1// Go support for Protocol Buffers - Google's data interchange format
2//
3// Copyright 2010 The Go Authors.  All rights reserved.
4// https://github.com/golang/protobuf
5//
6// Redistribution and use in source and binary forms, with or without
7// modification, are permitted provided that the following conditions are
8// met:
9//
10//     * Redistributions of source code must retain the above copyright
11// notice, this list of conditions and the following disclaimer.
12//     * Redistributions in binary form must reproduce the above
13// copyright notice, this list of conditions and the following disclaimer
14// in the documentation and/or other materials provided with the
15// distribution.
16//     * Neither the name of Google Inc. nor the names of its
17// contributors may be used to endorse or promote products derived from
18// this software without specific prior written permission.
19//
20// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
21// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
22// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
23// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
24// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
25// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
26// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
27// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
28// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
29// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
30// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
31
32package proto
33
34/*
35 * Routines for decoding protocol buffer data to construct in-memory representations.
36 */
37
38import (
39	"errors"
40	"fmt"
41	"io"
42)
43
44// errOverflow is returned when an integer is too large to be represented.
45var errOverflow = errors.New("proto: integer overflow")
46
47// ErrInternalBadWireType is returned by generated code when an incorrect
48// wire type is encountered. It does not get returned to user code.
49var ErrInternalBadWireType = errors.New("proto: internal error: bad wiretype for oneof")
50
51// DecodeVarint reads a varint-encoded integer from the slice.
52// It returns the integer and the number of bytes consumed, or
53// zero if there is not enough.
54// This is the format for the
55// int32, int64, uint32, uint64, bool, and enum
56// protocol buffer types.
57func DecodeVarint(buf []byte) (x uint64, n int) {
58	for shift := uint(0); shift < 64; shift += 7 {
59		if n >= len(buf) {
60			return 0, 0
61		}
62		b := uint64(buf[n])
63		n++
64		x |= (b & 0x7F) << shift
65		if (b & 0x80) == 0 {
66			return x, n
67		}
68	}
69
70	// The number is too large to represent in a 64-bit value.
71	return 0, 0
72}
73
74func (p *Buffer) decodeVarintSlow() (x uint64, err error) {
75	i := p.index
76	l := len(p.buf)
77
78	for shift := uint(0); shift < 64; shift += 7 {
79		if i >= l {
80			err = io.ErrUnexpectedEOF
81			return
82		}
83		b := p.buf[i]
84		i++
85		x |= (uint64(b) & 0x7F) << shift
86		if b < 0x80 {
87			p.index = i
88			return
89		}
90	}
91
92	// The number is too large to represent in a 64-bit value.
93	err = errOverflow
94	return
95}
96
97// DecodeVarint reads a varint-encoded integer from the Buffer.
98// This is the format for the
99// int32, int64, uint32, uint64, bool, and enum
100// protocol buffer types.
101func (p *Buffer) DecodeVarint() (x uint64, err error) {
102	i := p.index
103	buf := p.buf
104
105	if i >= len(buf) {
106		return 0, io.ErrUnexpectedEOF
107	} else if buf[i] < 0x80 {
108		p.index++
109		return uint64(buf[i]), nil
110	} else if len(buf)-i < 10 {
111		return p.decodeVarintSlow()
112	}
113
114	var b uint64
115	// we already checked the first byte
116	x = uint64(buf[i]) - 0x80
117	i++
118
119	b = uint64(buf[i])
120	i++
121	x += b << 7
122	if b&0x80 == 0 {
123		goto done
124	}
125	x -= 0x80 << 7
126
127	b = uint64(buf[i])
128	i++
129	x += b << 14
130	if b&0x80 == 0 {
131		goto done
132	}
133	x -= 0x80 << 14
134
135	b = uint64(buf[i])
136	i++
137	x += b << 21
138	if b&0x80 == 0 {
139		goto done
140	}
141	x -= 0x80 << 21
142
143	b = uint64(buf[i])
144	i++
145	x += b << 28
146	if b&0x80 == 0 {
147		goto done
148	}
149	x -= 0x80 << 28
150
151	b = uint64(buf[i])
152	i++
153	x += b << 35
154	if b&0x80 == 0 {
155		goto done
156	}
157	x -= 0x80 << 35
158
159	b = uint64(buf[i])
160	i++
161	x += b << 42
162	if b&0x80 == 0 {
163		goto done
164	}
165	x -= 0x80 << 42
166
167	b = uint64(buf[i])
168	i++
169	x += b << 49
170	if b&0x80 == 0 {
171		goto done
172	}
173	x -= 0x80 << 49
174
175	b = uint64(buf[i])
176	i++
177	x += b << 56
178	if b&0x80 == 0 {
179		goto done
180	}
181	x -= 0x80 << 56
182
183	b = uint64(buf[i])
184	i++
185	x += b << 63
186	if b&0x80 == 0 {
187		goto done
188	}
189	// x -= 0x80 << 63 // Always zero.
