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: jschorr@google.com (Joseph Schorr)
32 // Based on original Protocol Buffers design by
33 // Sanjay Ghemawat, Jeff Dean, and others.
34
35 #include <google/protobuf/text_format.h>
36
37 #include <float.h>
38 #include <math.h>
39 #include <stdio.h>
40 #include <algorithm>
41 #include <climits>
42 #include <limits>
43 #include <vector>
44
45 #include <google/protobuf/stubs/stringprintf.h>
46 #include <google/protobuf/any.h>
47 #include <google/protobuf/descriptor.pb.h>
48 #include <google/protobuf/io/strtod.h>
49 #include <google/protobuf/io/coded_stream.h>
50 #include <google/protobuf/io/tokenizer.h>
51 #include <google/protobuf/io/zero_copy_stream.h>
52 #include <google/protobuf/io/zero_copy_stream_impl.h>
53 #include <google/protobuf/descriptor.h>
54 #include <google/protobuf/dynamic_message.h>
55 #include <google/protobuf/map_field.h>
56 #include <google/protobuf/message.h>
57 #include <google/protobuf/repeated_field.h>
58 #include <google/protobuf/unknown_field_set.h>
59 #include <google/protobuf/wire_format_lite.h>
60 #include <google/protobuf/stubs/strutil.h>
61
62
63
64 #include <google/protobuf/stubs/map_util.h>
65 #include <google/protobuf/stubs/stl_util.h>
66
67 #include <google/protobuf/port_def.inc>
68
69
70 namespace google {
71 namespace protobuf {
72
73 namespace {
74
IsHexNumber(const std::string & str)75 inline bool IsHexNumber(const std::string& str) {
76 return (str.length() >= 2 && str[0] == '0' &&
77 (str[1] == 'x' || str[1] == 'X'));
78 }
79
IsOctNumber(const std::string & str)80 inline bool IsOctNumber(const std::string& str) {
81 return (str.length() >= 2 && str[0] == '0' &&
82 (str[1] >= '0' && str[1] < '8'));
83 }
84
85 } // namespace
86
DebugString() const87 std::string Message::DebugString() const {
88 std::string debug_string;
89
90 TextFormat::Printer printer;
91 printer.SetExpandAny(true);
92
93 printer.PrintToString(*this, &debug_string);
94
95 return debug_string;
96 }
97
ShortDebugString() const98 std::string Message::ShortDebugString() const {
99 std::string debug_string;
100
101 TextFormat::Printer printer;
102 printer.SetSingleLineMode(true);
103 printer.SetExpandAny(true);
104
105 printer.PrintToString(*this, &debug_string);
106 // Single line mode currently might have an extra space at the end.
107 if (debug_string.size() > 0 && debug_string[debug_string.size() - 1] == ' ') {
108 debug_string.resize(debug_string.size() - 1);
109 }
110
111 return debug_string;
112 }
113
Utf8DebugString() const114 std::string Message::Utf8DebugString() const {
115 std::string debug_string;
116
117 TextFormat::Printer printer;
118 printer.SetUseUtf8StringEscaping(true);
119 printer.SetExpandAny(true);
120
121 printer.PrintToString(*this, &debug_string);
122
123 return debug_string;
124 }
125
PrintDebugString() const126 void Message::PrintDebugString() const { printf("%s", DebugString().c_str()); }
127
128
129 // ===========================================================================
130 // Implementation of the parse information tree class.
ParseInfoTree()131 TextFormat::ParseInfoTree::ParseInfoTree() {}
132
~ParseInfoTree()133 TextFormat::ParseInfoTree::~ParseInfoTree() {
134 // Remove any nested information trees, as they are owned by this tree.
135 for (NestedMap::iterator it = nested_.begin(); it != nested_.end(); ++it) {
136 STLDeleteElements(&(it->second));
137 }
138 }
139
RecordLocation(const FieldDescriptor * field,TextFormat::ParseLocation location)140 void TextFormat::ParseInfoTree::RecordLocation(
141 const FieldDescriptor* field, TextFormat::ParseLocation location) {
142 locations_[field].push_back(location);
143 }
144
CreateNested(const FieldDescriptor * field)145 TextFormat::ParseInfoTree* TextFormat::ParseInfoTree::CreateNested(
146 const FieldDescriptor* field) {
147 // Owned by us in the map.
148 TextFormat::ParseInfoTree* instance = new TextFormat::ParseInfoTree();
149 std::vector<TextFormat::ParseInfoTree*>* trees = &nested_[field];
150 GOOGLE_CHECK(trees);
151 trees->push_back(instance);
152 return instance;
153 }
154
CheckFieldIndex(const FieldDescriptor * field,int index)155 void CheckFieldIndex(const FieldDescriptor* field, int index) {
156 if (field == nullptr) {
157 return;
158 }
159
160 if (field->is_repeated() && index == -1) {
161 GOOGLE_LOG(DFATAL) << "Index must be in range of repeated field values. "
162 << "Field: " << field->name();
163 } else if (!field->is_repeated() && index != -1) {
164 GOOGLE_LOG(DFATAL) << "Index must be -1 for singular fields."
165 << "Field: " << field->name();
166 }
167 }
168
GetLocation(const FieldDescriptor * field,int index) const169 TextFormat::ParseLocation TextFormat::ParseInfoTree::GetLocation(
170 const FieldDescriptor* field, int index) const {
171 CheckFieldIndex(field, index);
172 if (index == -1) {
173 index = 0;
174 }
175
176 const std::vector<TextFormat::ParseLocation>* locations =
177 FindOrNull(locations_, field);
178 if (locations == nullptr || index >= locations->size()) {
179 return TextFormat::ParseLocation();
180 }
181
182 return (*locations)[index];
183 }
184
GetTreeForNested(const FieldDescriptor * field,int index) const185 TextFormat::ParseInfoTree* TextFormat::ParseInfoTree::GetTreeForNested(
186 const FieldDescriptor* field, int index) const {
187 CheckFieldIndex(field, index);
188 if (index == -1) {
189 index = 0;
190 }
191
192 const std::vector<TextFormat::ParseInfoTree*>* trees =
193 FindOrNull(nested_, field);
194 if (trees == nullptr || index >= trees->size()) {
195 return nullptr;
196 }
197
198 return (*trees)[index];
199 }
200
201 namespace {
202 // These functions implement the behavior of the "default" TextFormat::Finder,
203 // they are defined as standalone to be called when finder_ is nullptr.
DefaultFinderFindExtension(Message * message,const std::string & name)204 const FieldDescriptor* DefaultFinderFindExtension(Message* message,
205 const std::string& name) {
206 const Descriptor* descriptor = message->GetDescriptor();
207 return descriptor->file()->pool()->FindExtensionByPrintableName(descriptor,
208 name);
209 }
210
DefaultFinderFindExtensionByNumber(const Descriptor * descriptor,int number)211 const FieldDescriptor* DefaultFinderFindExtensionByNumber(
212 const Descriptor* descriptor, int number) {
213 return descriptor->file()->pool()->FindExtensionByNumber(descriptor, number);
214 }
215
DefaultFinderFindAnyType(const Message & message,const std::string & prefix,const std::string & name)216 const Descriptor* DefaultFinderFindAnyType(const Message& message,
217 const std::string& prefix,
218 const std::string& name) {
219 if (prefix != internal::kTypeGoogleApisComPrefix &&
220 prefix != internal::kTypeGoogleProdComPrefix) {
221 return nullptr;
222 }
223 return message.GetDescriptor()->file()->pool()->FindMessageTypeByName(name);
224 }
225 } // namespace
226
227 // ===========================================================================
228 // Internal class for parsing an ASCII representation of a Protocol Message.
229 // This class makes use of the Protocol Message compiler's tokenizer found
230 // in //net/proto2/io/public/tokenizer.h. Note that class's Parse
231 // method is *not* thread-safe and should only be used in a single thread at
232 // a time.
233
234 // Makes code slightly more readable. The meaning of "DO(foo)" is
235 // "Execute foo and fail if it fails.", where failure is indicated by
236 // returning false. Borrowed from parser.cc (Thanks Kenton!).
237 #define DO(STATEMENT) \
238 if (STATEMENT) { \
239 } else { \
240 return false; \
241 }
242
243 class TextFormat::Parser::ParserImpl {
244 public:
245 // Determines if repeated values for non-repeated fields and
246 // oneofs are permitted, e.g., the string "foo: 1 foo: 2" for a
247 // required/optional field named "foo", or "baz: 1 qux: 2"
248 // where "baz" and "qux" are members of the same oneof.
249 enum SingularOverwritePolicy {
250 ALLOW_SINGULAR_OVERWRITES = 0, // the last value is retained
251 FORBID_SINGULAR_OVERWRITES = 1, // an error is issued
252 };
253
ParserImpl(const Descriptor * root_message_type,io::ZeroCopyInputStream * input_stream,io::ErrorCollector * error_collector,const TextFormat::Finder * finder,ParseInfoTree * parse_info_tree,SingularOverwritePolicy singular_overwrite_policy,bool allow_case_insensitive_field,bool allow_unknown_field,bool allow_unknown_extension,bool allow_unknown_enum,bool allow_field_number,bool allow_relaxed_whitespace,bool allow_partial,int recursion_limit)254 ParserImpl(const Descriptor* root_message_type,
255 io::ZeroCopyInputStream* input_stream,
256 io::ErrorCollector* error_collector,
257 const TextFormat::Finder* finder, ParseInfoTree* parse_info_tree,
258 SingularOverwritePolicy singular_overwrite_policy,
259 bool allow_case_insensitive_field, bool allow_unknown_field,
260 bool allow_unknown_extension, bool allow_unknown_enum,
261 bool allow_field_number, bool allow_relaxed_whitespace,
262 bool allow_partial, int recursion_limit)
263 : error_collector_(error_collector),
264 finder_(finder),
265 parse_info_tree_(parse_info_tree),
266 tokenizer_error_collector_(this),
267 tokenizer_(input_stream, &tokenizer_error_collector_),
268 root_message_type_(root_message_type),
269 singular_overwrite_policy_(singular_overwrite_policy),
270 allow_case_insensitive_field_(allow_case_insensitive_field),
271 allow_unknown_field_(allow_unknown_field),
272 allow_unknown_extension_(allow_unknown_extension),
273 allow_unknown_enum_(allow_unknown_enum),
274 allow_field_number_(allow_field_number),
275 allow_partial_(allow_partial),
276 recursion_limit_(recursion_limit),
277 had_errors_(false) {
278 // For backwards-compatibility with proto1, we need to allow the 'f' suffix
279 // for floats.
280 tokenizer_.set_allow_f_after_float(true);
281
282 // '#' starts a comment.
283 tokenizer_.set_comment_style(io::Tokenizer::SH_COMMENT_STYLE);
284
285 if (allow_relaxed_whitespace) {
286 tokenizer_.set_require_space_after_number(false);
287 tokenizer_.set_allow_multiline_strings(true);
288 }
289
290 // Consume the starting token.
291 tokenizer_.Next();
292 }
~ParserImpl()293 ~ParserImpl() {}
294
295 // Parses the ASCII representation specified in input and saves the
296 // information into the output pointer (a Message). Returns
297 // false if an error occurs (an error will also be logged to
298 // GOOGLE_LOG(ERROR)).
Parse(Message * output)299 bool Parse(Message* output) {
300 // Consume fields until we cannot do so anymore.
301 while (true) {
302 if (LookingAtType(io::Tokenizer::TYPE_END)) {
303 return !had_errors_;
304 }
305
306 DO(ConsumeField(output));
307 }
308 }
309
ParseField(const FieldDescriptor * field,Message * output)310 bool ParseField(const FieldDescriptor* field, Message* output) {
311 bool suc;
312 if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
313 suc = ConsumeFieldMessage(output, output->GetReflection(), field);
314 } else {
315 suc = ConsumeFieldValue(output, output->GetReflection(), field);
316 }
317 return suc && LookingAtType(io::Tokenizer::TYPE_END);
318 }
319
ReportError(int line,int col,const std::string & message)320 void ReportError(int line, int col, const std::string& message) {
321 had_errors_ = true;
322 if (error_collector_ == nullptr) {
323 if (line >= 0) {
324 GOOGLE_LOG(ERROR) << "Error parsing text-format "
325 << root_message_type_->full_name() << ": " << (line + 1)
326 << ":" << (col + 1) << ": " << message;
327 } else {
328 GOOGLE_LOG(ERROR) << "Error parsing text-format "
329 << root_message_type_->full_name() << ": " << message;
330 }
331 } else {
332 error_collector_->AddError(line, col, message);
333 }
334 }
335
ReportWarning(int line,int col,const std::string & message)336 void ReportWarning(int line, int col, const std::string& message) {
337 if (error_collector_ == nullptr) {
338 if (line >= 0) {
339 GOOGLE_LOG(WARNING) << "Warning parsing text-format "
340 << root_message_type_->full_name() << ": " << (line + 1)
341 << ":" << (col + 1) << ": " << message;
342 } else {
343 GOOGLE_LOG(WARNING) << "Warning parsing text-format "
344 << root_message_type_->full_name() << ": " << message;
345 }
346 } else {
347 error_collector_->AddWarning(line, col, message);
348 }
349 }
350
351 private:
352 GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(ParserImpl);
353
354 // Reports an error with the given message with information indicating
355 // the position (as derived from the current token).
