1 | // Protocol Buffers - Google's data interchange format |
2 | // Copyright 2008 Google Inc. All rights reserved. |
3 | // https://developers.google.com/protocol-buffers/ |
4 | // |
5 | // Redistribution and use in source and binary forms, with or without |
6 | // modification, are permitted provided that the following conditions are |
7 | // met: |
8 | // |
9 | // * Redistributions of source code must retain the above copyright |
10 | // notice, this list of conditions and the following disclaimer. |
11 | // * Redistributions in binary form must reproduce the above |
12 | // copyright notice, this list of conditions and the following disclaimer |
13 | // in the documentation and/or other materials provided with the |
14 | // distribution. |
15 | // * Neither the name of Google Inc. nor the names of its |
16 | // contributors may be used to endorse or promote products derived from |
17 | // this software without specific prior written permission. |
18 | // |
19 | // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
20 | // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
21 | // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
22 | // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
23 | // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
24 | // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
25 | // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
26 | // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
27 | // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
28 | // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
29 | // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
30 | |
31 | // Author: kenton@google.com (Kenton Varda) |
32 | // Based on original Protocol Buffers design by |
33 | // Sanjay Ghemawat, Jeff Dean, and others. |
34 | // |
35 | // Recursive descent FTW. |
36 | |
37 | #include <google/protobuf/compiler/parser.h> |
38 | |
39 | #include <float.h> |
40 | |
41 | #include <cstdint> |
42 | #include <limits> |
43 | #include <unordered_map> |
44 | #include <unordered_set> |
45 | |
46 | #include <google/protobuf/stubs/casts.h> |
47 | #include <google/protobuf/stubs/logging.h> |
48 | #include <google/protobuf/stubs/common.h> |
49 | #include <google/protobuf/stubs/strutil.h> |
50 | #include <google/protobuf/descriptor.h> |
51 | #include <google/protobuf/descriptor.pb.h> |
52 | #include <google/protobuf/io/tokenizer.h> |
53 | #include <google/protobuf/wire_format.h> |
54 | #include <google/protobuf/stubs/map_util.h> |
55 | #include <google/protobuf/stubs/hash.h> |
56 | |
57 | namespace google { |
58 | namespace protobuf { |
59 | namespace compiler { |
60 | |
61 | using internal::WireFormat; |
62 | |
63 | namespace { |
64 | |
65 | typedef std::unordered_map<std::string, FieldDescriptorProto::Type> TypeNameMap; |
66 | |
67 | const TypeNameMap& GetTypeNameTable() { |
68 | static auto* table = new auto([]() { |
69 | TypeNameMap result; |
70 | |
71 | result["double" ] = FieldDescriptorProto::TYPE_DOUBLE; |
72 | result["float" ] = FieldDescriptorProto::TYPE_FLOAT; |
73 | result["uint64" ] = FieldDescriptorProto::TYPE_UINT64; |
74 | result["fixed64" ] = FieldDescriptorProto::TYPE_FIXED64; |
75 | result["fixed32" ] = FieldDescriptorProto::TYPE_FIXED32; |
76 | result["bool" ] = FieldDescriptorProto::TYPE_BOOL; |
77 | result["string" ] = FieldDescriptorProto::TYPE_STRING; |
78 | result["group" ] = FieldDescriptorProto::TYPE_GROUP; |
79 | |
80 | result["bytes" ] = FieldDescriptorProto::TYPE_BYTES; |
81 | result["uint32" ] = FieldDescriptorProto::TYPE_UINT32; |
82 | result["sfixed32" ] = FieldDescriptorProto::TYPE_SFIXED32; |
83 | result["sfixed64" ] = FieldDescriptorProto::TYPE_SFIXED64; |
84 | result["int32" ] = FieldDescriptorProto::TYPE_INT32; |
85 | result["int64" ] = FieldDescriptorProto::TYPE_INT64; |
86 | result["sint32" ] = FieldDescriptorProto::TYPE_SINT32; |
87 | result["sint64" ] = FieldDescriptorProto::TYPE_SINT64; |
88 | |
89 | return result; |
90 | }()); |
91 | return *table; |
92 | } |
93 | |
94 | // Camel-case the field name and append "Entry" for generated map entry name. |
95 | // e.g. map<KeyType, ValueType> foo_map => FooMapEntry |
96 | std::string MapEntryName(const std::string& field_name) { |
97 | std::string result; |
98 | static const char kSuffix[] = "Entry" ; |
99 | result.reserve(res_arg: field_name.size() + sizeof(kSuffix)); |
100 | bool cap_next = true; |
101 | for (const char field_name_char : field_name) { |
102 | if (field_name_char == '_') { |
103 | cap_next = true; |
104 | } else if (cap_next) { |
105 | // Note: Do not use ctype.h due to locales. |
106 | if ('a' <= field_name_char && field_name_char <= 'z') { |
107 | result.push_back(c: field_name_char - 'a' + 'A'); |
108 | } else { |
109 | result.push_back(c: field_name_char); |
110 | } |
111 | cap_next = false; |
112 | } else { |
113 | result.push_back(c: field_name_char); |
114 | } |
115 | } |
116 | result.append(s: kSuffix); |
117 | return result; |
118 | } |
119 | |
120 | bool IsUppercase(char c) { return c >= 'A' && c <= 'Z'; } |
121 | |
122 | bool IsLowercase(char c) { return c >= 'a' && c <= 'z'; } |
123 | |
124 | bool IsNumber(char c) { return c >= '0' && c <= '9'; } |
125 | |
126 | bool IsUpperCamelCase(const std::string& name) { |
127 | if (name.empty()) { |
128 | return true; |
129 | } |
130 | // Name must start with an upper case character. |
131 | if (!IsUppercase(c: name[0])) { |
132 | return false; |
133 | } |
134 | // Must not contains underscore. |
135 | for (const char c : name) { |
136 | if (c == '_') { |
137 | return false; |
138 | } |
139 | } |
140 | return true; |
141 | } |
142 | |
143 | bool IsUpperUnderscore(const std::string& name) { |
144 | for (const char c : name) { |
145 | if (!IsUppercase(c) && c != '_' && !IsNumber(c)) { |
146 | return false; |
147 | } |
148 | } |
149 | return true; |
150 | } |
151 | |
152 | bool IsLowerUnderscore(const std::string& name) { |
153 | for (const char c : name) { |
154 | if (!IsLowercase(c) && c != '_' && !IsNumber(c)) { |
155 | return false; |
156 | } |
157 | } |
158 | return true; |
159 | } |
160 | |
161 | bool IsNumberFollowUnderscore(const std::string& name) { |
162 | for (int i = 1; i < name.length(); i++) { |
163 | const char c = name[i]; |
164 | if (IsNumber(c) && name[i - 1] == '_') { |
165 | return true; |
166 | } |
167 | } |
168 | return false; |
169 | } |
170 | |
171 | } // anonymous namespace |
172 | |
173 | // Makes code slightly more readable. The meaning of "DO(foo)" is |
174 | // "Execute foo and fail if it fails.", where failure is indicated by |
175 | // returning false. |
176 | #define DO(STATEMENT) \ |
177 | if (STATEMENT) { \ |
178 | } else \ |
179 | return false |
180 | |
181 | // =================================================================== |
182 | |
183 | Parser::Parser() |
184 | : input_(nullptr), |
185 | error_collector_(nullptr), |
186 | source_location_table_(nullptr), |
187 | had_errors_(false), |
188 | require_syntax_identifier_(false), |
189 | stop_after_syntax_identifier_(false) { |
190 | } |
191 | |
192 | Parser::~Parser() {} |
193 | |
194 | // =================================================================== |
195 | |
196 | inline bool Parser::LookingAt(const char* text) { |
197 | return input_->current().text == text; |
198 | } |
199 | |
200 | inline bool Parser::LookingAtType(io::Tokenizer::TokenType token_type) { |
201 | return input_->current().type == token_type; |
202 | } |
203 | |
204 | inline bool Parser::AtEnd() { return LookingAtType(token_type: io::Tokenizer::TYPE_END); } |
205 | |
206 | bool Parser::TryConsume(const char* text) { |
207 | if (LookingAt(text)) { |
208 | input_->Next(); |
209 | return true; |
210 | } else { |
211 | return false; |
212 | } |
213 | } |
214 | |
215 | bool Parser::Consume(const char* text, const char* error) { |
216 | if (TryConsume(text)) { |
217 | return true; |
218 | } else { |
219 | AddError(error); |
220 | return false; |
221 | } |
222 | } |
223 | |
224 | bool Parser::Consume(const char* text) { |
225 | std::string error = "Expected \"" + std::string(text) + "\"." ; |
226 | return Consume(text, error: error.c_str()); |
227 | } |
228 | |
229 | bool Parser::ConsumeIdentifier(std::string* output, const char* error) { |
230 | if (LookingAtType(token_type: io::Tokenizer::TYPE_IDENTIFIER)) { |
231 | *output = input_->current().text; |
232 | input_->Next(); |
233 | return true; |
234 | } else { |
235 | AddError(error); |
236 | return false; |
237 | } |
238 | } |
239 | |
240 | bool Parser::ConsumeInteger(int* output, const char* error) { |
241 | if (LookingAtType(token_type: io::Tokenizer::TYPE_INTEGER)) { |
242 | uint64_t value = 0; |
243 | if (!io::Tokenizer::ParseInteger(text: input_->current().text, |
244 | max_value: std::numeric_limits<int32_t>::max(), |
245 | output: &value)) { |
246 | AddError(error: "Integer out of range." ); |
247 | // We still return true because we did, in fact, parse an integer. |
248 | } |
249 | *output = value; |
250 | input_->Next(); |
251 | return true; |
252 | } else { |
253 | AddError(error); |
254 | return false; |
255 | } |
256 | } |
257 | |
258 | bool Parser::ConsumeSignedInteger(int* output, const char* error) { |
259 | bool is_negative = false; |
260 | uint64_t max_value = std::numeric_limits<int32_t>::max(); |
261 | if (TryConsume(text: "-" )) { |
262 | is_negative = true; |
263 | max_value += 1; |
264 | } |
265 | uint64_t value = 0; |
266 | DO(ConsumeInteger64(max_value, &value, error)); |
267 | if (is_negative) value *= -1; |
268 | *output = value; |
269 | return true; |
270 | } |
271 | |
272 | bool Parser::ConsumeInteger64(uint64_t max_value, uint64_t* output, |
273 | const char* error) { |
274 | if (LookingAtType(token_type: io::Tokenizer::TYPE_INTEGER)) { |
275 | if (!io::Tokenizer::ParseInteger(text: input_->current().text, max_value, |
276 | output)) { |
277 | AddError(error: "Integer out of range." ); |
278 | // We still return true because we did, in fact, parse an integer. |
279 | *output = 0; |
280 | } |
281 | input_->Next(); |
282 | return true; |
283 | } else { |
284 | AddError(error); |
285 | return false; |
286 | } |
287 | } |
288 | |
289 | bool Parser::(double* output, const char* error) { |
290 | if (LookingAtType(token_type: io::Tokenizer::TYPE_FLOAT)) { |
291 | *output = io::Tokenizer::ParseFloat(text: input_->current().text); |
292 | input_->Next(); |
293 | return true; |
294 | } else if (LookingAtType(token_type: io::Tokenizer::TYPE_INTEGER)) { |
295 | // Also accept integers. |
296 | uint64_t value = 0; |
297 | if (!io::Tokenizer::ParseInteger(text: input_->current().text, |
298 | max_value: std::numeric_limits<uint64_t>::max(), |
299 | output: &value)) { |
300 | AddError(error: "Integer out of range." ); |
301 | // We still return true because we did, in fact, parse a number. |
302 | } |
303 | *output = value; |
304 | input_->Next(); |
305 | return true; |
306 | } else if (LookingAt(text: "inf" )) { |
307 | *output = std::numeric_limits<double>::infinity(); |
308 | input_->Next(); |
309 | return true; |
310 | } else if (LookingAt(text: "nan" )) { |
311 | *output = std::numeric_limits<double>::quiet_NaN(); |
312 | input_->Next(); |
313 | return true; |
314 | } else { |
315 | AddError(error); |
316 | return false; |
317 | } |
318 | } |
319 | |
320 | bool Parser::ConsumeString(std::string* output, const char* error) { |
321 | if (LookingAtType(token_type: io::Tokenizer::TYPE_STRING)) { |
322 | io::Tokenizer::ParseString(text: input_->current().text, output); |
323 | input_->Next(); |
324 | // Allow C++ like concatenation of adjacent string tokens. |
325 | while (LookingAtType(token_type: io::Tokenizer::TYPE_STRING)) { |
326 | io::Tokenizer::ParseStringAppend(text: input_->current().text, output); |
327 | input_->Next(); |
328 | } |
329 | return true; |
330 | } else { |
331 | AddError(error); |
332 | return false; |
333 | } |
334 | } |
335 | |
336 | bool Parser::TryConsumeEndOfDeclaration(const char* text, |
337 | const LocationRecorder* location) { |
338 | if (LookingAt(text)) { |
339 | std::string leading, trailing; |
340 | std::vector<std::string> detached; |
341 | input_->NextWithComments(prev_trailing_comments: &trailing, detached_comments: &detached, next_leading_comments: &leading); |
