Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(62)

Side by Side Diff: third_party/protobuf/src/google/protobuf/util/message_differencer.cc

Issue 1291903002: Pull new version of protobuf sources. (Closed) Base URL: https://chromium.googlesource.com/chromium/src.git@master
Patch Set: Created 5 years, 4 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
OLDNEW
(Empty)
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 // This file defines static methods and classes for comparing Protocol
36 // Messages (see //google/protobuf/util/message_differencer.h for more
37 // information).
38
39 #include <google/protobuf/util/message_differencer.h>
40
41 #include <algorithm>
42 #include <memory>
43 #ifndef _SHARED_PTR_H
44 #include <google/protobuf/stubs/shared_ptr.h>
45 #endif
46 #include <utility>
47
48 #include <google/protobuf/stubs/common.h>
49 #include <google/protobuf/stubs/stringprintf.h>
50 #include <google/protobuf/any.h>
51 #include <google/protobuf/io/printer.h>
52 #include <google/protobuf/io/zero_copy_stream.h>
53 #include <google/protobuf/io/zero_copy_stream_impl.h>
54 #include <google/protobuf/dynamic_message.h>
55 #include <google/protobuf/text_format.h>
56 #include <google/protobuf/util/field_comparator.h>
57 #include <google/protobuf/stubs/strutil.h>
58
59 namespace google {
60 namespace protobuf {
61
62 namespace util {
63
64 // When comparing a repeated field as map, MultipleFieldMapKeyComparator can
65 // be used to specify multiple fields as key for key comparison.
66 // Two elements of a repeated field will be regarded as having the same key
67 // iff they have the same value for every specified key field.
68 // Note that you can also specify only one field as key.
69 class MessageDifferencer::MultipleFieldsMapKeyComparator
70 : public MessageDifferencer::MapKeyComparator {
71 public:
72 MultipleFieldsMapKeyComparator(
73 MessageDifferencer* message_differencer,
74 const vector<vector<const FieldDescriptor*> >& key_field_paths)
75 : message_differencer_(message_differencer),
76 key_field_paths_(key_field_paths) {
77 GOOGLE_CHECK(!key_field_paths_.empty());
78 for (int i = 0; i < key_field_paths_.size(); ++i) {
79 GOOGLE_CHECK(!key_field_paths_[i].empty());
80 }
81 }
82 MultipleFieldsMapKeyComparator(
83 MessageDifferencer* message_differencer,
84 const FieldDescriptor* key)
85 : message_differencer_(message_differencer) {
86 vector<const FieldDescriptor*> key_field_path;
87 key_field_path.push_back(key);
88 key_field_paths_.push_back(key_field_path);
89 }
90 virtual bool IsMatch(
91 const Message& message1,
92 const Message& message2,
93 const vector<SpecificField>& parent_fields) const {
94 for (int i = 0; i < key_field_paths_.size(); ++i) {
95 if (!IsMatchInternal(message1, message2, parent_fields,
96 key_field_paths_[i], 0)) {
97 return false;
98 }
99 }
100 return true;
101 }
102 private:
103 bool IsMatchInternal(
104 const Message& message1,
105 const Message& message2,
106 const vector<SpecificField>& parent_fields,
107 const vector<const FieldDescriptor*>& key_field_path,
108 int path_index) const {
109 const FieldDescriptor* field = key_field_path[path_index];
110 vector<SpecificField> current_parent_fields(parent_fields);
111 if (path_index == key_field_path.size() - 1) {
112 if (field->is_repeated()) {
113 if (!message_differencer_->CompareRepeatedField(
114 message1, message2, field, &current_parent_fields)) {
115 return false;
116 }
117 } else {
118 if (!message_differencer_->CompareFieldValueUsingParentFields(
119 message1, message2, field, -1, -1, &current_parent_fields)) {
120 return false;
121 }
122 }
123 return true;
124 } else {
125 const Reflection* reflection1 = message1.GetReflection();
126 const Reflection* reflection2 = message2.GetReflection();
127 bool has_field1 = reflection1->HasField(message1, field);
128 bool has_field2 = reflection2->HasField(message2, field);
129 if (!has_field1 && !has_field2) {
130 return true;
131 }
132 if (has_field1 != has_field2) {
133 return false;
134 }
135 SpecificField specific_field;
136 specific_field.field = field;
137 current_parent_fields.push_back(specific_field);
138 return IsMatchInternal(
139 reflection1->GetMessage(message1, field),
140 reflection2->GetMessage(message2, field),
141 current_parent_fields,
142 key_field_path,
143 path_index + 1);
144 }
145 }
146 MessageDifferencer* message_differencer_;
147 vector<vector<const FieldDescriptor*> > key_field_paths_;
148 GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(MultipleFieldsMapKeyComparator);
149 };
150
151 bool MessageDifferencer::Equals(const Message& message1,
152 const Message& message2) {
153 MessageDifferencer differencer;
154
155 return differencer.Compare(message1, message2);
156 }
157
158 bool MessageDifferencer::Equivalent(const Message& message1,
159 const Message& message2) {
160 MessageDifferencer differencer;
161 differencer.set_message_field_comparison(MessageDifferencer::EQUIVALENT);
162
163 return differencer.Compare(message1, message2);
164 }
165
166 bool MessageDifferencer::ApproximatelyEquals(const Message& message1,
167 const Message& message2) {
168 MessageDifferencer differencer;
169 differencer.set_float_comparison(
170 MessageDifferencer::APPROXIMATE);
171
172 return differencer.Compare(message1, message2);
173 }
174
175 bool MessageDifferencer::ApproximatelyEquivalent(const Message& message1,
176 const Message& message2) {
177 MessageDifferencer differencer;
178 differencer.set_message_field_comparison(MessageDifferencer::EQUIVALENT);
179 differencer.set_float_comparison(MessageDifferencer::APPROXIMATE);
180
181 return differencer.Compare(message1, message2);
182 }
183
184 // ===========================================================================
185
186 MessageDifferencer::MessageDifferencer()
187 : reporter_(NULL),
188 field_comparator_(NULL),
189 message_field_comparison_(EQUAL),
190 scope_(FULL),
191 repeated_field_comparison_(AS_LIST),
192 report_matches_(false),
193 output_string_(NULL) { }
194
195 MessageDifferencer::~MessageDifferencer() {
196 for (int i = 0; i < owned_key_comparators_.size(); ++i) {
197 delete owned_key_comparators_[i];
198 }
199 for (int i = 0; i < ignore_criteria_.size(); ++i) {
200 delete ignore_criteria_[i];
201 }
202 }
203
204 void MessageDifferencer::set_field_comparator(FieldComparator* comparator) {
205 GOOGLE_CHECK(comparator) << "Field comparator can't be NULL.";
206 field_comparator_ = comparator;
207 }
208
209 void MessageDifferencer::set_message_field_comparison(
210 MessageFieldComparison comparison) {
211 message_field_comparison_ = comparison;
212 }
213
214 void MessageDifferencer::set_scope(Scope scope) {
215 scope_ = scope;
216 }
217
218 MessageDifferencer::Scope MessageDifferencer::scope() {
219 return scope_;
220 }
221
222 void MessageDifferencer::set_float_comparison(FloatComparison comparison) {
223 default_field_comparator_.set_float_comparison(
224 comparison == EXACT ?
