Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 1 | // Copyright 2019 The Go Authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style. |
| 3 | // license that can be found in the LICENSE file. |
| 4 | |
| 5 | // Package messageset encodes and decodes the obsolete MessageSet wire format. |
| 6 | package messageset |
| 7 | |
| 8 | import ( |
| 9 | "google.golang.org/protobuf/internal/encoding/wire" |
| 10 | "google.golang.org/protobuf/internal/errors" |
| 11 | pref "google.golang.org/protobuf/reflect/protoreflect" |
Joe Tsai | 5ae10aa | 2019-07-11 18:23:08 -0700 | [diff] [blame] | 12 | preg "google.golang.org/protobuf/reflect/protoregistry" |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 13 | ) |
| 14 | |
| 15 | // The MessageSet wire format is equivalent to a message defiend as follows, |
| 16 | // where each Item defines an extension field with a field number of 'type_id' |
| 17 | // and content of 'message'. MessageSet extensions must be non-repeated message |
| 18 | // fields. |
| 19 | // |
| 20 | // message MessageSet { |
| 21 | // repeated group Item = 1 { |
| 22 | // required int32 type_id = 2; |
| 23 | // required string message = 3; |
| 24 | // } |
| 25 | // } |
| 26 | const ( |
| 27 | FieldItem = wire.Number(1) |
| 28 | FieldTypeID = wire.Number(2) |
| 29 | FieldMessage = wire.Number(3) |
| 30 | ) |
| 31 | |
Joe Tsai | 5ae10aa | 2019-07-11 18:23:08 -0700 | [diff] [blame] | 32 | // ExtensionName is the field name for extensions of MessageSet. |
| 33 | // |
| 34 | // A valid MessageSet extension must be of the form: |
| 35 | // message MyMessage { |
| 36 | // extend proto2.bridge.MessageSet { |
| 37 | // optional MyMessage message_set_extension = 1234; |
| 38 | // } |
| 39 | // ... |
| 40 | // } |
| 41 | const ExtensionName = "message_set_extension" |
| 42 | |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 43 | // IsMessageSet returns whether the message uses the MessageSet wire format. |
| 44 | func IsMessageSet(md pref.MessageDescriptor) bool { |
| 45 | xmd, ok := md.(interface{ IsMessageSet() bool }) |
| 46 | return ok && xmd.IsMessageSet() |
| 47 | } |
| 48 | |
Joe Tsai | 5ae10aa | 2019-07-11 18:23:08 -0700 | [diff] [blame] | 49 | // IsMessageSetExtension reports this field extends a MessageSet. |
| 50 | func IsMessageSetExtension(fd pref.FieldDescriptor) bool { |
| 51 | if fd.Name() != ExtensionName { |
| 52 | return false |
| 53 | } |
| 54 | if fd.FullName().Parent() != fd.Message().FullName() { |
| 55 | return false |
| 56 | } |
| 57 | return IsMessageSet(fd.ContainingMessage()) |
| 58 | } |
| 59 | |
| 60 | // FindMessageSetExtension locates a MessageSet extension field by name. |
| 61 | // In text and JSON formats, the extension name used is the message itself. |
| 62 | // The extension field name is derived by appending ExtensionName. |
| 63 | func FindMessageSetExtension(r preg.ExtensionTypeResolver, s pref.FullName) (pref.ExtensionType, error) { |
Damien Neil | 01b51b4 | 2020-01-17 13:40:51 -0800 | [diff] [blame^] | 64 | name := s.Append(ExtensionName) |
| 65 | xt, err := r.FindExtensionByName(name) |
Joe Tsai | 5ae10aa | 2019-07-11 18:23:08 -0700 | [diff] [blame] | 66 | if err != nil { |
Damien Neil | 01b51b4 | 2020-01-17 13:40:51 -0800 | [diff] [blame^] | 67 | if err == preg.NotFound { |
| 68 | return nil, err |
| 69 | } |
| 70 | return nil, errors.Wrap(err, "%q", name) |
Joe Tsai | 5ae10aa | 2019-07-11 18:23:08 -0700 | [diff] [blame] | 71 | } |
Damien Neil | 79bfdbe | 2019-08-28 11:08:22 -0700 | [diff] [blame] | 72 | if !IsMessageSetExtension(xt.TypeDescriptor()) { |
Joe Tsai | 5ae10aa | 2019-07-11 18:23:08 -0700 | [diff] [blame] | 73 | return nil, preg.NotFound |
| 74 | } |
