temporal | 40ee551 | 2008-07-10 02:12:20 +0000 | [diff] [blame] | 1 | // Protocol Buffers - Google's data interchange format |
| 2 | // Copyright 2008 Google Inc. |
| 3 | // http://code.google.com/p/protobuf/ |
| 4 | // |
| 5 | // Licensed under the Apache License, Version 2.0 (the "License"); |
| 6 | // you may not use this file except in compliance with the License. |
| 7 | // You may obtain a copy of the License at |
| 8 | // |
| 9 | // http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | // |
| 11 | // Unless required by applicable law or agreed to in writing, software |
| 12 | // distributed under the License is distributed on an "AS IS" BASIS, |
| 13 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | // See the License for the specific language governing permissions and |
| 15 | // limitations under the License. |
| 16 | |
| 17 | // Author: jschorr@google.com (Joseph Schorr) |
| 18 | // Based on original Protocol Buffers design by |
| 19 | // Sanjay Ghemawat, Jeff Dean, and others. |
| 20 | |
| 21 | #include <float.h> |
| 22 | #include <math.h> |
| 23 | #include <stack> |
| 24 | #include <limits> |
| 25 | |
| 26 | #include <google/protobuf/text_format.h> |
| 27 | |
| 28 | #include <google/protobuf/descriptor.h> |
| 29 | #include <google/protobuf/io/coded_stream.h> |
| 30 | #include <google/protobuf/io/zero_copy_stream.h> |
| 31 | #include <google/protobuf/io/zero_copy_stream_impl.h> |
| 32 | #include <google/protobuf/unknown_field_set.h> |
| 33 | #include <google/protobuf/descriptor.pb.h> |
| 34 | #include <google/protobuf/io/tokenizer.h> |
| 35 | #include <google/protobuf/stubs/strutil.h> |
| 36 | |
| 37 | namespace google { |
| 38 | namespace protobuf { |
| 39 | |
| 40 | string Message::DebugString() const { |
| 41 | string debug_string; |
| 42 | io::StringOutputStream output_stream(&debug_string); |
| 43 | |
| 44 | TextFormat::Print(*this, &output_stream); |
| 45 | |
| 46 | return debug_string; |
| 47 | } |
| 48 | |
| 49 | string Message::ShortDebugString() const { |
| 50 | // TODO(kenton): Make TextFormat support this natively instead of using |
| 51 | // DebugString() and munging the result. |
| 52 | string result = DebugString(); |
| 53 | |
| 54 | // Replace each contiguous range of whitespace (including newlines) with a |
| 55 | // single space. |
| 56 | for (int i = 0; i < result.size(); i++) { |
| 57 | int pos = i; |
| 58 | while (isspace(result[pos])) ++pos; |
| 59 | if (pos > i) result.replace(i, pos - i, " "); |
| 60 | } |
| 61 | |
| 62 | return result; |
| 63 | } |
| 64 | |
| 65 | void Message::PrintDebugString() const { |
| 66 | printf("%s", DebugString().c_str()); |
| 67 | } |
| 68 | |
| 69 | // =========================================================================== |
| 70 | // Internal class for parsing an ASCII representation of a Protocol Message. |
| 71 | // This class makes use of the Protocol Message compiler's tokenizer found |
| 72 | // in //google/protobuf/io/tokenizer.h. Note that class's Parse |
| 73 | // method is *not* thread-safe and should only be used in a single thread at |
| 74 | // a time. |
| 75 | |
| 76 | // Makes code slightly more readable. The meaning of "DO(foo)" is |
| 77 | // "Execute foo and fail if it fails.", where failure is indicated by |
| 78 | // returning false. Borrowed from parser.cc (Thanks Kenton!). |
| 79 | #define DO(STATEMENT) if (STATEMENT) {} else return false |
| 80 | |
| 81 | class TextFormat::ParserImpl { |
| 82 | public: |
| 83 | ParserImpl(io::ZeroCopyInputStream* input_stream, |
| 84 | io::ErrorCollector* error_collector) |
| 85 | : error_collector_(error_collector), |
| 86 | tokenizer_error_collector_(this), |
| 87 | tokenizer_(input_stream, &tokenizer_error_collector_), |
| 88 | root_message_type_(NULL) { |
| 89 | // For backwards-compatibility with proto1, we need to allow the 'f' suffix |
| 90 | // for floats. |
| 91 | tokenizer_.set_allow_f_after_float(true); |
| 92 | |
| 93 | // '#' starts a comment. |
| 94 | tokenizer_.set_comment_style(io::Tokenizer::SH_COMMENT_STYLE); |
| 95 | |
| 96 | // Consume the starting token. |
| 97 | tokenizer_.Next(); |
| 98 | } |
| 99 | ~ParserImpl() { } |
| 100 | |
| 101 | // Parses the ASCII representation specified in input and saves the |
| 102 | // information into the output pointer (a Message). Returns |
| 103 | // false if an error occurs (an error will also be logged to |
| 104 | // GOOGLE_LOG(ERROR)). |
| 105 | bool Parse(Message* output) { |
| 106 | Message::Reflection* reflection = output->GetReflection(); |