342 | |
343 | // Save the leading comments for next time, and recall the leading comments |
344 | // from last time. |
345 | leading.swap(s&: upcoming_doc_comments_); |
346 | |
347 | if (location != nullptr) { |
348 | upcoming_detached_comments_.swap(x&: detached); |
349 | location->AttachComments(leading: &leading, trailing: &trailing, detached_comments: &detached); |
350 | } else if (strcmp(s1: text, s2: "}" ) == 0) { |
351 | // If the current location is null and we are finishing the current scope, |
352 | // drop pending upcoming detached comments. |
353 | upcoming_detached_comments_.swap(x&: detached); |
354 | } else { |
355 | // Otherwise, append the new detached comments to the existing upcoming |
356 | // detached comments. |
357 | upcoming_detached_comments_.insert(position: upcoming_detached_comments_.end(), |
358 | first: detached.begin(), last: detached.end()); |
359 | } |
360 | |
361 | return true; |
362 | } else { |
363 | return false; |
364 | } |
365 | } |
366 | |
367 | bool Parser::ConsumeEndOfDeclaration(const char* text, |
368 | const LocationRecorder* location) { |
369 | if (TryConsumeEndOfDeclaration(text, location)) { |
370 | return true; |
371 | } else { |
372 | AddError(error: "Expected \"" + std::string(text) + "\"." ); |
373 | return false; |
374 | } |
375 | } |
376 | |
377 | // ------------------------------------------------------------------- |
378 | |
379 | void Parser::AddError(int line, int column, const std::string& error) { |
380 | if (error_collector_ != nullptr) { |
381 | error_collector_->AddError(line, column, message: error); |
382 | } |
383 | had_errors_ = true; |
384 | } |
385 | |
386 | void Parser::AddError(const std::string& error) { |
387 | AddError(line: input_->current().line, column: input_->current().column, error); |
388 | } |
389 | |
390 | void Parser::AddWarning(const std::string& warning) { |
391 | if (error_collector_ != nullptr) { |
392 | error_collector_->AddWarning(input_->current().line, |
393 | input_->current().column, warning); |
394 | } |
395 | } |
396 | |
397 | // ------------------------------------------------------------------- |
398 | |
399 | Parser::LocationRecorder::LocationRecorder(Parser* parser) |
400 | : parser_(parser), |
401 | source_code_info_(parser->source_code_info_), |
402 | location_(parser_->source_code_info_->add_location()) { |
403 | location_->add_span(value: parser_->input_->current().line); |
404 | location_->add_span(value: parser_->input_->current().column); |
405 | } |
406 | |
407 | Parser::LocationRecorder::LocationRecorder(const LocationRecorder& parent) { |
408 | Init(parent, source_code_info: parent.source_code_info_); |
409 | } |
410 | |
411 | Parser::LocationRecorder::LocationRecorder(const LocationRecorder& parent, |
412 | int path1, |
413 | SourceCodeInfo* source_code_info) { |
414 | Init(parent, source_code_info); |
415 | AddPath(path_component: path1); |
416 | } |
417 | |
418 | Parser::LocationRecorder::LocationRecorder(const LocationRecorder& parent, |
419 | int path1) { |
420 | Init(parent, source_code_info: parent.source_code_info_); |
421 | AddPath(path_component: path1); |
422 | } |
423 | |
424 | Parser::LocationRecorder::LocationRecorder(const LocationRecorder& parent, |
425 | int path1, int path2) { |
426 | Init(parent, source_code_info: parent.source_code_info_); |
427 | AddPath(path_component: path1); |
428 | AddPath(path_component: path2); |
429 | } |
430 | |
431 | void Parser::LocationRecorder::Init(const LocationRecorder& parent, |
432 | SourceCodeInfo* source_code_info) { |
433 | parser_ = parent.parser_; |
434 | source_code_info_ = source_code_info; |
435 | |
436 | location_ = source_code_info_->add_location(); |
437 | location_->mutable_path()->CopyFrom(other: parent.location_->path()); |
438 | |
439 | location_->add_span(value: parser_->input_->current().line); |
440 | location_->add_span(value: parser_->input_->current().column); |
441 | } |
442 | |
443 | Parser::LocationRecorder::~LocationRecorder() { |
444 | if (location_->span_size() <= 2) { |
445 | EndAt(token: parser_->input_->previous()); |
446 | } |
447 | } |
448 | |
449 | void Parser::LocationRecorder::AddPath(int path_component) { |
450 | location_->add_path(value: path_component); |
451 | } |
452 | |
453 | void Parser::LocationRecorder::StartAt(const io::Tokenizer::Token& token) { |
454 | location_->set_span(index: 0, value: token.line); |
455 | location_->set_span(index: 1, value: token.column); |
456 | } |
457 | |
458 | void Parser::LocationRecorder::StartAt(const LocationRecorder& other) { |
459 | location_->set_span(index: 0, value: other.location_->span(index: 0)); |
460 | location_->set_span(index: 1, value: other.location_->span(index: 1)); |
461 | } |
462 | |
463 | void Parser::LocationRecorder::EndAt(const io::Tokenizer::Token& token) { |
464 | if (token.line != location_->span(index: 0)) { |
465 | location_->add_span(value: token.line); |
466 | } |
467 | location_->add_span(value: token.end_column); |
468 | } |
469 | |
470 | void Parser::LocationRecorder::RecordLegacyLocation( |
471 | const Message* descriptor, |
472 | DescriptorPool::ErrorCollector::ErrorLocation location) { |
473 | if (parser_->source_location_table_ != nullptr) { |
474 | parser_->source_location_table_->Add( |
475 | descriptor, location, line: location_->span(index: 0), column: location_->span(index: 1)); |
476 | } |
477 | } |
478 | |
479 | void Parser::LocationRecorder::RecordLegacyImportLocation( |
480 | const Message* descriptor, const std::string& name) { |
481 | if (parser_->source_location_table_ != nullptr) { |
482 | parser_->source_location_table_->AddImport( |
483 | descriptor, name, line: location_->span(index: 0), column: location_->span(index: 1)); |
484 | } |
485 | } |
486 | |
487 | int Parser::LocationRecorder::CurrentPathSize() const { |
488 | return location_->path_size(); |
489 | } |
490 | |
491 | void Parser::LocationRecorder::( |
492 | std::string* leading, std::string* trailing, |
493 | std::vector<std::string>* ) const { |
494 | GOOGLE_CHECK(!location_->has_leading_comments()); |
495 | GOOGLE_CHECK(!location_->has_trailing_comments()); |
496 | |
497 | if (!leading->empty()) { |
498 | location_->mutable_leading_comments()->swap(s&: *leading); |
499 | } |
500 | if (!trailing->empty()) { |
501 | location_->mutable_trailing_comments()->swap(s&: *trailing); |
502 | } |
503 | for (int i = 0; i < detached_comments->size(); ++i) { |
504 | location_->add_leading_detached_comments()->swap(s&: (*detached_comments)[i]); |
505 | } |
506 | detached_comments->clear(); |
507 | } |
508 | |
509 | // ------------------------------------------------------------------- |
510 | |
511 | void Parser::SkipStatement() { |
512 | while (true) { |
513 | if (AtEnd()) { |
514 | return; |
515 | } else if (LookingAtType(token_type: io::Tokenizer::TYPE_SYMBOL)) { |
516 | if (TryConsumeEndOfDeclaration(text: ";" , location: nullptr)) { |
517 | return; |
518 | } else if (TryConsume(text: "{" )) { |
519 | SkipRestOfBlock(); |
520 | return; |
521 | } else if (LookingAt(text: "}" )) { |
522 | return; |
523 | } |
524 | } |
525 | input_->Next(); |
526 | } |
527 | } |
528 | |
529 | void Parser::SkipRestOfBlock() { |
530 | while (true) { |
531 | if (AtEnd()) { |
532 | return; |
533 | } else if (LookingAtType(token_type: io::Tokenizer::TYPE_SYMBOL)) { |
534 | if (TryConsumeEndOfDeclaration(text: "}" , location: nullptr)) { |
535 | return; |
536 | } else if (TryConsume(text: "{" )) { |
537 | SkipRestOfBlock(); |
538 | } |
539 | } |
540 | input_->Next(); |
541 | } |
542 | } |
543 | |
544 | // =================================================================== |
545 | |
546 | bool Parser::ValidateEnum(const EnumDescriptorProto* proto) { |
547 | bool has_allow_alias = false; |
548 | bool allow_alias = false; |
549 | |
550 | for (int i = 0; i < proto->options().uninterpreted_option_size(); i++) { |
551 | const UninterpretedOption option = proto->options().uninterpreted_option(index: i); |
552 | if (option.name_size() > 1) { |
553 | continue; |
554 | } |
555 | if (!option.name(index: 0).is_extension() && |
556 | option.name(index: 0).name_part() == "allow_alias" ) { |
557 | has_allow_alias = true; |
558 | if (option.identifier_value() == "true" ) { |
559 | allow_alias = true; |
560 | } |
561 | break; |
562 | } |
563 | } |
564 | |
565 | if (has_allow_alias && !allow_alias) { |
566 | std::string error = |
567 | "\"" + proto->name() + |
568 | "\" declares 'option allow_alias = false;' which has no effect. " |
569 | "Please remove the declaration." ; |
570 | // This needlessly clutters declarations with nops. |
571 | AddError(error); |
572 | return false; |
573 | } |
574 | |
575 | std::set<int> used_values; |
576 | bool has_duplicates = false; |
577 | for (int i = 0; i < proto->value_size(); ++i) { |
578 | const EnumValueDescriptorProto& enum_value = proto->value(index: i); |
579 | if (used_values.find(x: enum_value.number()) != used_values.end()) { |
580 | has_duplicates = true; |
581 | break; |
582 | } else { |
583 | used_values.insert(x: enum_value.number()); |
584 | } |
585 | } |
586 | if (allow_alias && !has_duplicates) { |
587 | std::string error = |
588 | "\"" + proto->name() + |
589 | "\" declares support for enum aliases but no enum values share field " |
590 | "numbers. Please remove the unnecessary 'option allow_alias = true;' " |
591 | "declaration." ; |
592 | // Generate an error if an enum declares support for duplicate enum values |
593 | // and does not use it protect future authors. |
594 | AddError(error); |
595 | return false; |
596 | } |
597 | |
598 | // Enforce that enum constants must be UPPER_CASE except in case of |
599 | // enum_alias. |
600 | if (!allow_alias) { |
601 | for (const auto& enum_value : proto->value()) { |
602 | if (!IsUpperUnderscore(name: enum_value.name())) { |
603 | AddWarning( |
604 | warning: "Enum constant should be in UPPER_CASE. Found: " + |
605 | enum_value.name() + |
606 | ". See https://developers.google.com/protocol-buffers/docs/style" ); |
607 | } |
608 | } |
609 | } |
610 | |
611 | return true; |
612 | } |
613 | |
614 | bool Parser::Parse(io::Tokenizer* input, FileDescriptorProto* file) { |
615 | input_ = input; |
616 | had_errors_ = false; |
617 | syntax_identifier_.clear(); |
618 | |
619 | // Note that |file| could be NULL at this point if |
620 | // stop_after_syntax_identifier_ is true. So, we conservatively allocate |
621 | // SourceCodeInfo on the stack, then swap it into the FileDescriptorProto |
622 | // later on. |
623 | SourceCodeInfo source_code_info; |
624 | source_code_info_ = &source_code_info; |
625 | |
626 | if (LookingAtType(token_type: io::Tokenizer::TYPE_START)) { |
627 | // Advance to first token. |
628 | input_->NextWithComments(prev_trailing_comments: nullptr, detached_comments: &upcoming_detached_comments_, |
629 | next_leading_comments: &upcoming_doc_comments_); |
630 | } |
631 | |
632 | { |
633 | LocationRecorder root_location(this); |
634 | root_location.RecordLegacyLocation(descriptor: file, |
635 | location: DescriptorPool::ErrorCollector::OTHER); |
636 | |
637 | if (require_syntax_identifier_ || LookingAt(text: "syntax" )) { |
638 | if (!ParseSyntaxIdentifier(parent: root_location)) { |
639 | // Don't attempt to parse the file if we didn't recognize the syntax |
640 | // identifier. |
641 | return false; |
642 | } |
643 | // Store the syntax into the file. |
644 | if (file != nullptr) file->set_syntax(syntax_identifier_); |
645 | } else if (!stop_after_syntax_identifier_) { |
646 | GOOGLE_LOG(WARNING) << "No syntax specified for the proto file: " << file->name() |
647 | << ". Please use 'syntax = \"proto2\";' " |
648 | << "or 'syntax = \"proto3\";' to specify a syntax " |
649 | << "version. (Defaulted to proto2 syntax.)" ; |
650 | syntax_identifier_ = "proto2" ; |
651 | } |
652 | |
653 | if (stop_after_syntax_identifier_) return !had_errors_; |
654 | |
655 | // Repeatedly parse statements until we reach the end of the file. |
656 | while (!AtEnd()) { |
657 | if (!ParseTopLevelStatement(file, root_location)) { |
658 | // This statement failed to parse. Skip it, but keep looping to parse |
659 | // other statements. |