225 DefaultFieldComparator::EXACT : DefaultFieldComparator::APPROXIMATE);
226 }
227
228 void MessageDifferencer::set_repeated_field_comparison(
229 RepeatedFieldComparison comparison) {
230 repeated_field_comparison_ = comparison;
231 }
232
233 void MessageDifferencer::TreatAsSet(const FieldDescriptor* field) {
234 GOOGLE_CHECK(field->is_repeated()) << "Field must be repeated: "
235 << field->full_name();
236 const MapKeyComparator* key_comparator = GetMapKeyComparator(field);
237 GOOGLE_CHECK(key_comparator == NULL)
238 << "Cannot treat this repeated field as both Map and Set for"
239 << " comparison. Field name is: " << field->full_name();
240 set_fields_.insert(field);
241 }
242
243 void MessageDifferencer::TreatAsMap(const FieldDescriptor* field,
244 const FieldDescriptor* key) {
245 GOOGLE_CHECK(field->is_repeated()) << "Field must be repeated: "
246 << field->full_name();
247 GOOGLE_CHECK_EQ(FieldDescriptor::CPPTYPE_MESSAGE, field->cpp_type())
248 << "Field has to be message type. Field name is: "
249 << field->full_name();
250 GOOGLE_CHECK(key->containing_type() == field->message_type())
251 << key->full_name()
252 << " must be a direct subfield within the repeated field "
253 << field->full_name() << ", not " << key->containing_type()->full_name();
254 GOOGLE_CHECK(set_fields_.find(field) == set_fields_.end())
255 << "Cannot treat this repeated field as both Map and Set for "
256 << "comparison.";
257 MapKeyComparator* key_comparator =
258 new MultipleFieldsMapKeyComparator(this, key);
259 owned_key_comparators_.push_back(key_comparator);
260 map_field_key_comparator_[field] = key_comparator;
261 }
262
263 void MessageDifferencer::TreatAsMapWithMultipleFieldsAsKey(
264 const FieldDescriptor* field,
265 const vector<const FieldDescriptor*>& key_fields) {
266 vector<vector<const FieldDescriptor*> > key_field_paths;
267 for (int i = 0; i < key_fields.size(); ++i) {
268 vector<const FieldDescriptor*> key_field_path;
269 key_field_path.push_back(key_fields[i]);
270 key_field_paths.push_back(key_field_path);
271 }
272 TreatAsMapWithMultipleFieldPathsAsKey(field, key_field_paths);
273 }
274
275 void MessageDifferencer::TreatAsMapWithMultipleFieldPathsAsKey(
276 const FieldDescriptor* field,
277 const vector<vector<const FieldDescriptor*> >& key_field_paths) {
278 GOOGLE_CHECK(field->is_repeated()) << "Field must be repeated: "
279 << field->full_name();
280 GOOGLE_CHECK_EQ(FieldDescriptor::CPPTYPE_MESSAGE, field->cpp_type())
281 << "Field has to be message type. Field name is: "
282 << field->full_name();
283 for (int i = 0; i < key_field_paths.size(); ++i) {
284 const vector<const FieldDescriptor*>& key_field_path = key_field_paths[i];
285 for (int j = 0; j < key_field_path.size(); ++j) {
286 const FieldDescriptor* parent_field =
287 j == 0 ? field : key_field_path[j - 1];
288 const FieldDescriptor* child_field = key_field_path[j];
289 GOOGLE_CHECK(child_field->containing_type() == parent_field->message_type( ))
290 << child_field->full_name()
291 << " must be a direct subfield within the field: "
292 << parent_field->full_name();
293 if (j != 0) {
294 GOOGLE_CHECK_EQ(FieldDescriptor::CPPTYPE_MESSAGE, parent_field->cpp_type ())
295 << parent_field->full_name() << " has to be of type message.";
296 GOOGLE_CHECK(!parent_field->is_repeated())
297 << parent_field->full_name() << " cannot be a repeated field.";
298 }
299 }
300 }
301 GOOGLE_CHECK(set_fields_.find(field) == set_fields_.end())
302 << "Cannot treat this repeated field as both Map and Set for "
303 << "comparison.";
304 MapKeyComparator* key_comparator =
305 new MultipleFieldsMapKeyComparator(this, key_field_paths);
306 owned_key_comparators_.push_back(key_comparator);
307 map_field_key_comparator_[field] = key_comparator;
308 }
309
310 void MessageDifferencer::TreatAsMapUsingKeyComparator(
311 const FieldDescriptor* field,
312 const MapKeyComparator* key_comparator) {
313 GOOGLE_CHECK(field->is_repeated()) << "Field must be repeated: "
314 << field->full_name();
315 GOOGLE_CHECK_EQ(FieldDescriptor::CPPTYPE_MESSAGE, field->cpp_type())
316 << "Field has to be message type. Field name is: "
317 << field->full_name();
318 GOOGLE_CHECK(set_fields_.find(field) == set_fields_.end())
319 << "Cannot treat this repeated field as both Map and Set for "
320 << "comparison.";
321 map_field_key_comparator_[field] = key_comparator;
322 }
323
324 void MessageDifferencer::AddIgnoreCriteria(IgnoreCriteria* ignore_criteria) {
325 ignore_criteria_.push_back(ignore_criteria);
326 }
327
328 void MessageDifferencer::IgnoreField(const FieldDescriptor* field) {
329 ignored_fields_.insert(field);
330 }
331
332 void MessageDifferencer::SetFractionAndMargin(const FieldDescriptor* field,
333 double fraction, double margin) {
334 default_field_comparator_.SetFractionAndMargin(field, fraction, margin);
335 }
336
337 void MessageDifferencer::ReportDifferencesToString(string* output) {
338 GOOGLE_DCHECK(output) << "Specified output string was NULL";
339
340 output_string_ = output;
341 output_string_->clear();
342 }
343
344 void MessageDifferencer::ReportDifferencesTo(Reporter* reporter) {
345 // If an output string is set, clear it to prevent
346 // it superceding the specified reporter.
347 if (output_string_) {
348 output_string_ = NULL;
349 }
350
351 reporter_ = reporter;
352 }
353
354 bool MessageDifferencer::FieldBefore(const FieldDescriptor* field1,
355 const FieldDescriptor* field2) {
356 // Handle sentinel values (i.e. make sure NULLs are always ordered
357 // at the end of the list).
358 if (field1 == NULL) {
359 return false;
360 }
361
362 if (field2 == NULL) {
363 return true;
364 }
365
366 // Always order fields by their tag number
367 return (field1->number() < field2->number());
368 }
369
370 bool MessageDifferencer::Compare(const Message& message1,
371 const Message& message2) {
372 vector<SpecificField> parent_fields;
373
374 bool result = false;
375
376 // Setup the internal reporter if need be.
377 if (output_string_) {
378 io::StringOutputStream output_stream(output_string_);
379 StreamReporter reporter(&output_stream);
380 reporter_ = &reporter;
381 result = Compare(message1, message2, &parent_fields);
382 reporter_ = NULL;
383 } else {
384 result = Compare(message1, message2, &parent_fields);
385 }
386
387 return result;
388 }
389
390 bool MessageDifferencer::CompareWithFields(
391 const Message& message1,
392 const Message& message2,
393 const vector<const FieldDescriptor*>& message1_fields_arg,
394 const vector<const FieldDescriptor*>& message2_fields_arg) {
395 if (message1.GetDescriptor() != message2.GetDescriptor()) {
396 GOOGLE_LOG(DFATAL) << "Comparison between two messages with different "
397 << "descriptors.";
398 return false;
399 }
400
401 vector<SpecificField> parent_fields;
402
403 bool result = false;
404
405 vector<const FieldDescriptor*> message1_fields(message1_fields_arg);
406 vector<const FieldDescriptor*> message2_fields(message2_fields_arg);
407
408 std::sort(message1_fields.begin(), message1_fields.end(), FieldBefore);
409 std::sort(message2_fields.begin(), message2_fields.end(), FieldBefore);
410 // Append NULL sentinel values.
411 message1_fields.push_back(NULL);
412 message2_fields.push_back(NULL);
413
414 // Setup the internal reporter if need be.