ReportError(const std::string & message)356 void ReportError(const std::string& message) {
357 ReportError(tokenizer_.current().line, tokenizer_.current().column,
358 message);
359 }
360
361 // Reports a warning with the given message with information indicating
362 // the position (as derived from the current token).
ReportWarning(const std::string & message)363 void ReportWarning(const std::string& message) {
364 ReportWarning(tokenizer_.current().line, tokenizer_.current().column,
365 message);
366 }
367
368 // Consumes the specified message with the given starting delimiter.
369 // This method checks to see that the end delimiter at the conclusion of
370 // the consumption matches the starting delimiter passed in here.
ConsumeMessage(Message * message,const std::string delimiter)371 bool ConsumeMessage(Message* message, const std::string delimiter) {
372 while (!LookingAt(">") && !LookingAt("}")) {
373 DO(ConsumeField(message));
374 }
375
376 // Confirm that we have a valid ending delimiter.
377 DO(Consume(delimiter));
378 return true;
379 }
380
381 // Consume either "<" or "{".
ConsumeMessageDelimiter(std::string * delimiter)382 bool ConsumeMessageDelimiter(std::string* delimiter) {
383 if (TryConsume("<")) {
384 *delimiter = ">";
385 } else {
386 DO(Consume("{"));
387 *delimiter = "}";
388 }
389 return true;
390 }
391
392
393 // Consumes the current field (as returned by the tokenizer) on the
394 // passed in message.
ConsumeField(Message * message)395 bool ConsumeField(Message* message) {
396 const Reflection* reflection = message->GetReflection();
397 const Descriptor* descriptor = message->GetDescriptor();
398
399 std::string field_name;
400 bool reserved_field = false;
401 const FieldDescriptor* field = nullptr;
402 int start_line = tokenizer_.current().line;
403 int start_column = tokenizer_.current().column;
404
405 const FieldDescriptor* any_type_url_field;
406 const FieldDescriptor* any_value_field;
407 if (internal::GetAnyFieldDescriptors(*message, &any_type_url_field,
408 &any_value_field) &&
409 TryConsume("[")) {
410 std::string full_type_name, prefix;
411 DO(ConsumeAnyTypeUrl(&full_type_name, &prefix));
412 DO(Consume("]"));
413 TryConsume(":"); // ':' is optional between message labels and values.
414 std::string serialized_value;
415 const Descriptor* value_descriptor =
416 finder_ ? finder_->FindAnyType(*message, prefix, full_type_name)
417 : DefaultFinderFindAnyType(*message, prefix, full_type_name);
418 if (value_descriptor == nullptr) {
419 ReportError("Could not find type \"" + prefix + full_type_name +
420 "\" stored in google.protobuf.Any.");
421 return false;
422 }
423 DO(ConsumeAnyValue(value_descriptor, &serialized_value));
424 if (singular_overwrite_policy_ == FORBID_SINGULAR_OVERWRITES) {
425 // Fail if any_type_url_field has already been specified.
426 if ((!any_type_url_field->is_repeated() &&
427 reflection->HasField(*message, any_type_url_field)) ||
428 (!any_value_field->is_repeated() &&
429 reflection->HasField(*message, any_value_field))) {
430 ReportError("Non-repeated Any specified multiple times.");
431 return false;
432 }
433 }
434 reflection->SetString(message, any_type_url_field,
435 std::string(prefix + full_type_name));
436 reflection->SetString(message, any_value_field, serialized_value);
437 return true;
438 }
439 if (TryConsume("[")) {
440 // Extension.
441 DO(ConsumeFullTypeName(&field_name));
442 DO(Consume("]"));
443
444 field = finder_ ? finder_->FindExtension(message, field_name)
445 : DefaultFinderFindExtension(message, field_name);
446
447 if (field == nullptr) {
448 if (!allow_unknown_field_ && !allow_unknown_extension_) {
449 ReportError("Extension \"" + field_name +
450 "\" is not defined or "
451 "is not an extension of \"" +
452 descriptor->full_name() + "\".");
453 return false;
454 } else {
455 ReportWarning("Ignoring extension \"" + field_name +
456 "\" which is not defined or is not an extension of \"" +
457 descriptor->full_name() + "\".");
458 }
459 }
460 } else {
461 DO(ConsumeIdentifier(&field_name));
462
463 int32 field_number;
464 if (allow_field_number_ &&
465 safe_strto32(field_name, &field_number)) {
466 if (descriptor->IsExtensionNumber(field_number)) {
467 field = finder_
468 ? finder_->FindExtensionByNumber(descriptor, field_number)
469 : DefaultFinderFindExtensionByNumber(descriptor,
470 field_number);
471 } else if (descriptor->IsReservedNumber(field_number)) {
472 reserved_field = true;
473 } else {
474 field = descriptor->FindFieldByNumber(field_number);
475 }
476 } else {
477 field = descriptor->FindFieldByName(field_name);
478 // Group names are expected to be capitalized as they appear in the
479 // .proto file, which actually matches their type names, not their
480 // field names.
481 if (field == nullptr) {
482 std::string lower_field_name = field_name;
483 LowerString(&lower_field_name);
484 field = descriptor->FindFieldByName(lower_field_name);
485 // If the case-insensitive match worked but the field is NOT a group,
486 if (field != nullptr &&
487 field->type() != FieldDescriptor::TYPE_GROUP) {
488 field = nullptr;
489 }
490 }
491 // Again, special-case group names as described above.
492 if (field != nullptr && field->type() == FieldDescriptor::TYPE_GROUP &&
493 field->message_type()->name() != field_name) {
494 field = nullptr;
495 }
496
497 if (field == nullptr && allow_case_insensitive_field_) {
498 std::string lower_field_name = field_name;
499 LowerString(&lower_field_name);
500 field = descriptor->FindFieldByLowercaseName(lower_field_name);
501 }
502
503 if (field == nullptr) {
504 reserved_field = descriptor->IsReservedName(field_name);
505 }
506 }
507
508 if (field == nullptr && !reserved_field) {
509 if (!allow_unknown_field_) {
510 ReportError("Message type \"" + descriptor->full_name() +
511 "\" has no field named \"" + field_name + "\".");
512 return false;
513 } else {
514 ReportWarning("Message type \"" + descriptor->full_name() +
515 "\" has no field named \"" + field_name + "\".");
516 }
517 }
518 }
519
520 // Skips unknown or reserved fields.
521 if (field == nullptr) {
522 GOOGLE_CHECK(allow_unknown_field_ || allow_unknown_extension_ || reserved_field);
523
524 // Try to guess the type of this field.
525 // If this field is not a message, there should be a ":" between the
526 // field name and the field value and also the field value should not
527 // start with "{" or "<" which indicates the beginning of a message body.
528 // If there is no ":" or there is a "{" or "<" after ":", this field has
529 // to be a message or the input is ill-formed.
530 if (TryConsume(":") && !LookingAt("{") && !LookingAt("<")) {
531 return SkipFieldValue();
532 } else {
533 return SkipFieldMessage();
534 }
535 }
536
537 if (singular_overwrite_policy_ == FORBID_SINGULAR_OVERWRITES) {
538 // Fail if the field is not repeated and it has already been specified.
539 if (!field->is_repeated() && reflection->HasField(*message, field)) {
540 ReportError("Non-repeated field \"" + field_name +
541 "\" is specified multiple times.");
542 return false;
543 }
544 // Fail if the field is a member of a oneof and another member has already
545 // been specified.
546 const OneofDescriptor* oneof = field->containing_oneof();
547 if (oneof != nullptr && reflection->HasOneof(*message, oneof)) {
548 const FieldDescriptor* other_field =
549 reflection->GetOneofFieldDescriptor(*message, oneof);
550 ReportError("Field \"" + field_name +
551 "\" is specified along with "
552 "field \"" +
553 other_field->name() +
554 "\", another member "
555 "of oneof \"" +
556 oneof->name() + "\".");
557 return false;
558 }
559 }
560
561 // Perform special handling for embedded message types.
562 if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
563 // ':' is optional here.
564 bool consumed_semicolon = TryConsume(":");
565 if (consumed_semicolon && field->options().weak() &&
566 LookingAtType(io::Tokenizer::TYPE_STRING)) {
567 // we are getting a bytes string for a weak field.
568 std::string tmp;
569 DO(ConsumeString(&tmp));
570 MessageFactory* factory =
571 finder_ ? finder_->FindExtensionFactory(field) : nullptr;
572 reflection->MutableMessage(message, field, factory)
573 ->ParseFromString(tmp);
574 goto label_skip_parsing;
575 }
576 } else {
577 // ':' is required here.
578 DO(Consume(":"));
579 }
580
581 if (field->is_repeated() && TryConsume("[")) {
582 // Short repeated format, e.g. "foo: [1, 2, 3]".
583 if (!TryConsume("]")) {
584 // "foo: []" is treated as empty.
585 while (true) {
586 if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
587 // Perform special handling for embedded message types.
588 DO(ConsumeFieldMessage(message, reflection, field));
589 } else {
590 DO(ConsumeFieldValue(message, reflection, field));
591 }
592 if (TryConsume("]")) {
593 break;
594 }
595 DO(Consume(","));
596 }
597 }
598 } else if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
599 DO(ConsumeFieldMessage(message, reflection, field));
600 } else {
601 DO(ConsumeFieldValue(message, reflection, field));
602 }
603 label_skip_parsing:
604 // For historical reasons, fields may optionally be separated by commas or
605 // semicolons.
606 TryConsume(";") || TryConsume(",");
607
608 if (field->options().deprecated()) {
609 ReportWarning("text format contains deprecated field \"" + field_name +
610 "\"");
611 }
612
613 // If a parse info tree exists, add the location for the parsed
614 // field.
615 if (parse_info_tree_ != nullptr) {
616 RecordLocation(parse_info_tree_, field,
617 ParseLocation(start_line, start_column));
618 }
619
620 return true;
621 }
622
623 // Skips the next field including the field's name and value.
SkipField()624 bool SkipField() {
625 if (TryConsume("[")) {
626 // Extension name or type URL.
627 DO(ConsumeTypeUrlOrFullTypeName());
628 DO(Consume("]"));
629 } else {
630 std::string field_name;
631 DO(ConsumeIdentifier(&field_name));
632 }
633
634 // Try to guess the type of this field.
635 // If this field is not a message, there should be a ":" between the
636 // field name and the field value and also the field value should not
637 // start with "{" or "<" which indicates the beginning of a message body.
638 // If there is no ":" or there is a "{" or "<" after ":", this field has
639 // to be a message or the input is ill-formed.
640 if (TryConsume(":") && !LookingAt("{") && !LookingAt("<")) {
641 DO(SkipFieldValue());
642 } else {
643 DO(SkipFieldMessage());
644 }
645 // For historical reasons, fields may optionally be separated by commas or
646 // semicolons.
647 TryConsume(";") || TryConsume(",");
648 return true;
649 }
650
ConsumeFieldMessage(Message * message,const Reflection * reflection,const FieldDescriptor * field)651 bool ConsumeFieldMessage(Message* message, const Reflection* reflection,
652 const FieldDescriptor* field) {
653 if (--recursion_limit_ < 0) {
654 ReportError("Message is too deep");
655 return false;
656 }
657 // If the parse information tree is not nullptr, create a nested one
658 // for the nested message.
659 ParseInfoTree* parent = parse_info_tree_;
660 if (parent != nullptr) {
661 parse_info_tree_ = CreateNested(parent, field);
662 }
663
664 std::string delimiter;
665 DO(ConsumeMessageDelimiter(&delimiter));
666 MessageFactory* factory =
667 finder_ ? finder_->FindExtensionFactory(field) : nullptr;
668 if (field->is_repeated()) {
669 DO(ConsumeMessage(reflection->AddMessage(message, field, factory),
670 delimiter));
671 } else {
672 DO(ConsumeMessage(reflection->MutableMessage(message, field, factory),
673 delimiter));
674 }
675
676 ++recursion_limit_;
677
678 // Reset the parse information tree.
679 parse_info_tree_ = parent;
680 return true;
681 }
682
683 // Skips the whole body of a message including the beginning delimiter and
684 // the ending delimiter.
SkipFieldMessage()685 bool SkipFieldMessage() {
686 std::string delimiter;
687 DO(ConsumeMessageDelimiter(&delimiter));
688 while (!LookingAt(">") && !LookingAt("}")) {
689 DO(SkipField());
690 }
691 DO(Consume(delimiter));
692 return true;
693 }
694
ConsumeFieldValue(Message * message,const Reflection * reflection,const FieldDescriptor * field)695 bool ConsumeFieldValue(Message* message, const Reflection* reflection,
696 const FieldDescriptor* field) {
697 // Define an easy to use macro for setting fields. This macro checks
698 // to see if the field is repeated (in which case we need to use the Add
699 // methods or not (in which case we need to use the Set methods).