660 | SkipStatement(); |
661 | |
662 | if (LookingAt(text: "}" )) { |
663 | AddError(error: "Unmatched \"}\"." ); |
664 | input_->NextWithComments(prev_trailing_comments: nullptr, detached_comments: &upcoming_detached_comments_, |
665 | next_leading_comments: &upcoming_doc_comments_); |
666 | } |
667 | } |
668 | } |
669 | } |
670 | |
671 | input_ = nullptr; |
672 | source_code_info_ = nullptr; |
673 | assert(file != nullptr); |
674 | source_code_info.Swap(other: file->mutable_source_code_info()); |
675 | return !had_errors_; |
676 | } |
677 | |
678 | bool Parser::ParseSyntaxIdentifier(const LocationRecorder& parent) { |
679 | LocationRecorder syntax_location(parent, |
680 | FileDescriptorProto::kSyntaxFieldNumber); |
681 | DO(Consume( |
682 | "syntax" , |
683 | "File must begin with a syntax statement, e.g. 'syntax = \"proto2\";'." )); |
684 | DO(Consume("=" )); |
685 | io::Tokenizer::Token syntax_token = input_->current(); |
686 | std::string syntax; |
687 | DO(ConsumeString(&syntax, "Expected syntax identifier." )); |
688 | DO(ConsumeEndOfDeclaration(";" , &syntax_location)); |
689 | |
690 | syntax_identifier_ = syntax; |
691 | |
692 | if (syntax != "proto2" && syntax != "proto3" && |
693 | !stop_after_syntax_identifier_) { |
694 | AddError(line: syntax_token.line, column: syntax_token.column, |
695 | error: "Unrecognized syntax identifier \"" + syntax + |
696 | "\". This parser " |
697 | "only recognizes \"proto2\" and \"proto3\"." ); |
698 | return false; |
699 | } |
700 | |
701 | return true; |
702 | } |
703 | |
704 | bool Parser::ParseTopLevelStatement(FileDescriptorProto* file, |
705 | const LocationRecorder& root_location) { |
706 | if (TryConsumeEndOfDeclaration(text: ";" , location: nullptr)) { |
707 | // empty statement; ignore |
708 | return true; |
709 | } else if (LookingAt(text: "message" )) { |
710 | LocationRecorder location(root_location, |
711 | FileDescriptorProto::kMessageTypeFieldNumber, |
712 | file->message_type_size()); |
713 | return ParseMessageDefinition(message: file->add_message_type(), message_location: location, containing_file: file); |
714 | } else if (LookingAt(text: "enum" )) { |
715 | LocationRecorder location(root_location, |
716 | FileDescriptorProto::kEnumTypeFieldNumber, |
717 | file->enum_type_size()); |
718 | return ParseEnumDefinition(enum_type: file->add_enum_type(), enum_location: location, containing_file: file); |
719 | } else if (LookingAt(text: "service" )) { |
720 | LocationRecorder location(root_location, |
721 | FileDescriptorProto::kServiceFieldNumber, |
722 | file->service_size()); |
723 | return ParseServiceDefinition(service: file->add_service(), service_location: location, containing_file: file); |
724 | } else if (LookingAt(text: "extend" )) { |
725 | LocationRecorder location(root_location, |
726 | FileDescriptorProto::kExtensionFieldNumber); |
727 | return ParseExtend( |
728 | extensions: file->mutable_extension(), messages: file->mutable_message_type(), parent_location: root_location, |
729 | location_field_number_for_nested_type: FileDescriptorProto::kMessageTypeFieldNumber, extend_location: location, containing_file: file); |
730 | } else if (LookingAt(text: "import" )) { |
731 | return ParseImport(dependency: file->mutable_dependency(), |
732 | public_dependency: file->mutable_public_dependency(), |
733 | weak_dependency: file->mutable_weak_dependency(), root_location, containing_file: file); |
734 | } else if (LookingAt(text: "package" )) { |
735 | return ParsePackage(file, root_location, containing_file: file); |
736 | } else if (LookingAt(text: "option" )) { |
737 | LocationRecorder location(root_location, |
738 | FileDescriptorProto::kOptionsFieldNumber); |
739 | return ParseOption(options: file->mutable_options(), options_location: location, containing_file: file, |
740 | style: OPTION_STATEMENT); |
741 | } else { |
742 | AddError(error: "Expected top-level statement (e.g. \"message\")." ); |
743 | return false; |
744 | } |
745 | } |
746 | |
747 | // ------------------------------------------------------------------- |
748 | // Messages |
749 | |
750 | bool Parser::ParseMessageDefinition( |
751 | DescriptorProto* message, const LocationRecorder& message_location, |
752 | const FileDescriptorProto* containing_file) { |
753 | DO(Consume("message" )); |
754 | { |
755 | LocationRecorder location(message_location, |
756 | DescriptorProto::kNameFieldNumber); |
757 | location.RecordLegacyLocation(descriptor: message, |
758 | location: DescriptorPool::ErrorCollector::NAME); |
759 | DO(ConsumeIdentifier(message->mutable_name(), "Expected message name." )); |
760 | if (!IsUpperCamelCase(name: message->name())) { |
761 | AddWarning( |
762 | warning: "Message name should be in UpperCamelCase. Found: " + |
763 | message->name() + |
764 | ". See https://developers.google.com/protocol-buffers/docs/style" ); |
765 | } |
766 | } |
767 | DO(ParseMessageBlock(message, message_location, containing_file)); |
768 | |
769 | if (syntax_identifier_ == "proto3" ) { |
770 | // Add synthetic one-field oneofs for optional fields, except messages which |
771 | // already have presence in proto3. |
772 | // |
773 | // We have to make sure the oneof names don't conflict with any other |
774 | // field or oneof. |
775 | std::unordered_set<std::string> names; |
776 | for (const auto& field : message->field()) { |
777 | names.insert(x: field.name()); |
778 | } |
779 | for (const auto& oneof : message->oneof_decl()) { |
780 | names.insert(x: oneof.name()); |
781 | } |
782 | |
783 | for (auto& field : *message->mutable_field()) { |
784 | if (field.proto3_optional()) { |
785 | std::string oneof_name = field.name(); |
786 | |
787 | // Prepend 'XXXXX_' until we are no longer conflicting. |
788 | // Avoid prepending a double-underscore because such names are |
789 | // reserved in C++. |
790 | if (oneof_name.empty() || oneof_name[0] != '_') { |
791 | oneof_name = '_' + oneof_name; |
792 | } |
793 | while (names.count(x: oneof_name) > 0) { |
794 | oneof_name = 'X' + oneof_name; |
795 | } |
796 | |
797 | names.insert(x: oneof_name); |
798 | field.set_oneof_index(message->oneof_decl_size()); |
799 | OneofDescriptorProto* oneof = message->add_oneof_decl(); |
800 | oneof->set_name(oneof_name); |
801 | } |
802 | } |
803 | } |
804 | |
805 | return true; |
806 | } |
807 | |
808 | namespace { |
809 | |
810 | const int kMaxRangeSentinel = -1; |
811 | |
812 | bool IsMessageSetWireFormatMessage(const DescriptorProto& message) { |
813 | const MessageOptions& options = message.options(); |
814 | for (int i = 0; i < options.uninterpreted_option_size(); ++i) { |
815 | const UninterpretedOption& uninterpreted = options.uninterpreted_option(index: i); |
816 | if (uninterpreted.name_size() == 1 && |
817 | uninterpreted.name(index: 0).name_part() == "message_set_wire_format" && |
818 | uninterpreted.identifier_value() == "true" ) { |
819 | return true; |
820 | } |
821 | } |
822 | return false; |
823 | } |
824 | |
825 | // Modifies any extension ranges that specified 'max' as the end of the |
826 | // extension range, and sets them to the type-specific maximum. The actual max |
827 | // tag number can only be determined after all options have been parsed. |
828 | void AdjustExtensionRangesWithMaxEndNumber(DescriptorProto* message) { |
829 | const bool is_message_set = IsMessageSetWireFormatMessage(message: *message); |
830 | const int max_extension_number = is_message_set |
831 | ? std::numeric_limits<int32_t>::max() |
832 | : FieldDescriptor::kMaxNumber + 1; |
833 | for (int i = 0; i < message->extension_range_size(); ++i) { |
834 | if (message->extension_range(index: i).end() == kMaxRangeSentinel) { |
835 | message->mutable_extension_range(index: i)->set_end(max_extension_number); |
836 | } |
837 | } |
838 | } |
839 | |
840 | // Modifies any reserved ranges that specified 'max' as the end of the |
841 | // reserved range, and sets them to the type-specific maximum. The actual max |
842 | // tag number can only be determined after all options have been parsed. |
843 | void AdjustReservedRangesWithMaxEndNumber(DescriptorProto* message) { |
844 | const bool is_message_set = IsMessageSetWireFormatMessage(message: *message); |
845 | const int max_field_number = is_message_set |
846 | ? std::numeric_limits<int32_t>::max() |
847 | : FieldDescriptor::kMaxNumber + 1; |
848 | for (int i = 0; i < message->reserved_range_size(); ++i) { |
849 | if (message->reserved_range(index: i).end() == kMaxRangeSentinel) { |
850 | message->mutable_reserved_range(index: i)->set_end(max_field_number); |
851 | } |
852 | } |
853 | } |
854 | |
855 | } // namespace |
856 | |
857 | bool Parser::ParseMessageBlock(DescriptorProto* message, |
858 | const LocationRecorder& message_location, |
859 | const FileDescriptorProto* containing_file) { |
860 | DO(ConsumeEndOfDeclaration("{" , &message_location)); |
861 | |
862 | while (!TryConsumeEndOfDeclaration(text: "}" , location: nullptr)) { |
863 | if (AtEnd()) { |
864 | AddError(error: "Reached end of input in message definition (missing '}')." ); |
865 | return false; |
866 | } |
867 | |
868 | if (!ParseMessageStatement(message, message_location, containing_file)) { |
869 | // This statement failed to parse. Skip it, but keep looping to parse |
870 | // other statements. |
871 | SkipStatement(); |
872 | } |
873 | } |
874 | |
875 | if (message->extension_range_size() > 0) { |
876 | AdjustExtensionRangesWithMaxEndNumber(message); |
877 | } |
878 | if (message->reserved_range_size() > 0) { |
879 | AdjustReservedRangesWithMaxEndNumber(message); |
880 | } |
881 | return true; |
882 | } |
883 | |
884 | bool Parser::ParseMessageStatement(DescriptorProto* message, |
885 | const LocationRecorder& message_location, |
886 | const FileDescriptorProto* containing_file) { |
887 | if (TryConsumeEndOfDeclaration(text: ";" , location: nullptr)) { |
888 | // empty statement; ignore |
889 | return true; |
890 | } else if (LookingAt(text: "message" )) { |
891 | LocationRecorder location(message_location, |
892 | DescriptorProto::kNestedTypeFieldNumber, |
893 | message->nested_type_size()); |
894 | return ParseMessageDefinition(message: message->add_nested_type(), message_location: location, |
895 | containing_file); |
896 | } else if (LookingAt(text: "enum" )) { |
897 | LocationRecorder location(message_location, |
898 | DescriptorProto::kEnumTypeFieldNumber, |
899 | message->enum_type_size()); |
900 | return ParseEnumDefinition(enum_type: message->add_enum_type(), enum_location: location, |
901 | containing_file); |
902 | } else if (LookingAt(text: "extensions" )) { |
903 | LocationRecorder location(message_location, |
904 | DescriptorProto::kExtensionRangeFieldNumber); |
905 | return ParseExtensions(message, extensions_location: location, containing_file); |
906 | } else if (LookingAt(text: "reserved" )) { |
907 | return ParseReserved(message, message_location); |
908 | } else if (LookingAt(text: "extend" )) { |
909 | LocationRecorder location(message_location, |
910 | DescriptorProto::kExtensionFieldNumber); |
911 | return ParseExtend(extensions: message->mutable_extension(), |
912 | messages: message->mutable_nested_type(), parent_location: message_location, |
913 | location_field_number_for_nested_type: DescriptorProto::kNestedTypeFieldNumber, extend_location: location, |
914 | containing_file); |
915 | } else if (LookingAt(text: "option" )) { |
916 | LocationRecorder location(message_location, |
917 | DescriptorProto::kOptionsFieldNumber); |
918 | return ParseOption(options: message->mutable_options(), options_location: location, containing_file, |
919 | style: OPTION_STATEMENT); |
920 | } else if (LookingAt(text: "oneof" )) { |
921 | int oneof_index = message->oneof_decl_size(); |
922 | LocationRecorder oneof_location( |
923 | message_location, DescriptorProto::kOneofDeclFieldNumber, oneof_index); |
924 | |
925 | return ParseOneof(oneof_decl: message->add_oneof_decl(), containing_type: message, oneof_index, |
926 | oneof_location, containing_type_location: message_location, containing_file); |
927 | } else { |
928 | LocationRecorder location(message_location, |
929 | DescriptorProto::kFieldFieldNumber, |
930 | message->field_size()); |
