David K. Bainbridge | bd6b288 | 2021-08-26 13:31:02 +0000 | [diff] [blame] | 1 | // Copyright 2018 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 text |
| 6 | |
| 7 | import ( |
| 8 | "bytes" |
| 9 | "fmt" |
| 10 | "math" |
| 11 | "strconv" |
| 12 | "strings" |
| 13 | |
| 14 | "google.golang.org/protobuf/internal/flags" |
| 15 | ) |
| 16 | |
| 17 | // Kind represents a token kind expressible in the textproto format. |
| 18 | type Kind uint8 |
| 19 | |
| 20 | // Kind values. |
| 21 | const ( |
| 22 | Invalid Kind = iota |
| 23 | EOF |
| 24 | Name // Name indicates the field name. |
| 25 | Scalar // Scalar are scalar values, e.g. "string", 47, ENUM_LITERAL, true. |
| 26 | MessageOpen |
| 27 | MessageClose |
| 28 | ListOpen |
| 29 | ListClose |
| 30 | |
| 31 | // comma and semi-colon are only for parsing in between values and should not be exposed. |
| 32 | comma |
| 33 | semicolon |
| 34 | |
| 35 | // bof indicates beginning of file, which is the default token |
| 36 | // kind at the beginning of parsing. |
| 37 | bof = Invalid |
| 38 | ) |
| 39 | |
| 40 | func (t Kind) String() string { |
| 41 | switch t { |
| 42 | case Invalid: |
| 43 | return "<invalid>" |
| 44 | case EOF: |
| 45 | return "eof" |
| 46 | case Scalar: |
| 47 | return "scalar" |
| 48 | case Name: |
| 49 | return "name" |
| 50 | case MessageOpen: |
| 51 | return "{" |
| 52 | case MessageClose: |
| 53 | return "}" |
| 54 | case ListOpen: |
| 55 | return "[" |
| 56 | case ListClose: |
| 57 | return "]" |
| 58 | case comma: |
| 59 | return "," |
| 60 | case semicolon: |
| 61 | return ";" |
| 62 | default: |
| 63 | return fmt.Sprintf("<invalid:%v>", uint8(t)) |
| 64 | } |
| 65 | } |
| 66 | |
| 67 | // NameKind represents different types of field names. |
| 68 | type NameKind uint8 |
| 69 | |
| 70 | // NameKind values. |
| 71 | const ( |
| 72 | IdentName NameKind = iota + 1 |
| 73 | TypeName |
| 74 | FieldNumber |
| 75 | ) |
| 76 | |
| 77 | func (t NameKind) String() string { |
| 78 | switch t { |
| 79 | case IdentName: |
| 80 | return "IdentName" |
| 81 | case TypeName: |
| 82 | return "TypeName" |
| 83 | case FieldNumber: |
| 84 | return "FieldNumber" |
| 85 | default: |
| 86 | return fmt.Sprintf("<invalid:%v>", uint8(t)) |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | // Bit mask in Token.attrs to indicate if a Name token is followed by the |
| 91 | // separator char ':'. The field name separator char is optional for message |
| 92 | // field or repeated message field, but required for all other types. Decoder |
| 93 | // simply indicates whether a Name token is followed by separator or not. It is |
| 94 | // up to the prototext package to validate. |
| 95 | const hasSeparator = 1 << 7 |
| 96 | |
| 97 | // Scalar value types. |
| 98 | const ( |
| 99 | numberValue = iota + 1 |
| 100 | stringValue |
| 101 | literalValue |
| 102 | ) |
| 103 | |
| 104 | // Bit mask in Token.numAttrs to indicate that the number is a negative. |
| 105 | const isNegative = 1 << 7 |
| 106 | |
| 107 | // Token provides a parsed token kind and value. Values are provided by the |
| 108 | // different accessor methods. |
| 109 | type Token struct { |