700 #define SET_FIELD(CPPTYPE, VALUE) \
701 if (field->is_repeated()) { \
702 reflection->Add##CPPTYPE(message, field, VALUE); \
703 } else { \
704 reflection->Set##CPPTYPE(message, field, VALUE); \
705 }
706
707 switch (field->cpp_type()) {
708 case FieldDescriptor::CPPTYPE_INT32: {
709 int64 value;
710 DO(ConsumeSignedInteger(&value, kint32max));
711 SET_FIELD(Int32, static_cast<int32>(value));
712 break;
713 }
714
715 case FieldDescriptor::CPPTYPE_UINT32: {
716 uint64 value;
717 DO(ConsumeUnsignedInteger(&value, kuint32max));
718 SET_FIELD(UInt32, static_cast<uint32>(value));
719 break;
720 }
721
722 case FieldDescriptor::CPPTYPE_INT64: {
723 int64 value;
724 DO(ConsumeSignedInteger(&value, kint64max));
725 SET_FIELD(Int64, value);
726 break;
727 }
728
729 case FieldDescriptor::CPPTYPE_UINT64: {
730 uint64 value;
731 DO(ConsumeUnsignedInteger(&value, kuint64max));
732 SET_FIELD(UInt64, value);
733 break;
734 }
735
736 case FieldDescriptor::CPPTYPE_FLOAT: {
737 double value;
738 DO(ConsumeDouble(&value));
739 SET_FIELD(Float, io::SafeDoubleToFloat(value));
740 break;
741 }
742
743 case FieldDescriptor::CPPTYPE_DOUBLE: {
744 double value;
745 DO(ConsumeDouble(&value));
746 SET_FIELD(Double, value);
747 break;
748 }
749
750 case FieldDescriptor::CPPTYPE_STRING: {
751 std::string value;
752 DO(ConsumeString(&value));
753 SET_FIELD(String, value);
754 break;
755 }
756
757 case FieldDescriptor::CPPTYPE_BOOL: {
758 if (LookingAtType(io::Tokenizer::TYPE_INTEGER)) {
759 uint64 value;
760 DO(ConsumeUnsignedInteger(&value, 1));
761 SET_FIELD(Bool, value);
762 } else {
763 std::string value;
764 DO(ConsumeIdentifier(&value));
765 if (value == "true" || value == "True" || value == "t") {
766 SET_FIELD(Bool, true);
767 } else if (value == "false" || value == "False" || value == "f") {
768 SET_FIELD(Bool, false);
769 } else {
770 ReportError("Invalid value for boolean field \"" + field->name() +
771 "\". Value: \"" + value + "\".");
772 return false;
773 }
774 }
775 break;
776 }
777
778 case FieldDescriptor::CPPTYPE_ENUM: {
779 std::string value;
780 int64 int_value = kint64max;
781 const EnumDescriptor* enum_type = field->enum_type();
782 const EnumValueDescriptor* enum_value = nullptr;
783
784 if (LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) {
785 DO(ConsumeIdentifier(&value));
786 // Find the enumeration value.
787 enum_value = enum_type->FindValueByName(value);
788
789 } else if (LookingAt("-") ||
790 LookingAtType(io::Tokenizer::TYPE_INTEGER)) {
791 DO(ConsumeSignedInteger(&int_value, kint32max));
792 value = StrCat(int_value); // for error reporting
793 enum_value = enum_type->FindValueByNumber(int_value);
794 } else {
795 ReportError("Expected integer or identifier, got: " +
796 tokenizer_.current().text);
797 return false;
798 }
799
800 if (enum_value == nullptr) {
801 if (int_value != kint64max &&
802 reflection->SupportsUnknownEnumValues()) {
803 SET_FIELD(EnumValue, int_value);
804 return true;
805 } else if (!allow_unknown_enum_) {
806 ReportError("Unknown enumeration value of \"" + value +
807 "\" for "
808 "field \"" +
809 field->name() + "\".");
810 return false;
811 } else {
812 ReportWarning("Unknown enumeration value of \"" + value +
813 "\" for "
814 "field \"" +
815 field->name() + "\".");
816 return true;
817 }
818 }
819
820 SET_FIELD(Enum, enum_value);
821 break;
822 }
823
824 case FieldDescriptor::CPPTYPE_MESSAGE: {
825 // We should never get here. Put here instead of a default
826 // so that if new types are added, we get a nice compiler warning.
827 GOOGLE_LOG(FATAL) << "Reached an unintended state: CPPTYPE_MESSAGE";
828 break;
829 }
830 }
831 #undef SET_FIELD
832 return true;
833 }
834
SkipFieldValue()835 bool SkipFieldValue() {
836 if (LookingAtType(io::Tokenizer::TYPE_STRING)) {
837 while (LookingAtType(io::Tokenizer::TYPE_STRING)) {
838 tokenizer_.Next();
839 }
840 return true;
841 }
842 if (TryConsume("[")) {
843 while (true) {
844 if (!LookingAt("{") && !LookingAt("<")) {
845 DO(SkipFieldValue());
846 } else {
847 DO(SkipFieldMessage());
848 }
849 if (TryConsume("]")) {
850 break;
851 }
852 DO(Consume(","));
853 }
854 return true;
855 }
856 // Possible field values other than string:
857 // 12345 => TYPE_INTEGER
858 // -12345 => TYPE_SYMBOL + TYPE_INTEGER
859 // 1.2345 => TYPE_FLOAT
860 // -1.2345 => TYPE_SYMBOL + TYPE_FLOAT
861 // inf => TYPE_IDENTIFIER
862 // -inf => TYPE_SYMBOL + TYPE_IDENTIFIER
863 // TYPE_INTEGER => TYPE_IDENTIFIER
864 // Divides them into two group, one with TYPE_SYMBOL
865 // and the other without:
866 // Group one:
867 // 12345 => TYPE_INTEGER
868 // 1.2345 => TYPE_FLOAT
869 // inf => TYPE_IDENTIFIER
870 // TYPE_INTEGER => TYPE_IDENTIFIER
871 // Group two:
872 // -12345 => TYPE_SYMBOL + TYPE_INTEGER
873 // -1.2345 => TYPE_SYMBOL + TYPE_FLOAT
874 // -inf => TYPE_SYMBOL + TYPE_IDENTIFIER
875 // As we can see, the field value consists of an optional '-' and one of
876 // TYPE_INTEGER, TYPE_FLOAT and TYPE_IDENTIFIER.
877 bool has_minus = TryConsume("-");
878 if (!LookingAtType(io::Tokenizer::TYPE_INTEGER) &&
879 !LookingAtType(io::Tokenizer::TYPE_FLOAT) &&
880 !LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) {
881 std::string text = tokenizer_.current().text;
882 ReportError("Cannot skip field value, unexpected token: " + text);
883 return false;
884 }
885 // Combination of '-' and TYPE_IDENTIFIER may result in an invalid field
886 // value while other combinations all generate valid values.
887 // We check if the value of this combination is valid here.
888 // TYPE_IDENTIFIER after a '-' should be one of the float values listed
889 // below:
890 // inf, inff, infinity, nan
891 if (has_minus && LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) {
892 std::string text = tokenizer_.current().text;
893 LowerString(&text);
894 if (text != "inf" &&
895 text != "infinity" && text != "nan") {
896 ReportError("Invalid float number: " + text);
897 return false;
898 }
899 }
900 tokenizer_.Next();
901 return true;
902 }
903
904 // Returns true if the current token's text is equal to that specified.
LookingAt(const std::string & text)905 bool LookingAt(const std::string& text) {
906 return tokenizer_.current().text == text;
907 }
908
909 // Returns true if the current token's type is equal to that specified.
LookingAtType(io::Tokenizer::TokenType token_type)910 bool LookingAtType(io::Tokenizer::TokenType token_type) {
911 return tokenizer_.current().type == token_type;
912 }
913
914 // Consumes an identifier and saves its value in the identifier parameter.
915 // Returns false if the token is not of type IDENTFIER.
ConsumeIdentifier(std::string * identifier)916 bool ConsumeIdentifier(std::string* identifier) {
917 if (LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) {
918 *identifier = tokenizer_.current().text;
919 tokenizer_.Next();
920 return true;
921 }
922
923 // If allow_field_numer_ or allow_unknown_field_ is true, we should able
924 // to parse integer identifiers.
925 if ((allow_field_number_ || allow_unknown_field_ ||
926 allow_unknown_extension_) &&
927 LookingAtType(io::Tokenizer::TYPE_INTEGER)) {
928 *identifier = tokenizer_.current().text;
929 tokenizer_.Next();
930 return true;
931 }
932
933 ReportError("Expected identifier, got: " + tokenizer_.current().text);
934 return false;
935 }
936
937 // Consume a string of form "<id1>.<id2>....<idN>".
ConsumeFullTypeName(std::string * name)938 bool ConsumeFullTypeName(std::string* name) {
939 DO(ConsumeIdentifier(name));
940 while (TryConsume(".")) {
941 std::string part;
942 DO(ConsumeIdentifier(&part));
943 *name += ".";
944 *name += part;
945 }
946 return true;
947 }
948
ConsumeTypeUrlOrFullTypeName()949 bool ConsumeTypeUrlOrFullTypeName() {
950 std::string discarded;
951 DO(ConsumeIdentifier(&discarded));
952 while (TryConsume(".") || TryConsume("/")) {
953 DO(ConsumeIdentifier(&discarded));
954 }
955 return true;
956 }
957
958 // Consumes a string and saves its value in the text parameter.
959 // Returns false if the token is not of type STRING.
ConsumeString(std::string * text)960 bool ConsumeString(std::string* text) {
961 if (!LookingAtType(io::Tokenizer::TYPE_STRING)) {
962 ReportError("Expected string, got: " + tokenizer_.current().text);
963 return false;
964 }
965
966 text->clear();
967 while (LookingAtType(io::Tokenizer::TYPE_STRING)) {
968 io::Tokenizer::ParseStringAppend(tokenizer_.current().text, text);
969
970 tokenizer_.Next();
971 }
972
973 return true;
974 }
975
976 // Consumes a uint64 and saves its value in the value parameter.
977 // Returns false if the token is not of type INTEGER.
ConsumeUnsignedInteger(uint64 * value,uint64 max_value)978 bool ConsumeUnsignedInteger(uint64* value, uint64 max_value) {
979 if (!LookingAtType(io::Tokenizer::TYPE_INTEGER)) {
980 ReportError("Expected integer, got: " + tokenizer_.current().text);
981 return false;
982 }
983
984 if (!io::Tokenizer::ParseInteger(tokenizer_.current().text, max_value,
985 value)) {
986 ReportError("Integer out of range (" + tokenizer_.current().text + ")");
987 return false;
988 }
989
990 tokenizer_.Next();
991 return true;
992 }
993
994 // Consumes an int64 and saves its value in the value parameter.
995 // Note that since the tokenizer does not support negative numbers,
996 // we actually may consume an additional token (for the minus sign) in this
997 // method. Returns false if the token is not an integer
998 // (signed or otherwise).
ConsumeSignedInteger(int64 * value,uint64 max_value)999 bool ConsumeSignedInteger(int64* value, uint64 max_value) {
1000 bool negative = false;
1001
1002 if (TryConsume("-")) {
1003 negative = true;
1004 // Two's complement always allows one more negative integer than
1005 // positive.
1006 ++max_value;
1007 }
1008
1009 uint64 unsigned_value;
1010
1011 DO(ConsumeUnsignedInteger(&unsigned_value, max_value));
1012
1013 if (negative) {
1014 if ((static_cast<uint64>(kint64max) + 1) == unsigned_value) {
1015 *value = kint64min;
1016 } else {
1017 *value = -static_cast<int64>(unsigned_value);
1018 }
1019 } else {
1020 *value = static_cast<int64>(unsigned_value);
1021 }
1022
1023 return true;
1024 }
1025
1026 // Consumes a double and saves its value in the value parameter.
1027 // Accepts decimal numbers only, rejects hex or oct numbers.
ConsumeUnsignedDecimalAsDouble(double * value,uint64 max_value)1028 bool ConsumeUnsignedDecimalAsDouble(double* value, uint64 max_value) {
1029 if (!LookingAtType(io::Tokenizer::TYPE_INTEGER)) {
1030 ReportError("Expected integer, got: " + tokenizer_.current().text);
1031 return false;
1032 }
1033
1034 const std::string& text = tokenizer_.current().text;
1035 if (IsHexNumber(text) || IsOctNumber(text)) {
1036 ReportError("Expect a decimal number, got: " + text);
1037 return false;
1038 }
1039
1040 uint64 uint64_value;
1041 if (io::Tokenizer::ParseInteger(text, max_value, &uint64_value)) {
1042 *value = static_cast<double>(uint64_value);
1043 } else {
1044 // Uint64 overflow, attempt to parse as a double instead.
1045 *value = io::Tokenizer::ParseFloat(text);
1046 }
1047
1048 tokenizer_.Next();
1049 return true;
1050 }
1051
1052 // Consumes a double and saves its value in the value parameter.
1053 // Note that since the tokenizer does not support negative numbers,
1054 // we actually may consume an additional token (for the minus sign) in this
1055 // method. Returns false if the token is not a double
1056 // (signed or otherwise).
ConsumeDouble(double * value)1057 bool ConsumeDouble(double* value) {
1058 bool negative = false;
1059
1060 if (TryConsume("-")) {
1061 negative = true;
1062 }
1063
1064 // A double can actually be an integer, according to the tokenizer.
1065 // Therefore, we must check both cases here.
1066 if (LookingAtType(io::Tokenizer::TYPE_INTEGER)) {
1067 // We have found an integer value for the double.