| 107 | const Descriptor* descriptor = output->GetDescriptor(); |
| 108 | root_message_type_ = descriptor; |
| 109 | |
| 110 | // Consume fields until we cannot do so anymore. |
| 111 | while(true) { |
| 112 | if (LookingAtType(io::Tokenizer::TYPE_END)) { |
| 113 | return true; |
| 114 | } |
| 115 | |
| 116 | DO(ConsumeField(reflection, descriptor)); |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | void ReportError(int line, int col, const string& message) { |
| 121 | if (error_collector_ == NULL) { |
| 122 | if (line >= 0) { |
| 123 | GOOGLE_LOG(ERROR) << "Error parsing text-format " |
| 124 | << root_message_type_->full_name() |
| 125 | << ": " << (line + 1) << ":" |
| 126 | << (col + 1) << ": " << message; |
| 127 | } else { |
| 128 | GOOGLE_LOG(ERROR) << "Error parsing text-format " |
| 129 | << root_message_type_->full_name() |
| 130 | << ": " << message; |
| 131 | } |
| 132 | } else { |
| 133 | error_collector_->AddError(line, col, message); |
| 134 | } |
| 135 | } |
| 136 | |
| 137 | private: |
| 138 | GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(ParserImpl); |
| 139 | |
| 140 | // Reports an error with the given message with information indicating |
| 141 | // the position (as derived from the current token). |
| 142 | void ReportError(const string& message) { |
| 143 | ReportError(tokenizer_.current().line, tokenizer_.current().column, |
| 144 | message); |
| 145 | } |
| 146 | |
| 147 | // Consumes the specified message with the given starting delimeter. |
| 148 | // This method checks to see that the end delimeter at the conclusion of |
| 149 | // the consumption matches the starting delimeter passed in here. |
| 150 | bool ConsumeMessage(Message* message, const string delimeter) { |
| 151 | Message::Reflection* reflection = message->GetReflection(); |
| 152 | const Descriptor* descriptor = message->GetDescriptor(); |
| 153 | |
| 154 | while (!LookingAt(">") && !LookingAt("}")) { |
| 155 | DO(ConsumeField(reflection, descriptor)); |
| 156 | } |
| 157 | |
| 158 | // Confirm that we have a valid ending delimeter. |
| 159 | DO(Consume(delimeter)); |
| 160 | |
| 161 | return true; |
| 162 | } |
| 163 | |
| 164 | // Consumes the current field (as returned by the tokenizer) on the |
| 165 | // passed in message. |
| 166 | bool ConsumeField(Message::Reflection* reflection, |
| 167 | const Descriptor* descriptor) { |
| 168 | string field_name; |
| 169 | |
| 170 | const FieldDescriptor* field = NULL; |
| 171 | |
| 172 | if (TryConsume("[")) { |
| 173 | // Extension. |
| 174 | DO(ConsumeIdentifier(&field_name)); |
| 175 | while (TryConsume(".")) { |
| 176 | string part; |
| 177 | DO(ConsumeIdentifier(&part)); |
| 178 | field_name += "."; |
| 179 | field_name += part; |
| 180 | } |
| 181 | DO(Consume("]")); |
| 182 | |
| 183 | field = reflection->FindKnownExtensionByName(field_name); |
| 184 | |
| 185 | if (field == NULL) { |
| 186 | ReportError("Extension \"" + field_name + "\" is not defined or " |
| 187 | "is not an extension of \"" + |
| 188 | descriptor->full_name() + "\"."); |
| 189 | return false; |
| 190 | } |
| 191 | } else { |
| 192 | DO(ConsumeIdentifier(&field_name)); |
| 193 | |
| 194 | field = descriptor->FindFieldByName(field_name); |
| 195 | // Group names are expected to be capitalized as they appear in the |
| 196 | // .proto file, which actually matches their type names, not their field |
| 197 | // names. |
| 198 | if (field == NULL) { |
| 199 | string lower_field_name = field_name; |
| 200 | LowerString(&lower_field_name); |
| 201 | field = descriptor->FindFieldByName(lower_field_name); |
| 202 | // If the case-insensitive match worked but the field is NOT a group, |
| 203 | if (field != NULL && field->type() != FieldDescriptor::TYPE_GROUP) { |
| 204 | field = NULL; |
| 205 | } |
| 206 | } |
| 207 | // Again, special-case group names as described above. |
| 208 | if (field != NULL && field->type() == FieldDescriptor::TYPE_GROUP |
| 209 | && field->message_type()->name() != field_name) { |
| 210 | field = NULL; |
| 211 | } |
| 212 | |
| 213 | if (field == NULL) { |
| 214 | ReportError("Message type \"" + descriptor->full_name() + |
| 215 | "\" has no field named \"" + field_name + "\"."); |
| 216 | return false; |
| 217 | } |
| 218 | } |
| 219 | |
| 220 | // Perform special handling for embedded message types. |
| 221 | if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) { |
| 222 | string delimeter; |
| 223 | |
| 224 | // ':' is optional here. |
| 225 | TryConsume(":"); |
| 226 | |
| 227 | if (TryConsume("<")) { |