| 75 | return xt, nil |
| 76 | } |
| 77 | |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 78 | // SizeField returns the size of a MessageSet item field containing an extension |
| 79 | // with the given field number, not counting the contents of the message subfield. |
| 80 | func SizeField(num wire.Number) int { |
| 81 | return 2*wire.SizeTag(FieldItem) + wire.SizeTag(FieldTypeID) + wire.SizeVarint(uint64(num)) |
| 82 | } |
| 83 | |
Damien Neil | ce3384c | 2019-11-06 13:18:28 -0800 | [diff] [blame] | 84 | // Unmarshal parses a MessageSet. |
| 85 | // |
| 86 | // It calls fn with the type ID and value of each item in the MessageSet. |
| 87 | // Unknown fields are discarded. |
| 88 | // |
| 89 | // If wantLen is true, the item values include the varint length prefix. |
| 90 | // This is ugly, but simplifies the fast-path decoder in internal/impl. |
| 91 | func Unmarshal(b []byte, wantLen bool, fn func(typeID wire.Number, value []byte) error) error { |
| 92 | for len(b) > 0 { |
| 93 | num, wtyp, n := wire.ConsumeTag(b) |
| 94 | if n < 0 { |
| 95 | return wire.ParseError(n) |
| 96 | } |
| 97 | b = b[n:] |
| 98 | if num != FieldItem || wtyp != wire.StartGroupType { |
| 99 | n := wire.ConsumeFieldValue(num, wtyp, b) |
| 100 | if n < 0 { |
| 101 | return wire.ParseError(n) |
| 102 | } |
| 103 | b = b[n:] |
| 104 | continue |
| 105 | } |
Damien Neil | 9afe9bb | 2020-02-07 10:06:53 -0800 | [diff] [blame] | 106 | typeID, value, n, err := ConsumeFieldValue(b, wantLen) |
Damien Neil | ce3384c | 2019-11-06 13:18:28 -0800 | [diff] [blame] | 107 | if err != nil { |
| 108 | return err |
| 109 | } |
| 110 | b = b[n:] |
| 111 | if typeID == 0 { |
| 112 | continue |
| 113 | } |
| 114 | if err := fn(typeID, value); err != nil { |
| 115 | return err |
| 116 | } |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 117 | } |
Damien Neil | ce3384c | 2019-11-06 13:18:28 -0800 | [diff] [blame] | 118 | return nil |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 119 | } |
| 120 | |
Damien Neil | 9afe9bb | 2020-02-07 10:06:53 -0800 | [diff] [blame] | 121 | // ConsumeFieldValue parses b as a MessageSet item field value until and including |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 122 | // the trailing end group marker. It assumes the start group tag has already been parsed. |
| 123 | // It returns the contents of the type_id and message subfields and the total |
| 124 | // item length. |
| 125 | // |
| 126 | // If wantLen is true, the returned message value includes the length prefix. |
Damien Neil | 9afe9bb | 2020-02-07 10:06:53 -0800 | [diff] [blame] | 127 | func ConsumeFieldValue(b []byte, wantLen bool) (typeid wire.Number, message []byte, n int, err error) { |
Damien Neil | 302cb32 | 2019-06-19 15:22:13 -0700 | [diff] [blame] | 128 | ilen := len(b) |
| 129 | for { |
| 130 | num, wtyp, n := wire.ConsumeTag(b) |
| 131 | if n < 0 { |
| 132 | return 0, nil, 0, wire.ParseError(n) |
| 133 | } |
| 134 | b = b[n:] |
| 135 | switch { |
| 136 | case num == FieldItem && wtyp == wire.EndGroupType: |
| 137 | if wantLen && len(message) == 0 { |
| 138 | // The message field was missing, which should never happen. |
| 139 | // Be prepared for this case anyway. |
| 140 | message = wire.AppendVarint(message, 0) |
| 141 | } |
| 142 | return typeid, message, ilen - len(b), nil |
| 143 | case num == FieldTypeID && wtyp == wire.VarintType: |
| 144 | v, n := wire.ConsumeVarint(b) |
| 145 | if n < 0 { |
| 146 | return 0, nil, 0, wire.ParseError(n) |
| 147 | } |
| 148 | b = b[n:] |
| 149 | typeid = wire.Number(v) |
| 150 | case num == FieldMessage && wtyp == wire.BytesType: |
| 151 | m, n := wire.ConsumeBytes(b) |
| 152 | if n < 0 { |
| 153 | return 0, nil, 0, wire.ParseError(n) |