1068 DO(ConsumeUnsignedDecimalAsDouble(value, kuint64max));
1069 } else if (LookingAtType(io::Tokenizer::TYPE_FLOAT)) {
1070 // We have found a float value for the double.
1071 *value = io::Tokenizer::ParseFloat(tokenizer_.current().text);
1072
1073 // Mark the current token as consumed.
1074 tokenizer_.Next();
1075 } else if (LookingAtType(io::Tokenizer::TYPE_IDENTIFIER)) {
1076 std::string text = tokenizer_.current().text;
1077 LowerString(&text);
1078 if (text == "inf" ||
1079 text == "infinity") {
1080 *value = std::numeric_limits<double>::infinity();
1081 tokenizer_.Next();
1082 } else if (text == "nan") {
1083 *value = std::numeric_limits<double>::quiet_NaN();
1084 tokenizer_.Next();
1085 } else {
1086 ReportError("Expected double, got: " + text);
1087 return false;
1088 }
1089 } else {
1090 ReportError("Expected double, got: " + tokenizer_.current().text);
1091 return false;
1092 }
1093
1094 if (negative) {
1095 *value = -*value;
1096 }
1097
1098 return true;
1099 }
1100
1101 // Consumes Any::type_url value, of form "type.googleapis.com/full.type.Name"
1102 // or "type.googleprod.com/full.type.Name"
ConsumeAnyTypeUrl(std::string * full_type_name,std::string * prefix)1103 bool ConsumeAnyTypeUrl(std::string* full_type_name, std::string* prefix) {
1104 // TODO(saito) Extend Consume() to consume multiple tokens at once, so that
1105 // this code can be written as just DO(Consume(kGoogleApisTypePrefix)).
1106 DO(ConsumeIdentifier(prefix));
1107 while (TryConsume(".")) {
1108 std::string url;
1109 DO(ConsumeIdentifier(&url));
1110 *prefix += "." + url;
1111 }
1112 DO(Consume("/"));
1113 *prefix += "/";
1114 DO(ConsumeFullTypeName(full_type_name));
1115
1116 return true;
1117 }
1118
1119 // A helper function for reconstructing Any::value. Consumes a text of
1120 // full_type_name, then serializes it into serialized_value.
ConsumeAnyValue(const Descriptor * value_descriptor,std::string * serialized_value)1121 bool ConsumeAnyValue(const Descriptor* value_descriptor,
1122 std::string* serialized_value) {
1123 DynamicMessageFactory factory;
1124 const Message* value_prototype = factory.GetPrototype(value_descriptor);
1125 if (value_prototype == nullptr) {
1126 return false;
1127 }
1128 std::unique_ptr<Message> value(value_prototype->New());
1129 std::string sub_delimiter;
1130 DO(ConsumeMessageDelimiter(&sub_delimiter));
1131 DO(ConsumeMessage(value.get(), sub_delimiter));
1132
1133 if (allow_partial_) {
1134 value->AppendPartialToString(serialized_value);
1135 } else {
1136 if (!value->IsInitialized()) {
1137 ReportError(
1138 "Value of type \"" + value_descriptor->full_name() +
1139 "\" stored in google.protobuf.Any has missing required fields");
1140 return false;
1141 }
1142 value->AppendToString(serialized_value);
1143 }
1144 return true;
1145 }
1146
1147 // Consumes a token and confirms that it matches that specified in the
1148 // value parameter. Returns false if the token found does not match that
1149 // which was specified.
Consume(const std::string & value)1150 bool Consume(const std::string& value) {
1151 const std::string& current_value = tokenizer_.current().text;
1152
1153 if (current_value != value) {
1154 ReportError("Expected \"" + value + "\", found \"" + current_value +
1155 "\".");
1156 return false;
1157 }
1158
1159 tokenizer_.Next();
1160
1161 return true;
1162 }
1163
1164 // Attempts to consume the supplied value. Returns false if a the
1165 // token found does not match the value specified.
TryConsume(const std::string & value)1166 bool TryConsume(const std::string& value) {
1167 if (tokenizer_.current().text == value) {
1168 tokenizer_.Next();
1169 return true;
1170 } else {
1171 return false;
1172 }
1173 }
1174
1175 // An internal instance of the Tokenizer's error collector, used to
1176 // collect any base-level parse errors and feed them to the ParserImpl.
1177 class ParserErrorCollector : public io::ErrorCollector {
1178 public:
ParserErrorCollector(TextFormat::Parser::ParserImpl * parser)1179 explicit ParserErrorCollector(TextFormat::Parser::ParserImpl* parser)
1180 : parser_(parser) {}
1181
~ParserErrorCollector()1182 ~ParserErrorCollector() override {}
1183
AddError(int line,int column,const std::string & message)1184 void AddError(int line, int column, const std::string& message) override {
1185 parser_->ReportError(line, column, message);
1186 }
1187
AddWarning(int line,int column,const std::string & message)1188 void AddWarning(int line, int column, const std::string& message) override {
1189 parser_->ReportWarning(line, column, message);
1190 }
1191
1192 private:
1193 GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(ParserErrorCollector);
1194 TextFormat::Parser::ParserImpl* parser_;
1195 };
1196
1197 io::ErrorCollector* error_collector_;
1198 const TextFormat::Finder* finder_;
1199 ParseInfoTree* parse_info_tree_;
1200 ParserErrorCollector tokenizer_error_collector_;
1201 io::Tokenizer tokenizer_;
1202 const Descriptor* root_message_type_;
1203 SingularOverwritePolicy singular_overwrite_policy_;
1204 const bool allow_case_insensitive_field_;
1205 const bool allow_unknown_field_;
1206 const bool allow_unknown_extension_;
1207 const bool allow_unknown_enum_;
1208 const bool allow_field_number_;
1209 const bool allow_partial_;
1210 int recursion_limit_;
1211 bool had_errors_;
1212 };
1213
1214 // ===========================================================================
1215 // Internal class for writing text to the io::ZeroCopyOutputStream. Adapted
1216 // from the Printer found in //net/proto2/io/public/printer.h
1217 class TextFormat::Printer::TextGenerator
1218 : public TextFormat::BaseTextGenerator {
1219 public:
TextGenerator(io::ZeroCopyOutputStream * output,int initial_indent_level)1220 explicit TextGenerator(io::ZeroCopyOutputStream* output,
1221 int initial_indent_level)
1222 : output_(output),
1223 buffer_(nullptr),
1224 buffer_size_(0),
1225 at_start_of_line_(true),
1226 failed_(false),
1227 indent_level_(initial_indent_level),
1228 initial_indent_level_(initial_indent_level) {}
1229
~TextGenerator()1230 ~TextGenerator() {
1231 // Only BackUp() if we're sure we've successfully called Next() at least
1232 // once.
1233 if (!failed_ && buffer_size_ > 0) {
1234 output_->BackUp(buffer_size_);
1235 }
1236 }
1237
1238 // Indent text by two spaces. After calling Indent(), two spaces will be
1239 // inserted at the beginning of each line of text. Indent() may be called
1240 // multiple times to produce deeper indents.
Indent()1241 void Indent() override { ++indent_level_; }
1242
1243 // Reduces the current indent level by two spaces, or crashes if the indent
1244 // level is zero.
Outdent()1245 void Outdent() override {
1246 if (indent_level_ == 0 || indent_level_ < initial_indent_level_) {
1247 GOOGLE_LOG(DFATAL) << " Outdent() without matching Indent().";
1248 return;
1249 }
1250
1251 --indent_level_;
1252 }
1253
1254 // Print text to the output stream.
Print(const char * text,size_t size)1255 void Print(const char* text, size_t size) override {
1256 if (indent_level_ > 0) {
1257 size_t pos = 0; // The number of bytes we've written so far.
1258 for (size_t i = 0; i < size; i++) {
1259 if (text[i] == '\n') {
1260 // Saw newline. If there is more text, we may need to insert an
1261 // indent here. So, write what we have so far, including the '\n'.
1262 Write(text + pos, i - pos + 1);
1263 pos = i + 1;
1264
1265 // Setting this true will cause the next Write() to insert an indent
1266 // first.
1267 at_start_of_line_ = true;
1268 }
1269 }
1270 // Write the rest.
1271 Write(text + pos, size - pos);
1272 } else {
1273 Write(text, size);
1274 if (size > 0 && text[size - 1] == '\n') {
1275 at_start_of_line_ = true;
1276 }
1277 }
1278 }
1279
1280 // True if any write to the underlying stream failed. (We don't just
1281 // crash in this case because this is an I/O failure, not a programming
1282 // error.)
failed() const1283 bool failed() const { return failed_; }
1284
1285 private:
1286 GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(TextGenerator);
1287
Write(const char * data,size_t size)1288 void Write(const char* data, size_t size) {
1289 if (failed_) return;
1290 if (size == 0) return;
1291
1292 if (at_start_of_line_) {
1293 // Insert an indent.
1294 at_start_of_line_ = false;
1295 WriteIndent();
1296 if (failed_) return;
1297 }
1298
1299 while (size > buffer_size_) {
1300 // Data exceeds space in the buffer. Copy what we can and request a
1301 // new buffer.
1302 if (buffer_size_ > 0) {
1303 memcpy(buffer_, data, buffer_size_);
1304 data += buffer_size_;
1305 size -= buffer_size_;
1306 }
1307 void* void_buffer = nullptr;
1308 failed_ = !output_->Next(&void_buffer, &buffer_size_);
1309 if (failed_) return;
1310 buffer_ = reinterpret_cast<char*>(void_buffer);
1311 }
1312
1313 // Buffer is big enough to receive the data; copy it.
1314 memcpy(buffer_, data, size);
1315 buffer_ += size;
1316 buffer_size_ -= size;
1317 }
1318
WriteIndent()1319 void WriteIndent() {
1320 if (indent_level_ == 0) {
1321 return;
1322 }
1323 GOOGLE_DCHECK(!failed_);
1324 int size = 2 * indent_level_;
1325
1326 while (size > buffer_size_) {
1327 // Data exceeds space in the buffer. Write what we can and request a new
1328 // buffer.
1329 if (buffer_size_ > 0) {
1330 memset(buffer_, ' ', buffer_size_);
1331 }
1332 size -= buffer_size_;
1333 void* void_buffer;
1334 failed_ = !output_->Next(&void_buffer, &buffer_size_);
1335 if (failed_) return;
1336 buffer_ = reinterpret_cast<char*>(void_buffer);
1337 }
1338
1339 // Buffer is big enough to receive the data; copy it.
1340 memset(buffer_, ' ', size);
1341 buffer_ += size;
1342 buffer_size_ -= size;
1343 }
1344
1345 io::ZeroCopyOutputStream* const output_;
1346 char* buffer_;
1347 int buffer_size_;
1348 bool at_start_of_line_;
1349 bool failed_;
1350
1351 int indent_level_;
1352 int initial_indent_level_;
1353 };
1354
1355 // ===========================================================================
1356 // Implementation of the default Finder for extensions.