931 | return ParseMessageField( |
932 | field: message->add_field(), messages: message->mutable_nested_type(), parent_location: message_location, |
933 | location_field_number_for_nested_type: DescriptorProto::kNestedTypeFieldNumber, field_location: location, containing_file); |
934 | } |
935 | } |
936 | |
937 | bool Parser::ParseMessageField(FieldDescriptorProto* field, |
938 | RepeatedPtrField<DescriptorProto>* messages, |
939 | const LocationRecorder& parent_location, |
940 | int location_field_number_for_nested_type, |
941 | const LocationRecorder& field_location, |
942 | const FileDescriptorProto* containing_file) { |
943 | { |
944 | FieldDescriptorProto::Label label; |
945 | if (ParseLabel(label: &label, field_location)) { |
946 | field->set_label(label); |
947 | if (label == FieldDescriptorProto::LABEL_OPTIONAL && |
948 | syntax_identifier_ == "proto3" ) { |
949 | field->set_proto3_optional(true); |
950 | } |
951 | } |
952 | } |
953 | |
954 | return ParseMessageFieldNoLabel(field, messages, parent_location, |
955 | location_field_number_for_nested_type, |
956 | field_location, containing_file); |
957 | } |
958 | |
959 | bool Parser::ParseMessageFieldNoLabel( |
960 | FieldDescriptorProto* field, RepeatedPtrField<DescriptorProto>* messages, |
961 | const LocationRecorder& parent_location, |
962 | int location_field_number_for_nested_type, |
963 | const LocationRecorder& field_location, |
964 | const FileDescriptorProto* containing_file) { |
965 | MapField map_field; |
966 | // Parse type. |
967 | { |
968 | LocationRecorder location(field_location); // add path later |
969 | location.RecordLegacyLocation(descriptor: field, location: DescriptorPool::ErrorCollector::TYPE); |
970 | |
971 | bool type_parsed = false; |
972 | FieldDescriptorProto::Type type = FieldDescriptorProto::TYPE_INT32; |
973 | std::string type_name; |
974 | |
975 | // Special case map field. We only treat the field as a map field if the |
976 | // field type name starts with the word "map" with a following "<". |
977 | if (TryConsume(text: "map" )) { |
978 | if (LookingAt(text: "<" )) { |
979 | map_field.is_map_field = true; |
980 | DO(ParseMapType(&map_field, field, location)); |
981 | } else { |
982 | // False positive |
983 | type_parsed = true; |
984 | type_name = "map" ; |
985 | } |
986 | } |
987 | if (!map_field.is_map_field) { |
988 | // Handle the case where no explicit label is given for a non-map field. |
989 | if (!field->has_label() && DefaultToOptionalFields()) { |
990 | field->set_label(FieldDescriptorProto::LABEL_OPTIONAL); |
991 | } |
992 | if (!field->has_label()) { |
993 | AddError(error: "Expected \"required\", \"optional\", or \"repeated\"." ); |
994 | // We can actually reasonably recover here by just assuming the user |
995 | // forgot the label altogether. |
996 | field->set_label(FieldDescriptorProto::LABEL_OPTIONAL); |
997 | } |
998 | |
999 | // Handle the case where the actual type is a message or enum named |
1000 | // "map", which we already consumed in the code above. |
1001 | if (!type_parsed) { |
1002 | DO(ParseType(&type, &type_name)); |
1003 | } |
1004 | if (type_name.empty()) { |
1005 | location.AddPath(path_component: FieldDescriptorProto::kTypeFieldNumber); |
1006 | field->set_type(type); |
1007 | } else { |
1008 | location.AddPath(path_component: FieldDescriptorProto::kTypeNameFieldNumber); |
1009 | field->set_type_name(type_name); |
1010 | } |
1011 | } |
1012 | } |
1013 | |
1014 | // Parse name and '='. |
1015 | io::Tokenizer::Token name_token = input_->current(); |
1016 | { |
1017 | LocationRecorder location(field_location, |
1018 | FieldDescriptorProto::kNameFieldNumber); |
1019 | location.RecordLegacyLocation(descriptor: field, location: DescriptorPool::ErrorCollector::NAME); |
1020 | DO(ConsumeIdentifier(field->mutable_name(), "Expected field name." )); |
1021 | |
1022 | if (!IsLowerUnderscore(name: field->name())) { |
1023 | AddWarning( |
1024 | warning: "Field name should be lowercase. Found: " + field->name() + |
1025 | ". See: https://developers.google.com/protocol-buffers/docs/style" ); |
1026 | } |
1027 | if (IsNumberFollowUnderscore(name: field->name())) { |
1028 | AddWarning( |
1029 | warning: "Number should not come right after an underscore. Found: " + |
1030 | field->name() + |
1031 | ". See: https://developers.google.com/protocol-buffers/docs/style" ); |
1032 | } |
1033 | } |
1034 | DO(Consume("=" , "Missing field number." )); |
1035 | |
1036 | // Parse field number. |
1037 | { |
1038 | LocationRecorder location(field_location, |
1039 | FieldDescriptorProto::kNumberFieldNumber); |
1040 | location.RecordLegacyLocation(descriptor: field, |
1041 | location: DescriptorPool::ErrorCollector::NUMBER); |
1042 | int number; |
1043 | DO(ConsumeInteger(&number, "Expected field number." )); |
1044 | field->set_number(number); |
1045 | } |
1046 | |
1047 | // Parse options. |
1048 | DO(ParseFieldOptions(field, field_location, containing_file)); |
1049 | |
1050 | // Deal with groups. |
1051 | if (field->has_type() && field->type() == FieldDescriptorProto::TYPE_GROUP) { |
1052 | // Awkward: Since a group declares both a message type and a field, we |
1053 | // have to create overlapping locations. |
1054 | LocationRecorder group_location(parent_location); |
1055 | group_location.StartAt(other: field_location); |
1056 | group_location.AddPath(path_component: location_field_number_for_nested_type); |
1057 | group_location.AddPath(path_component: messages->size()); |
1058 | |
1059 | DescriptorProto* group = messages->Add(); |
1060 | group->set_name(field->name()); |
1061 | |
1062 | // Record name location to match the field name's location. |
1063 | { |
1064 | LocationRecorder location(group_location, |
1065 | DescriptorProto::kNameFieldNumber); |
1066 | location.StartAt(token: name_token); |
1067 | location.EndAt(token: name_token); |
1068 | location.RecordLegacyLocation(descriptor: group, |
1069 | location: DescriptorPool::ErrorCollector::NAME); |
1070 | } |
1071 | |
1072 | // The field's type_name also comes from the name. Confusing! |
1073 | { |
1074 | LocationRecorder location(field_location, |
1075 | FieldDescriptorProto::kTypeNameFieldNumber); |
1076 | location.StartAt(token: name_token); |
1077 | location.EndAt(token: name_token); |
1078 | } |
1079 | |
1080 | // As a hack for backwards-compatibility, we force the group name to start |
1081 | // with a capital letter and lower-case the field name. New code should |
1082 | // not use groups; it should use nested messages. |
1083 | if (group->name()[0] < 'A' || 'Z' < group->name()[0]) { |
1084 | AddError(line: name_token.line, column: name_token.column, |
1085 | error: "Group names must start with a capital letter." ); |
1086 | } |
1087 | LowerString(s: field->mutable_name()); |
1088 | |
1089 | field->set_type_name(group->name()); |
1090 | if (LookingAt(text: "{" )) { |
1091 | DO(ParseMessageBlock(group, group_location, containing_file)); |
1092 | } else { |
1093 | AddError(error: "Missing group body." ); |
1094 | return false; |
1095 | } |
1096 | } else { |
1097 | DO(ConsumeEndOfDeclaration(";" , &field_location)); |
1098 | } |
1099 | |
1100 | // Create a map entry type if this is a map field. |
1101 | if (map_field.is_map_field) { |
1102 | GenerateMapEntry(map_field, field, messages); |
1103 | } |
1104 | |
1105 | return true; |
1106 | } |
1107 | |
1108 | bool Parser::ParseMapType(MapField* map_field, FieldDescriptorProto* field, |
1109 | LocationRecorder& type_name_location) { |
1110 | if (field->has_oneof_index()) { |
1111 | AddError(error: "Map fields are not allowed in oneofs." ); |
1112 | return false; |
1113 | } |
1114 | if (field->has_label()) { |
1115 | AddError( |
1116 | error: "Field labels (required/optional/repeated) are not allowed on " |
1117 | "map fields." ); |
1118 | return false; |
1119 | } |
1120 | if (field->has_extendee()) { |
1121 | AddError(error: "Map fields are not allowed to be extensions." ); |
1122 | return false; |
1123 | } |
1124 | field->set_label(FieldDescriptorProto::LABEL_REPEATED); |
1125 | DO(Consume("<" )); |
1126 | DO(ParseType(&map_field->key_type, &map_field->key_type_name)); |
1127 | DO(Consume("," )); |
1128 | DO(ParseType(&map_field->value_type, &map_field->value_type_name)); |
1129 | DO(Consume(">" )); |
1130 | // Defer setting of the type name of the map field until the |
1131 | // field name is parsed. Add the source location though. |
1132 | type_name_location.AddPath(path_component: FieldDescriptorProto::kTypeNameFieldNumber); |
1133 | return true; |
1134 | } |
1135 | |
1136 | void Parser::GenerateMapEntry(const MapField& map_field, |
1137 | FieldDescriptorProto* field, |
1138 | RepeatedPtrField<DescriptorProto>* messages) { |
1139 | DescriptorProto* entry = messages->Add(); |
1140 | std::string entry_name = MapEntryName(field_name: field->name()); |
1141 | field->set_type_name(entry_name); |
1142 | entry->set_name(entry_name); |
1143 | entry->mutable_options()->set_map_entry(true); |
1144 | FieldDescriptorProto* key_field = entry->add_field(); |
1145 | key_field->set_name("key" ); |
1146 | key_field->set_label(FieldDescriptorProto::LABEL_OPTIONAL); |
1147 | key_field->set_number(1); |
1148 | if (map_field.key_type_name.empty()) { |
1149 | key_field->set_type(map_field.key_type); |
1150 | } else { |
1151 | key_field->set_type_name(map_field.key_type_name); |
1152 | } |
1153 | FieldDescriptorProto* value_field = entry->add_field(); |
1154 | value_field->set_name("value" ); |
1155 | value_field->set_label(FieldDescriptorProto::LABEL_OPTIONAL); |
1156 | value_field->set_number(2); |
1157 | if (map_field.value_type_name.empty()) { |
1158 | value_field->set_type(map_field.value_type); |
1159 | } else { |
1160 | value_field->set_type_name(map_field.value_type_name); |
1161 | } |
1162 | // Propagate the "enforce_utf8" option to key and value fields if they |
1163 | // are strings. This helps simplify the implementation of code generators |
1164 | // and also reflection-based parsing code. |
1165 | // |
1166 | // The following definition: |
1167 | // message Foo { |
1168 | // map<string, string> value = 1 [enforce_utf8 = false]; |
1169 | // } |
1170 | // will be interpreted as: |
1171 | // message Foo { |
1172 | // message ValueEntry { |
1173 | // option map_entry = true; |
1174 | // string key = 1 [enforce_utf8 = false]; |
1175 | // string value = 2 [enforce_utf8 = false]; |
1176 | // } |
1177 | // repeated ValueEntry value = 1 [enforce_utf8 = false]; |
1178 | // } |
1179 | // |
1180 | // TODO(xiaofeng): Remove this when the "enforce_utf8" option is removed |
1181 | // from protocol compiler. |
1182 | for (int i = 0; i < field->options().uninterpreted_option_size(); ++i) { |
1183 | const UninterpretedOption& option = |
1184 | field->options().uninterpreted_option(index: i); |
1185 | if (option.name_size() == 1 && |
1186 | option.name(index: 0).name_part() == "enforce_utf8" && |
1187 | !option.name(index: 0).is_extension()) { |
1188 | if (key_field->type() == FieldDescriptorProto::TYPE_STRING) { |
1189 | key_field->mutable_options()->add_uninterpreted_option()->CopyFrom( |
1190 | from: option); |
1191 | } |
1192 | if (value_field->type() == FieldDescriptorProto::TYPE_STRING) { |
1193 | value_field->mutable_options()->add_uninterpreted_option()->CopyFrom( |
1194 | from: option); |
1195 | } |
1196 | } |
1197 | } |
1198 | } |
1199 | |
1200 | bool Parser::ParseFieldOptions(FieldDescriptorProto* field, |
1201 | const LocationRecorder& field_location, |
1202 | const FileDescriptorProto* containing_file) { |
1203 | if (!LookingAt(text: "[" )) return true; |
1204 | |
1205 | LocationRecorder location(field_location, |
1206 | FieldDescriptorProto::kOptionsFieldNumber); |
1207 | |
1208 | DO(Consume("[" )); |
1209 | |
1210 | // Parse field options. |
1211 | do { |
1212 | if (LookingAt(text: "default" )) { |
1213 | // We intentionally pass field_location rather than location here, since |
1214 | // the default value is not actually an option. |
1215 | DO(ParseDefaultAssignment(field, field_location, containing_file)); |
1216 | } else if (LookingAt(text: "json_name" )) { |
1217 | // Like default value, this "json_name" is not an actual option. |
1218 | DO(ParseJsonName(field, field_location, containing_file)); |