| 154 | } |
| 155 | if message == nil { |
| 156 | if wantLen { |
| 157 | message = b[:n] |
| 158 | } else { |
| 159 | message = m |
| 160 | } |
| 161 | } else { |
| 162 | // This case should never happen in practice, but handle it for |
| 163 | // correctness: The MessageSet item contains multiple message |
| 164 | // fields, which need to be merged. |
| 165 | // |
| 166 | // In the case where we're returning the length, this becomes |
| 167 | // quite inefficient since we need to strip the length off |
| 168 | // the existing data and reconstruct it with the combined length. |
| 169 | if wantLen { |
| 170 | _, nn := wire.ConsumeVarint(message) |
| 171 | m0 := message[nn:] |
| 172 | message = message[:0] |
| 173 | message = wire.AppendVarint(message, uint64(len(m0)+len(m))) |
| 174 | message = append(message, m0...) |
| 175 | message = append(message, m...) |
| 176 | } else { |
| 177 | message = append(message, m...) |
| 178 | } |
| 179 | } |
| 180 | b = b[n:] |
| 181 | } |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | // AppendFieldStart appends the start of a MessageSet item field containing |
| 186 | // an extension with the given number. The caller must add the message |
| 187 | // subfield (including the tag). |
| 188 | func AppendFieldStart(b []byte, num wire.Number) []byte { |
| 189 | b = wire.AppendTag(b, FieldItem, wire.StartGroupType) |
| 190 | b = wire.AppendTag(b, FieldTypeID, wire.VarintType) |
| 191 | b = wire.AppendVarint(b, uint64(num)) |
| 192 | return b |
| 193 | } |
| 194 | |
| 195 | // AppendFieldEnd appends the trailing end group marker for a MessageSet item field. |
| 196 | func AppendFieldEnd(b []byte) []byte { |
| 197 | return wire.AppendTag(b, FieldItem, wire.EndGroupType) |
| 198 | } |
Damien Neil | ce3384c | 2019-11-06 13:18:28 -0800 | [diff] [blame] | 199 | |
| 200 | // SizeUnknown returns the size of an unknown fields section in MessageSet format. |
| 201 | // |
| 202 | // See AppendUnknown. |
| 203 | func SizeUnknown(unknown []byte) (size int) { |
| 204 | for len(unknown) > 0 { |
| 205 | num, typ, n := wire.ConsumeTag(unknown) |
| 206 | if n < 0 || typ != wire.BytesType { |
| 207 | return 0 |
| 208 | } |
| 209 | unknown = unknown[n:] |
| 210 | _, n = wire.ConsumeBytes(unknown) |
| 211 | if n < 0 { |
| 212 | return 0 |
| 213 | } |
| 214 | unknown = unknown[n:] |
| 215 | size += SizeField(num) + wire.SizeTag(FieldMessage) + n |
| 216 | } |
| 217 | return size |
| 218 | } |
| 219 | |
| 220 | // AppendUnknown appends unknown fields to b in MessageSet format. |
| 221 | // |
| 222 | // For historic reasons, unresolved items in a MessageSet are stored in a |
| 223 | // message's unknown fields section in non-MessageSet format. That is, an |
| 224 | // unknown item with typeID T and value V appears in the unknown fields as |
| 225 | // a field with number T and value V. |
| 226 | // |
| 227 | // This function converts the unknown fields back into MessageSet form. |
| 228 | func AppendUnknown(b, unknown []byte) ([]byte, error) { |
| 229 | for len(unknown) > 0 { |
| 230 | num, typ, n := wire.ConsumeTag(unknown) |
| 231 | if n < 0 || typ != wire.BytesType { |
| 232 | return nil, errors.New("invalid data in message set unknown fields") |
| 233 | } |
| 234 | unknown = unknown[n:] |
| 235 | _, n = wire.ConsumeBytes(unknown) |
| 236 | if n < 0 { |
| 237 | return nil, errors.New("invalid data in message set unknown fields") |
| 238 | } |
| 239 | b = AppendFieldStart(b, num) |
| 240 | b = wire.AppendTag(b, FieldMessage, wire.BytesType) |
| 241 | b = append(b, unknown[:n]...) |
| 242 | b = AppendFieldEnd(b) |
| 243 | unknown = unknown[n:] |
| 244 | } |
| 245 | return b, nil |
| 246 | } |