~Finder()1357 TextFormat::Finder::~Finder() {}
1358
FindExtension(Message * message,const std::string & name) const1359 const FieldDescriptor* TextFormat::Finder::FindExtension(
1360 Message* message, const std::string& name) const {
1361 return DefaultFinderFindExtension(message, name);
1362 }
1363
FindExtensionByNumber(const Descriptor * descriptor,int number) const1364 const FieldDescriptor* TextFormat::Finder::FindExtensionByNumber(
1365 const Descriptor* descriptor, int number) const {
1366 return DefaultFinderFindExtensionByNumber(descriptor, number);
1367 }
1368
FindAnyType(const Message & message,const std::string & prefix,const std::string & name) const1369 const Descriptor* TextFormat::Finder::FindAnyType(
1370 const Message& message, const std::string& prefix,
1371 const std::string& name) const {
1372 return DefaultFinderFindAnyType(message, prefix, name);
1373 }
1374
FindExtensionFactory(const FieldDescriptor * field) const1375 MessageFactory* TextFormat::Finder::FindExtensionFactory(
1376 const FieldDescriptor* field) const {
1377 return nullptr;
1378 }
1379
1380 // ===========================================================================
1381
Parser()1382 TextFormat::Parser::Parser()
1383 : error_collector_(nullptr),
1384 finder_(nullptr),
1385 parse_info_tree_(nullptr),
1386 allow_partial_(false),
1387 allow_case_insensitive_field_(false),
1388 allow_unknown_field_(false),
1389 allow_unknown_extension_(false),
1390 allow_unknown_enum_(false),
1391 allow_field_number_(false),
1392 allow_relaxed_whitespace_(false),
1393 allow_singular_overwrites_(false),
1394 recursion_limit_(std::numeric_limits<int>::max()) {}
1395
~Parser()1396 TextFormat::Parser::~Parser() {}
1397
1398 namespace {
1399
CheckParseInputSize(StringPiece input,io::ErrorCollector * error_collector)1400 bool CheckParseInputSize(StringPiece input,
1401 io::ErrorCollector* error_collector) {
1402 if (input.size() > INT_MAX) {
1403 error_collector->AddError(
1404 -1, 0,
1405 StrCat("Input size too large: ", static_cast<int64>(input.size()),
1406 " bytes", " > ", INT_MAX, " bytes."));
1407 return false;
1408 }
1409 return true;
1410 }
1411
1412 } // namespace
1413
Parse(io::ZeroCopyInputStream * input,Message * output)1414 bool TextFormat::Parser::Parse(io::ZeroCopyInputStream* input,
1415 Message* output) {
1416 output->Clear();
1417
1418 ParserImpl::SingularOverwritePolicy overwrites_policy =
1419 allow_singular_overwrites_ ? ParserImpl::ALLOW_SINGULAR_OVERWRITES
1420 : ParserImpl::FORBID_SINGULAR_OVERWRITES;
1421
1422 ParserImpl parser(output->GetDescriptor(), input, error_collector_, finder_,
1423 parse_info_tree_, overwrites_policy,
1424 allow_case_insensitive_field_, allow_unknown_field_,
1425 allow_unknown_extension_, allow_unknown_enum_,
1426 allow_field_number_, allow_relaxed_whitespace_,
1427 allow_partial_, recursion_limit_);
1428 return MergeUsingImpl(input, output, &parser);
1429 }
1430
ParseFromString(const std::string & input,Message * output)1431 bool TextFormat::Parser::ParseFromString(const std::string& input,
1432 Message* output) {
1433 DO(CheckParseInputSize(input, error_collector_));
1434 io::ArrayInputStream input_stream(input.data(), input.size());
1435 return Parse(&input_stream, output);
1436 }
1437
1438
Merge(io::ZeroCopyInputStream * input,Message * output)1439 bool TextFormat::Parser::Merge(io::ZeroCopyInputStream* input,
1440 Message* output) {
1441 ParserImpl parser(output->GetDescriptor(), input, error_collector_, finder_,
1442 parse_info_tree_, ParserImpl::ALLOW_SINGULAR_OVERWRITES,
1443 allow_case_insensitive_field_, allow_unknown_field_,
1444 allow_unknown_extension_, allow_unknown_enum_,
1445 allow_field_number_, allow_relaxed_whitespace_,
1446 allow_partial_, recursion_limit_);
1447 return MergeUsingImpl(input, output, &parser);
1448 }
1449
MergeFromString(const std::string & input,Message * output)1450 bool TextFormat::Parser::MergeFromString(const std::string& input,
1451 Message* output) {
1452 DO(CheckParseInputSize(input, error_collector_));
1453 io::ArrayInputStream input_stream(input.data(), input.size());
1454 return Merge(&input_stream, output);
1455 }
1456
1457
MergeUsingImpl(io::ZeroCopyInputStream *,Message * output,ParserImpl * parser_impl)1458 bool TextFormat::Parser::MergeUsingImpl(io::ZeroCopyInputStream* /* input */,
1459 Message* output,
1460 ParserImpl* parser_impl) {
1461 if (!parser_impl->Parse(output)) return false;
1462 if (!allow_partial_ && !output->IsInitialized()) {
1463 std::vector<std::string> missing_fields;
1464 output->FindInitializationErrors(&missing_fields);
1465 parser_impl->ReportError(-1, 0,
1466 "Message missing required fields: " +
1467 Join(missing_fields, ", "));
1468 return false;
1469 }
1470 return true;
1471 }
1472
ParseFieldValueFromString(const std::string & input,const FieldDescriptor * field,Message * output)1473 bool TextFormat::Parser::ParseFieldValueFromString(const std::string& input,
1474 const FieldDescriptor* field,
1475 Message* output) {
1476 io::ArrayInputStream input_stream(input.data(), input.size());
1477 ParserImpl parser(
1478 output->GetDescriptor(), &input_stream, error_collector_, finder_,
1479 parse_info_tree_, ParserImpl::ALLOW_SINGULAR_OVERWRITES,
1480 allow_case_insensitive_field_, allow_unknown_field_,
1481 allow_unknown_extension_, allow_unknown_enum_, allow_field_number_,
1482 allow_relaxed_whitespace_, allow_partial_, recursion_limit_);
1483 return parser.ParseField(field, output);
1484 }
1485
Parse(io::ZeroCopyInputStream * input,Message * output)1486 /* static */ bool TextFormat::Parse(io::ZeroCopyInputStream* input,
1487 Message* output) {
1488 return Parser().Parse(input, output);
1489 }
1490
Merge(io::ZeroCopyInputStream * input,Message * output)1491 /* static */ bool TextFormat::Merge(io::ZeroCopyInputStream* input,
1492 Message* output) {
1493 return Parser().Merge(input, output);
1494 }
1495
ParseFromString(const std::string & input,Message * output)1496 /* static */ bool TextFormat::ParseFromString(const std::string& input,
1497 Message* output) {
1498 return Parser().ParseFromString(input, output);
1499 }
1500
MergeFromString(const std::string & input,Message * output)1501 /* static */ bool TextFormat::MergeFromString(const std::string& input,
1502 Message* output) {
1503 return Parser().MergeFromString(input, output);
1504 }
1505
1506
1507 #undef DO
1508
1509 // ===========================================================================
1510
~BaseTextGenerator()1511 TextFormat::BaseTextGenerator::~BaseTextGenerator() {}
1512
1513 namespace {
1514
1515 // A BaseTextGenerator that writes to a string.
1516 class StringBaseTextGenerator : public TextFormat::BaseTextGenerator {
1517 public:
Print(const char * text,size_t size)1518 void Print(const char* text, size_t size) override {
1519 output_.append(text, size);
1520 }
1521
1522 // Some compilers do not support ref-qualifiers even in C++11 mode.
1523 // Disable the optimization for now and revisit it later.
1524 #if 0 // LANG_CXX11
1525 std::string Consume() && { return std::move(output_); }
1526 #else // !LANG_CXX11
Get()1527 const std::string& Get() { return output_; }
1528 #endif // LANG_CXX11
1529
1530 private:
1531 std::string output_;
1532 };
1533
1534 } // namespace
1535
1536 // The default implementation for FieldValuePrinter. We just delegate the
1537 // implementation to the default FastFieldValuePrinter to avoid duplicating the
1538 // logic.
FieldValuePrinter()1539 TextFormat::FieldValuePrinter::FieldValuePrinter() {}
~FieldValuePrinter()1540 TextFormat::FieldValuePrinter::~FieldValuePrinter() {}
1541
1542 #if 0 // LANG_CXX11
1543 #define FORWARD_IMPL(fn, ...) \
1544 StringBaseTextGenerator generator; \
1545 delegate_.fn(__VA_ARGS__, &generator); \
1546 return std::move(generator).Consume()
1547 #else // !LANG_CXX11
1548 #define FORWARD_IMPL(fn, ...) \
1549 StringBaseTextGenerator generator; \
1550 delegate_.fn(__VA_ARGS__, &generator); \
1551 return generator.Get()
1552 #endif // LANG_CXX11
1553
PrintBool(bool val) const1554 std::string TextFormat::FieldValuePrinter::PrintBool(bool val) const {
1555 FORWARD_IMPL(PrintBool, val);
1556 }
PrintInt32(int32 val) const1557 std::string TextFormat::FieldValuePrinter::PrintInt32(int32 val) const {
1558 FORWARD_IMPL(PrintInt32, val);
1559 }
PrintUInt32(uint32 val) const1560 std::string TextFormat::FieldValuePrinter::PrintUInt32(uint32 val) const {
1561 FORWARD_IMPL(PrintUInt32, val);
1562 }
PrintInt64(int64 val) const1563 std::string TextFormat::FieldValuePrinter::PrintInt64(int64 val) const {
1564 FORWARD_IMPL(PrintInt64, val);
1565 }
PrintUInt64(uint64 val) const1566 std::string TextFormat::FieldValuePrinter::PrintUInt64(uint64 val) const {
1567 FORWARD_IMPL(PrintUInt64, val);
1568 }
PrintFloat(float val) const1569 std::string TextFormat::FieldValuePrinter::PrintFloat(float val) const {
1570 FORWARD_IMPL(PrintFloat, val);
1571 }
PrintDouble(double val) const1572 std::string TextFormat::FieldValuePrinter::PrintDouble(double val) const {
1573 FORWARD_IMPL(PrintDouble, val);
1574 }
PrintString(const std::string & val) const1575 std::string TextFormat::FieldValuePrinter::PrintString(
1576 const std::string& val) const {
1577 FORWARD_IMPL(PrintString, val);
1578 }
PrintBytes(const std::string & val) const1579 std::string TextFormat::FieldValuePrinter::PrintBytes(
1580 const std::string& val) const {
1581 return PrintString(val);
1582 }
PrintEnum(int32 val,const std::string & name) const1583 std::string TextFormat::FieldValuePrinter::PrintEnum(
1584 int32 val, const std::string& name) const {
1585 FORWARD_IMPL(PrintEnum, val, name);
1586 }
PrintFieldName(const Message & message,const Reflection * reflection,const FieldDescriptor * field) const1587 std::string TextFormat::FieldValuePrinter::PrintFieldName(
1588 const Message& message, const Reflection* reflection,
1589 const FieldDescriptor* field) const {
1590 FORWARD_IMPL(PrintFieldName, message, reflection, field);
1591 }
PrintMessageStart(const Message & message,int field_index,int field_count,bool single_line_mode) const1592 std::string TextFormat::FieldValuePrinter::PrintMessageStart(
1593 const Message& message, int field_index, int field_count,
1594 bool single_line_mode) const {
1595 FORWARD_IMPL(PrintMessageStart, message, field_index, field_count,
1596 single_line_mode);
1597 }
PrintMessageEnd(const Message & message,int field_index,int field_count,bool single_line_mode) const1598 std::string TextFormat::FieldValuePrinter::PrintMessageEnd(
1599 const Message& message, int field_index, int field_count,
1600 bool single_line_mode) const {
1601 FORWARD_IMPL(PrintMessageEnd, message, field_index, field_count,
1602 single_line_mode);
1603 }
1604 #undef FORWARD_IMPL
1605
FastFieldValuePrinter()1606 TextFormat::FastFieldValuePrinter::FastFieldValuePrinter() {}
~FastFieldValuePrinter()1607 TextFormat::FastFieldValuePrinter::~FastFieldValuePrinter() {}
PrintBool(bool val,BaseTextGenerator * generator) const1608 void TextFormat::FastFieldValuePrinter::PrintBool(
1609 bool val, BaseTextGenerator* generator) const {
1610 if (val) {
1611 generator->PrintLiteral("true");
1612 } else {
1613 generator->PrintLiteral("false");
1614 }
1615 }
PrintInt32(int32 val,BaseTextGenerator * generator) const1616 void TextFormat::FastFieldValuePrinter::PrintInt32(
1617 int32 val, BaseTextGenerator* generator) const {
1618 generator->PrintString(StrCat(val));
1619 }
PrintUInt32(uint32 val,BaseTextGenerator * generator) const1620 void TextFormat::FastFieldValuePrinter::PrintUInt32(
1621 uint32 val, BaseTextGenerator* generator) const {
1622 generator->PrintString(StrCat(val));
1623 }
PrintInt64(int64 val,BaseTextGenerator * generator) const1624 void TextFormat::FastFieldValuePrinter::PrintInt64(
1625 int64 val, BaseTextGenerator* generator) const {
1626 generator->PrintString(StrCat(val));
1627 }
PrintUInt64(uint64 val,BaseTextGenerator * generator) const1628 void TextFormat::FastFieldValuePrinter::PrintUInt64(
1629 uint64 val, BaseTextGenerator* generator) const {
1630 generator->PrintString(StrCat(val));
1631 }
PrintFloat(float val,BaseTextGenerator * generator) const1632 void TextFormat::FastFieldValuePrinter::PrintFloat(
1633 float val, BaseTextGenerator* generator) const {
1634 generator->PrintString(SimpleFtoa(val));
1635 }
PrintDouble(double val,BaseTextGenerator * generator) const1636 void TextFormat::FastFieldValuePrinter::PrintDouble(
1637 double val, BaseTextGenerator* generator) const {
1638 generator->PrintString(SimpleDtoa(val));
1639 }
PrintEnum(int32 val,const std::string & name,BaseTextGenerator * generator) const1640 void TextFormat::FastFieldValuePrinter::PrintEnum(
1641 int32 val, const std::string& name, BaseTextGenerator* generator) const {
1642 generator->PrintString(name);
1643 }
1644
PrintString(const std::string & val,BaseTextGenerator * generator) const1645 void TextFormat::FastFieldValuePrinter::PrintString(
1646 const std::string& val, BaseTextGenerator* generator) const {
1647 generator->PrintLiteral("\"");
1648 generator->PrintString(CEscape(val));
1649 generator->PrintLiteral("\"");
1650 }
PrintBytes(const std::string & val,BaseTextGenerator * generator) const1651 void TextFormat::FastFieldValuePrinter::PrintBytes(
1652 const std::string& val, BaseTextGenerator* generator) const {
1653 PrintString(val, generator);
1654 }
PrintFieldName(const Message & message,int field_index,int field_count,const Reflection * reflection,const FieldDescriptor * field,BaseTextGenerator * generator) const1655 void TextFormat::FastFieldValuePrinter::PrintFieldName(
1656 const Message& message, int field_index, int field_count,
1657 const Reflection* reflection, const FieldDescriptor* field,
1658 BaseTextGenerator* generator) const {
1659 PrintFieldName(message, reflection, field, generator);
1660 }
PrintFieldName(const Message & message,const Reflection * reflection,const FieldDescriptor * field,BaseTextGenerator * generator) const1661 void TextFormat::FastFieldValuePrinter::PrintFieldName(
1662 const Message& message, const Reflection* reflection,
1663 const FieldDescriptor* field, BaseTextGenerator* generator) const {
1664 if (field->is_extension()) {
1665 generator->PrintLiteral("[");
1666 generator->PrintString(field->PrintableNameForExtension());
1667 generator->PrintLiteral("]");
1668 } else if (field->type() == FieldDescriptor::TYPE_GROUP) {
1669 // Groups must be serialized with their original capitalization.