1219 | } else { |
1220 | DO(ParseOption(field->mutable_options(), location, containing_file, |
1221 | OPTION_ASSIGNMENT)); |
1222 | } |
1223 | } while (TryConsume(text: "," )); |
1224 | |
1225 | DO(Consume("]" )); |
1226 | return true; |
1227 | } |
1228 | |
1229 | bool Parser::ParseDefaultAssignment( |
1230 | FieldDescriptorProto* field, const LocationRecorder& field_location, |
1231 | const FileDescriptorProto* containing_file) { |
1232 | if (field->has_default_value()) { |
1233 | AddError(error: "Already set option \"default\"." ); |
1234 | field->clear_default_value(); |
1235 | } |
1236 | |
1237 | DO(Consume("default" )); |
1238 | DO(Consume("=" )); |
1239 | |
1240 | LocationRecorder location(field_location, |
1241 | FieldDescriptorProto::kDefaultValueFieldNumber); |
1242 | location.RecordLegacyLocation(descriptor: field, |
1243 | location: DescriptorPool::ErrorCollector::DEFAULT_VALUE); |
1244 | std::string* default_value = field->mutable_default_value(); |
1245 | |
1246 | if (!field->has_type()) { |
1247 | // The field has a type name, but we don't know if it is a message or an |
1248 | // enum yet. (If it were a primitive type, |field| would have a type set |
1249 | // already.) In this case, simply take the current string as the default |
1250 | // value; we will catch the error later if it is not a valid enum value. |
1251 | // (N.B. that we do not check whether the current token is an identifier: |
1252 | // doing so throws strange errors when the user mistypes a primitive |
1253 | // typename and we assume it's an enum. E.g.: "optional int foo = 1 [default |
1254 | // = 42]". In such a case the fundamental error is really that "int" is not |
1255 | // a type, not that "42" is not an identifier. See b/12533582.) |
1256 | *default_value = input_->current().text; |
1257 | input_->Next(); |
1258 | return true; |
1259 | } |
1260 | |
1261 | switch (field->type()) { |
1262 | case FieldDescriptorProto::TYPE_INT32: |
1263 | case FieldDescriptorProto::TYPE_INT64: |
1264 | case FieldDescriptorProto::TYPE_SINT32: |
1265 | case FieldDescriptorProto::TYPE_SINT64: |
1266 | case FieldDescriptorProto::TYPE_SFIXED32: |
1267 | case FieldDescriptorProto::TYPE_SFIXED64: { |
1268 | uint64_t max_value = std::numeric_limits<int64_t>::max(); |
1269 | if (field->type() == FieldDescriptorProto::TYPE_INT32 || |
1270 | field->type() == FieldDescriptorProto::TYPE_SINT32 || |
1271 | field->type() == FieldDescriptorProto::TYPE_SFIXED32) { |
1272 | max_value = std::numeric_limits<int32_t>::max(); |
1273 | } |
1274 | |
1275 | // These types can be negative. |
1276 | if (TryConsume(text: "-" )) { |
1277 | default_value->append(s: "-" ); |
1278 | // Two's complement always has one more negative value than positive. |
1279 | ++max_value; |
1280 | } |
1281 | // Parse the integer to verify that it is not out-of-range. |
1282 | uint64_t value; |
1283 | DO(ConsumeInteger64(max_value, &value, |
1284 | "Expected integer for field default value." )); |
1285 | // And stringify it again. |
1286 | default_value->append(str: StrCat(a: value)); |
1287 | break; |
1288 | } |
1289 | |
1290 | case FieldDescriptorProto::TYPE_UINT32: |
1291 | case FieldDescriptorProto::TYPE_UINT64: |
1292 | case FieldDescriptorProto::TYPE_FIXED32: |
1293 | case FieldDescriptorProto::TYPE_FIXED64: { |
1294 | uint64_t max_value = std::numeric_limits<uint64_t>::max(); |
1295 | if (field->type() == FieldDescriptorProto::TYPE_UINT32 || |
1296 | field->type() == FieldDescriptorProto::TYPE_FIXED32) { |
1297 | max_value = std::numeric_limits<uint32_t>::max(); |
1298 | } |
1299 | |
1300 | // Numeric, not negative. |
1301 | if (TryConsume(text: "-" )) { |
1302 | AddError(error: "Unsigned field can't have negative default value." ); |
1303 | } |
1304 | // Parse the integer to verify that it is not out-of-range. |
1305 | uint64_t value; |
1306 | DO(ConsumeInteger64(max_value, &value, |
1307 | "Expected integer for field default value." )); |
1308 | // And stringify it again. |
1309 | default_value->append(str: StrCat(a: value)); |
1310 | break; |
1311 | } |
1312 | |
1313 | case FieldDescriptorProto::TYPE_FLOAT: |
1314 | case FieldDescriptorProto::TYPE_DOUBLE: |
1315 | // These types can be negative. |
1316 | if (TryConsume(text: "-" )) { |
1317 | default_value->append(s: "-" ); |
1318 | } |
1319 | // Parse the integer because we have to convert hex integers to decimal |
1320 | // floats. |
1321 | double value; |
1322 | DO(ConsumeNumber(&value, "Expected number." )); |
1323 | // And stringify it again. |
1324 | default_value->append(str: SimpleDtoa(value)); |
1325 | break; |
1326 | |
1327 | case FieldDescriptorProto::TYPE_BOOL: |
1328 | if (TryConsume(text: "true" )) { |
1329 | default_value->assign(s: "true" ); |
1330 | } else if (TryConsume(text: "false" )) { |
1331 | default_value->assign(s: "false" ); |
1332 | } else { |
1333 | AddError(error: "Expected \"true\" or \"false\"." ); |
1334 | return false; |
1335 | } |
1336 | break; |
1337 | |
1338 | case FieldDescriptorProto::TYPE_STRING: |
1339 | // Note: When file option java_string_check_utf8 is true, if a |
1340 | // non-string representation (eg byte[]) is later supported, it must |
1341 | // be checked for UTF-8-ness. |
1342 | DO(ConsumeString(default_value, |
1343 | "Expected string for field default " |
1344 | "value." )); |
1345 | break; |
1346 | |
1347 | case FieldDescriptorProto::TYPE_BYTES: |
1348 | DO(ConsumeString(default_value, "Expected string." )); |
1349 | *default_value = CEscape(src: *default_value); |
1350 | break; |
1351 | |
1352 | case FieldDescriptorProto::TYPE_ENUM: |
1353 | DO(ConsumeIdentifier(default_value, |
1354 | "Expected enum identifier for field " |
1355 | "default value." )); |
1356 | break; |
1357 | |
1358 | case FieldDescriptorProto::TYPE_MESSAGE: |
1359 | case FieldDescriptorProto::TYPE_GROUP: |
1360 | AddError(error: "Messages can't have default values." ); |
1361 | return false; |
1362 | } |
1363 | |
1364 | return true; |
1365 | } |
1366 | |
1367 | bool Parser::ParseJsonName(FieldDescriptorProto* field, |
1368 | const LocationRecorder& field_location, |
1369 | const FileDescriptorProto* containing_file) { |
1370 | if (field->has_json_name()) { |
1371 | AddError(error: "Already set option \"json_name\"." ); |
1372 | field->clear_json_name(); |
1373 | } |
1374 | |
1375 | LocationRecorder location(field_location, |
1376 | FieldDescriptorProto::kJsonNameFieldNumber); |
1377 | location.RecordLegacyLocation(descriptor: field, |
1378 | location: DescriptorPool::ErrorCollector::OPTION_NAME); |
1379 | |
1380 | DO(Consume("json_name" )); |
1381 | DO(Consume("=" )); |
1382 | |
1383 | LocationRecorder value_location(location); |
1384 | value_location.RecordLegacyLocation( |
1385 | descriptor: field, location: DescriptorPool::ErrorCollector::OPTION_VALUE); |
1386 | |
1387 | DO(ConsumeString(field->mutable_json_name(), |
1388 | "Expected string for JSON name." )); |
1389 | return true; |
1390 | } |
1391 | |
1392 | bool Parser::ParseOptionNamePart(UninterpretedOption* uninterpreted_option, |
1393 | const LocationRecorder& part_location, |
1394 | const FileDescriptorProto* containing_file) { |
1395 | UninterpretedOption::NamePart* name = uninterpreted_option->add_name(); |
1396 | std::string identifier; // We parse identifiers into this string. |
1397 | if (LookingAt(text: "(" )) { // This is an extension. |
1398 | DO(Consume("(" )); |
1399 | |
1400 | { |
1401 | LocationRecorder location( |
1402 | part_location, UninterpretedOption::NamePart::kNamePartFieldNumber); |
1403 | // An extension name consists of dot-separated identifiers, and may begin |
1404 | // with a dot. |
1405 | if (LookingAtType(token_type: io::Tokenizer::TYPE_IDENTIFIER)) { |
1406 | DO(ConsumeIdentifier(&identifier, "Expected identifier." )); |
1407 | name->mutable_name_part()->append(str: identifier); |
1408 | } |
1409 | while (LookingAt(text: "." )) { |
1410 | DO(Consume("." )); |
1411 | name->mutable_name_part()->append(s: "." ); |
1412 | DO(ConsumeIdentifier(&identifier, "Expected identifier." )); |
1413 | name->mutable_name_part()->append(str: identifier); |
1414 | } |
1415 | } |
1416 | |
1417 | DO(Consume(")" )); |
1418 | name->set_is_extension(true); |
1419 | } else { // This is a regular field. |
1420 | LocationRecorder location( |
1421 | part_location, UninterpretedOption::NamePart::kNamePartFieldNumber); |
1422 | DO(ConsumeIdentifier(&identifier, "Expected identifier." )); |
1423 | name->mutable_name_part()->append(str: identifier); |
1424 | name->set_is_extension(false); |
1425 | } |
1426 | return true; |
1427 | } |
1428 | |
1429 | bool Parser::ParseUninterpretedBlock(std::string* value) { |
1430 | // Note that enclosing braces are not added to *value. |
1431 | // We do NOT use ConsumeEndOfStatement for this brace because it's delimiting |
1432 | // an expression, not a block of statements. |
1433 | DO(Consume("{" )); |
1434 | int brace_depth = 1; |
1435 | while (!AtEnd()) { |
1436 | if (LookingAt(text: "{" )) { |
1437 | brace_depth++; |
1438 | } else if (LookingAt(text: "}" )) { |
1439 | brace_depth--; |
1440 | if (brace_depth == 0) { |
1441 | input_->Next(); |
1442 | return true; |
1443 | } |
1444 | } |
1445 | // TODO(sanjay): Interpret line/column numbers to preserve formatting |
1446 | if (!value->empty()) value->push_back(c: ' '); |
1447 | value->append(str: input_->current().text); |
1448 | input_->Next(); |
1449 | } |
1450 | AddError(error: "Unexpected end of stream while parsing aggregate value." ); |
1451 | return false; |
1452 | } |
1453 | |
1454 | // We don't interpret the option here. Instead we store it in an |
1455 | // UninterpretedOption, to be interpreted later. |
1456 | bool Parser::ParseOption(Message* options, |
1457 | const LocationRecorder& options_location, |
1458 | const FileDescriptorProto* containing_file, |
1459 | OptionStyle style) { |
1460 | // Create an entry in the uninterpreted_option field. |
1461 | const FieldDescriptor* uninterpreted_option_field = |
1462 | options->GetDescriptor()->FindFieldByName(name: "uninterpreted_option" ); |
1463 | GOOGLE_CHECK(uninterpreted_option_field != nullptr) |
1464 | << "No field named \"uninterpreted_option\" in the Options proto." ; |
1465 | |
1466 | const Reflection* reflection = options->GetReflection(); |
1467 | |
1468 | LocationRecorder location( |
1469 | options_location, uninterpreted_option_field->number(), |
1470 | reflection->FieldSize(message: *options, field: uninterpreted_option_field)); |
1471 | |
1472 | if (style == OPTION_STATEMENT) { |
1473 | DO(Consume("option" )); |
1474 | } |
1475 | |
1476 | UninterpretedOption* uninterpreted_option = |
1477 | down_cast<UninterpretedOption*>(f: options->GetReflection()->AddMessage( |
1478 | message: options, field: uninterpreted_option_field)); |
1479 | |
1480 | // Parse dot-separated name. |
1481 | { |
1482 | LocationRecorder name_location(location, |
1483 | UninterpretedOption::kNameFieldNumber); |
1484 | name_location.RecordLegacyLocation( |
1485 | descriptor: uninterpreted_option, location: DescriptorPool::ErrorCollector::OPTION_NAME); |
1486 | |
1487 | { |
1488 | LocationRecorder part_location(name_location, |
1489 | uninterpreted_option->name_size()); |
1490 | DO(ParseOptionNamePart(uninterpreted_option, part_location, |
1491 | containing_file)); |
1492 | } |
1493 | |
1494 | while (LookingAt(text: "." )) { |
1495 | DO(Consume("." )); |
1496 | LocationRecorder part_location(name_location, |
1497 | uninterpreted_option->name_size()); |
1498 | DO(ParseOptionNamePart(uninterpreted_option, part_location, |
1499 | containing_file)); |
1500 | } |
1501 | } |
1502 | |
1503 | DO(Consume("=" )); |
1504 | |
1505 | { |
1506 | LocationRecorder value_location(location); |
1507 | value_location.RecordLegacyLocation( |
1508 | descriptor: uninterpreted_option, location: DescriptorPool::ErrorCollector::OPTION_VALUE); |
1509 | |
1510 | // All values are a single token, except for negative numbers, which consist |
1511 | // of a single '-' symbol, followed by a positive number. |
1512 | bool is_negative = TryConsume(text: "-" ); |
1513 | |
1514 | switch (input_->current().type) { |
1515 | case io::Tokenizer::TYPE_START: |