190
191	return 0, errOverflow
192
193done:
194	p.index = i
195	return x, nil
196}
197
198// DecodeFixed64 reads a 64-bit integer from the Buffer.
199// This is the format for the
200// fixed64, sfixed64, and double protocol buffer types.
201func (p *Buffer) DecodeFixed64() (x uint64, err error) {
202	// x, err already 0
203	i := p.index + 8
204	if i < 0 || i > len(p.buf) {
205		err = io.ErrUnexpectedEOF
206		return
207	}
208	p.index = i
209
210	x = uint64(p.buf[i-8])
211	x |= uint64(p.buf[i-7]) << 8
212	x |= uint64(p.buf[i-6]) << 16
213	x |= uint64(p.buf[i-5]) << 24
214	x |= uint64(p.buf[i-4]) << 32
215	x |= uint64(p.buf[i-3]) << 40
216	x |= uint64(p.buf[i-2]) << 48
217	x |= uint64(p.buf[i-1]) << 56
218	return
219}
220
221// DecodeFixed32 reads a 32-bit integer from the Buffer.
222// This is the format for the
223// fixed32, sfixed32, and float protocol buffer types.
224func (p *Buffer) DecodeFixed32() (x uint64, err error) {
225	// x, err already 0
226	i := p.index + 4
227	if i < 0 || i > len(p.buf) {
228		err = io.ErrUnexpectedEOF
229		return
230	}
231	p.index = i
232
233	x = uint64(p.buf[i-4])
234	x |= uint64(p.buf[i-3]) << 8
235	x |= uint64(p.buf[i-2]) << 16
236	x |= uint64(p.buf[i-1]) << 24
237	return
238}
239
240// DecodeZigzag64 reads a zigzag-encoded 64-bit integer
241// from the Buffer.
242// This is the format used for the sint64 protocol buffer type.
243func (p *Buffer) DecodeZigzag64() (x uint64, err error) {
244	x, err = p.DecodeVarint()
245	if err != nil {
246		return
247	}
248	x = (x >> 1) ^ uint64((int64(x&1)<<63)>>63)
249	return
250}
251
252// DecodeZigzag32 reads a zigzag-encoded 32-bit integer
253// from  the Buffer.
254// This is the format used for the sint32 protocol buffer type.
255func (p *Buffer) DecodeZigzag32() (x uint64, err error) {
256	x, err = p.DecodeVarint()
257	if err != nil {
258		return
259	}
260	x = uint64((uint32(x) >> 1) ^ uint32((int32(x&1)<<31)>>31))
261	return
262}
263
264// DecodeRawBytes reads a count-delimited byte buffer from the Buffer.
265// This is the format used for the bytes protocol buffer
266// type and for embedded messages.
267func (p *Buffer) DecodeRawBytes(alloc bool) (buf []byte, err error) {
268	n, err := p.DecodeVarint()
269	if err != nil {
270		return nil, err
271	}
272
273	nb := int(n)
274	if nb < 0 {
275		return nil, fmt.Errorf("proto: bad byte length %d", nb)
276	}
277	end := p.index + nb
278	if end < p.index || end > len(p.buf) {
279		return nil, io.ErrUnexpectedEOF
280	}
281
282	if !alloc {
283		// todo: check if can get more uses of alloc=false
284		buf = p.buf[p.index:end]
285		p.index += nb
286		return
287	}
288
289	buf = make([]byte, nb)
290	copy(buf, p.buf[p.index:])
291	p.index += nb
292	return
293}
294
295// DecodeStringBytes reads an encoded string from the Buffer.
296// This is the format used for the proto2 string type.
297func (p *Buffer) DecodeStringBytes() (s string, err error) {
298	buf, err := p.DecodeRawBytes(false)
299	if err != nil {
300		return
301	}
302	return string(buf), nil
303}
304
305// Unmarshaler is the interface representing objects that can
306// unmarshal themselves.  The argument points to data that may be
307// overwritten, so implementations should not keep references to the
308// buffer.
309// Unmarshal implementations should not clear the receiver.
310// Any unmarshaled data should be merged into the receiver.
311// Callers of Unmarshal that do not want to retain existing data
312// should Reset the receiver before calling Unmarshal.