415 if (output_string_) {
416 io::StringOutputStream output_stream(output_string_);
417 StreamReporter reporter(&output_stream);
418 reporter_ = &reporter;
419 result = CompareRequestedFieldsUsingSettings(
420 message1, message2, message1_fields, message2_fields, &parent_fields);
421 reporter_ = NULL;
422 } else {
423 result = CompareRequestedFieldsUsingSettings(
424 message1, message2, message1_fields, message2_fields, &parent_fields);
425 }
426
427 return result;
428 }
429
430 bool MessageDifferencer::Compare(
431 const Message& message1,
432 const Message& message2,
433 vector<SpecificField>* parent_fields) {
434 const Descriptor* descriptor1 = message1.GetDescriptor();
435 const Descriptor* descriptor2 = message2.GetDescriptor();
436 if (descriptor1 != descriptor2) {
437 GOOGLE_LOG(DFATAL) << "Comparison between two messages with different "
438 << "descriptors.";
439 return false;
440 }
441 // Expand google.protobuf.Any payload if possible.
442 if (descriptor1->full_name() == internal::kAnyFullTypeName) {
443 google::protobuf::scoped_ptr<Message> data1;
444 google::protobuf::scoped_ptr<Message> data2;
445 if (UnpackAny(message1, &data1) && UnpackAny(message2, &data2)) {
446 return Compare(*data1, *data2, parent_fields);
447 }
448 }
449 const Reflection* reflection1 = message1.GetReflection();
450 const Reflection* reflection2 = message2.GetReflection();
451
452 // Retrieve all the set fields, including extensions.
453 vector<const FieldDescriptor*> message1_fields;
454 vector<const FieldDescriptor*> message2_fields;
455
456 reflection1->ListFields(message1, &message1_fields);
457 reflection2->ListFields(message2, &message2_fields);
458
459 // Add sentinel values to deal with the
460 // case where the number of the fields in
461 // each list are different.
462 message1_fields.push_back(NULL);
463 message2_fields.push_back(NULL);
464
465 bool unknown_compare_result = true;
466 // Ignore unknown fields in EQUIVALENT mode
467 if (message_field_comparison_ != EQUIVALENT) {
468 const google::protobuf::UnknownFieldSet* unknown_field_set1 =
469 &reflection1->GetUnknownFields(message1);
470 const google::protobuf::UnknownFieldSet* unknown_field_set2 =
471 &reflection2->GetUnknownFields(message2);
472 if (!CompareUnknownFields(message1, message2,
473 *unknown_field_set1, *unknown_field_set2,
474 parent_fields)) {
475 if (reporter_ == NULL) {
476 return false;
477 };
478 unknown_compare_result = false;
479 }
480 }
481
482 return CompareRequestedFieldsUsingSettings(
483 message1, message2,
484 message1_fields, message2_fields,
485 parent_fields) && unknown_compare_result;
486 }
487
488 bool MessageDifferencer::CompareRequestedFieldsUsingSettings(
489 const Message& message1,
490 const Message& message2,
491 const vector<const FieldDescriptor*>& message1_fields,
492 const vector<const FieldDescriptor*>& message2_fields,
493 vector<SpecificField>* parent_fields) {
494 if (scope_ == FULL) {
495 if (message_field_comparison_ == EQUIVALENT) {
496 // We need to merge the field lists of both messages (i.e.
497 // we are merely checking for a difference in field values,
498 // rather than the addition or deletion of fields).
499 vector<const FieldDescriptor*> fields_union;
500 CombineFields(message1_fields, FULL, message2_fields, FULL,
501 &fields_union);
502 return CompareWithFieldsInternal(message1, message2, fields_union,
503 fields_union, parent_fields);
504 } else {
505 // Simple equality comparison, use the unaltered field lists.
506 return CompareWithFieldsInternal(message1, message2, message1_fields,
507 message2_fields, parent_fields);
508 }
509 } else {
510 if (message_field_comparison_ == EQUIVALENT) {
511 // We use the list of fields for message1 for both messages when
512 // comparing. This way, extra fields in message2 are ignored,
513 // and missing fields in message2 use their default value.
514 return CompareWithFieldsInternal(message1, message2, message1_fields,
515 message1_fields, parent_fields);
516 } else {
517 // We need to consider the full list of fields for message1
518 // but only the intersection for message2. This way, any fields
519 // only present in message2 will be ignored, but any fields only
520 // present in message1 will be marked as a difference.
521 vector<const FieldDescriptor*> fields_intersection;
522 CombineFields(message1_fields, PARTIAL, message2_fields, PARTIAL,
523 &fields_intersection);
524 return CompareWithFieldsInternal(message1, message2, message1_fields,
525 fields_intersection, parent_fields);
526 }
527 }
528 }
529
530 void MessageDifferencer::CombineFields(
531 const vector<const FieldDescriptor*>& fields1,
532 Scope fields1_scope,
533 const vector<const FieldDescriptor*>& fields2,
534 Scope fields2_scope,
535 vector<const FieldDescriptor*>* combined_fields) {
536
537 int index1 = 0;
538 int index2 = 0;
539
540 while (index1 < fields1.size() && index2 < fields2.size()) {
541 const FieldDescriptor* field1 = fields1[index1];
542 const FieldDescriptor* field2 = fields2[index2];
543
544 if (FieldBefore(field1, field2)) {
545 if (fields1_scope == FULL) {
546 combined_fields->push_back(fields1[index1]);
547 }
548 ++index1;
549 } else if (FieldBefore(field2, field1)) {
550 if (fields2_scope == FULL) {
551 combined_fields->push_back(fields2[index2]);
552 }
553 ++index2;
554 } else {
555 combined_fields->push_back(fields1[index1]);
556 ++index1;
557 ++index2;
558 }
559 }
560 }
561
562 bool MessageDifferencer::CompareWithFieldsInternal(
563 const Message& message1,
564 const Message& message2,
565 const vector<const FieldDescriptor*>& message1_fields,
566 const vector<const FieldDescriptor*>& message2_fields,
567 vector<SpecificField>* parent_fields) {
568 bool isDifferent = false;
569 int field_index1 = 0;
570 int field_index2 = 0;
571
572 const Reflection* reflection1 = message1.GetReflection();
573 const Reflection* reflection2 = message2.GetReflection();
574
575 while (true) {
576 const FieldDescriptor* field1 = message1_fields[field_index1];
577 const FieldDescriptor* field2 = message2_fields[field_index2];
578
579 // Once we have reached sentinel values, we are done the comparison.
580 if (field1 == NULL && field2 == NULL) {
581 break;
582 }
583
584 // Check for differences in the field itself.
585 if (FieldBefore(field1, field2)) {
586 // Field 1 is not in the field list for message 2.
587 if (IsIgnored(message1, message2, field1, *parent_fields)) {
588 // We are ignoring field1. Report the ignore and move on to
589 // the next field in message1_fields.
590 if (reporter_ != NULL) {
591 SpecificField specific_field;
592 specific_field.field = field1;
593
594 parent_fields->push_back(specific_field);
595 reporter_->ReportIgnored(message1, message2, *parent_fields);
596 parent_fields->pop_back();
597 }
598 ++field_index1;
599 continue;
600 }
601
602 if (reporter_ != NULL) {
603 int count = field1->is_repeated() ?
604 reflection1->FieldSize(message1, field1) : 1;
605
606 for (int i = 0; i < count; ++i) {
607 SpecificField specific_field;
608 specific_field.field = field1;
609 specific_field.index = field1->is_repeated() ? i : -1;
610
611 parent_fields->push_back(specific_field);
612 reporter_->ReportDeleted(message1, message2, *parent_fields);
613 parent_fields->pop_back();
614 }
615
616 isDifferent = true;
617 } else {
618 return false;
619 }
620
621 ++field_index1;
622 continue;
623 } else if (FieldBefore(field2, field1)) {
624 // Field 2 is not in the field list for message 1.