1670 generator->PrintString(field->message_type()->name());
1671 } else {
1672 generator->PrintString(field->name());
1673 }
1674 }
PrintMessageStart(const Message & message,int field_index,int field_count,bool single_line_mode,BaseTextGenerator * generator) const1675 void TextFormat::FastFieldValuePrinter::PrintMessageStart(
1676 const Message& message, int field_index, int field_count,
1677 bool single_line_mode, BaseTextGenerator* generator) const {
1678 if (single_line_mode) {
1679 generator->PrintLiteral(" { ");
1680 } else {
1681 generator->PrintLiteral(" {\n");
1682 }
1683 }
PrintMessageEnd(const Message & message,int field_index,int field_count,bool single_line_mode,BaseTextGenerator * generator) const1684 void TextFormat::FastFieldValuePrinter::PrintMessageEnd(
1685 const Message& message, int field_index, int field_count,
1686 bool single_line_mode, BaseTextGenerator* generator) const {
1687 if (single_line_mode) {
1688 generator->PrintLiteral("} ");
1689 } else {
1690 generator->PrintLiteral("}\n");
1691 }
1692 }
1693
1694 namespace {
1695
1696 // A legacy compatibility wrapper. Takes ownership of the delegate.
1697 class FieldValuePrinterWrapper : public TextFormat::FastFieldValuePrinter {
1698 public:
FieldValuePrinterWrapper(const TextFormat::FieldValuePrinter * delegate)1699 explicit FieldValuePrinterWrapper(
1700 const TextFormat::FieldValuePrinter* delegate)
1701 : delegate_(delegate) {}
1702
SetDelegate(const TextFormat::FieldValuePrinter * delegate)1703 void SetDelegate(const TextFormat::FieldValuePrinter* delegate) {
1704 delegate_.reset(delegate);
1705 }
1706
PrintBool(bool val,TextFormat::BaseTextGenerator * generator) const1707 void PrintBool(bool val,
1708 TextFormat::BaseTextGenerator* generator) const override {
1709 generator->PrintString(delegate_->PrintBool(val));
1710 }
PrintInt32(int32 val,TextFormat::BaseTextGenerator * generator) const1711 void PrintInt32(int32 val,
1712 TextFormat::BaseTextGenerator* generator) const override {
1713 generator->PrintString(delegate_->PrintInt32(val));
1714 }
PrintUInt32(uint32 val,TextFormat::BaseTextGenerator * generator) const1715 void PrintUInt32(uint32 val,
1716 TextFormat::BaseTextGenerator* generator) const override {
1717 generator->PrintString(delegate_->PrintUInt32(val));
1718 }
PrintInt64(int64 val,TextFormat::BaseTextGenerator * generator) const1719 void PrintInt64(int64 val,
1720 TextFormat::BaseTextGenerator* generator) const override {
1721 generator->PrintString(delegate_->PrintInt64(val));
1722 }
PrintUInt64(uint64 val,TextFormat::BaseTextGenerator * generator) const1723 void PrintUInt64(uint64 val,
1724 TextFormat::BaseTextGenerator* generator) const override {
1725 generator->PrintString(delegate_->PrintUInt64(val));
1726 }
PrintFloat(float val,TextFormat::BaseTextGenerator * generator) const1727 void PrintFloat(float val,
1728 TextFormat::BaseTextGenerator* generator) const override {
1729 generator->PrintString(delegate_->PrintFloat(val));
1730 }
PrintDouble(double val,TextFormat::BaseTextGenerator * generator) const1731 void PrintDouble(double val,
1732 TextFormat::BaseTextGenerator* generator) const override {
1733 generator->PrintString(delegate_->PrintDouble(val));
1734 }
PrintString(const std::string & val,TextFormat::BaseTextGenerator * generator) const1735 void PrintString(const std::string& val,
1736 TextFormat::BaseTextGenerator* generator) const override {
1737 generator->PrintString(delegate_->PrintString(val));
1738 }
PrintBytes(const std::string & val,TextFormat::BaseTextGenerator * generator) const1739 void PrintBytes(const std::string& val,
1740 TextFormat::BaseTextGenerator* generator) const override {
1741 generator->PrintString(delegate_->PrintBytes(val));
1742 }
PrintEnum(int32 val,const std::string & name,TextFormat::BaseTextGenerator * generator) const1743 void PrintEnum(int32 val, const std::string& name,
1744 TextFormat::BaseTextGenerator* generator) const override {
1745 generator->PrintString(delegate_->PrintEnum(val, name));
1746 }
PrintFieldName(const Message & message,int field_index,int field_count,const Reflection * reflection,const FieldDescriptor * field,TextFormat::BaseTextGenerator * generator) const1747 void PrintFieldName(const Message& message, int field_index, int field_count,
1748 const Reflection* reflection,
1749 const FieldDescriptor* field,
1750 TextFormat::BaseTextGenerator* generator) const override {
1751 generator->PrintString(
1752 delegate_->PrintFieldName(message, reflection, field));
1753 }
PrintFieldName(const Message & message,const Reflection * reflection,const FieldDescriptor * field,TextFormat::BaseTextGenerator * generator) const1754 void PrintFieldName(const Message& message, const Reflection* reflection,
1755 const FieldDescriptor* field,
1756 TextFormat::BaseTextGenerator* generator) const override {
1757 generator->PrintString(
1758 delegate_->PrintFieldName(message, reflection, field));
1759 }
PrintMessageStart(const Message & message,int field_index,int field_count,bool single_line_mode,TextFormat::BaseTextGenerator * generator) const1760 void PrintMessageStart(
1761 const Message& message, int field_index, int field_count,
1762 bool single_line_mode,
1763 TextFormat::BaseTextGenerator* generator) const override {
1764 generator->PrintString(delegate_->PrintMessageStart(
1765 message, field_index, field_count, single_line_mode));
1766 }
PrintMessageEnd(const Message & message,int field_index,int field_count,bool single_line_mode,TextFormat::BaseTextGenerator * generator) const1767 void PrintMessageEnd(
1768 const Message& message, int field_index, int field_count,
1769 bool single_line_mode,
1770 TextFormat::BaseTextGenerator* generator) const override {
1771 generator->PrintString(delegate_->PrintMessageEnd(
1772 message, field_index, field_count, single_line_mode));
1773 }
1774
1775 private:
1776 std::unique_ptr<const TextFormat::FieldValuePrinter> delegate_;
1777 };
1778
1779 // Our own specialization: for UTF8 escaped strings.
1780 class FastFieldValuePrinterUtf8Escaping
1781 : public TextFormat::FastFieldValuePrinter {
1782 public:
PrintString(const std::string & val,TextFormat::BaseTextGenerator * generator) const1783 void PrintString(const std::string& val,
1784 TextFormat::BaseTextGenerator* generator) const override {
1785 generator->PrintLiteral("\"");
1786 generator->PrintString(strings::Utf8SafeCEscape(val));
1787 generator->PrintLiteral("\"");
1788 }
PrintBytes(const std::string & val,TextFormat::BaseTextGenerator * generator) const1789 void PrintBytes(const std::string& val,
1790 TextFormat::BaseTextGenerator* generator) const override {
1791 return FastFieldValuePrinter::PrintString(val, generator);
1792 }
1793 };
1794
1795 } // namespace
1796
Printer()1797 TextFormat::Printer::Printer()
1798 : initial_indent_level_(0),
1799 single_line_mode_(false),
1800 use_field_number_(false),
1801 use_short_repeated_primitives_(false),
1802 hide_unknown_fields_(false),
1803 print_message_fields_in_index_order_(false),
1804 expand_any_(false),
1805 truncate_string_field_longer_than_(0LL),
1806 finder_(nullptr) {
1807 SetUseUtf8StringEscaping(false);
1808 }
1809
~Printer()1810 TextFormat::Printer::~Printer() {
1811 STLDeleteValues(&custom_printers_);
1812 STLDeleteValues(&custom_message_printers_);
1813 }
1814
SetUseUtf8StringEscaping(bool as_utf8)1815 void TextFormat::Printer::SetUseUtf8StringEscaping(bool as_utf8) {
1816 SetDefaultFieldValuePrinter(as_utf8 ? new FastFieldValuePrinterUtf8Escaping()
1817 : new FastFieldValuePrinter());
1818 }
1819
SetDefaultFieldValuePrinter(const FieldValuePrinter * printer)1820 void TextFormat::Printer::SetDefaultFieldValuePrinter(
1821 const FieldValuePrinter* printer) {
1822 default_field_value_printer_.reset(new FieldValuePrinterWrapper(printer));
1823 }
1824
SetDefaultFieldValuePrinter(const FastFieldValuePrinter * printer)1825 void TextFormat::Printer::SetDefaultFieldValuePrinter(
1826 const FastFieldValuePrinter* printer) {
1827 default_field_value_printer_.reset(printer);
1828 }
1829
RegisterFieldValuePrinter(const FieldDescriptor * field,const FieldValuePrinter * printer)1830 bool TextFormat::Printer::RegisterFieldValuePrinter(
1831 const FieldDescriptor* field, const FieldValuePrinter* printer) {
1832 if (field == nullptr || printer == nullptr) {
1833 return false;
1834 }
1835 FieldValuePrinterWrapper* const wrapper =
1836 new FieldValuePrinterWrapper(nullptr);
1837 if (custom_printers_.insert(std::make_pair(field, wrapper)).second) {
1838 wrapper->SetDelegate(printer);
1839 return true;
1840 } else {
1841 delete wrapper;
1842 return false;
1843 }
1844 }
1845
RegisterFieldValuePrinter(const FieldDescriptor * field,const FastFieldValuePrinter * printer)1846 bool TextFormat::Printer::RegisterFieldValuePrinter(
1847 const FieldDescriptor* field, const FastFieldValuePrinter* printer) {
1848 return field != nullptr && printer != nullptr &&
1849 custom_printers_.insert(std::make_pair(field, printer)).second;
1850 }
1851
RegisterMessagePrinter(const Descriptor * descriptor,const MessagePrinter * printer)1852 bool TextFormat::Printer::RegisterMessagePrinter(
1853 const Descriptor* descriptor, const MessagePrinter* printer) {
1854 return descriptor != nullptr && printer != nullptr &&
1855 custom_message_printers_.insert(std::make_pair(descriptor, printer))
1856 .second;
1857 }
1858
PrintToString(const Message & message,std::string * output) const1859 bool TextFormat::Printer::PrintToString(const Message& message,
1860 std::string* output) const {
1861 GOOGLE_DCHECK(output) << "output specified is nullptr";
1862
1863 output->clear();
1864 io::StringOutputStream output_stream(output);
1865
1866 return Print(message, &output_stream);
1867 }
1868
PrintUnknownFieldsToString(const UnknownFieldSet & unknown_fields,std::string * output) const1869 bool TextFormat::Printer::PrintUnknownFieldsToString(
1870 const UnknownFieldSet& unknown_fields, std::string* output) const {
1871 GOOGLE_DCHECK(output) << "output specified is nullptr";
1872
1873 output->clear();
1874 io::StringOutputStream output_stream(output);
1875 return PrintUnknownFields(unknown_fields, &output_stream);
1876 }
1877
Print(const Message & message,io::ZeroCopyOutputStream * output) const1878 bool TextFormat::Printer::Print(const Message& message,
1879 io::ZeroCopyOutputStream* output) const {
1880 TextGenerator generator(output, initial_indent_level_);
1881
1882 Print(message, &generator);
1883
1884 // Output false if the generator failed internally.
1885 return !generator.failed();
1886 }
1887
PrintUnknownFields(const UnknownFieldSet & unknown_fields,io::ZeroCopyOutputStream * output) const1888 bool TextFormat::Printer::PrintUnknownFields(
1889 const UnknownFieldSet& unknown_fields,
1890 io::ZeroCopyOutputStream* output) const {
1891 TextGenerator generator(output, initial_indent_level_);
1892
1893 PrintUnknownFields(unknown_fields, &generator);
1894
1895 // Output false if the generator failed internally.
1896 return !generator.failed();
1897 }
1898
1899 namespace {
1900 // Comparison functor for sorting FieldDescriptors by field index.
1901 // Normal fields have higher precedence than extensions.
1902 struct FieldIndexSorter {
operator ()google::protobuf::__anona09921bc0611::FieldIndexSorter1903 bool operator()(const FieldDescriptor* left,
1904 const FieldDescriptor* right) const {
1905 if (left->is_extension() && right->is_extension()) {
1906 return left->number() < right->number();
1907 } else if (left->is_extension()) {
1908 return false;
1909 } else if (right->is_extension()) {
1910 return true;
1911 } else {
1912 return left->index() < right->index();
1913 }
1914 }
1915 };
1916
1917 } // namespace
1918
PrintAny(const Message & message,TextGenerator * generator) const1919 bool TextFormat::Printer::PrintAny(const Message& message,
1920 TextGenerator* generator) const {
1921 const FieldDescriptor* type_url_field;
1922 const FieldDescriptor* value_field;
1923 if (!internal::GetAnyFieldDescriptors(message, &type_url_field,
1924 &value_field)) {
1925 return false;
1926 }
1927
1928 const Reflection* reflection = message.GetReflection();
1929
1930 // Extract the full type name from the type_url field.