| 110 | // Kind of the Token object. |
| 111 | kind Kind |
| 112 | // attrs contains metadata for the following Kinds: |
| 113 | // Name: hasSeparator bit and one of NameKind. |
| 114 | // Scalar: one of numberValue, stringValue, literalValue. |
| 115 | attrs uint8 |
| 116 | // numAttrs contains metadata for numberValue: |
| 117 | // - highest bit is whether negative or positive. |
| 118 | // - lower bits indicate one of numDec, numHex, numOct, numFloat. |
| 119 | numAttrs uint8 |
| 120 | // pos provides the position of the token in the original input. |
| 121 | pos int |
| 122 | // raw bytes of the serialized token. |
| 123 | // This is a subslice into the original input. |
| 124 | raw []byte |
| 125 | // str contains parsed string for the following: |
| 126 | // - stringValue of Scalar kind |
| 127 | // - numberValue of Scalar kind |
| 128 | // - TypeName of Name kind |
| 129 | str string |
| 130 | } |
| 131 | |
| 132 | // Kind returns the token kind. |
| 133 | func (t Token) Kind() Kind { |
| 134 | return t.kind |
| 135 | } |
| 136 | |
| 137 | // RawString returns the read value in string. |
| 138 | func (t Token) RawString() string { |
| 139 | return string(t.raw) |
| 140 | } |
| 141 | |
| 142 | // Pos returns the token position from the input. |
| 143 | func (t Token) Pos() int { |
| 144 | return t.pos |
| 145 | } |
| 146 | |
| 147 | // NameKind returns IdentName, TypeName or FieldNumber. |
| 148 | // It panics if type is not Name. |
| 149 | func (t Token) NameKind() NameKind { |
| 150 | if t.kind == Name { |
| 151 | return NameKind(t.attrs &^ hasSeparator) |
| 152 | } |
| 153 | panic(fmt.Sprintf("Token is not a Name type: %s", t.kind)) |
| 154 | } |
| 155 | |
| 156 | // HasSeparator returns true if the field name is followed by the separator char |
| 157 | // ':', else false. It panics if type is not Name. |
| 158 | func (t Token) HasSeparator() bool { |
| 159 | if t.kind == Name { |
| 160 | return t.attrs&hasSeparator != 0 |
| 161 | } |
| 162 | panic(fmt.Sprintf("Token is not a Name type: %s", t.kind)) |
| 163 | } |
| 164 | |
| 165 | // IdentName returns the value for IdentName type. |
| 166 | func (t Token) IdentName() string { |
| 167 | if t.kind == Name && t.attrs&uint8(IdentName) != 0 { |
| 168 | return string(t.raw) |
| 169 | } |
| 170 | panic(fmt.Sprintf("Token is not an IdentName: %s:%s", t.kind, NameKind(t.attrs&^hasSeparator))) |
| 171 | } |
| 172 | |
| 173 | // TypeName returns the value for TypeName type. |
| 174 | func (t Token) TypeName() string { |
| 175 | if t.kind == Name && t.attrs&uint8(TypeName) != 0 { |
| 176 | return t.str |
| 177 | } |
| 178 | panic(fmt.Sprintf("Token is not a TypeName: %s:%s", t.kind, NameKind(t.attrs&^hasSeparator))) |
| 179 | } |
| 180 | |
| 181 | // FieldNumber returns the value for FieldNumber type. It returns a |
| 182 | // non-negative int32 value. Caller will still need to validate for the correct |
| 183 | // field number range. |
| 184 | func (t Token) FieldNumber() int32 { |
| 185 | if t.kind != Name || t.attrs&uint8(FieldNumber) == 0 { |
| 186 | panic(fmt.Sprintf("Token is not a FieldNumber: %s:%s", t.kind, NameKind(t.attrs&^hasSeparator))) |