625 if (IsIgnored(message1, message2, field2, *parent_fields)) {
626 // We are ignoring field2. Report the ignore and move on to
627 // the next field in message2_fields.
628 if (reporter_ != NULL) {
629 SpecificField specific_field;
630 specific_field.field = field2;
631
632 parent_fields->push_back(specific_field);
633 reporter_->ReportIgnored(message1, message2, *parent_fields);
634 parent_fields->pop_back();
635 }
636 ++field_index2;
637 continue;
638 }
639
640 if (reporter_ != NULL) {
641 int count = field2->is_repeated() ?
642 reflection2->FieldSize(message2, field2) : 1;
643
644 for (int i = 0; i < count; ++i) {
645 SpecificField specific_field;
646 specific_field.field = field2;
647 specific_field.index = field2->is_repeated() ? i : -1;
648 specific_field.new_index = specific_field.index;
649
650 parent_fields->push_back(specific_field);
651 reporter_->ReportAdded(message1, message2, *parent_fields);
652 parent_fields->pop_back();
653 }
654
655 isDifferent = true;
656 } else {
657 return false;
658 }
659
660 ++field_index2;
661 continue;
662 }
663
664 // By this point, field1 and field2 are guarenteed to point to the same
665 // field, so we can now compare the values.
666 if (IsIgnored(message1, message2, field1, *parent_fields)) {
667 // Ignore this field. Report and move on.
668 if (reporter_ != NULL) {
669 SpecificField specific_field;
670 specific_field.field = field1;
671
672 parent_fields->push_back(specific_field);
673 reporter_->ReportIgnored(message1, message2, *parent_fields);
674 parent_fields->pop_back();
675 }
676
677 ++field_index1;
678 ++field_index2;
679 continue;
680 }
681
682 bool fieldDifferent = false;
683 if (field1->is_repeated()) {
684 fieldDifferent = !CompareRepeatedField(message1, message2, field1,
685 parent_fields);
686 if (fieldDifferent) {
687 if (reporter_ == NULL) return false;
688 isDifferent = true;
689 }
690 } else {
691 fieldDifferent = !CompareFieldValueUsingParentFields(
692 message1, message2, field1, -1, -1, parent_fields);
693
694 // If we have found differences, either report them or terminate if
695 // no reporter is present.
696 if (fieldDifferent && reporter_ == NULL) {
697 return false;
698 }
699
700 if (reporter_ != NULL) {
701 SpecificField specific_field;
702 specific_field.field = field1;
703 parent_fields->push_back(specific_field);
704 if (fieldDifferent) {
705 reporter_->ReportModified(message1, message2, *parent_fields);
706 isDifferent = true;
707 } else if (report_matches_) {
708 reporter_->ReportMatched(message1, message2, *parent_fields);
709 }
710 parent_fields->pop_back();
711 }
712 }
713 // Increment the field indicies.
714 ++field_index1;
715 ++field_index2;
716 }
717
718 return !isDifferent;
719 }
720
721 bool MessageDifferencer::IsMatch(const FieldDescriptor* repeated_field,
722 const MapKeyComparator* key_comparator,
723 const Message* message1,
724 const Message* message2,
725 const vector<SpecificField>& parent_fields,
726 int index1, int index2) {
727 vector<SpecificField> current_parent_fields(parent_fields);
728 if (repeated_field->cpp_type() != FieldDescriptor::CPPTYPE_MESSAGE) {
729 return CompareFieldValueUsingParentFields(
730 *message1, *message2, repeated_field, index1, index2,
731 &current_parent_fields);
732 }
733 // Back up the Reporter and output_string_. They will be reset in the
734 // following code.
735 Reporter* backup_reporter = reporter_;
736 string* output_string = output_string_;
737 reporter_ = NULL;
738 output_string_ = NULL;
739 bool match;
740
741 if (key_comparator == NULL) {
742 match = CompareFieldValueUsingParentFields(
743 *message1, *message2, repeated_field, index1, index2,
744 &current_parent_fields);
745 } else {
746 const Reflection* reflection1 = message1->GetReflection();
747 const Reflection* reflection2 = message2->GetReflection();
748 const Message& m1 =
749 reflection1->GetRepeatedMessage(*message1, repeated_field, index1);
750 const Message& m2 =
751 reflection2->GetRepeatedMessage(*message2, repeated_field, index2);
752 SpecificField specific_field;
753 specific_field.field = repeated_field;
754 current_parent_fields.push_back(specific_field);
755 match = key_comparator->IsMatch(m1, m2, current_parent_fields);
756 }
757
758 reporter_ = backup_reporter;
759 output_string_ = output_string;
760 return match;
761 }
762
763 bool MessageDifferencer::CompareRepeatedField(
764 const Message& message1,
765 const Message& message2,
766 const FieldDescriptor* repeated_field,
767 vector<SpecificField>* parent_fields) {
768 // the input FieldDescriptor is guaranteed to be repeated field.
769 const Reflection* reflection1 = message1.GetReflection();
770 const Reflection* reflection2 = message2.GetReflection();
771 const int count1 = reflection1->FieldSize(message1, repeated_field);
772 const int count2 = reflection2->FieldSize(message2, repeated_field);
773 const bool treated_as_subset = IsTreatedAsSubset(repeated_field);
774
775 // If the field is not treated as subset and no detailed reports is needed,
776 // we do a quick check on the number of the elements to avoid unnecessary
777 // comparison.
778 if (count1 != count2 && reporter_ == NULL && !treated_as_subset) {
779 return false;
780 }
781 // A match can never be found if message1 has more items than message2.
782 if (count1 > count2 && reporter_ == NULL) {
783 return false;
784 }
785
786 // These two list are used for store the index of the correspondent
787 // element in peer repeated field.
788 vector<int> match_list1;
789 vector<int> match_list2;
790
791 // Try to match indices of the repeated fields. Return false if match fails
792 // and there's no detailed report needed.
793 if (!MatchRepeatedFieldIndices(message1, message2, repeated_field,
794 *parent_fields, &match_list1, &match_list2) &&
795 reporter_ == NULL) {
796 return false;
797 }
798
799 bool fieldDifferent = false;
800 SpecificField specific_field;
801 specific_field.field = repeated_field;
802
803 // At this point, we have already matched pairs of fields (with the reporting
804 // to be done later). Now to check if the paired elements are different.
805 for (int i = 0; i < count1; i++) {
806 if (match_list1[i] == -1) continue;
807 specific_field.index = i;
808 specific_field.new_index = match_list1[i];
809
810 const bool result = CompareFieldValueUsingParentFields(
811 message1, message2, repeated_field, i, specific_field.new_index,
812 parent_fields);
813
814 // If we have found differences, either report them or terminate if
815 // no reporter is present. Note that ReportModified, ReportMoved, and
816 // ReportMatched are all mutually exclusive.
817 if (!result) {
818 if (reporter_ == NULL) return false;
819 parent_fields->push_back(specific_field);
820 reporter_->ReportModified(message1, message2, *parent_fields);
821 parent_fields->pop_back();
822 fieldDifferent = true;
823 } else if (reporter_ != NULL &&
824 specific_field.index != specific_field.new_index) {
825 parent_fields->push_back(specific_field);
826 reporter_->ReportMoved(message1, message2, *parent_fields);
827 parent_fields->pop_back();
828 } else if (report_matches_ && reporter_ != NULL) {
829 parent_fields->push_back(specific_field);
830 reporter_->ReportMatched(message1, message2, *parent_fields);
831 parent_fields->pop_back();
832 }
833 }
834
835 // Report any remaining additions or deletions.