1931 const std::string& type_url = reflection->GetString(message, type_url_field);
1932 std::string url_prefix;
1933 std::string full_type_name;
1934 if (!internal::ParseAnyTypeUrl(type_url, &url_prefix, &full_type_name)) {
1935 return false;
1936 }
1937
1938 // Print the "value" in text.
1939 const Descriptor* value_descriptor =
1940 finder_ ? finder_->FindAnyType(message, url_prefix, full_type_name)
1941 : DefaultFinderFindAnyType(message, url_prefix, full_type_name);
1942 if (value_descriptor == nullptr) {
1943 GOOGLE_LOG(WARNING) << "Proto type " << type_url << " not found";
1944 return false;
1945 }
1946 DynamicMessageFactory factory;
1947 std::unique_ptr<Message> value_message(
1948 factory.GetPrototype(value_descriptor)->New());
1949 std::string serialized_value = reflection->GetString(message, value_field);
1950 if (!value_message->ParseFromString(serialized_value)) {
1951 GOOGLE_LOG(WARNING) << type_url << ": failed to parse contents";
1952 return false;
1953 }
1954 generator->PrintLiteral("[");
1955 generator->PrintString(type_url);
1956 generator->PrintLiteral("]");
1957 const FastFieldValuePrinter* printer = FindWithDefault(
1958 custom_printers_, value_field, default_field_value_printer_.get());
1959 printer->PrintMessageStart(message, -1, 0, single_line_mode_, generator);
1960 generator->Indent();
1961 Print(*value_message, generator);
1962 generator->Outdent();
1963 printer->PrintMessageEnd(message, -1, 0, single_line_mode_, generator);
1964 return true;
1965 }
1966
Print(const Message & message,TextGenerator * generator) const1967 void TextFormat::Printer::Print(const Message& message,
1968 TextGenerator* generator) const {
1969 const Reflection* reflection = message.GetReflection();
1970 if (!reflection) {
1971 // This message does not provide any way to describe its structure.
1972 // Parse it again in an UnknownFieldSet, and display this instead.
1973 UnknownFieldSet unknown_fields;
1974 {
1975 std::string serialized = message.SerializeAsString();
1976 io::ArrayInputStream input(serialized.data(), serialized.size());
1977 unknown_fields.ParseFromZeroCopyStream(&input);
1978 }
1979 PrintUnknownFields(unknown_fields, generator);
1980 return;
1981 }
1982 const Descriptor* descriptor = message.GetDescriptor();
1983 auto itr = custom_message_printers_.find(descriptor);
1984 if (itr != custom_message_printers_.end()) {
1985 itr->second->Print(message, single_line_mode_, generator);
1986 return;
1987 }
1988 if (descriptor->full_name() == internal::kAnyFullTypeName && expand_any_ &&
1989 PrintAny(message, generator)) {
1990 return;
1991 }
1992 std::vector<const FieldDescriptor*> fields;
1993 if (descriptor->options().map_entry()) {
1994 fields.push_back(descriptor->field(0));
1995 fields.push_back(descriptor->field(1));
1996 } else {
1997 reflection->ListFields(message, &fields);
1998 }
1999
2000 if (print_message_fields_in_index_order_) {
2001 std::sort(fields.begin(), fields.end(), FieldIndexSorter());
2002 }
2003 for (int i = 0; i < fields.size(); i++) {
2004 PrintField(message, reflection, fields[i], generator);
2005 }
2006 if (!hide_unknown_fields_) {
2007 PrintUnknownFields(reflection->GetUnknownFields(message), generator);
2008 }
2009 }
2010
PrintFieldValueToString(const Message & message,const FieldDescriptor * field,int index,std::string * output) const2011 void TextFormat::Printer::PrintFieldValueToString(const Message& message,
2012 const FieldDescriptor* field,
2013 int index,
2014 std::string* output) const {
2015 GOOGLE_DCHECK(output) << "output specified is nullptr";
2016
2017 output->clear();
2018 io::StringOutputStream output_stream(output);
2019 TextGenerator generator(&output_stream, initial_indent_level_);
2020
2021 PrintFieldValue(message, message.GetReflection(), field, index, &generator);
2022 }
2023
2024 class MapEntryMessageComparator {
2025 public:
MapEntryMessageComparator(const Descriptor * descriptor)2026 explicit MapEntryMessageComparator(const Descriptor* descriptor)
2027 : field_(descriptor->field(0)) {}
2028
operator ()(const Message * a,const Message * b)2029 bool operator()(const Message* a, const Message* b) {
2030 const Reflection* reflection = a->GetReflection();
2031 switch (field_->cpp_type()) {
2032 case FieldDescriptor::CPPTYPE_BOOL: {
2033 bool first = reflection->GetBool(*a, field_);
2034 bool second = reflection->GetBool(*b, field_);
2035 return first < second;
2036 }
2037 case FieldDescriptor::CPPTYPE_INT32: {
2038 int32 first = reflection->GetInt32(*a, field_);
2039 int32 second = reflection->GetInt32(*b, field_);
2040 return first < second;
2041 }
2042 case FieldDescriptor::CPPTYPE_INT64: {
2043 int64 first = reflection->GetInt64(*a, field_);
2044 int64 second = reflection->GetInt64(*b, field_);
2045 return first < second;
2046 }
2047 case FieldDescriptor::CPPTYPE_UINT32: {
2048 uint32 first = reflection->GetUInt32(*a, field_);
2049 uint32 second = reflection->GetUInt32(*b, field_);
2050 return first < second;
2051 }
2052 case FieldDescriptor::CPPTYPE_UINT64: {
2053 uint64 first = reflection->GetUInt64(*a, field_);
2054 uint64 second = reflection->GetUInt64(*b, field_);
2055 return first < second;
2056 }
2057 case FieldDescriptor::CPPTYPE_STRING: {
2058 std::string first = reflection->GetString(*a, field_);
2059 std::string second = reflection->GetString(*b, field_);
2060 return first < second;
2061 }
2062 default:
2063 GOOGLE_LOG(DFATAL) << "Invalid key for map field.";
2064 return true;
2065 }
2066 }
2067
2068 private:
2069 const FieldDescriptor* field_;
2070 };
2071
2072 namespace internal {
2073 class MapFieldPrinterHelper {
2074 public:
2075 // DynamicMapSorter::Sort cannot be used because it enfores syncing with
2076 // repeated field.
2077 static bool SortMap(const Message& message, const Reflection* reflection,
2078 const FieldDescriptor* field, MessageFactory* factory,
2079 std::vector<const Message*>* sorted_map_field);
2080 static void CopyKey(const MapKey& key, Message* message,
2081 const FieldDescriptor* field_desc);
2082 static void CopyValue(const MapValueRef& value, Message* message,
2083 const FieldDescriptor* field_desc);
2084 };
2085
2086 // Returns true if elements contained in sorted_map_field need to be released.
SortMap(const Message & message,const Reflection * reflection,const FieldDescriptor * field,MessageFactory * factory,std::vector<const Message * > * sorted_map_field)2087 bool MapFieldPrinterHelper::SortMap(
2088 const Message& message, const Reflection* reflection,
2089 const FieldDescriptor* field, MessageFactory* factory,
2090 std::vector<const Message*>* sorted_map_field) {
2091 bool need_release = false;
2092 const MapFieldBase& base = *reflection->GetMapData(message, field);
2093
2094 if (base.IsRepeatedFieldValid()) {
2095 const RepeatedPtrField<Message>& map_field =
2096 reflection->GetRepeatedPtrField<Message>(message, field);
2097 for (int i = 0; i < map_field.size(); ++i) {
2098 sorted_map_field->push_back(
2099 const_cast<RepeatedPtrField<Message>*>(&map_field)->Mutable(i));
2100 }
2101 } else {
2102 // TODO(teboring): For performance, instead of creating map entry message
2103 // for each element, just store map keys and sort them.
2104 const Descriptor* map_entry_desc = field->message_type();
2105 const Message* prototype = factory->GetPrototype(map_entry_desc);
2106 for (MapIterator iter =
2107 reflection->MapBegin(const_cast<Message*>(&message), field);
2108 iter != reflection->MapEnd(const_cast<Message*>(&message), field);
2109 ++iter) {
2110 Message* map_entry_message = prototype->New();
2111 CopyKey(iter.GetKey(), map_entry_message, map_entry_desc->field(0));
2112 CopyValue(iter.GetValueRef(), map_entry_message,
2113 map_entry_desc->field(1));
2114 sorted_map_field->push_back(map_entry_message);
2115 }
2116 need_release = true;
2117 }
2118
2119 MapEntryMessageComparator comparator(field->message_type());
2120 std::stable_sort(sorted_map_field->begin(), sorted_map_field->end(),
2121 comparator);
2122 return need_release;
2123 }
2124
CopyKey(const MapKey & key,Message * message,const FieldDescriptor * field_desc)2125 void MapFieldPrinterHelper::CopyKey(const MapKey& key, Message* message,
2126 const FieldDescriptor* field_desc) {
2127 const Reflection* reflection = message->GetReflection();
2128 switch (field_desc->cpp_type()) {
2129 case FieldDescriptor::CPPTYPE_DOUBLE:
2130 case FieldDescriptor::CPPTYPE_FLOAT:
2131 case FieldDescriptor::CPPTYPE_ENUM:
2132 case FieldDescriptor::CPPTYPE_MESSAGE:
2133 GOOGLE_LOG(ERROR) << "Not supported.";
2134 break;
2135 case FieldDescriptor::CPPTYPE_STRING:
2136 reflection->SetString(message, field_desc, key.GetStringValue());
2137 return;
2138 case FieldDescriptor::CPPTYPE_INT64:
2139 reflection->SetInt64(message, field_desc, key.GetInt64Value());
2140 return;
2141 case FieldDescriptor::CPPTYPE_INT32:
2142 reflection->SetInt32(message, field_desc, key.GetInt32Value());
2143 return;
2144 case FieldDescriptor::CPPTYPE_UINT64:
2145 reflection->SetUInt64(message, field_desc, key.GetUInt64Value());
2146 return;
2147 case FieldDescriptor::CPPTYPE_UINT32:
2148 reflection->SetUInt32(message, field_desc, key.GetUInt32Value());
2149 return;
2150 case FieldDescriptor::CPPTYPE_BOOL:
2151 reflection->SetBool(message, field_desc, key.GetBoolValue());
2152 return;
2153 }
2154 }
2155
CopyValue(const MapValueRef & value,Message * message,const FieldDescriptor * field_desc)2156 void MapFieldPrinterHelper::CopyValue(const MapValueRef& value,
2157 Message* message,
2158 const FieldDescriptor* field_desc) {
2159 const Reflection* reflection = message->GetReflection();
2160 switch (field_desc->cpp_type()) {
2161 case FieldDescriptor::CPPTYPE_DOUBLE:
2162 reflection->SetDouble(message, field_desc, value.GetDoubleValue());
2163 return;
2164 case FieldDescriptor::CPPTYPE_FLOAT:
2165 reflection->SetFloat(message, field_desc, value.GetFloatValue());
2166 return;
2167 case FieldDescriptor::CPPTYPE_ENUM:
2168 reflection->SetEnumValue(message, field_desc, value.GetEnumValue());
2169 return;
2170 case FieldDescriptor::CPPTYPE_MESSAGE: {
2171 Message* sub_message = value.GetMessageValue().New();
2172 sub_message->CopyFrom(value.GetMessageValue());
2173 reflection->SetAllocatedMessage(message, sub_message, field_desc);
2174 return;
2175 }
2176 case FieldDescriptor::CPPTYPE_STRING:
2177 reflection->SetString(message, field_desc, value.GetStringValue());
2178 return;
2179 case FieldDescriptor::CPPTYPE_INT64:
2180 reflection->SetInt64(message, field_desc, value.GetInt64Value());
2181 return;
2182 case FieldDescriptor::CPPTYPE_INT32:
2183 reflection->SetInt32(message, field_desc, value.GetInt32Value());
2184 return;
2185 case FieldDescriptor::CPPTYPE_UINT64:
2186 reflection->SetUInt64(message, field_desc, value.GetUInt64Value());
2187 return;
2188 case FieldDescriptor::CPPTYPE_UINT32:
2189 reflection->SetUInt32(message, field_desc, value.GetUInt32Value());
2190 return;
2191 case FieldDescriptor::CPPTYPE_BOOL:
2192 reflection->SetBool(message, field_desc, value.GetBoolValue());
2193 return;
2194 }
2195 }
2196 } // namespace internal
2197
PrintField(const Message & message,const Reflection * reflection,const FieldDescriptor * field,TextGenerator * generator) const2198 void TextFormat::Printer::PrintField(const Message& message,
2199 const Reflection* reflection,
2200 const FieldDescriptor* field,
2201 TextGenerator* generator) const {
2202 if (use_short_repeated_primitives_ && field->is_repeated() &&
2203 field->cpp_type() != FieldDescriptor::CPPTYPE_STRING &&
2204 field->cpp_type() != FieldDescriptor::CPPTYPE_MESSAGE) {
2205 PrintShortRepeatedField(message, reflection, field, generator);
2206 return;
2207 }
2208
2209 int count = 0;
2210
2211 if (field->is_repeated()) {
2212 count = reflection->FieldSize(message, field);
2213 } else if (reflection->HasField(message, field) ||
2214 field->containing_type()->options().map_entry()) {
2215 count = 1;
2216 }
2217
2218 DynamicMessageFactory factory;
2219 std::vector<const Message*> sorted_map_field;
2220 bool need_release = false;
2221 bool is_map = field->is_map();
2222 if (is_map) {
2223 need_release = internal::MapFieldPrinterHelper::SortMap(
2224 message, reflection, field, &factory, &sorted_map_field);
2225 }
2226
2227 for (int j = 0; j < count; ++j) {
2228 const int field_index = field->is_repeated() ? j : -1;
2229
2230 PrintFieldName(message, field_index, count, reflection, field, generator);
2231
2232 if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
2233 const FastFieldValuePrinter* printer = FindWithDefault(
2234 custom_printers_, field, default_field_value_printer_.get());
2235 const Message& sub_message =
2236 field->is_repeated()
2237 ? (is_map ? *sorted_map_field[j]
2238 : reflection->GetRepeatedMessage(message, field, j))
2239 : reflection->GetMessage(message, field);
2240 printer->PrintMessageStart(sub_message, field_index, count,
2241 single_line_mode_, generator);
2242 generator->Indent();
2243 Print(sub_message, generator);
2244 generator->Outdent();
2245 printer->PrintMessageEnd(sub_message, field_index, count,
2246 single_line_mode_, generator);
2247 } else {
2248 generator->PrintLiteral(": ");
2249 // Write the field value.