| 187 | } |
| 188 | // Following should not return an error as it had already been called right |
| 189 | // before this Token was constructed. |
| 190 | num, _ := strconv.ParseInt(string(t.raw), 10, 32) |
| 191 | return int32(num) |
| 192 | } |
| 193 | |
| 194 | // String returns the string value for a Scalar type. |
| 195 | func (t Token) String() (string, bool) { |
| 196 | if t.kind != Scalar || t.attrs != stringValue { |
| 197 | return "", false |
| 198 | } |
| 199 | return t.str, true |
| 200 | } |
| 201 | |
| 202 | // Enum returns the literal value for a Scalar type for use as enum literals. |
| 203 | func (t Token) Enum() (string, bool) { |
| 204 | if t.kind != Scalar || t.attrs != literalValue || (len(t.raw) > 0 && t.raw[0] == '-') { |
| 205 | return "", false |
| 206 | } |
| 207 | return string(t.raw), true |
| 208 | } |
| 209 | |
| 210 | // Bool returns the bool value for a Scalar type. |
| 211 | func (t Token) Bool() (bool, bool) { |
| 212 | if t.kind != Scalar { |
| 213 | return false, false |
| 214 | } |
| 215 | switch t.attrs { |
| 216 | case literalValue: |
| 217 | if b, ok := boolLits[string(t.raw)]; ok { |
| 218 | return b, true |
| 219 | } |
| 220 | case numberValue: |
| 221 | // Unsigned integer representation of 0 or 1 is permitted: 00, 0x0, 01, |
| 222 | // 0x1, etc. |
| 223 | n, err := strconv.ParseUint(t.str, 0, 64) |
| 224 | if err == nil { |
| 225 | switch n { |
| 226 | case 0: |
| 227 | return false, true |
| 228 | case 1: |
| 229 | return true, true |
| 230 | } |
| 231 | } |
| 232 | } |
| 233 | return false, false |
| 234 | } |
| 235 | |
| 236 | // These exact boolean literals are the ones supported in C++. |
| 237 | var boolLits = map[string]bool{ |
| 238 | "t": true, |
| 239 | "true": true, |
| 240 | "True": true, |
| 241 | "f": false, |
| 242 | "false": false, |
| 243 | "False": false, |
| 244 | } |
| 245 | |
| 246 | // Uint64 returns the uint64 value for a Scalar type. |
| 247 | func (t Token) Uint64() (uint64, bool) { |
| 248 | if t.kind != Scalar || t.attrs != numberValue || |
| 249 | t.numAttrs&isNegative > 0 || t.numAttrs&numFloat > 0 { |
| 250 | return 0, false |
| 251 | } |
| 252 | n, err := strconv.ParseUint(t.str, 0, 64) |
| 253 | if err != nil { |
| 254 | return 0, false |
| 255 | } |
| 256 | return n, true |
| 257 | } |
| 258 | |
| 259 | // Uint32 returns the uint32 value for a Scalar type. |
| 260 | func (t Token) Uint32() (uint32, bool) { |
| 261 | if t.kind != Scalar || t.attrs != numberValue || |
| 262 | t.numAttrs&isNegative > 0 || t.numAttrs&numFloat > 0 { |
| 263 | return 0, false |
| 264 | } |
| 265 | n, err := strconv.ParseUint(t.str, 0, 32) |
| 266 | if err != nil { |
| 267 | return 0, false |
| 268 | } |
| 269 | return uint32(n), true |
| 270 | } |
| 271 | |
| 272 | // Int64 returns the int64 value for a Scalar type. |
| 273 | func (t Token) Int64() (int64, bool) { |
| 274 | if t.kind != Scalar || t.attrs != numberValue || t.numAttrs&numFloat > 0 { |
| 275 | return 0, false |
| 276 | } |
| 277 | if n, err := strconv.ParseInt(t.str, 0, 64); err == nil { |
| 278 | return n, true |
| 279 | } |
| 280 | // C++ accepts large positive hex numbers as negative values. |