| 228 | delimeter = ">"; |
| 229 | } else { |
| 230 | DO(Consume("{")); |
| 231 | delimeter = "}"; |
| 232 | } |
| 233 | |
| 234 | if (field->is_repeated()) { |
| 235 | DO(ConsumeMessage(reflection->AddMessage(field), delimeter)); |
| 236 | } else { |
| 237 | DO(ConsumeMessage(reflection->MutableMessage(field), delimeter)); |
| 238 | } |
| 239 | } else { |
| 240 | DO(Consume(":")); |
| 241 | DO(ConsumeFieldValue(reflection, field)); |
| 242 | } |
| 243 | |
| 244 | return true; |
| 245 | } |
| 246 | |
| 247 | bool ConsumeFieldValue(Message::Reflection* reflection, |
| 248 | const FieldDescriptor* field) { |
| 249 | |
| 250 | // Define an easy to use macro for setting fields. This macro checks |
| 251 | // to see if the field is repeated (in which case we need to use the Add |
| 252 | // methods or not (in which case we need to use the Set methods). |
| 253 | #define SET_FIELD(CPPTYPE, VALUE) \ |
| 254 | if (field->is_repeated()) { \ |
| 255 | reflection->Add##CPPTYPE(field, VALUE); \ |
| 256 | } else { \ |
| 257 | reflection->Set##CPPTYPE(field, VALUE); \ |
| 258 | } \ |
| 259 | |
| 260 | switch(field->cpp_type()) { |
| 261 | case FieldDescriptor::CPPTYPE_INT32: { |
| 262 | int64 value; |
| 263 | DO(ConsumeSignedInteger(&value, kint32max)); |
| 264 | SET_FIELD(Int32, static_cast<int32>(value)); |
| 265 | break; |
| 266 | } |
| 267 | |
| 268 | case FieldDescriptor::CPPTYPE_UINT32: { |
| 269 | uint64 value; |
| 270 | DO(ConsumeUnsignedInteger(&value, kuint32max)); |
| 271 | SET_FIELD(UInt32, static_cast<uint32>(value)); |
| 272 | break; |
| 273 | } |
| 274 | |
| 275 | case FieldDescriptor::CPPTYPE_INT64: { |
| 276 | int64 value; |
| 277 | DO(ConsumeSignedInteger(&value, kint64max)); |
| 278 | SET_FIELD(Int64, value); |
| 279 | break; |
| 280 | } |
| 281 | |
| 282 | case FieldDescriptor::CPPTYPE_UINT64: { |
| 283 | uint64 value; |
| 284 | DO(ConsumeUnsignedInteger(&value, kuint64max)); |
| 285 | SET_FIELD(UInt64, value); |
| 286 | break; |
| 287 | } |
| 288 | |
| 289 | case FieldDescriptor::CPPTYPE_FLOAT: { |
| 290 | double value; |
| 291 | DO(ConsumeDouble(&value)); |
| 292 | SET_FIELD(Float, static_cast<float>(value)); |
| 293 | break; |
| 294 | } |
| 295 | |
| 296 | case FieldDescriptor::CPPTYPE_DOUBLE: { |
| 297 | double value; |
| 298 | DO(ConsumeDouble(&value)); |
| 299 | SET_FIELD(Double, value); |
| 300 | break; |
| 301 | } |
| 302 | |
| 303 | case FieldDescriptor::CPPTYPE_STRING: { |
| 304 | string value; |
| 305 | DO(ConsumeString(&value)); |
| 306 | SET_FIELD(String, value); |
| 307 | break; |
| 308 | } |
| 309 | |
| 310 | case FieldDescriptor::CPPTYPE_BOOL: { |
| 311 | string value; |
| 312 | DO(ConsumeIdentifier(&value)); |
| 313 | |
| 314 | if (value == "true") { |
| 315 | SET_FIELD(Bool, true); |
| 316 | } else if (value == "false") { |
| 317 | SET_FIELD(Bool, false); |
| 318 | } else { |
| 319 | ReportError("Invalid value for boolean field \"" + field->name() |
| 320 | + "\". Value: \"" + value + "\"."); |
| 321 | return false; |
| 322 | } |
| 323 | break; |
| 324 | } |
| 325 | |
| 326 | case FieldDescriptor::CPPTYPE_ENUM: { |
| 327 | string value; |
| 328 | DO(ConsumeIdentifier(&value)); |
| 329 | |
| 330 | // Find the enumeration value. |
| 331 | const EnumDescriptor* enum_type = field->enum_type(); |
| 332 | const EnumValueDescriptor* enum_value |
| 333 | = enum_type->FindValueByName(value); |
| 334 | |
| 335 | if (enum_value == NULL) { |
| 336 | ReportError("Unknown enumeration value of \"" + value + "\" for " |
| 337 | "field \"" + field->name() + "\"."); |
| 338 | return false; |
| 339 | } |
| 340 | |
| 341 | SET_FIELD(Enum, enum_value); |
| 342 | break; |
| 343 | } |
| 344 | |
| 345 | case FieldDescriptor::CPPTYPE_MESSAGE: { |
| 346 | // We should never get here. Put here instead of a default |
| 347 | // so that if new types are added, we get a nice compiler warning. |
| 348 | GOOGLE_LOG(FATAL) << "Reached an unintended state: CPPTYPE_MESSAGE"; |
| 349 | break; |
| 350 | } |
| 351 | } |
| 352 | #undef SET_FIELD |
| 353 | return true; |
| 354 | } |
| 355 | |
| 356 | // Returns true if the current token's text is equal to that specified. |
| 357 | bool LookingAt(const string& text) { |
| 358 | return tokenizer_.current().text == text; |
| 359 | } |
| 360 | |
| 361 | // Returns true if the current token's type is equal to that specified. |
| 362 | bool LookingAtType(io::Tokenizer::TokenType token_type) { |
| 363 | return tokenizer_.current().type == token_type; |
| 364 | } |
| 365 | |
| 366 | // Consumes an identifier and saves its value in the identifier parameter. |