1516 | GOOGLE_LOG(FATAL) << "Trying to read value before any tokens have been read." ; |
1517 | return false; |
1518 | |
1519 | case io::Tokenizer::TYPE_END: |
1520 | AddError(error: "Unexpected end of stream while parsing option value." ); |
1521 | return false; |
1522 | |
1523 | case io::Tokenizer::TYPE_WHITESPACE: |
1524 | case io::Tokenizer::TYPE_NEWLINE: |
1525 | GOOGLE_CHECK(!input_->report_whitespace() && !input_->report_newlines()) |
1526 | << "Whitespace tokens were not requested." ; |
1527 | GOOGLE_LOG(FATAL) << "Tokenizer reported whitespace." ; |
1528 | return false; |
1529 | |
1530 | case io::Tokenizer::TYPE_IDENTIFIER: { |
1531 | value_location.AddPath( |
1532 | path_component: UninterpretedOption::kIdentifierValueFieldNumber); |
1533 | if (is_negative) { |
1534 | AddError(error: "Invalid '-' symbol before identifier." ); |
1535 | return false; |
1536 | } |
1537 | std::string value; |
1538 | DO(ConsumeIdentifier(&value, "Expected identifier." )); |
1539 | uninterpreted_option->set_identifier_value(value); |
1540 | break; |
1541 | } |
1542 | |
1543 | case io::Tokenizer::TYPE_INTEGER: { |
1544 | uint64_t value; |
1545 | uint64_t max_value = |
1546 | is_negative |
1547 | ? static_cast<uint64_t>(std::numeric_limits<int64_t>::max()) + 1 |
1548 | : std::numeric_limits<uint64_t>::max(); |
1549 | DO(ConsumeInteger64(max_value, &value, "Expected integer." )); |
1550 | if (is_negative) { |
1551 | value_location.AddPath( |
1552 | path_component: UninterpretedOption::kNegativeIntValueFieldNumber); |
1553 | uninterpreted_option->set_negative_int_value( |
1554 | static_cast<int64_t>(0 - value)); |
1555 | } else { |
1556 | value_location.AddPath( |
1557 | path_component: UninterpretedOption::kPositiveIntValueFieldNumber); |
1558 | uninterpreted_option->set_positive_int_value(value); |
1559 | } |
1560 | break; |
1561 | } |
1562 | |
1563 | case io::Tokenizer::TYPE_FLOAT: { |
1564 | value_location.AddPath(path_component: UninterpretedOption::kDoubleValueFieldNumber); |
1565 | double value; |
1566 | DO(ConsumeNumber(&value, "Expected number." )); |
1567 | uninterpreted_option->set_double_value(is_negative ? -value : value); |
1568 | break; |
1569 | } |
1570 | |
1571 | case io::Tokenizer::TYPE_STRING: { |
1572 | value_location.AddPath(path_component: UninterpretedOption::kStringValueFieldNumber); |
1573 | if (is_negative) { |
1574 | AddError(error: "Invalid '-' symbol before string." ); |
1575 | return false; |
1576 | } |
1577 | std::string value; |
1578 | DO(ConsumeString(&value, "Expected string." )); |
1579 | uninterpreted_option->set_string_value(value); |
1580 | break; |
1581 | } |
1582 | |
1583 | case io::Tokenizer::TYPE_SYMBOL: |
1584 | if (LookingAt(text: "{" )) { |
1585 | value_location.AddPath( |
1586 | path_component: UninterpretedOption::kAggregateValueFieldNumber); |
1587 | DO(ParseUninterpretedBlock( |
1588 | uninterpreted_option->mutable_aggregate_value())); |
1589 | } else { |
1590 | AddError(error: "Expected option value." ); |
1591 | return false; |
1592 | } |
1593 | break; |
1594 | } |
1595 | } |
1596 | |
1597 | if (style == OPTION_STATEMENT) { |
1598 | DO(ConsumeEndOfDeclaration(";" , &location)); |
1599 | } |
1600 | |
1601 | return true; |
1602 | } |
1603 | |
1604 | bool Parser::ParseExtensions(DescriptorProto* message, |
1605 | const LocationRecorder& extensions_location, |
1606 | const FileDescriptorProto* containing_file) { |
1607 | // Parse the declaration. |
1608 | DO(Consume("extensions" )); |
1609 | |
1610 | int old_range_size = message->extension_range_size(); |
1611 | |
1612 | do { |
1613 | // Note that kExtensionRangeFieldNumber was already pushed by the parent. |
1614 | LocationRecorder location(extensions_location, |
1615 | message->extension_range_size()); |
1616 | |
1617 | DescriptorProto::ExtensionRange* range = message->add_extension_range(); |
1618 | location.RecordLegacyLocation(descriptor: range, |
1619 | location: DescriptorPool::ErrorCollector::NUMBER); |
1620 | |
1621 | int start, end; |
1622 | io::Tokenizer::Token start_token; |
1623 | |
1624 | { |
1625 | LocationRecorder start_location( |
1626 | location, DescriptorProto::ExtensionRange::kStartFieldNumber); |
1627 | start_token = input_->current(); |
1628 | DO(ConsumeInteger(&start, "Expected field number range." )); |
1629 | } |
1630 | |
1631 | if (TryConsume(text: "to" )) { |
1632 | LocationRecorder end_location( |
1633 | location, DescriptorProto::ExtensionRange::kEndFieldNumber); |
1634 | if (TryConsume(text: "max" )) { |
1635 | // Set to the sentinel value - 1 since we increment the value below. |
1636 | // The actual value of the end of the range should be set with |
1637 | // AdjustExtensionRangesWithMaxEndNumber. |
1638 | end = kMaxRangeSentinel - 1; |
1639 | } else { |
1640 | DO(ConsumeInteger(&end, "Expected integer." )); |
1641 | } |
1642 | } else { |
1643 | LocationRecorder end_location( |
1644 | location, DescriptorProto::ExtensionRange::kEndFieldNumber); |
1645 | end_location.StartAt(token: start_token); |
1646 | end_location.EndAt(token: start_token); |
1647 | end = start; |
1648 | } |
1649 | |
1650 | // Users like to specify inclusive ranges, but in code we like the end |
1651 | // number to be exclusive. |
1652 | ++end; |
1653 | |
1654 | range->set_start(start); |
1655 | range->set_end(end); |
1656 | } while (TryConsume(text: "," )); |
1657 | |
1658 | if (LookingAt(text: "[" )) { |
1659 | int range_number_index = extensions_location.CurrentPathSize(); |
1660 | SourceCodeInfo info; |
1661 | |
1662 | // Parse extension range options in the first range. |
1663 | ExtensionRangeOptions* options = |
1664 | message->mutable_extension_range(index: old_range_size)->mutable_options(); |
1665 | |
1666 | { |
1667 | LocationRecorder index_location( |
1668 | extensions_location, 0 /* we fill this in w/ actual index below */, |
1669 | &info); |
1670 | LocationRecorder location( |
1671 | index_location, DescriptorProto::ExtensionRange::kOptionsFieldNumber); |
1672 | DO(Consume("[" )); |
1673 | |
1674 | do { |
1675 | DO(ParseOption(options, location, containing_file, OPTION_ASSIGNMENT)); |
1676 | } while (TryConsume(text: "," )); |
1677 | |
1678 | DO(Consume("]" )); |
1679 | } |
1680 | |
1681 | // Then copy the extension range options to all of the other ranges we've |
1682 | // parsed. |
1683 | for (int i = old_range_size + 1; i < message->extension_range_size(); i++) { |
1684 | message->mutable_extension_range(index: i)->mutable_options()->CopyFrom( |
1685 | from: *options); |
1686 | } |
1687 | // and copy source locations to the other ranges, too |
1688 | for (int i = old_range_size; i < message->extension_range_size(); i++) { |
1689 | for (int j = 0; j < info.location_size(); j++) { |
1690 | if (info.location(index: j).path_size() == range_number_index + 1) { |
1691 | // this location's path is up to the extension range index, but |
1692 | // doesn't include options; so it's redundant with location above |
1693 | continue; |
1694 | } |
1695 | SourceCodeInfo_Location* dest = source_code_info_->add_location(); |
1696 | *dest = info.location(index: j); |
1697 | dest->set_path(index: range_number_index, value: i); |
1698 | } |
1699 | } |
1700 | } |
1701 | |
1702 | DO(ConsumeEndOfDeclaration(";" , &extensions_location)); |
1703 | return true; |
1704 | } |
1705 | |
1706 | // This is similar to extension range parsing, except that it accepts field |
1707 | // name literals. |
1708 | bool Parser::ParseReserved(DescriptorProto* message, |
1709 | const LocationRecorder& message_location) { |
1710 | io::Tokenizer::Token start_token = input_->current(); |
1711 | // Parse the declaration. |
1712 | DO(Consume("reserved" )); |
1713 | if (LookingAtType(token_type: io::Tokenizer::TYPE_STRING)) { |
1714 | LocationRecorder location(message_location, |
1715 | DescriptorProto::kReservedNameFieldNumber); |
1716 | location.StartAt(token: start_token); |
1717 | return ParseReservedNames(message, parent_location: location); |
1718 | } else { |
1719 | LocationRecorder location(message_location, |
1720 | DescriptorProto::kReservedRangeFieldNumber); |
1721 | location.StartAt(token: start_token); |
1722 | return ParseReservedNumbers(message, parent_location: location); |
1723 | } |
1724 | } |
1725 | |
1726 | bool Parser::ParseReservedNames(DescriptorProto* message, |
1727 | const LocationRecorder& parent_location) { |
1728 | do { |
1729 | LocationRecorder location(parent_location, message->reserved_name_size()); |
1730 | DO(ConsumeString(message->add_reserved_name(), "Expected field name." )); |
1731 | } while (TryConsume(text: "," )); |
1732 | DO(ConsumeEndOfDeclaration(";" , &parent_location)); |
1733 | return true; |
1734 | } |
1735 | |
1736 | bool Parser::ParseReservedNumbers(DescriptorProto* message, |
1737 | const LocationRecorder& parent_location) { |
1738 | bool first = true; |
1739 | do { |
1740 | LocationRecorder location(parent_location, message->reserved_range_size()); |
1741 | |
1742 | DescriptorProto::ReservedRange* range = message->add_reserved_range(); |
1743 | int start, end; |
1744 | io::Tokenizer::Token start_token; |
1745 | { |
1746 | LocationRecorder start_location( |
1747 | location, DescriptorProto::ReservedRange::kStartFieldNumber); |
1748 | start_token = input_->current(); |
1749 | DO(ConsumeInteger(&start, (first ? "Expected field name or number range." |
1750 | : "Expected field number range." ))); |
1751 | } |
1752 | |
1753 | if (TryConsume(text: "to" )) { |
1754 | LocationRecorder end_location( |
1755 | location, DescriptorProto::ReservedRange::kEndFieldNumber); |
1756 | if (TryConsume(text: "max" )) { |
1757 | // Set to the sentinel value - 1 since we increment the value below. |
1758 | // The actual value of the end of the range should be set with |
1759 | // AdjustExtensionRangesWithMaxEndNumber. |
1760 | end = kMaxRangeSentinel - 1; |
1761 | } else { |
1762 | DO(ConsumeInteger(&end, "Expected integer." )); |
1763 | } |
1764 | } else { |
1765 | LocationRecorder end_location( |
1766 | location, DescriptorProto::ReservedRange::kEndFieldNumber); |
1767 | end_location.StartAt(token: start_token); |
1768 | end_location.EndAt(token: start_token); |
1769 | end = start; |
1770 | } |
1771 | |
1772 | // Users like to specify inclusive ranges, but in code we like the end |
1773 | // number to be exclusive. |
1774 | ++end; |
1775 | |
1776 | range->set_start(start); |
1777 | range->set_end(end); |
1778 | first = false; |
1779 | } while (TryConsume(text: "," )); |
1780 | |
1781 | DO(ConsumeEndOfDeclaration(";" , &parent_location)); |
1782 | return true; |
1783 | } |
1784 | |
1785 | bool Parser::ParseReserved(EnumDescriptorProto* message, |
1786 | const LocationRecorder& message_location) { |
1787 | io::Tokenizer::Token start_token = input_->current(); |
1788 | // Parse the declaration. |
1789 | DO(Consume("reserved" )); |
1790 | if (LookingAtType(token_type: io::Tokenizer::TYPE_STRING)) { |
1791 | LocationRecorder location(message_location, |
1792 | EnumDescriptorProto::kReservedNameFieldNumber); |
1793 | location.StartAt(token: start_token); |
1794 | return ParseReservedNames(message, parent_location: location); |
1795 | } else { |
1796 | LocationRecorder location(message_location, |
1797 | EnumDescriptorProto::kReservedRangeFieldNumber); |
1798 | location.StartAt(token: start_token); |
1799 | return ParseReservedNumbers(message, parent_location: location); |
1800 | } |
1801 | } |
1802 | |
1803 | bool Parser::ParseReservedNames(EnumDescriptorProto* message, |
1804 | const LocationRecorder& parent_location) { |
1805 | do { |
1806 | LocationRecorder location(parent_location, message->reserved_name_size()); |
1807 | DO(ConsumeString(message->add_reserved_name(), "Expected enum value." )); |
1808 | } while (TryConsume(text: "," )); |
1809 | DO(ConsumeEndOfDeclaration(";" , &parent_location)); |
1810 | return true; |
1811 | } |
1812 | |
1813 | bool Parser::ParseReservedNumbers(EnumDescriptorProto* message, |
1814 | const LocationRecorder& parent_location) { |
1815 | bool first = true; |
1816 | do { |
1817 | LocationRecorder location(parent_location, message->reserved_range_size()); |
1818 | |
1819 | EnumDescriptorProto::EnumReservedRange* range = |
1820 | message->add_reserved_range(); |
1821 | int start, end; |