836 for (int i = 0; i < count2; ++i) {
837 if (match_list2[i] != -1) continue;
838 if (!treated_as_subset) {
839 fieldDifferent = true;
840 }
841
842 if (reporter_ == NULL) continue;
843 specific_field.index = i;
844 specific_field.new_index = i;
845 parent_fields->push_back(specific_field);
846 reporter_->ReportAdded(message1, message2, *parent_fields);
847 parent_fields->pop_back();
848 }
849
850 for (int i = 0; i < count1; ++i) {
851 if (match_list1[i] != -1) continue;
852 specific_field.index = i;
853 parent_fields->push_back(specific_field);
854 reporter_->ReportDeleted(message1, message2, *parent_fields);
855 parent_fields->pop_back();
856 fieldDifferent = true;
857 }
858 return !fieldDifferent;
859 }
860
861 bool MessageDifferencer::CompareFieldValue(const Message& message1,
862 const Message& message2,
863 const FieldDescriptor* field,
864 int index1,
865 int index2) {
866 return CompareFieldValueUsingParentFields(message1, message2, field, index1,
867 index2, NULL);
868 }
869
870 bool MessageDifferencer::CompareFieldValueUsingParentFields(
871 const Message& message1, const Message& message2,
872 const FieldDescriptor* field, int index1, int index2,
873 vector<SpecificField>* parent_fields) {
874 FieldContext field_context(parent_fields);
875 FieldComparator::ComparisonResult result = GetFieldComparisonResult(
876 message1, message2, field, index1, index2, &field_context);
877
878 if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE &&
879 result == FieldComparator::RECURSE) {
880 // Get the nested messages and compare them using one of the Compare
881 // methods.
882 const Reflection* reflection1 = message1.GetReflection();
883 const Reflection* reflection2 = message2.GetReflection();
884 const Message& m1 = field->is_repeated() ?
885 reflection1->GetRepeatedMessage(message1, field, index1) :
886 reflection1->GetMessage(message1, field);
887 const Message& m2 = field->is_repeated() ?
888 reflection2->GetRepeatedMessage(message2, field, index2) :
889 reflection2->GetMessage(message2, field);
890
891 // parent_fields is used in calls to Reporter methods.
892 if (parent_fields != NULL) {
893 // Append currently compared field to the end of parent_fields.
894 SpecificField specific_field;
895 specific_field.field = field;
896 specific_field.index = index1;
897 specific_field.new_index = index2;
898 parent_fields->push_back(specific_field);
899 const bool compare_result = Compare(m1, m2, parent_fields);
900 parent_fields->pop_back();
901 return compare_result;
902 } else {
903 // Recreates parent_fields as if m1 and m2 had no parents.
904 return Compare(m1, m2);
905 }
906 } else {
907 return (result == FieldComparator::SAME);
908 }
909 }
910
911 bool MessageDifferencer::CheckPathChanged(
912 const vector<SpecificField>& field_path) {
913 for (int i = 0; i < field_path.size(); ++i) {
914 if (field_path[i].index != field_path[i].new_index) return true;
915 }
916 return false;
917 }
918
919 bool MessageDifferencer::IsTreatedAsSet(const FieldDescriptor* field) {
920 if (!field->is_repeated()) return false;
921 if (field->is_map()) return true;
922 if (repeated_field_comparison_ == AS_SET) return true;
923 return (set_fields_.find(field) != set_fields_.end());
924 }
925
926 bool MessageDifferencer::IsTreatedAsSubset(const FieldDescriptor* field) {
927 return scope_ == PARTIAL &&
928 (IsTreatedAsSet(field) || GetMapKeyComparator(field) != NULL);
929 }
930
931 bool MessageDifferencer::IsIgnored(
932 const Message& message1,
933 const Message& message2,
934 const FieldDescriptor* field,
935 const vector<SpecificField>& parent_fields) {
936 if (ignored_fields_.find(field) != ignored_fields_.end()) {
937 return true;
938 }
939 for (int i = 0; i < ignore_criteria_.size(); ++i) {
940 if (ignore_criteria_[i]->IsIgnored(message1, message2, field,
941 parent_fields)) {
942 return true;
943 }
944 }
945 return false;
946 }
947
948 const MessageDifferencer::MapKeyComparator* MessageDifferencer
949 ::GetMapKeyComparator(const FieldDescriptor* field) {
950 if (!field->is_repeated()) return NULL;
951 if (map_field_key_comparator_.find(field) !=
952 map_field_key_comparator_.end()) {
953 return map_field_key_comparator_[field];
954 }
955 return NULL;
956 }
957
958 namespace {
959
960 typedef pair<int, const UnknownField*> IndexUnknownFieldPair;
961
962 struct UnknownFieldOrdering {
963 inline bool operator()(const IndexUnknownFieldPair& a,
964 const IndexUnknownFieldPair& b) const {
965 if (a.second->number() < b.second->number()) return true;
966 if (a.second->number() > b.second->number()) return false;
967 return a.second->type() < b.second->type();
968 }
969 };
970
971 } // namespace
972
973 bool MessageDifferencer::UnpackAny(const Message& any,
974 google::protobuf::scoped_ptr<Message>* data) {
975 const Reflection* reflection = any.GetReflection();
976 const FieldDescriptor* type_url_field;
977 const FieldDescriptor* value_field;
978 if (!internal::GetAnyFieldDescriptors(any, &type_url_field, &value_field)) {
979 return false;
980 }
981 const string& type_url = reflection->GetString(any, type_url_field);
982 string full_type_name;
983 if (!internal::ParseAnyTypeUrl(type_url, &full_type_name)) {
984 return false;
985 }
986
987 const google::protobuf::Descriptor* desc =
988 any.GetDescriptor()->file()->pool()->FindMessageTypeByName(
989 full_type_name);
990 if (desc == NULL) {
991 GOOGLE_LOG(ERROR) << "Proto type '" << full_type_name << "' not found";
992 return false;
993 }
994
995 if (dynamic_message_factory_ == NULL) {
996 dynamic_message_factory_.reset(new DynamicMessageFactory());
997 }
998 data->reset(dynamic_message_factory_->GetPrototype(desc)->New());
999 string serialized_value = reflection->GetString(any, value_field);
1000 if (!(*data)->ParseFromString(serialized_value)) {
1001 GOOGLE_LOG(ERROR) << "Failed to parse value for " << full_type_name;
1002 return false;
1003 }
1004 return true;
1005 }
1006
1007 bool MessageDifferencer::CompareUnknownFields(
1008 const Message& message1, const Message& message2,
1009 const google::protobuf::UnknownFieldSet& unknown_field_set1,
1010 const google::protobuf::UnknownFieldSet& unknown_field_set2,
1011 vector<SpecificField>* parent_field) {
1012 // Ignore unknown fields in EQUIVALENT mode.
1013 if (message_field_comparison_ == EQUIVALENT) return true;
1014
1015 if (unknown_field_set1.empty() && unknown_field_set2.empty()) {
1016 return true;
1017 }
1018
1019 bool is_different = false;
1020
1021 // We first sort the unknown fields by field number and type (in other words,
1022 // in tag order), making sure to preserve ordering of values with the same
1023 // tag. This allows us to report only meaningful differences between the
1024 // two sets -- that is, differing values for the same tag. We use
1025 // IndexUnknownFieldPairs to keep track of the field's original index for
1026 // reporting purposes.
1027 vector<IndexUnknownFieldPair> fields1; // unknown_field_set1, sorted
1028 vector<IndexUnknownFieldPair> fields2; // unknown_field_set2, sorted
1029 fields1.reserve(unknown_field_set1.field_count());
1030 fields2.reserve(unknown_field_set2.field_count());
1031
1032 for (int i = 0; i < unknown_field_set1.field_count(); i++) {
1033 fields1.push_back(std::make_pair(i, &unknown_field_set1.field(i)));
1034 }
1035 for (int i = 0; i < unknown_field_set2.field_count(); i++) {
1036 fields2.push_back(std::make_pair(i, &unknown_field_set2.field(i)));
1037 }
1038
1039 UnknownFieldOrdering is_before;
1040 std::stable_sort(fields1.begin(), fields1.end(), is_before);
1041 std::stable_sort(fields2.begin(), fields2.end(), is_before);
1042
1043 // In order to fill in SpecificField::index, we have to keep track of how
1044 // many values we've seen with the same field number and type.