| 367 | // Returns false if the token is not of type IDENTFIER. |
| 368 | bool ConsumeIdentifier(string* identifier) { |
| 369 | if (!LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) { |
| 370 | ReportError("Expected identifier."); |
| 371 | return false; |
| 372 | } |
| 373 | |
| 374 | *identifier = tokenizer_.current().text; |
| 375 | |
| 376 | tokenizer_.Next(); |
| 377 | return true; |
| 378 | } |
| 379 | |
| 380 | // Consumes a string and saves its value in the text parameter. |
| 381 | // Returns false if the token is not of type STRING. |
| 382 | bool ConsumeString(string* text) { |
| 383 | if (!LookingAtType(io::Tokenizer::TYPE_STRING)) { |
| 384 | ReportError("Expected string."); |
| 385 | return false; |
| 386 | } |
| 387 | |
| 388 | io::Tokenizer::ParseString(tokenizer_.current().text, text); |
| 389 | |
| 390 | tokenizer_.Next(); |
| 391 | return true; |
| 392 | } |
| 393 | |
| 394 | // Consumes a uint64 and saves its value in the value parameter. |
| 395 | // Returns false if the token is not of type INTEGER. |
| 396 | bool ConsumeUnsignedInteger(uint64* value, uint64 max_value) { |
| 397 | if (!LookingAtType(io::Tokenizer::TYPE_INTEGER)) { |
| 398 | ReportError("Expected integer."); |
| 399 | return false; |
| 400 | } |
| 401 | |
| 402 | if (!io::Tokenizer::ParseInteger(tokenizer_.current().text, |
| 403 | max_value, value)) { |
| 404 | ReportError("Integer out of range."); |
| 405 | return false; |
| 406 | } |
| 407 | |
| 408 | tokenizer_.Next(); |
| 409 | return true; |
| 410 | } |
| 411 | |
| 412 | // Consumes an int64 and saves its value in the value parameter. |
| 413 | // Note that since the tokenizer does not support negative numbers, |
| 414 | // we actually may consume an additional token (for the minus sign) in this |
| 415 | // method. Returns false if the token is not an integer |
| 416 | // (signed or otherwise). |
| 417 | bool ConsumeSignedInteger(int64* value, uint64 max_value) { |
| 418 | bool negative = false; |
| 419 | |
| 420 | if (TryConsume("-")) { |
| 421 | negative = true; |
| 422 | // Two's complement always allows one more negative integer than |
| 423 | // positive. |
| 424 | ++max_value; |
| 425 | } |
| 426 | |
| 427 | uint64 unsigned_value; |
| 428 | |
| 429 | DO(ConsumeUnsignedInteger(&unsigned_value, max_value)); |
| 430 | |
| 431 | *value = static_cast<int64>(unsigned_value); |
| 432 | |
| 433 | if (negative) { |
| 434 | *value = -*value; |
| 435 | } |
| 436 | |
| 437 | return true; |
| 438 | } |
| 439 | |
| 440 | // Consumes a double and saves its value in the value parameter. |
| 441 | // Note that since the tokenizer does not support negative numbers, |
| 442 | // we actually may consume an additional token (for the minus sign) in this |
| 443 | // method. Returns false if the token is not a double |
| 444 | // (signed or otherwise). |
| 445 | bool ConsumeDouble(double* value) { |
| 446 | bool negative = false; |
| 447 | |
| 448 | if (TryConsume("-")) { |
| 449 | negative = true; |
| 450 | } |
| 451 | |
| 452 | // A double can actually be an integer, according to the tokenizer. |
| 453 | // Therefore, we must check both cases here. |
| 454 | if (LookingAtType(io::Tokenizer::TYPE_INTEGER)) { |
| 455 | // We have found an integer value for the double. |
| 456 | uint64 integer_value; |
| 457 | DO(ConsumeUnsignedInteger(&integer_value, kuint64max)); |
| 458 | |
| 459 | *value = static_cast<double>(integer_value); |
| 460 | } else if (LookingAtType(io::Tokenizer::TYPE_FLOAT)) { |
| 461 | // We have found a float value for the double. |
| 462 | *value = io::Tokenizer::ParseFloat(tokenizer_.current().text); |
| 463 | |
| 464 | // Mark the current token as consumed. |
| 465 | tokenizer_.Next(); |
| 466 | } else if (LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) { |
| 467 | string text = tokenizer_.current().text; |
| 468 | LowerString(&text); |
| 469 | if (text == "inf" || text == "infinity") { |
| 470 | *value = std::numeric_limits<double>::infinity(); |
| 471 | tokenizer_.Next(); |
| 472 | } else if (text == "nan") { |
| 473 | *value = std::numeric_limits<double>::quiet_NaN(); |
| 474 | tokenizer_.Next(); |
| 475 | } else { |
| 476 | ReportError("Expected double."); |
| 477 | return false; |
| 478 | } |
| 479 | } else { |
| 480 | ReportError("Expected double."); |
| 481 | return false; |
| 482 | } |
| 483 | |
| 484 | if (negative) { |
| 485 | *value = -*value; |
| 486 | } |
| 487 | |
| 488 | return true; |
| 489 | } |
| 490 | |