1822 | io::Tokenizer::Token start_token; |
1823 | { |
1824 | LocationRecorder start_location( |
1825 | location, EnumDescriptorProto::EnumReservedRange::kStartFieldNumber); |
1826 | start_token = input_->current(); |
1827 | DO(ConsumeSignedInteger(&start, |
1828 | (first ? "Expected enum value or number range." |
1829 | : "Expected enum number range." ))); |
1830 | } |
1831 | |
1832 | if (TryConsume(text: "to" )) { |
1833 | LocationRecorder end_location( |
1834 | location, EnumDescriptorProto::EnumReservedRange::kEndFieldNumber); |
1835 | if (TryConsume(text: "max" )) { |
1836 | // This is in the enum descriptor path, which doesn't have the message |
1837 | // set duality to fix up, so it doesn't integrate with the sentinel. |
1838 | end = INT_MAX; |
1839 | } else { |
1840 | DO(ConsumeSignedInteger(&end, "Expected integer." )); |
1841 | } |
1842 | } else { |
1843 | LocationRecorder end_location( |
1844 | location, EnumDescriptorProto::EnumReservedRange::kEndFieldNumber); |
1845 | end_location.StartAt(token: start_token); |
1846 | end_location.EndAt(token: start_token); |
1847 | end = start; |
1848 | } |
1849 | |
1850 | range->set_start(start); |
1851 | range->set_end(end); |
1852 | first = false; |
1853 | } while (TryConsume(text: "," )); |
1854 | |
1855 | DO(ConsumeEndOfDeclaration(";" , &parent_location)); |
1856 | return true; |
1857 | } |
1858 | |
1859 | bool Parser::ParseExtend(RepeatedPtrField<FieldDescriptorProto>* extensions, |
1860 | RepeatedPtrField<DescriptorProto>* messages, |
1861 | const LocationRecorder& parent_location, |
1862 | int location_field_number_for_nested_type, |
1863 | const LocationRecorder& extend_location, |
1864 | const FileDescriptorProto* containing_file) { |
1865 | DO(Consume("extend" )); |
1866 | |
1867 | // Parse the extendee type. |
1868 | io::Tokenizer::Token extendee_start = input_->current(); |
1869 | std::string extendee; |
1870 | DO(ParseUserDefinedType(&extendee)); |
1871 | io::Tokenizer::Token extendee_end = input_->previous(); |
1872 | |
1873 | // Parse the block. |
1874 | DO(ConsumeEndOfDeclaration("{" , &extend_location)); |
1875 | |
1876 | bool is_first = true; |
1877 | |
1878 | do { |
1879 | if (AtEnd()) { |
1880 | AddError(error: "Reached end of input in extend definition (missing '}')." ); |
1881 | return false; |
1882 | } |
1883 | |
1884 | // Note that kExtensionFieldNumber was already pushed by the parent. |
1885 | LocationRecorder location(extend_location, extensions->size()); |
1886 | |
1887 | FieldDescriptorProto* field = extensions->Add(); |
1888 | |
1889 | { |
1890 | LocationRecorder extendee_location( |
1891 | location, FieldDescriptorProto::kExtendeeFieldNumber); |
1892 | extendee_location.StartAt(token: extendee_start); |
1893 | extendee_location.EndAt(token: extendee_end); |
1894 | |
1895 | if (is_first) { |
1896 | extendee_location.RecordLegacyLocation( |
1897 | descriptor: field, location: DescriptorPool::ErrorCollector::EXTENDEE); |
1898 | is_first = false; |
1899 | } |
1900 | } |
1901 | |
1902 | field->set_extendee(extendee); |
1903 | |
1904 | if (!ParseMessageField(field, messages, parent_location, |
1905 | location_field_number_for_nested_type, field_location: location, |
1906 | containing_file)) { |
1907 | // This statement failed to parse. Skip it, but keep looping to parse |
1908 | // other statements. |
1909 | SkipStatement(); |
1910 | } |
1911 | } while (!TryConsumeEndOfDeclaration(text: "}" , location: nullptr)); |
1912 | |
1913 | return true; |
1914 | } |
1915 | |
1916 | bool Parser::ParseOneof(OneofDescriptorProto* oneof_decl, |
1917 | DescriptorProto* containing_type, int oneof_index, |
1918 | const LocationRecorder& oneof_location, |
1919 | const LocationRecorder& containing_type_location, |
1920 | const FileDescriptorProto* containing_file) { |
1921 | DO(Consume("oneof" )); |
1922 | |
1923 | { |
1924 | LocationRecorder name_location(oneof_location, |
1925 | OneofDescriptorProto::kNameFieldNumber); |
1926 | DO(ConsumeIdentifier(oneof_decl->mutable_name(), "Expected oneof name." )); |
1927 | } |
1928 | |
1929 | DO(ConsumeEndOfDeclaration("{" , &oneof_location)); |
1930 | |
1931 | do { |
1932 | if (AtEnd()) { |
1933 | AddError(error: "Reached end of input in oneof definition (missing '}')." ); |
1934 | return false; |
1935 | } |
1936 | |
1937 | if (LookingAt(text: "option" )) { |
1938 | LocationRecorder option_location( |
1939 | oneof_location, OneofDescriptorProto::kOptionsFieldNumber); |
1940 | if (!ParseOption(options: oneof_decl->mutable_options(), options_location: option_location, |
1941 | containing_file, style: OPTION_STATEMENT)) { |
1942 | return false; |
1943 | } |
1944 | continue; |
1945 | } |
1946 | |
1947 | // Print a nice error if the user accidentally tries to place a label |
1948 | // on an individual member of a oneof. |
1949 | if (LookingAt(text: "required" ) || LookingAt(text: "optional" ) || |
1950 | LookingAt(text: "repeated" )) { |
1951 | AddError( |
1952 | error: "Fields in oneofs must not have labels (required / optional " |
1953 | "/ repeated)." ); |
1954 | // We can continue parsing here because we understand what the user |
1955 | // meant. The error report will still make parsing fail overall. |
1956 | input_->Next(); |
1957 | } |
1958 | |
1959 | LocationRecorder field_location(containing_type_location, |
1960 | DescriptorProto::kFieldFieldNumber, |
1961 | containing_type->field_size()); |
1962 | |
1963 | FieldDescriptorProto* field = containing_type->add_field(); |
1964 | field->set_label(FieldDescriptorProto::LABEL_OPTIONAL); |
1965 | field->set_oneof_index(oneof_index); |
1966 | |
1967 | if (!ParseMessageFieldNoLabel(field, messages: containing_type->mutable_nested_type(), |
1968 | parent_location: containing_type_location, |
1969 | location_field_number_for_nested_type: DescriptorProto::kNestedTypeFieldNumber, |
1970 | field_location, containing_file)) { |
1971 | // This statement failed to parse. Skip it, but keep looping to parse |
1972 | // other statements. |
1973 | SkipStatement(); |
1974 | } |
1975 | } while (!TryConsumeEndOfDeclaration(text: "}" , location: nullptr)); |
1976 | |
1977 | return true; |
1978 | } |
1979 | |
1980 | // ------------------------------------------------------------------- |
1981 | // Enums |
1982 | |
1983 | bool Parser::ParseEnumDefinition(EnumDescriptorProto* enum_type, |
1984 | const LocationRecorder& enum_location, |
1985 | const FileDescriptorProto* containing_file) { |
1986 | DO(Consume("enum" )); |
1987 | |
1988 | { |
1989 | LocationRecorder location(enum_location, |
1990 | EnumDescriptorProto::kNameFieldNumber); |
1991 | location.RecordLegacyLocation(descriptor: enum_type, |
1992 | location: DescriptorPool::ErrorCollector::NAME); |
1993 | DO(ConsumeIdentifier(enum_type->mutable_name(), "Expected enum name." )); |
1994 | } |
1995 | |
1996 | DO(ParseEnumBlock(enum_type, enum_location, containing_file)); |
1997 | |
1998 | DO(ValidateEnum(enum_type)); |
1999 | |
2000 | return true; |
2001 | } |
2002 | |
2003 | bool Parser::ParseEnumBlock(EnumDescriptorProto* enum_type, |
2004 | const LocationRecorder& enum_location, |
2005 | const FileDescriptorProto* containing_file) { |
2006 | DO(ConsumeEndOfDeclaration("{" , &enum_location)); |
2007 | |
2008 | while (!TryConsumeEndOfDeclaration(text: "}" , location: nullptr)) { |
2009 | if (AtEnd()) { |
2010 | AddError(error: "Reached end of input in enum definition (missing '}')." ); |
2011 | return false; |
2012 | } |
2013 | |
2014 | if (!ParseEnumStatement(message: enum_type, enum_location, containing_file)) { |
2015 | // This statement failed to parse. Skip it, but keep looping to parse |
2016 | // other statements. |
2017 | SkipStatement(); |
2018 | } |
2019 | } |
2020 | |
2021 | return true; |
2022 | } |
2023 | |
2024 | bool Parser::ParseEnumStatement(EnumDescriptorProto* enum_type, |
2025 | const LocationRecorder& enum_location, |
2026 | const FileDescriptorProto* containing_file) { |
2027 | if (TryConsumeEndOfDeclaration(text: ";" , location: nullptr)) { |
2028 | // empty statement; ignore |
2029 | return true; |
2030 | } else if (LookingAt(text: "option" )) { |
2031 | LocationRecorder location(enum_location, |
2032 | EnumDescriptorProto::kOptionsFieldNumber); |
2033 | return ParseOption(options: enum_type->mutable_options(), options_location: location, containing_file, |
2034 | style: OPTION_STATEMENT); |
2035 | } else if (LookingAt(text: "reserved" )) { |
2036 | return ParseReserved(message: enum_type, message_location: enum_location); |
2037 | } else { |
2038 | LocationRecorder location(enum_location, |
2039 | EnumDescriptorProto::kValueFieldNumber, |
2040 | enum_type->value_size()); |
2041 | return ParseEnumConstant(enum_value: enum_type->add_value(), enum_value_location: location, containing_file); |
2042 | } |
2043 | } |
2044 | |
2045 | bool Parser::ParseEnumConstant(EnumValueDescriptorProto* enum_value, |
2046 | const LocationRecorder& enum_value_location, |
2047 | const FileDescriptorProto* containing_file) { |
2048 | // Parse name. |
2049 | { |
2050 | LocationRecorder location(enum_value_location, |
2051 | EnumValueDescriptorProto::kNameFieldNumber); |
2052 | location.RecordLegacyLocation(descriptor: enum_value, |
2053 | location: DescriptorPool::ErrorCollector::NAME); |
2054 | DO(ConsumeIdentifier(enum_value->mutable_name(), |
2055 | "Expected enum constant name." )); |
2056 | } |
2057 | |
2058 | DO(Consume("=" , "Missing numeric value for enum constant." )); |
2059 | |
2060 | // Parse value. |
2061 | { |
2062 | LocationRecorder location(enum_value_location, |
2063 | EnumValueDescriptorProto::kNumberFieldNumber); |
2064 | location.RecordLegacyLocation(descriptor: enum_value, |
2065 | location: DescriptorPool::ErrorCollector::NUMBER); |
2066 | |
2067 | int number; |
2068 | DO(ConsumeSignedInteger(&number, "Expected integer." )); |
2069 | enum_value->set_number(number); |
2070 | } |
2071 | |
2072 | DO(ParseEnumConstantOptions(enum_value, enum_value_location, |
2073 | containing_file)); |
2074 | |
2075 | DO(ConsumeEndOfDeclaration(";" , &enum_value_location)); |
2076 | |
2077 | return true; |
2078 | } |
2079 | |
2080 | bool Parser::ParseEnumConstantOptions( |
2081 | EnumValueDescriptorProto* value, |
2082 | const LocationRecorder& enum_value_location, |
2083 | const FileDescriptorProto* containing_file) { |
2084 | if (!LookingAt(text: "[" )) return true; |
2085 | |
2086 | LocationRecorder location(enum_value_location, |
2087 | EnumValueDescriptorProto::kOptionsFieldNumber); |
2088 | |
2089 | DO(Consume("[" )); |
2090 | |
2091 | do { |
2092 | DO(ParseOption(value->mutable_options(), location, containing_file, |
2093 | OPTION_ASSIGNMENT)); |
2094 | } while (TryConsume(text: "," )); |
2095 | |
2096 | DO(Consume("]" )); |
2097 | return true; |
2098 | } |
2099 | |
2100 | // ------------------------------------------------------------------- |
2101 | // Services |
2102 | |
2103 | bool Parser::ParseServiceDefinition( |
2104 | ServiceDescriptorProto* service, const LocationRecorder& service_location, |
2105 | const FileDescriptorProto* containing_file) { |
2106 | DO(Consume("service" )); |
2107 | |
2108 | { |
2109 | LocationRecorder location(service_location, |
2110 | ServiceDescriptorProto::kNameFieldNumber); |
2111 | location.RecordLegacyLocation(descriptor: service, |
2112 | location: DescriptorPool::ErrorCollector::NAME); |
2113 | DO(ConsumeIdentifier(service->mutable_name(), "Expected service name." )); |
2114 | } |
2115 | |
2116 | DO(ParseServiceBlock(service, service_location, containing_file)); |
2117 | return true; |
2118 | } |
2119 | |
2120 | bool Parser::ParseServiceBlock(ServiceDescriptorProto* service, |
2121 | const LocationRecorder& service_location, |
2122 | const FileDescriptorProto* containing_file) { |
2123 | DO(ConsumeEndOfDeclaration("{" , &service_location)); |
2124 | |
2125 | while (!TryConsumeEndOfDeclaration(text: "}" , location: nullptr)) { |
2126 | if (AtEnd()) { |
2127 | AddError(error: "Reached end of input in service definition (missing '}')." ); |
2128 | return false; |
2129 | } |
2130 | |
2131 | if (!ParseServiceStatement(message: service, service_location, containing_file)) { |
2132 | // This statement failed to parse. Skip it, but keep looping to parse |