1045 // current_repeated points at the first field in this range, and
1046 // current_repeated_start{1,2} are the indexes of the first field in the
1047 // range within fields1 and fields2.
1048 const UnknownField* current_repeated = NULL;
1049 int current_repeated_start1 = 0;
1050 int current_repeated_start2 = 0;
1051
1052 // Now that we have two sorted lists, we can detect fields which appear only
1053 // in one list or the other by traversing them simultaneously.
1054 int index1 = 0;
1055 int index2 = 0;
1056 while (index1 < fields1.size() || index2 < fields2.size()) {
1057 enum { ADDITION, DELETION, MODIFICATION, COMPARE_GROUPS,
1058 NO_CHANGE } change_type;
1059
1060 // focus_field is the field we're currently reporting on. (In the case
1061 // of a modification, it's the field on the left side.)
1062 const UnknownField* focus_field;
1063 bool match = false;
1064
1065 if (index2 == fields2.size() ||
1066 (index1 < fields1.size() &&
1067 is_before(fields1[index1], fields2[index2]))) {
1068 // fields1[index1] is not present in fields2.
1069 change_type = DELETION;
1070 focus_field = fields1[index1].second;
1071 } else if (index1 == fields1.size() ||
1072 is_before(fields2[index2], fields1[index1])) {
1073 // fields2[index2] is not present in fields1.
1074 if (scope_ == PARTIAL) {
1075 // Ignore.
1076 ++index2;
1077 continue;
1078 }
1079 change_type = ADDITION;
1080 focus_field = fields2[index2].second;
1081 } else {
1082 // Field type and number are the same. See if the values differ.
1083 change_type = MODIFICATION;
1084 focus_field = fields1[index1].second;
1085
1086 switch (focus_field->type()) {
1087 case UnknownField::TYPE_VARINT:
1088 match = fields1[index1].second->varint() ==
1089 fields2[index2].second->varint();
1090 break;
1091 case UnknownField::TYPE_FIXED32:
1092 match = fields1[index1].second->fixed32() ==
1093 fields2[index2].second->fixed32();
1094 break;
1095 case UnknownField::TYPE_FIXED64:
1096 match = fields1[index1].second->fixed64() ==
1097 fields2[index2].second->fixed64();
1098 break;
1099 case UnknownField::TYPE_LENGTH_DELIMITED:
1100 match = fields1[index1].second->length_delimited() ==
1101 fields2[index2].second->length_delimited();
1102 break;
1103 case UnknownField::TYPE_GROUP:
1104 // We must deal with this later, after building the SpecificField.
1105 change_type = COMPARE_GROUPS;
1106 break;
1107 }
1108 if (match && change_type != COMPARE_GROUPS) {
1109 change_type = NO_CHANGE;
1110 }
1111 }
1112
1113 if (current_repeated == NULL ||
1114 focus_field->number() != current_repeated->number() ||
1115 focus_field->type() != current_repeated->type()) {
1116 // We've started a new repeated field.
1117 current_repeated = focus_field;
1118 current_repeated_start1 = index1;
1119 current_repeated_start2 = index2;
1120 }
1121
1122 if (change_type == NO_CHANGE && reporter_ == NULL) {
1123 // Fields were already compared and matched and we have no reporter.
1124 ++index1;
1125 ++index2;
1126 continue;
1127 }
1128
1129 if (change_type == ADDITION || change_type == DELETION ||
1130 change_type == MODIFICATION) {
1131 if (reporter_ == NULL) {
1132 // We found a difference and we have no reproter.
1133 return false;
1134 }
1135 is_different = true;
1136 }
1137
1138 // Build the SpecificField. This is slightly complicated.
1139 SpecificField specific_field;
1140 specific_field.unknown_field_number = focus_field->number();
1141 specific_field.unknown_field_type = focus_field->type();
1142
1143 specific_field.unknown_field_set1 = &unknown_field_set1;
1144 specific_field.unknown_field_set2 = &unknown_field_set2;
1145
1146 if (change_type != ADDITION) {
1147 specific_field.unknown_field_index1 = fields1[index1].first;
1148 }
1149 if (change_type != DELETION) {
1150 specific_field.unknown_field_index2 = fields2[index2].first;
1151 }
1152
1153 // Calculate the field index.
1154 if (change_type == ADDITION) {
1155 specific_field.index = index2 - current_repeated_start2;
1156 specific_field.new_index = index2 - current_repeated_start2;
1157 } else {
1158 specific_field.index = index1 - current_repeated_start1;
1159 specific_field.new_index = index2 - current_repeated_start2;
1160 }
1161
1162 parent_field->push_back(specific_field);
1163
1164 switch (change_type) {
1165 case ADDITION:
1166 reporter_->ReportAdded(message1, message2, *parent_field);
1167 ++index2;
1168 break;
1169 case DELETION:
1170 reporter_->ReportDeleted(message1, message2, *parent_field);
1171 ++index1;
1172 break;
1173 case MODIFICATION:
1174 reporter_->ReportModified(message1, message2, *parent_field);
1175 ++index1;
1176 ++index2;
1177 break;
1178 case COMPARE_GROUPS:
1179 if (!CompareUnknownFields(message1, message2,
1180 fields1[index1].second->group(),
1181 fields2[index2].second->group(),
1182 parent_field)) {
1183 if (reporter_ == NULL) return false;
1184 is_different = true;
1185 reporter_->ReportModified(message1, message2, *parent_field);
1186 }
1187 ++index1;
1188 ++index2;
1189 break;
1190 case NO_CHANGE:
1191 ++index1;
1192 ++index2;
1193 if (report_matches_) {
1194 reporter_->ReportMatched(message1, message2, *parent_field);
1195 }
1196 }
1197
1198 parent_field->pop_back();
1199 }
1200
1201 return !is_different;
1202 }
1203
1204 namespace {
1205
1206 // Find maximum bipartite matching using the argumenting path algorithm.
1207 class MaximumMatcher {
1208 public:
1209 typedef ResultCallback2<bool, int, int> NodeMatchCallback;
1210 // MaximumMatcher takes ownership of the passed in callback and uses it to
1211 // determine whether a node on the left side of the bipartial graph matches
1212 // a node on the right side. count1 is the number of nodes on the left side
1213 // of the graph and count2 to is the number of nodes on the right side.
1214 // Every node is referred to using 0-based indices.
1215 // If a maximum match is found, the result will be stored in match_list1 and
1216 // match_list2. match_list1[i] == j means the i-th node on the left side is
1217 // matched to the j-th node on the right side and match_list2[x] == y means
1218 // the x-th node on the right side is matched to y-th node on the left side.
1219 // match_list1[i] == -1 means the node is not matched. Same with match_list2.
1220 MaximumMatcher(int count1, int count2, NodeMatchCallback* callback,
1221 vector<int>* match_list1, vector<int>* match_list2);
1222 // Find a maximum match and return the number of matched node pairs.
1223 // If early_return is true, this method will return 0 immediately when it
1224 // finds that not all nodes on the left side can be matched.
1225 int FindMaximumMatch(bool early_return);
1226 private:
1227 // Determines whether the node on the left side of the bipartial graph
1228 // matches the one on the right side.
1229 bool Match(int left, int right);
1230 // Find an argumenting path starting from the node v on the left side. If a
1231 // path can be found, update match_list2_ to reflect the path and return
1232 // true.