| 491 | // Consumes a token and confirms that it matches that specified in the |
| 492 | // value parameter. Returns false if the token found does not match that |
| 493 | // which was specified. |
| 494 | bool Consume(const string& value) { |
| 495 | const string& current_value = tokenizer_.current().text; |
| 496 | |
| 497 | if (current_value != value) { |
| 498 | ReportError("Expected \"" + value + "\", found \"" + current_value |
| 499 | + "\"."); |
| 500 | return false; |
| 501 | } |
| 502 | |
| 503 | tokenizer_.Next(); |
| 504 | |
| 505 | return true; |
| 506 | } |
| 507 | |
| 508 | // Attempts to consume the supplied value. Returns false if a the |
| 509 | // token found does not match the value specified. |
| 510 | bool TryConsume(const string& value) { |
| 511 | if (tokenizer_.current().text == value) { |
| 512 | tokenizer_.Next(); |
| 513 | return true; |
| 514 | } else { |
| 515 | return false; |
| 516 | } |
| 517 | } |
| 518 | |
| 519 | // An internal instance of the Tokenizer's error collector, used to |
| 520 | // collect any base-level parse errors and feed them to the ParserImpl. |
| 521 | class ParserErrorCollector : public io::ErrorCollector { |
| 522 | public: |
| 523 | explicit ParserErrorCollector(TextFormat::ParserImpl* parser) : |
| 524 | parser_(parser) { } |
| 525 | |
| 526 | virtual ~ParserErrorCollector() { }; |
| 527 | |
| 528 | virtual void AddError(int line, int column, const string& message) { |
| 529 | parser_->ReportError(line, column, message); |
| 530 | } |
| 531 | |
| 532 | private: |
| 533 | GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(ParserErrorCollector); |
| 534 | TextFormat::ParserImpl* parser_; |
| 535 | }; |
| 536 | |
| 537 | io::ErrorCollector* error_collector_; |
| 538 | ParserErrorCollector tokenizer_error_collector_; |
| 539 | io::Tokenizer tokenizer_; |
| 540 | const Descriptor* root_message_type_; |
| 541 | }; |
| 542 | |
| 543 | #undef DO |
| 544 | |
| 545 | // =========================================================================== |
| 546 | // Internal class for writing text to the io::ZeroCopyOutputStream. Adapted |
| 547 | // from the Printer found in //google/protobuf/io/printer.h |
| 548 | class TextFormat::TextGenerator { |
| 549 | public: |
| 550 | explicit TextGenerator(io::ZeroCopyOutputStream* output) |
| 551 | : output_(output), |
| 552 | buffer_(NULL), |
| 553 | buffer_size_(0), |
| 554 | at_start_of_line_(true), |
| 555 | failed_(false), |
| 556 | indent_("") { |
| 557 | } |
| 558 | |
| 559 | ~TextGenerator() { |
| 560 | // Only BackUp() if we're sure we've successfully called Next() at least |
| 561 | // once. |
| 562 | if (buffer_size_ > 0) { |
| 563 | output_->BackUp(buffer_size_); |
| 564 | } |
| 565 | } |
| 566 | |
| 567 | // Indent text by two spaces. After calling Indent(), two spaces will be |
| 568 | // inserted at the beginning of each line of text. Indent() may be called |
| 569 | // multiple times to produce deeper indents. |
| 570 | void Indent() { |
| 571 | indent_ += " "; |
| 572 | } |
| 573 | |
| 574 | // Reduces the current indent level by two spaces, or crashes if the indent |
| 575 | // level is zero. |
| 576 | void Outdent() { |
| 577 | if (indent_.empty()) { |
| 578 | GOOGLE_LOG(DFATAL) << " Outdent() without matching Indent()."; |
| 579 | return; |
| 580 | } |
| 581 | |
| 582 | indent_.resize(indent_.size() - 2); |
| 583 | } |
| 584 | |
| 585 | // Print text to the output stream. |
| 586 | void Print(const string& str) { |
| 587 | Print(str.c_str()); |
| 588 | } |
| 589 | |
| 590 | // Print text to the output stream. |
| 591 | void Print(const char* text) { |
| 592 | int size = strlen(text); |
| 593 | int pos = 0; // The number of bytes we've written so far. |
| 594 | |
| 595 | for (int i = 0; i < size; i++) { |
| 596 | if (text[i] == '\n') { |
| 597 | // Saw newline. If there is more text, we may need to insert an indent |
| 598 | // here. So, write what we have so far, including the '\n'. |
| 599 | Write(text + pos, i - pos + 1); |
| 600 | pos = i + 1; |
| 601 | |
| 602 | // Setting this true will cause the next Write() to insert an indent |
| 603 | // first. |
| 604 | at_start_of_line_ = true; |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | // Write the rest. |
| 609 | Write(text + pos, size - pos); |
| 610 | } |
| 611 | |
| 612 | // True if any write to the underlying stream failed. (We don't just |
| 613 | // crash in this case because this is an I/O failure, not a programming |
| 614 | // error.) |
| 615 | bool failed() const { return failed_; } |
| 616 | |
| 617 | private: |