313type Unmarshaler interface {
314	Unmarshal([]byte) error
315}
316
317// newUnmarshaler is the interface representing objects that can
318// unmarshal themselves. The semantics are identical to Unmarshaler.
319//
320// This exists to support protoc-gen-go generated messages.
321// The proto package will stop type-asserting to this interface in the future.
322//
323// DO NOT DEPEND ON THIS.
324type newUnmarshaler interface {
325	XXX_Unmarshal([]byte) error
326}
327
328// Unmarshal parses the protocol buffer representation in buf and places the
329// decoded result in pb.  If the struct underlying pb does not match
330// the data in buf, the results can be unpredictable.
331//
332// Unmarshal resets pb before starting to unmarshal, so any
333// existing data in pb is always removed. Use UnmarshalMerge
334// to preserve and append to existing data.
335func Unmarshal(buf []byte, pb Message) error {
336	pb.Reset()
337	if u, ok := pb.(newUnmarshaler); ok {
338		return u.XXX_Unmarshal(buf)
339	}
340	if u, ok := pb.(Unmarshaler); ok {
341		return u.Unmarshal(buf)
342	}
343	return NewBuffer(buf).Unmarshal(pb)
344}
345
346// UnmarshalMerge parses the protocol buffer representation in buf and
347// writes the decoded result to pb.  If the struct underlying pb does not match
348// the data in buf, the results can be unpredictable.
349//
350// UnmarshalMerge merges into existing data in pb.
351// Most code should use Unmarshal instead.
352func UnmarshalMerge(buf []byte, pb Message) error {
353	if u, ok := pb.(newUnmarshaler); ok {
354		return u.XXX_Unmarshal(buf)
355	}
356	if u, ok := pb.(Unmarshaler); ok {
357		// NOTE: The history of proto have unfortunately been inconsistent
358		// whether Unmarshaler should or should not implicitly clear itself.
359		// Some implementations do, most do not.
360		// Thus, calling this here may or may not do what people want.
361		//
362		// See https://github.com/golang/protobuf/issues/424
363		return u.Unmarshal(buf)
364	}
365	return NewBuffer(buf).Unmarshal(pb)
366}
367
368// DecodeMessage reads a count-delimited message from the Buffer.
369func (p *Buffer) DecodeMessage(pb Message) error {
370	enc, err := p.DecodeRawBytes(false)
371	if err != nil {
372		return err
373	}
374	return NewBuffer(enc).Unmarshal(pb)
375}
376
377// DecodeGroup reads a tag-delimited group from the Buffer.
378// StartGroup tag is already consumed. This function consumes
379// EndGroup tag.
380func (p *Buffer) DecodeGroup(pb Message) error {
381	b := p.buf[p.index:]
382	x, y := findEndGroup(b)
383	if x < 0 {
384		return io.ErrUnexpectedEOF
385	}
386	err := Unmarshal(b[:x], pb)
387	p.index += y
388	return err
389}
390
391// Unmarshal parses the protocol buffer representation in the
392// Buffer and places the decoded result in pb.  If the struct
393// underlying pb does not match the data in the buffer, the results can be
394// unpredictable.
395//
396// Unlike proto.Unmarshal, this does not reset pb before starting to unmarshal.
397func (p *Buffer) Unmarshal(pb Message) error {
398	// If the object can unmarshal itself, let it.
399	if u, ok := pb.(newUnmarshaler); ok {
400		err := u.XXX_Unmarshal(p.buf[p.index:])
401		p.index = len(p.buf)
402		return err
403	}
404	if u, ok := pb.(Unmarshaler); ok {
405		// NOTE: The history of proto have unfortunately been inconsistent
406		// whether Unmarshaler should or should not implicitly clear itself.
407		// Some implementations do, most do not.
408		// Thus, calling this here may or may not do what people want.
409		//
410		// See https://github.com/golang/protobuf/issues/424
411		err := u.Unmarshal(p.buf[p.index:])
412		p.index = len(p.buf)
413		return err
414	}
415
416	// Slow workaround for messages that aren't Unmarshalers.
417	// This includes some hand-coded .pb.go files and
418	// bootstrap protos.
419	// TODO: fix all of those and then add Unmarshal to
420	// the Message interface. Then:
421	// The cast above and code below can be deleted.
422	// The old unmarshaler can be deleted.
423	// Clients can call Unmarshal directly (can already do that, actually).
424	var info InternalMessageInfo
425	err := info.Unmarshal(pb, p.buf[p.index:])
426	p.index = len(p.buf)
427	return err
428}
429