2133 | // other statements. |
2134 | SkipStatement(); |
2135 | } |
2136 | } |
2137 | |
2138 | return true; |
2139 | } |
2140 | |
2141 | bool Parser::ParseServiceStatement(ServiceDescriptorProto* service, |
2142 | const LocationRecorder& service_location, |
2143 | const FileDescriptorProto* containing_file) { |
2144 | if (TryConsumeEndOfDeclaration(text: ";" , location: nullptr)) { |
2145 | // empty statement; ignore |
2146 | return true; |
2147 | } else if (LookingAt(text: "option" )) { |
2148 | LocationRecorder location(service_location, |
2149 | ServiceDescriptorProto::kOptionsFieldNumber); |
2150 | return ParseOption(options: service->mutable_options(), options_location: location, containing_file, |
2151 | style: OPTION_STATEMENT); |
2152 | } else { |
2153 | LocationRecorder location(service_location, |
2154 | ServiceDescriptorProto::kMethodFieldNumber, |
2155 | service->method_size()); |
2156 | return ParseServiceMethod(method: service->add_method(), method_location: location, containing_file); |
2157 | } |
2158 | } |
2159 | |
2160 | bool Parser::ParseServiceMethod(MethodDescriptorProto* method, |
2161 | const LocationRecorder& method_location, |
2162 | const FileDescriptorProto* containing_file) { |
2163 | DO(Consume("rpc" )); |
2164 | |
2165 | { |
2166 | LocationRecorder location(method_location, |
2167 | MethodDescriptorProto::kNameFieldNumber); |
2168 | location.RecordLegacyLocation(descriptor: method, location: DescriptorPool::ErrorCollector::NAME); |
2169 | DO(ConsumeIdentifier(method->mutable_name(), "Expected method name." )); |
2170 | } |
2171 | |
2172 | // Parse input type. |
2173 | DO(Consume("(" )); |
2174 | { |
2175 | if (LookingAt(text: "stream" )) { |
2176 | LocationRecorder location( |
2177 | method_location, MethodDescriptorProto::kClientStreamingFieldNumber); |
2178 | location.RecordLegacyLocation(descriptor: method, |
2179 | location: DescriptorPool::ErrorCollector::OTHER); |
2180 | method->set_client_streaming(true); |
2181 | DO(Consume("stream" )); |
2182 | } |
2183 | LocationRecorder location(method_location, |
2184 | MethodDescriptorProto::kInputTypeFieldNumber); |
2185 | location.RecordLegacyLocation(descriptor: method, |
2186 | location: DescriptorPool::ErrorCollector::INPUT_TYPE); |
2187 | DO(ParseUserDefinedType(method->mutable_input_type())); |
2188 | } |
2189 | DO(Consume(")" )); |
2190 | |
2191 | // Parse output type. |
2192 | DO(Consume("returns" )); |
2193 | DO(Consume("(" )); |
2194 | { |
2195 | if (LookingAt(text: "stream" )) { |
2196 | LocationRecorder location( |
2197 | method_location, MethodDescriptorProto::kServerStreamingFieldNumber); |
2198 | location.RecordLegacyLocation(descriptor: method, |
2199 | location: DescriptorPool::ErrorCollector::OTHER); |
2200 | DO(Consume("stream" )); |
2201 | method->set_server_streaming(true); |
2202 | } |
2203 | LocationRecorder location(method_location, |
2204 | MethodDescriptorProto::kOutputTypeFieldNumber); |
2205 | location.RecordLegacyLocation(descriptor: method, |
2206 | location: DescriptorPool::ErrorCollector::OUTPUT_TYPE); |
2207 | DO(ParseUserDefinedType(method->mutable_output_type())); |
2208 | } |
2209 | DO(Consume(")" )); |
2210 | |
2211 | if (LookingAt(text: "{" )) { |
2212 | // Options! |
2213 | DO(ParseMethodOptions(method_location, containing_file, |
2214 | MethodDescriptorProto::kOptionsFieldNumber, |
2215 | method->mutable_options())); |
2216 | } else { |
2217 | DO(ConsumeEndOfDeclaration(";" , &method_location)); |
2218 | } |
2219 | |
2220 | return true; |
2221 | } |
2222 | |
2223 | bool Parser::ParseMethodOptions(const LocationRecorder& parent_location, |
2224 | const FileDescriptorProto* containing_file, |
2225 | const int optionsFieldNumber, |
2226 | Message* mutable_options) { |
2227 | // Options! |
2228 | ConsumeEndOfDeclaration(text: "{" , location: &parent_location); |
2229 | while (!TryConsumeEndOfDeclaration(text: "}" , location: nullptr)) { |
2230 | if (AtEnd()) { |
2231 | AddError(error: "Reached end of input in method options (missing '}')." ); |
2232 | return false; |
2233 | } |
2234 | |
2235 | if (TryConsumeEndOfDeclaration(text: ";" , location: nullptr)) { |
2236 | // empty statement; ignore |
2237 | } else { |
2238 | LocationRecorder location(parent_location, optionsFieldNumber); |
2239 | if (!ParseOption(options: mutable_options, options_location: location, containing_file, |
2240 | style: OPTION_STATEMENT)) { |
2241 | // This statement failed to parse. Skip it, but keep looping to |
2242 | // parse other statements. |
2243 | SkipStatement(); |
2244 | } |
2245 | } |
2246 | } |
2247 | |
2248 | return true; |
2249 | } |
2250 | |
2251 | // ------------------------------------------------------------------- |
2252 | |
2253 | bool Parser::ParseLabel(FieldDescriptorProto::Label* label, |
2254 | const LocationRecorder& field_location) { |
2255 | if (!LookingAt(text: "optional" ) && !LookingAt(text: "repeated" ) && |
2256 | !LookingAt(text: "required" )) { |
2257 | return false; |
2258 | } |
2259 | LocationRecorder location(field_location, |
2260 | FieldDescriptorProto::kLabelFieldNumber); |
2261 | if (TryConsume(text: "optional" )) { |
2262 | *label = FieldDescriptorProto::LABEL_OPTIONAL; |
2263 | } else if (TryConsume(text: "repeated" )) { |
2264 | *label = FieldDescriptorProto::LABEL_REPEATED; |
2265 | } else { |
2266 | Consume(text: "required" ); |
2267 | *label = FieldDescriptorProto::LABEL_REQUIRED; |
2268 | } |
2269 | return true; |
2270 | } |
2271 | |
2272 | bool Parser::ParseType(FieldDescriptorProto::Type* type, |
2273 | std::string* type_name) { |
2274 | const auto& type_names_table = GetTypeNameTable(); |
2275 | auto iter = type_names_table.find(x: input_->current().text); |
2276 | if (iter != type_names_table.end()) { |
2277 | *type = iter->second; |
2278 | input_->Next(); |
2279 | } else { |
2280 | DO(ParseUserDefinedType(type_name)); |
2281 | } |
2282 | return true; |
2283 | } |
2284 | |
2285 | bool Parser::ParseUserDefinedType(std::string* type_name) { |
2286 | type_name->clear(); |
2287 | |
2288 | const auto& type_names_table = GetTypeNameTable(); |
2289 | auto iter = type_names_table.find(x: input_->current().text); |
2290 | if (iter != type_names_table.end()) { |
2291 | // Note: The only place enum types are allowed is for field types, but |
2292 | // if we are parsing a field type then we would not get here because |
2293 | // primitives are allowed there as well. So this error message doesn't |
2294 | // need to account for enums. |
2295 | AddError(error: "Expected message type." ); |
2296 | |
2297 | // Pretend to accept this type so that we can go on parsing. |
2298 | *type_name = input_->current().text; |
2299 | input_->Next(); |
2300 | return true; |
2301 | } |
2302 | |
2303 | // A leading "." means the name is fully-qualified. |
2304 | if (TryConsume(text: "." )) type_name->append(s: "." ); |
2305 | |
2306 | // Consume the first part of the name. |
2307 | std::string identifier; |
2308 | DO(ConsumeIdentifier(&identifier, "Expected type name." )); |
2309 | type_name->append(str: identifier); |
2310 | |
2311 | // Consume more parts. |
2312 | while (TryConsume(text: "." )) { |
2313 | type_name->append(s: "." ); |
2314 | DO(ConsumeIdentifier(&identifier, "Expected identifier." )); |
2315 | type_name->append(str: identifier); |
2316 | } |
2317 | |
2318 | return true; |
2319 | } |
2320 | |
2321 | // =================================================================== |
2322 | |
2323 | bool Parser::ParsePackage(FileDescriptorProto* file, |
2324 | const LocationRecorder& root_location, |
2325 | const FileDescriptorProto* containing_file) { |
2326 | if (file->has_package()) { |
2327 | AddError(error: "Multiple package definitions." ); |
2328 | // Don't append the new package to the old one. Just replace it. Not |
2329 | // that it really matters since this is an error anyway. |
2330 | file->clear_package(); |
2331 | } |
2332 | |
2333 | LocationRecorder location(root_location, |
2334 | FileDescriptorProto::kPackageFieldNumber); |
2335 | location.RecordLegacyLocation(descriptor: file, location: DescriptorPool::ErrorCollector::NAME); |
2336 | |
2337 | DO(Consume("package" )); |
2338 | |
2339 | while (true) { |
2340 | std::string identifier; |
2341 | DO(ConsumeIdentifier(&identifier, "Expected identifier." )); |
2342 | file->mutable_package()->append(str: identifier); |
2343 | if (!TryConsume(text: "." )) break; |
2344 | file->mutable_package()->append(s: "." ); |
2345 | } |
2346 | |
2347 | DO(ConsumeEndOfDeclaration(";" , &location)); |
2348 | |
2349 | return true; |
2350 | } |
2351 | |
2352 | bool Parser::ParseImport(RepeatedPtrField<std::string>* dependency, |
2353 | RepeatedField<int32_t>* public_dependency, |
2354 | RepeatedField<int32_t>* weak_dependency, |
2355 | const LocationRecorder& root_location, |
2356 | const FileDescriptorProto* containing_file) { |
2357 | LocationRecorder location(root_location, |
2358 | FileDescriptorProto::kDependencyFieldNumber, |
2359 | dependency->size()); |
2360 | |
2361 | DO(Consume("import" )); |
2362 | |
2363 | if (LookingAt(text: "public" )) { |
2364 | LocationRecorder public_location( |
2365 | root_location, FileDescriptorProto::kPublicDependencyFieldNumber, |
2366 | public_dependency->size()); |
2367 | DO(Consume("public" )); |
2368 | *public_dependency->Add() = dependency->size(); |
2369 | } else if (LookingAt(text: "weak" )) { |
2370 | LocationRecorder weak_location( |
2371 | root_location, FileDescriptorProto::kWeakDependencyFieldNumber, |
2372 | weak_dependency->size()); |
2373 | weak_location.RecordLegacyImportLocation(descriptor: containing_file, name: "weak" ); |
2374 | DO(Consume("weak" )); |
2375 | *weak_dependency->Add() = dependency->size(); |
2376 | } |
2377 | |
2378 | std::string import_file; |
2379 | DO(ConsumeString(&import_file, |
2380 | "Expected a string naming the file to import." )); |
2381 | *dependency->Add() = import_file; |
2382 | location.RecordLegacyImportLocation(descriptor: containing_file, name: import_file); |
2383 | |
2384 | DO(ConsumeEndOfDeclaration(";" , &location)); |
2385 | |
2386 | return true; |
2387 | } |
2388 | |
2389 | // =================================================================== |
2390 | |
2391 | SourceLocationTable::SourceLocationTable() {} |
2392 | SourceLocationTable::~SourceLocationTable() {} |
2393 | |
2394 | bool SourceLocationTable::Find( |
2395 | const Message* descriptor, |
2396 | DescriptorPool::ErrorCollector::ErrorLocation location, int* line, |
2397 | int* column) const { |
2398 | const std::pair<int, int>* result = |
2399 | FindOrNull(collection: location_map_, key: std::make_pair(x&: descriptor, y&: location)); |
2400 | if (result == nullptr) { |
2401 | *line = -1; |
2402 | *column = 0; |
2403 | return false; |
2404 | } else { |
2405 | *line = result->first; |
2406 | *column = result->second; |
2407 | return true; |
2408 | } |
2409 | } |
2410 | |
2411 | bool SourceLocationTable::FindImport(const Message* descriptor, |
2412 | const std::string& name, int* line, |
2413 | int* column) const { |
2414 | const std::pair<int, int>* result = |
2415 | FindOrNull(collection: import_location_map_, key: std::make_pair(x&: descriptor, y: name)); |
2416 | if (result == nullptr) { |
2417 | *line = -1; |
2418 | *column = 0; |
2419 | return false; |
2420 | } else { |
2421 | *line = result->first; |
2422 | *column = result->second; |
2423 | return true; |
2424 | } |
2425 | } |
2426 | |
2427 | void SourceLocationTable::Add( |
2428 | const Message* descriptor, |
2429 | DescriptorPool::ErrorCollector::ErrorLocation location, int line, |
2430 | int column) { |
2431 | location_map_[std::make_pair(x&: descriptor, y&: location)] = |
2432 | std::make_pair(x&: line, y&: column); |
2433 | } |
2434 | |
2435 | void SourceLocationTable::AddImport(const Message* descriptor, |
2436 | const std::string& name, int line, |
2437 | int column) { |
2438 | import_location_map_[std::make_pair(x&: descriptor, y: name)] = |
2439 | std::make_pair(x&: line, y&: column); |
2440 | } |
2441 | |
2442 | void SourceLocationTable::Clear() { location_map_.clear(); } |
2443 | |
2444 | } // namespace compiler |
2445 | } // namespace protobuf |
2446 | } // namespace google |
2447 | |