| 618 | GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(TextGenerator); |
| 619 | |
| 620 | void Write(const char* data, int size) { |
| 621 | if (failed_) return; |
| 622 | if (size == 0) return; |
| 623 | |
| 624 | if (at_start_of_line_) { |
| 625 | // Insert an indent. |
| 626 | at_start_of_line_ = false; |
| 627 | Write(indent_.data(), indent_.size()); |
| 628 | if (failed_) return; |
| 629 | } |
| 630 | |
| 631 | while (size > buffer_size_) { |
| 632 | // Data exceeds space in the buffer. Copy what we can and request a |
| 633 | // new buffer. |
| 634 | memcpy(buffer_, data, buffer_size_); |
| 635 | data += buffer_size_; |
| 636 | size -= buffer_size_; |
| 637 | void* void_buffer; |
| 638 | failed_ = !output_->Next(&void_buffer, &buffer_size_); |
| 639 | if (failed_) return; |
| 640 | buffer_ = reinterpret_cast<char*>(void_buffer); |
| 641 | } |
| 642 | |
| 643 | // Buffer is big enough to receive the data; copy it. |
| 644 | memcpy(buffer_, data, size); |
| 645 | buffer_ += size; |
| 646 | buffer_size_ -= size; |
| 647 | } |
| 648 | |
| 649 | io::ZeroCopyOutputStream* const output_; |
| 650 | char* buffer_; |
| 651 | int buffer_size_; |
| 652 | bool at_start_of_line_; |
| 653 | bool failed_; |
| 654 | |
| 655 | string indent_; |
| 656 | }; |
| 657 | |
| 658 | // =========================================================================== |
| 659 | |
| 660 | TextFormat::Parser::Parser() |
| 661 | : error_collector_(NULL), |
| 662 | allow_partial_(false) {} |
| 663 | |
| 664 | TextFormat::Parser::~Parser() {} |
| 665 | |
| 666 | bool TextFormat::Parser::Parse(io::ZeroCopyInputStream* input, |
| 667 | Message* output) { |
| 668 | output->Clear(); |
| 669 | return Merge(input, output); |
| 670 | } |
| 671 | |
| 672 | bool TextFormat::Parser::ParseFromString(const string& input, |
| 673 | Message* output) { |
| 674 | io::ArrayInputStream input_stream(input.data(), input.size()); |
| 675 | return Parse(&input_stream, output); |
| 676 | } |
| 677 | |
| 678 | bool TextFormat::Parser::Merge(io::ZeroCopyInputStream* input, |
| 679 | Message* output) { |
| 680 | ParserImpl parser(input, error_collector_); |
| 681 | if (!parser.Parse(output)) return false; |
| 682 | if (!allow_partial_ && !output->IsInitialized()) { |
| 683 | vector<string> missing_fields; |
| 684 | output->FindInitializationErrors(&missing_fields); |
| 685 | parser.ReportError(-1, 0, "Message missing required fields: " + |
| 686 | JoinStrings(missing_fields, ", ")); |
| 687 | return false; |
| 688 | } |
| 689 | return true; |
| 690 | } |
| 691 | |
| 692 | bool TextFormat::Parser::MergeFromString(const string& input, |
| 693 | Message* output) { |
| 694 | io::ArrayInputStream input_stream(input.data(), input.size()); |
| 695 | return Merge(&input_stream, output); |
| 696 | } |
| 697 | |
| 698 | |
| 699 | /* static */ bool TextFormat::Parse(io::ZeroCopyInputStream* input, |
| 700 | Message* output) { |
| 701 | return Parser().Parse(input, output); |
| 702 | } |
| 703 | |
| 704 | /* static */ bool TextFormat::Merge(io::ZeroCopyInputStream* input, |
| 705 | Message* output) { |
| 706 | return Parser().Merge(input, output); |
| 707 | } |
| 708 | |
| 709 | /* static */ bool TextFormat::ParseFromString(const string& input, |
| 710 | Message* output) { |
| 711 | return Parser().ParseFromString(input, output); |
| 712 | } |
| 713 | |
| 714 | /* static */ bool TextFormat::MergeFromString(const string& input, |
| 715 | Message* output) { |
| 716 | return Parser().MergeFromString(input, output); |
| 717 | } |
| 718 | |
| 719 | /* static */ bool TextFormat::PrintToString(const Message& message, |
| 720 | string* output) { |
| 721 | GOOGLE_DCHECK(output) << "output specified is NULL"; |
| 722 | |
| 723 | output->clear(); |
| 724 | io::StringOutputStream output_stream(output); |
| 725 | |
| 726 | bool result = Print(message, &output_stream); |
| 727 | |
| 728 | return result; |
| 729 | } |
| 730 | |
temporal | a0f27fc | 2008-08-06 01:12:21 +0000 | [diff] [blame^] | 731 | /* static */ bool TextFormat::PrintUnknownFieldsToString( |
| 732 | const UnknownFieldSet& unknown_fields, |
| 733 | string* output) { |
| 734 | GOOGLE_DCHECK(output) << "output specified is NULL"; |
| 735 | |
| 736 | output->clear(); |
| 737 | io::StringOutputStream output_stream(output); |
| 738 | return PrintUnknownFields(unknown_fields, &output_stream); |
| 739 | } |
| 740 | |
temporal | 40ee551 | 2008-07-10 02:12:20 +0000 | [diff] [blame] | 741 | /* static */ bool TextFormat::Print(const Message& message, |