1233 bool FindArgumentPathDFS(int v, vector<bool>* visited);
1234
1235 int count1_;
1236 int count2_;
1237 google::protobuf::scoped_ptr<NodeMatchCallback> match_callback_;
1238 map<pair<int, int>, bool> cached_match_results_;
1239 vector<int>* match_list1_;
1240 vector<int>* match_list2_;
1241 GOOGLE_DISALLOW_EVIL_CONSTRUCTORS(MaximumMatcher);
1242 };
1243
1244 MaximumMatcher::MaximumMatcher(int count1, int count2,
1245 NodeMatchCallback* callback,
1246 vector<int>* match_list1,
1247 vector<int>* match_list2)
1248 : count1_(count1), count2_(count2), match_callback_(callback),
1249 match_list1_(match_list1), match_list2_(match_list2) {
1250 match_list1_->assign(count1, -1);
1251 match_list2_->assign(count2, -1);
1252 }
1253
1254 int MaximumMatcher::FindMaximumMatch(bool early_return) {
1255 int result = 0;
1256 for (int i = 0; i < count1_; ++i) {
1257 vector<bool> visited(count1_);
1258 if (FindArgumentPathDFS(i, &visited)) {
1259 ++result;
1260 } else if (early_return) {
1261 return 0;
1262 }
1263 }
1264 // Backfill match_list1_ as we only filled match_list2_ when finding
1265 // argumenting pathes.
1266 for (int i = 0; i < count2_; ++i) {
1267 if ((*match_list2_)[i] != -1) {
1268 (*match_list1_)[(*match_list2_)[i]] = i;
1269 }
1270 }
1271 return result;
1272 }
1273
1274 bool MaximumMatcher::Match(int left, int right) {
1275 pair<int, int> p(left, right);
1276 map<pair<int, int>, bool>::iterator it = cached_match_results_.find(p);
1277 if (it != cached_match_results_.end()) {
1278 return it->second;
1279 }
1280 cached_match_results_[p] = match_callback_->Run(left, right);
1281 return cached_match_results_[p];
1282 }
1283
1284 bool MaximumMatcher::FindArgumentPathDFS(int v, vector<bool>* visited) {
1285 (*visited)[v] = true;
1286 // We try to match those un-matched nodes on the right side first. This is
1287 // the step that the navie greedy matching algorithm uses. In the best cases
1288 // where the greedy algorithm can find a maximum matching, we will always
1289 // find a match in this step and the performance will be identical to the
1290 // greedy algorithm.
1291 for (int i = 0; i < count2_; ++i) {
1292 int matched = (*match_list2_)[i];
1293 if (matched == -1 && Match(v, i)) {
1294 (*match_list2_)[i] = v;
1295 return true;
1296 }
1297 }
1298 // Then we try those already matched nodes and see if we can find an
1299 // alternaive match for the node matched to them.
1300 // The greedy algorithm will stop before this and fail to produce the
1301 // correct result.
1302 for (int i = 0; i < count2_; ++i) {
1303 int matched = (*match_list2_)[i];
1304 if (matched != -1 && Match(v, i)) {
1305 if (!(*visited)[matched] && FindArgumentPathDFS(matched, visited)) {
1306 (*match_list2_)[i] = v;
1307 return true;
1308 }
1309 }
1310 }
1311 return false;
1312 }
1313
1314 } // namespace
1315
1316 bool MessageDifferencer::MatchRepeatedFieldIndices(
1317 const Message& message1,
1318 const Message& message2,
1319 const FieldDescriptor* repeated_field,
1320 const vector<SpecificField>& parent_fields,
1321 vector<int>* match_list1,
1322 vector<int>* match_list2) {
1323 const int count1 =
1324 message1.GetReflection()->FieldSize(message1, repeated_field);
1325 const int count2 =
1326 message2.GetReflection()->FieldSize(message2, repeated_field);
1327 const MapKeyComparator* key_comparator = GetMapKeyComparator(repeated_field);
1328
1329 match_list1->assign(count1, -1);
1330 match_list2->assign(count2, -1);
1331
1332 SpecificField specific_field;
1333 specific_field.field = repeated_field;
1334
1335 bool success = true;
1336 // Find potential match if this is a special repeated field.
1337 if (key_comparator != NULL || IsTreatedAsSet(repeated_field)) {
1338 if (scope_ == PARTIAL) {
1339 // When partial matching is enabled, Compare(a, b) && Compare(a, c)
1340 // doesn't neccessarily imply Compare(b, c). Therefore a naive greedy
1341 // algorithm will fail to find a maximum matching.
1342 // Here we use the argumenting path algorithm.
1343 MaximumMatcher::NodeMatchCallback* callback = NewPermanentCallback(
1344 this, &MessageDifferencer::IsMatch, repeated_field, key_comparator,
1345 &message1, &message2, parent_fields);
1346 MaximumMatcher matcher(count1, count2, callback, match_list1,
1347 match_list2);
1348 // If diff info is not needed, we should end the matching process as
1349 // soon as possible if not all items can be matched.
1350 bool early_return = (reporter_ == NULL);
1351 int match_count = matcher.FindMaximumMatch(early_return);
1352 if (match_count != count1 && reporter_ == NULL) return false;
1353 success = success && (match_count == count1);
1354 } else {
1355 for (int i = 0; i < count1; ++i) {
1356 // Indicates any matched elements for this repeated field.
1357 bool match = false;
1358
1359 specific_field.index = i;
1360 specific_field.new_index = i;
1361
1362 for (int j = 0; j < count2; j++) {
1363 if (match_list2->at(j) != -1) continue;
1364 specific_field.index = i;
1365 specific_field.new_index = j;
1366
1367 match = IsMatch(repeated_field, key_comparator,
1368 &message1, &message2, parent_fields, i, j);
1369
1370 if (match) {
1371 match_list1->at(specific_field.index) = specific_field.new_index;
1372 match_list2->at(specific_field.new_index) = specific_field.index;
1373 break;
1374 }
1375 }
1376 if (!match && reporter_ == NULL) return false;
1377 success = success && match;
1378 }
1379 }
1380 } else {
1381 // If this field should be treated as list, just label the match_list.
1382 for (int i = 0; i < count1 && i < count2; i++) {
1383 match_list1->at(i) = i;
1384 match_list2->at(i) = i;
1385 }
1386 }
1387
1388 return success;
1389 }
1390
1391 FieldComparator::ComparisonResult MessageDifferencer::GetFieldComparisonResult(
1392 const Message& message1, const Message& message2,
1393 const FieldDescriptor* field, int index1, int index2,
1394 const FieldContext* field_context) {
1395 FieldComparator* comparator = field_comparator_ != NULL ?
1396 field_comparator_ : &default_field_comparator_;
1397 return comparator->Compare(message1, message2, field,
1398 index1, index2, field_context);
1399 }
1400
1401 // ===========================================================================
1402
1403 MessageDifferencer::Reporter::Reporter() { }
1404 MessageDifferencer::Reporter::~Reporter() {}
1405
1406 // ===========================================================================
1407
1408 MessageDifferencer::MapKeyComparator::MapKeyComparator() {}
1409 MessageDifferencer::MapKeyComparator::~MapKeyComparator() {}
1410
1411 // ===========================================================================
1412
1413 MessageDifferencer::IgnoreCriteria::IgnoreCriteria() {}
1414 MessageDifferencer::IgnoreCriteria::~IgnoreCriteria() {}
1415
1416 // ===========================================================================
1417
1418 // Note that the printer's delimiter is not used, because if we are given a
1419 // printer, we don't know its delimiter.