2250 PrintFieldValue(message, reflection, field, field_index, generator);
2251 if (single_line_mode_) {
2252 generator->PrintLiteral(" ");
2253 } else {
2254 generator->PrintLiteral("\n");
2255 }
2256 }
2257 }
2258
2259 if (need_release) {
2260 for (int j = 0; j < sorted_map_field.size(); ++j) {
2261 delete sorted_map_field[j];
2262 }
2263 }
2264 }
2265
PrintShortRepeatedField(const Message & message,const Reflection * reflection,const FieldDescriptor * field,TextGenerator * generator) const2266 void TextFormat::Printer::PrintShortRepeatedField(
2267 const Message& message, const Reflection* reflection,
2268 const FieldDescriptor* field, TextGenerator* generator) const {
2269 // Print primitive repeated field in short form.
2270 int size = reflection->FieldSize(message, field);
2271 PrintFieldName(message, /*field_index=*/-1, /*field_count=*/size, reflection,
2272 field, generator);
2273 generator->PrintLiteral(": [");
2274 for (int i = 0; i < size; i++) {
2275 if (i > 0) generator->PrintLiteral(", ");
2276 PrintFieldValue(message, reflection, field, i, generator);
2277 }
2278 if (single_line_mode_) {
2279 generator->PrintLiteral("] ");
2280 } else {
2281 generator->PrintLiteral("]\n");
2282 }
2283 }
2284
PrintFieldName(const Message & message,int field_index,int field_count,const Reflection * reflection,const FieldDescriptor * field,TextGenerator * generator) const2285 void TextFormat::Printer::PrintFieldName(const Message& message,
2286 int field_index, int field_count,
2287 const Reflection* reflection,
2288 const FieldDescriptor* field,
2289 TextGenerator* generator) const {
2290 // if use_field_number_ is true, prints field number instead
2291 // of field name.
2292 if (use_field_number_) {
2293 generator->PrintString(StrCat(field->number()));
2294 return;
2295 }
2296
2297 const FastFieldValuePrinter* printer = FindWithDefault(
2298 custom_printers_, field, default_field_value_printer_.get());
2299 printer->PrintFieldName(message, field_index, field_count, reflection, field,
2300 generator);
2301 }
2302
PrintFieldValue(const Message & message,const Reflection * reflection,const FieldDescriptor * field,int index,TextGenerator * generator) const2303 void TextFormat::Printer::PrintFieldValue(const Message& message,
2304 const Reflection* reflection,
2305 const FieldDescriptor* field,
2306 int index,
2307 TextGenerator* generator) const {
2308 GOOGLE_DCHECK(field->is_repeated() || (index == -1))
2309 << "Index must be -1 for non-repeated fields";
2310
2311 const FastFieldValuePrinter* printer = FindWithDefault(
2312 custom_printers_, field, default_field_value_printer_.get());
2313
2314 switch (field->cpp_type()) {
2315 #define OUTPUT_FIELD(CPPTYPE, METHOD) \
2316 case FieldDescriptor::CPPTYPE_##CPPTYPE: \
2317 printer->Print##METHOD( \
2318 field->is_repeated() \
2319 ? reflection->GetRepeated##METHOD(message, field, index) \
2320 : reflection->Get##METHOD(message, field), \
2321 generator); \
2322 break
2323
2324 OUTPUT_FIELD(INT32, Int32);
2325 OUTPUT_FIELD(INT64, Int64);
2326 OUTPUT_FIELD(UINT32, UInt32);
2327 OUTPUT_FIELD(UINT64, UInt64);
2328 OUTPUT_FIELD(FLOAT, Float);
2329 OUTPUT_FIELD(DOUBLE, Double);
2330 OUTPUT_FIELD(BOOL, Bool);
2331 #undef OUTPUT_FIELD
2332
2333 case FieldDescriptor::CPPTYPE_STRING: {
2334 std::string scratch;
2335 const std::string& value =
2336 field->is_repeated()
2337 ? reflection->GetRepeatedStringReference(message, field, index,
2338 &scratch)
2339 : reflection->GetStringReference(message, field, &scratch);
2340 const std::string* value_to_print = &value;
2341 std::string truncated_value;
2342 if (truncate_string_field_longer_than_ > 0 &&
2343 truncate_string_field_longer_than_ < value.size()) {
2344 truncated_value = value.substr(0, truncate_string_field_longer_than_) +
2345 "...<truncated>...";
2346 value_to_print = &truncated_value;
2347 }
2348 if (field->type() == FieldDescriptor::TYPE_STRING) {
2349 printer->PrintString(*value_to_print, generator);
2350 } else {
2351 GOOGLE_DCHECK_EQ(field->type(), FieldDescriptor::TYPE_BYTES);
2352 printer->PrintBytes(*value_to_print, generator);
2353 }
2354 break;
2355 }
2356
2357 case FieldDescriptor::CPPTYPE_ENUM: {
2358 int enum_value =
2359 field->is_repeated()
2360 ? reflection->GetRepeatedEnumValue(message, field, index)
2361 : reflection->GetEnumValue(message, field);
2362 const EnumValueDescriptor* enum_desc =
2363 field->enum_type()->FindValueByNumber(enum_value);
2364 if (enum_desc != nullptr) {
2365 printer->PrintEnum(enum_value, enum_desc->name(), generator);
2366 } else {
2367 // Ordinarily, enum_desc should not be null, because proto2 has the
2368 // invariant that set enum field values must be in-range, but with the
2369 // new integer-based API for enums (or the RepeatedField<int> loophole),
2370 // it is possible for the user to force an unknown integer value. So we
2371 // simply use the integer value itself as the enum value name in this
2372 // case.
2373 printer->PrintEnum(enum_value, StringPrintf("%d", enum_value),
2374 generator);
2375 }
2376 break;
2377 }
2378
2379 case FieldDescriptor::CPPTYPE_MESSAGE:
2380 Print(field->is_repeated()
2381 ? reflection->GetRepeatedMessage(message, field, index)
2382 : reflection->GetMessage(message, field),
2383 generator);
2384 break;
2385 }
2386 }
2387
Print(const Message & message,io::ZeroCopyOutputStream * output)2388 /* static */ bool TextFormat::Print(const Message& message,
2389 io::ZeroCopyOutputStream* output) {
2390 return Printer().Print(message, output);
2391 }
2392
PrintUnknownFields(const UnknownFieldSet & unknown_fields,io::ZeroCopyOutputStream * output)2393 /* static */ bool TextFormat::PrintUnknownFields(
2394 const UnknownFieldSet& unknown_fields, io::ZeroCopyOutputStream* output) {
2395 return Printer().PrintUnknownFields(unknown_fields, output);
2396 }
2397
PrintToString(const Message & message,std::string * output)2398 /* static */ bool TextFormat::PrintToString(const Message& message,
2399 std::string* output) {
2400 return Printer().PrintToString(message, output);
2401 }
2402
PrintUnknownFieldsToString(const UnknownFieldSet & unknown_fields,std::string * output)2403 /* static */ bool TextFormat::PrintUnknownFieldsToString(
2404 const UnknownFieldSet& unknown_fields, std::string* output) {
2405 return Printer().PrintUnknownFieldsToString(unknown_fields, output);
2406 }
2407
PrintFieldValueToString(const Message & message,const FieldDescriptor * field,int index,std::string * output)2408 /* static */ void TextFormat::PrintFieldValueToString(
2409 const Message& message, const FieldDescriptor* field, int index,
2410 std::string* output) {
2411 return Printer().PrintFieldValueToString(message, field, index, output);
2412 }
2413
ParseFieldValueFromString(const std::string & input,const FieldDescriptor * field,Message * message)2414 /* static */ bool TextFormat::ParseFieldValueFromString(
2415 const std::string& input, const FieldDescriptor* field, Message* message) {
2416 return Parser().ParseFieldValueFromString(input, field, message);
2417 }
2418
PrintUnknownFields(const UnknownFieldSet & unknown_fields,TextGenerator * generator) const2419 void TextFormat::Printer::PrintUnknownFields(
2420 const UnknownFieldSet& unknown_fields, TextGenerator* generator) const {
2421 for (int i = 0; i < unknown_fields.field_count(); i++) {
2422 const UnknownField& field = unknown_fields.field(i);
2423 std::string field_number = StrCat(field.number());
2424
2425 switch (field.type()) {
2426 case UnknownField::TYPE_VARINT:
2427 generator->PrintString(field_number);
2428 generator->PrintLiteral(": ");
2429 generator->PrintString(StrCat(field.varint()));
2430 if (single_line_mode_) {
2431 generator->PrintLiteral(" ");
2432 } else {
2433 generator->PrintLiteral("\n");
2434 }
2435 break;
2436 case UnknownField::TYPE_FIXED32: {
2437 generator->PrintString(field_number);
2438 generator->PrintLiteral(": 0x");
2439 generator->PrintString(
2440 StrCat(strings::Hex(field.fixed32(), strings::ZERO_PAD_8)));
2441 if (single_line_mode_) {
2442 generator->PrintLiteral(" ");
2443 } else {
2444 generator->PrintLiteral("\n");
2445 }
2446 break;
2447 }
2448 case UnknownField::TYPE_FIXED64: {
2449 generator->PrintString(field_number);
2450 generator->PrintLiteral(": 0x");
2451 generator->PrintString(
2452 StrCat(strings::Hex(field.fixed64(), strings::ZERO_PAD_16)));
2453 if (single_line_mode_) {
2454 generator->PrintLiteral(" ");
2455 } else {
2456 generator->PrintLiteral("\n");
2457 }
2458 break;
2459 }
2460 case UnknownField::TYPE_LENGTH_DELIMITED: {
2461 generator->PrintString(field_number);
2462 const std::string& value = field.length_delimited();
2463 UnknownFieldSet embedded_unknown_fields;
2464 if (!value.empty() && embedded_unknown_fields.ParseFromString(value)) {
2465 // This field is parseable as a Message.
2466 // So it is probably an embedded message.
2467 if (single_line_mode_) {
2468 generator->PrintLiteral(" { ");
2469 } else {
2470 generator->PrintLiteral(" {\n");
2471 generator->Indent();
2472 }
2473 PrintUnknownFields(embedded_unknown_fields, generator);
2474 if (single_line_mode_) {
2475 generator->PrintLiteral("} ");
2476 } else {
2477 generator->Outdent();
2478 generator->PrintLiteral("}\n");
2479 }
2480 } else {
2481 // This field is not parseable as a Message.
2482 // So it is probably just a plain string.
2483 generator->PrintLiteral(": \"");
2484 generator->PrintString(CEscape(value));
2485 if (single_line_mode_) {
2486 generator->PrintLiteral("\" ");
2487 } else {
2488 generator->PrintLiteral("\"\n");
2489 }
2490 }
2491 break;
2492 }
2493 case UnknownField::TYPE_GROUP:
2494 generator->PrintString(field_number);
2495 if (single_line_mode_) {
2496 generator->PrintLiteral(" { ");
2497 } else {
2498 generator->PrintLiteral(" {\n");
2499 generator->Indent();
2500 }
2501 PrintUnknownFields(field.group(), generator);
2502 if (single_line_mode_) {
2503 generator->PrintLiteral("} ");
2504 } else {
2505 generator->Outdent();
2506 generator->PrintLiteral("}\n");
2507 }
2508 break;
2509 }
2510 }
2511 }
2512
2513 } // namespace protobuf
2514 } // namespace google
2515