder(r io.Reader) io.Reader { return &decoder{r: r} }
|
2009-10-20 10:04:09 -06:00
|
|
|
|
|
|
|
type decoder struct {
|
|
|
|
err os.Error;
|
|
|
|
readErr os.Error;
|
|
|
|
r io.Reader;
|
|
|
|
end bool; // saw end of message
|
|
|
|
buf [1024]byte; // leftover input
|
|
|
|
nbuf int;
|
|
|
|
out []byte; // leftover decoded output
|
|
|
|
outbuf [1024]byte;
|
|
|
|
}
|
|
|
|
|
|
|
|
func (d *decoder) Read(p []byte) (n int, err os.Error) {
|
|
|
|
if len(p) == 0 {
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0, nil
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
if d.err != nil {
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0, d.err
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
for {
|
|
|
|
// Copy leftover output from last decode.
|
|
|
|
if len(d.out) > 0 {
|
2009-11-18 16:24:24 -07:00
|
|
|
n = copy(p, d.out);
|
2009-11-20 12:45:05 -07:00
|
|
|
d.out = d.out[n:];
|
2009-10-20 10:04:09 -06:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Decode leftover input from last read.
|
|
|
|
var nn, nsrc, ndst int;
|
|
|
|
if d.nbuf > 0 {
|
2009-11-09 22:13:17 -07:00
|
|
|
ndst, nsrc, d.err = Decode(&d.outbuf, d.buf[0:d.nbuf], d.readErr != nil);
|
2009-10-20 10:04:09 -06:00
|
|
|
if ndst > 0 {
|
|
|
|
d.out = d.outbuf[0:ndst];
|
2009-11-18 16:24:24 -07:00
|
|
|
d.nbuf = copy(&d.buf, d.buf[nsrc:d.nbuf]);
|
2009-10-20 10:04:09 -06:00
|
|
|
continue; // copy out and return
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Out of input, out of decoded output. Check errors.
|
|
|
|
if d.err != nil {
|
2009-11-09 13:07:39 -07:00
|
|
|
return 0, d.err
|
2009-10-20 10:04:09 -06:00
|
|
|
}
|
|
|
|
if d.readErr != nil {
|
|
|
|
d.err = d.readErr;
|
|
|
|
return 0, d.err;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Read more data.
|
2009-11-20 12:45:05 -07:00
|
|
|
nn, d.readErr = d.r.Read(d.buf[d.nbuf:]);
|
2009-10-20 10:04:09 -06:00
|
|
|
d.nbuf += nn;
|
|
|
|
}
|
|
|
|
panic("unreachable");
|
|
|
|
}
|