| 742 | io::ZeroCopyOutputStream* output) { |
| 743 | TextGenerator generator(output); |
| 744 | |
| 745 | Print(message.GetDescriptor(), message.GetReflection(), generator); |
| 746 | |
| 747 | // Output false if the generator failed internally. |
| 748 | return !generator.failed(); |
| 749 | } |
| 750 | |
temporal | a0f27fc | 2008-08-06 01:12:21 +0000 | [diff] [blame^] | 751 | /* static */ bool TextFormat::PrintUnknownFields( |
| 752 | const UnknownFieldSet& unknown_fields, |
| 753 | io::ZeroCopyOutputStream* output) { |
| 754 | TextGenerator generator(output); |
| 755 | |
| 756 | PrintUnknownFields(unknown_fields, generator); |
| 757 | |
| 758 | // Output false if the generator failed internally. |
| 759 | return !generator.failed(); |
| 760 | } |
| 761 | |
temporal | 40ee551 | 2008-07-10 02:12:20 +0000 | [diff] [blame] | 762 | /* static */ void TextFormat::Print(const Descriptor* descriptor, |
| 763 | const Message::Reflection* message, |
| 764 | TextGenerator& generator) { |
| 765 | vector<const FieldDescriptor*> fields; |
| 766 | message->ListFields(&fields); |
| 767 | for (int i = 0; i < fields.size(); i++) { |
| 768 | PrintField(fields[i], message, generator); |
| 769 | } |
| 770 | PrintUnknownFields(message->GetUnknownFields(), generator); |
| 771 | } |
| 772 | |
| 773 | /* static */ void TextFormat::PrintFieldValueToString( |
| 774 | const Message& message, |
| 775 | const FieldDescriptor* field, |
| 776 | int index, |
| 777 | string* output) { |
| 778 | |
| 779 | GOOGLE_DCHECK(output) << "output specified is NULL"; |
| 780 | |
| 781 | output->clear(); |
| 782 | io::StringOutputStream output_stream(output); |
| 783 | TextGenerator generator(&output_stream); |
| 784 | |
| 785 | PrintFieldValue(message.GetReflection(), field, index, generator); |
| 786 | } |
| 787 | |
| 788 | /* static */ void TextFormat::PrintField(const FieldDescriptor* field, |
| 789 | const Message::Reflection* message, |
| 790 | TextGenerator& generator) { |
| 791 | int count = 0; |
| 792 | |
| 793 | if (field->is_repeated()) { |
| 794 | count = message->FieldSize(field); |
| 795 | } else if (message->HasField(field)) { |
| 796 | count = 1; |
| 797 | } |
| 798 | |
| 799 | for (int j = 0; j < count; ++j) { |
| 800 | if (field->is_extension()) { |
| 801 | generator.Print("["); |
| 802 | // We special-case MessageSet elements for compatibility with proto1. |
| 803 | if (field->containing_type()->options().message_set_wire_format() |
| 804 | && field->type() == FieldDescriptor::TYPE_MESSAGE |
| 805 | && field->is_optional() |
| 806 | && field->extension_scope() == field->message_type()) { |
| 807 | generator.Print(field->message_type()->full_name()); |
| 808 | } else { |
| 809 | generator.Print(field->full_name()); |
| 810 | } |
| 811 | generator.Print("]"); |
| 812 | } else { |
| 813 | if (field->type() == FieldDescriptor::TYPE_GROUP) { |
| 814 | // Groups must be serialized with their original capitalization. |
| 815 | generator.Print(field->message_type()->name()); |
| 816 | } else { |
| 817 | generator.Print(field->name()); |
| 818 | } |
| 819 | } |
| 820 | |
| 821 | if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) { |
| 822 | generator.Print(" {\n"); |
| 823 | generator.Indent(); |
| 824 | } else { |
| 825 | generator.Print(": "); |
| 826 | } |
| 827 | |
| 828 | // Write the field value. |
| 829 | int field_index = j; |
| 830 | if (!field->is_repeated()) { |
| 831 | field_index = -1; |
| 832 | } |
| 833 | |
| 834 | PrintFieldValue(message, field, field_index, generator); |
| 835 | |
| 836 | if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) { |
| 837 | generator.Outdent(); |
| 838 | generator.Print("}"); |
| 839 | } |
| 840 | |
| 841 | generator.Print("\n"); |
| 842 | } |
| 843 | } |
| 844 | |
| 845 | /* static */ void TextFormat::PrintFieldValue( |
| 846 | const Message::Reflection* reflection, |
| 847 | const FieldDescriptor* field, |
| 848 | int index, |
| 849 | TextGenerator& generator) { |
| 850 | GOOGLE_DCHECK(field->is_repeated() || (index == -1)) |
| 851 | << "Index must be -1 for non-repeated fields"; |
| 852 | |
| 853 | switch (field->cpp_type()) { |
| 854 | #define OUTPUT_FIELD(CPPTYPE, METHOD, TO_STRING) \ |
| 855 | case FieldDescriptor::CPPTYPE_##CPPTYPE: \ |
| 856 | generator.Print(TO_STRING(field->is_repeated() ? \ |
| 857 | reflection->GetRepeated##METHOD(field, index) : \ |
| 858 | reflection->Get##METHOD(field))); \ |
| 859 | break; \ |
| 860 | |
| 861 | OUTPUT_FIELD( INT32, Int32, SimpleItoa); |
| 862 | OUTPUT_FIELD( INT64, Int64, SimpleItoa); |