| 281 | // This feature is here for proto1 backwards compatibility purposes. |
| 282 | if flags.ProtoLegacy && (t.numAttrs == numHex) { |
| 283 | if n, err := strconv.ParseUint(t.str, 0, 64); err == nil { |
| 284 | return int64(n), true |
| 285 | } |
| 286 | } |
| 287 | return 0, false |
| 288 | } |
| 289 | |
| 290 | // Int32 returns the int32 value for a Scalar type. |
| 291 | func (t Token) Int32() (int32, bool) { |
| 292 | if t.kind != Scalar || t.attrs != numberValue || t.numAttrs&numFloat > 0 { |
| 293 | return 0, false |
| 294 | } |
| 295 | if n, err := strconv.ParseInt(t.str, 0, 32); err == nil { |
| 296 | return int32(n), true |
| 297 | } |
| 298 | // C++ accepts large positive hex numbers as negative values. |
| 299 | // This feature is here for proto1 backwards compatibility purposes. |
| 300 | if flags.ProtoLegacy && (t.numAttrs == numHex) { |
| 301 | if n, err := strconv.ParseUint(t.str, 0, 32); err == nil { |
| 302 | return int32(n), true |
| 303 | } |
| 304 | } |
| 305 | return 0, false |
| 306 | } |
| 307 | |
| 308 | // Float64 returns the float64 value for a Scalar type. |
| 309 | func (t Token) Float64() (float64, bool) { |
| 310 | if t.kind != Scalar { |
| 311 | return 0, false |
| 312 | } |
| 313 | switch t.attrs { |
| 314 | case literalValue: |
| 315 | if f, ok := floatLits[strings.ToLower(string(t.raw))]; ok { |
| 316 | return f, true |
| 317 | } |
| 318 | case numberValue: |
| 319 | n, err := strconv.ParseFloat(t.str, 64) |
| 320 | if err == nil { |
| 321 | return n, true |
| 322 | } |
| 323 | nerr := err.(*strconv.NumError) |
| 324 | if nerr.Err == strconv.ErrRange { |
| 325 | return n, true |
| 326 | } |
| 327 | } |
| 328 | return 0, false |
| 329 | } |
| 330 | |
| 331 | // Float32 returns the float32 value for a Scalar type. |
| 332 | func (t Token) Float32() (float32, bool) { |
| 333 | if t.kind != Scalar { |
| 334 | return 0, false |
| 335 | } |
| 336 | switch t.attrs { |
| 337 | case literalValue: |
| 338 | if f, ok := floatLits[strings.ToLower(string(t.raw))]; ok { |
| 339 | return float32(f), true |
| 340 | } |
| 341 | case numberValue: |
| 342 | n, err := strconv.ParseFloat(t.str, 64) |
| 343 | if err == nil { |
| 344 | // Overflows are treated as (-)infinity. |
| 345 | return float32(n), true |
| 346 | } |
| 347 | nerr := err.(*strconv.NumError) |
| 348 | if nerr.Err == strconv.ErrRange { |
| 349 | return float32(n), true |
| 350 | } |
| 351 | } |
| 352 | return 0, false |
| 353 | } |
| 354 | |
| 355 | // These are the supported float literals which C++ permits case-insensitive |
| 356 | // variants of these. |
| 357 | var floatLits = map[string]float64{ |
| 358 | "nan": math.NaN(), |
| 359 | "inf": math.Inf(1), |
| 360 | "infinity": math.Inf(1), |
| 361 | "-inf": math.Inf(-1), |
| 362 | "-infinity": math.Inf(-1), |
| 363 | } |
| 364 | |
| 365 | // TokenEquals returns true if given Tokens are equal, else false. |
| 366 | func TokenEquals(x, y Token) bool { |
| 367 | return x.kind == y.kind && |
| 368 | x.attrs == y.attrs && |
| 369 | x.numAttrs == y.numAttrs && |
| 370 | x.pos == y.pos && |
| 371 | bytes.Equal(x.raw, y.raw) && |
| 372 | x.str == y.str |
| 373 | } |