1420 MessageDifferencer::StreamReporter::StreamReporter(
1421 io::ZeroCopyOutputStream* output) : printer_(new io::Printer(output, '$')),
1422 delete_printer_(true),
1423 report_modified_aggregates_(false) { }
1424
1425 MessageDifferencer::StreamReporter::StreamReporter(
1426 io::Printer* printer) : printer_(printer),
1427 delete_printer_(false),
1428 report_modified_aggregates_(false) { }
1429
1430 MessageDifferencer::StreamReporter::~StreamReporter() {
1431 if (delete_printer_) delete printer_;
1432 }
1433
1434 void MessageDifferencer::StreamReporter::PrintPath(
1435 const vector<SpecificField>& field_path, bool left_side) {
1436 for (int i = 0; i < field_path.size(); ++i) {
1437 if (i > 0) {
1438 printer_->Print(".");
1439 }
1440
1441 SpecificField specific_field = field_path[i];
1442
1443 if (specific_field.field != NULL) {
1444 if (specific_field.field->is_extension()) {
1445 printer_->Print("($name$)", "name",
1446 specific_field.field->full_name());
1447 } else {
1448 printer_->PrintRaw(specific_field.field->name());
1449 }
1450 } else {
1451 printer_->PrintRaw(SimpleItoa(specific_field.unknown_field_number));
1452 }
1453 if (left_side && specific_field.index >= 0) {
1454 printer_->Print("[$name$]", "name", SimpleItoa(specific_field.index));
1455 }
1456 if (!left_side && specific_field.new_index >= 0) {
1457 printer_->Print("[$name$]", "name", SimpleItoa(specific_field.new_index));
1458 }
1459 }
1460 }
1461
1462 void MessageDifferencer::
1463 StreamReporter::PrintValue(const Message& message,
1464 const vector<SpecificField>& field_path,
1465 bool left_side) {
1466 const SpecificField& specific_field = field_path.back();
1467 const FieldDescriptor* field = specific_field.field;
1468 if (field != NULL) {
1469 string output;
1470 int index = left_side ? specific_field.index : specific_field.new_index;
1471 if (field->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
1472 const Reflection* reflection = message.GetReflection();
1473 const Message& field_message = field->is_repeated() ?
1474 reflection->GetRepeatedMessage(message, field, index) :
1475 reflection->GetMessage(message, field);
1476 output = field_message.ShortDebugString();
1477 if (output.empty()) {
1478 printer_->Print("{ }");
1479 } else {
1480 printer_->Print("{ $name$ }", "name", output);
1481 }
1482 } else {
1483 TextFormat::PrintFieldValueToString(message, field, index, &output);
1484 printer_->PrintRaw(output);
1485 }
1486 } else {
1487 const UnknownFieldSet* unknown_fields =
1488 (left_side ?
1489 specific_field.unknown_field_set1 :
1490 specific_field.unknown_field_set2);
1491 const UnknownField* unknown_field = &unknown_fields->field(
1492 left_side ?
1493 specific_field.unknown_field_index1 :
1494 specific_field.unknown_field_index2);
1495 PrintUnknownFieldValue(unknown_field);
1496 }
1497 }
1498
1499 void MessageDifferencer::
1500 StreamReporter::PrintUnknownFieldValue(const UnknownField* unknown_field) {
1501 GOOGLE_CHECK(unknown_field != NULL) << " Cannot print NULL unknown_field.";
1502
1503 string output;
1504 switch (unknown_field->type()) {
1505 case UnknownField::TYPE_VARINT:
1506 output = SimpleItoa(unknown_field->varint());
1507 break;
1508 case UnknownField::TYPE_FIXED32:
1509 output = StrCat("0x", strings::Hex(unknown_field->fixed32(),
1510 strings::ZERO_PAD_8));
1511 break;
1512 case UnknownField::TYPE_FIXED64:
1513 output = StrCat("0x", strings::Hex(unknown_field->fixed64(),
1514 strings::ZERO_PAD_16));
1515 break;
1516 case UnknownField::TYPE_LENGTH_DELIMITED:
1517 output = StringPrintf("\"%s\"",
1518 CEscape(unknown_field->length_delimited()).c_str());
1519 break;
1520 case UnknownField::TYPE_GROUP:
1521 // TODO(kenton): Print the contents of the group like we do for
1522 // messages. Requires an equivalent of ShortDebugString() for
1523 // UnknownFieldSet.
1524 output = "{ ... }";
1525 break;
1526 }
1527 printer_->PrintRaw(output);
1528 }
1529
1530 void MessageDifferencer::StreamReporter::Print(const string& str) {
1531 printer_->Print(str.c_str());
1532 }
1533
1534 void MessageDifferencer::StreamReporter::ReportAdded(
1535 const Message& message1,
1536 const Message& message2,
1537 const vector<SpecificField>& field_path) {
1538 printer_->Print("added: ");
1539 PrintPath(field_path, false);
1540 printer_->Print(": ");
1541 PrintValue(message2, field_path, false);
1542 printer_->Print("\n"); // Print for newlines.
1543 }
1544
1545 void MessageDifferencer::StreamReporter::ReportDeleted(
1546 const Message& message1,
1547 const Message& message2,
1548 const vector<SpecificField>& field_path) {
1549 printer_->Print("deleted: ");
1550 PrintPath(field_path, true);
1551 printer_->Print(": ");
1552 PrintValue(message1, field_path, true);
1553 printer_->Print("\n"); // Print for newlines
1554 }
1555
1556 void MessageDifferencer::StreamReporter::ReportModified(
1557 const Message& message1,
1558 const Message& message2,
1559 const vector<SpecificField>& field_path) {
1560 if (!report_modified_aggregates_ && field_path.back().field == NULL) {
1561 if (field_path.back().unknown_field_type == UnknownField::TYPE_GROUP) {
1562 // Any changes to the subfields have already been printed.
1563 return;
1564 }
1565 } else if (!report_modified_aggregates_) {
1566 if (field_path.back().field->cpp_type() ==
1567 FieldDescriptor::CPPTYPE_MESSAGE) {
1568 // Any changes to the subfields have already been printed.
1569 return;
1570 }
1571 }
1572
1573 printer_->Print("modified: ");
1574 PrintPath(field_path, true);
1575 if (CheckPathChanged(field_path)) {
1576 printer_->Print(" -> ");
1577 PrintPath(field_path, false);
1578 }
1579 printer_->Print(": ");
1580 PrintValue(message1, field_path, true);
1581 printer_->Print(" -> ");
1582 PrintValue(message2, field_path, false);
1583 printer_->Print("\n"); // Print for newlines.
1584 }
1585
1586 void MessageDifferencer::StreamReporter::ReportMoved(
1587 const Message& message1,
1588 const Message& message2,
1589 const vector<SpecificField>& field_path) {
1590 printer_->Print("moved: ");
1591 PrintPath(field_path, true);
1592 printer_->Print(" -> ");
1593 PrintPath(field_path, false);
1594 printer_->Print(" : ");
1595 PrintValue(message1, field_path, true);
1596 printer_->Print("\n"); // Print for newlines.
1597 }
1598
1599 void MessageDifferencer::StreamReporter::ReportMatched(
1600 const Message& message1,
1601 const Message& message2,
1602 const vector<SpecificField>& field_path) {
1603 printer_->Print("matched: ");
1604 PrintPath(field_path, true);
1605 if (CheckPathChanged(field_path)) {
1606 printer_->Print(" -> ");
1607 PrintPath(field_path, false);
1608 }
1609 printer_->Print(" : ");
1610 PrintValue(message1, field_path, true);
1611 printer_->Print("\n"); // Print for newlines.
1612 }
1613
1614 void MessageDifferencer::StreamReporter::ReportIgnored(
1615 const Message& message1,
1616 const Message& message2,
1617 const vector<SpecificField>& field_path) {
1618 printer_->Print("ignored: ");
1619 PrintPath(field_path, true);
1620 if (CheckPathChanged(field_path)) {
1621 printer_->Print(" -> ");
1622 PrintPath(field_path, false);
1623 }
1624 printer_->Print("\n"); // Print for newlines.
1625 }
1626
1627 } // namespace util
1628 } // namespace protobuf
1629 } // namespace google
OLDNEW

Powered by Google App Engine
This is Rietveld 408576698