| 863 | OUTPUT_FIELD(UINT32, UInt32, SimpleItoa); |
| 864 | OUTPUT_FIELD(UINT64, UInt64, SimpleItoa); |
| 865 | OUTPUT_FIELD( FLOAT, Float, SimpleFtoa); |
| 866 | OUTPUT_FIELD(DOUBLE, Double, SimpleDtoa); |
| 867 | #undef OUTPUT_FIELD |
| 868 | |
| 869 | case FieldDescriptor::CPPTYPE_STRING: { |
| 870 | string scratch; |
| 871 | const string& value = field->is_repeated() ? |
| 872 | reflection->GetRepeatedStringReference(field, index, &scratch) : |
| 873 | reflection->GetStringReference(field, &scratch); |
| 874 | |
| 875 | generator.Print("\""); |
| 876 | generator.Print(CEscape(value)); |
| 877 | generator.Print("\""); |
| 878 | |
| 879 | break; |
| 880 | } |
| 881 | |
| 882 | case FieldDescriptor::CPPTYPE_BOOL: |
| 883 | if (field->is_repeated()) { |
| 884 | generator.Print(reflection->GetRepeatedBool(field, index) |
| 885 | ? "true" : "false"); |
| 886 | } else { |
| 887 | generator.Print(reflection->GetBool(field) ? "true" : "false"); |
| 888 | } |
| 889 | break; |
| 890 | |
| 891 | case FieldDescriptor::CPPTYPE_ENUM: |
| 892 | generator.Print(field->is_repeated() ? |
| 893 | reflection->GetRepeatedEnum(field, index)->name() |
| 894 | : reflection->GetEnum(field)->name()); |
| 895 | break; |
| 896 | |
| 897 | case FieldDescriptor::CPPTYPE_MESSAGE: |
| 898 | Print(field->message_type(), |
| 899 | field->is_repeated() ? |
| 900 | reflection->GetRepeatedMessage(field, index).GetReflection() |
| 901 | : reflection->GetMessage(field).GetReflection(), generator); |
| 902 | break; |
| 903 | } |
| 904 | } |
| 905 | |
| 906 | // Prints an integer as hex with a fixed number of digits dependent on the |
| 907 | // integer type. |
| 908 | template<typename IntType> |
| 909 | static string PaddedHex(IntType value) { |
| 910 | string result; |
| 911 | result.reserve(sizeof(value) * 2); |
| 912 | for (int i = sizeof(value) * 2 - 1; i >= 0; i--) { |
| 913 | result.push_back(int_to_hex_digit(value >> (i*4) & 0x0F)); |
| 914 | } |
| 915 | return result; |
| 916 | } |
| 917 | |
| 918 | /* static */ void TextFormat::PrintUnknownFields( |
| 919 | const UnknownFieldSet& unknown_fields, TextGenerator& generator) { |
| 920 | for (int i = 0; i < unknown_fields.field_count(); i++) { |
| 921 | const UnknownField& field = unknown_fields.field(i); |
| 922 | string field_number = SimpleItoa(field.number()); |
| 923 | |
| 924 | for (int j = 0; j < field.varint_size(); j++) { |
| 925 | generator.Print(field_number); |
| 926 | generator.Print(": "); |
| 927 | generator.Print(SimpleItoa(field.varint(j))); |
| 928 | generator.Print("\n"); |
| 929 | } |
| 930 | for (int j = 0; j < field.fixed32_size(); j++) { |
| 931 | generator.Print(field_number); |
| 932 | generator.Print(": 0x"); |
| 933 | char buffer[kFastToBufferSize]; |
| 934 | generator.Print(FastHex32ToBuffer(field.fixed32(j), buffer)); |
| 935 | generator.Print("\n"); |
| 936 | } |
| 937 | for (int j = 0; j < field.fixed64_size(); j++) { |
| 938 | generator.Print(field_number); |
| 939 | generator.Print(": 0x"); |
| 940 | char buffer[kFastToBufferSize]; |
| 941 | generator.Print(FastHex64ToBuffer(field.fixed64(j), buffer)); |
| 942 | generator.Print("\n"); |
| 943 | } |
| 944 | for (int j = 0; j < field.length_delimited_size(); j++) { |
| 945 | generator.Print(field_number); |
temporal | a0f27fc | 2008-08-06 01:12:21 +0000 | [diff] [blame^] | 946 | const string& value = field.length_delimited(j); |
| 947 | UnknownFieldSet embedded_unknown_fields; |
| 948 | if (!value.empty() && embedded_unknown_fields.ParseFromString(value)) { |
| 949 | // This field is parseable as a Message. |
| 950 | // So it is probably an embedded message. |
| 951 | generator.Print(" {\n"); |
| 952 | generator.Indent(); |
| 953 | PrintUnknownFields(embedded_unknown_fields, generator); |
| 954 | generator.Outdent(); |
| 955 | generator.Print("}\n"); |
| 956 | } else { |
| 957 | // This field is not parseable as a Message. |
| 958 | // So it is probably just a plain string. |
| 959 | generator.Print(": \""); |
| 960 | generator.Print(CEscape(value)); |
| 961 | generator.Print("\"\n"); |
| 962 | } |
temporal | 40ee551 | 2008-07-10 02:12:20 +0000 | [diff] [blame] | 963 | } |
| 964 | for (int j = 0; j < field.group_size(); j++) { |
| 965 | generator.Print(field_number); |
| 966 | generator.Print(" {\n"); |
| 967 | generator.Indent(); |
| 968 | PrintUnknownFields(field.group(j), generator); |
| 969 | generator.Outdent(); |
| 970 | generator.Print("}\n"); |
| 971 | } |
| 972 | } |
| 973 | } |
| 974 | |
| 975 | } // namespace protobuf |
| 976 | } // namespace google |