OLD | NEW |
1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "chrome/renderer/safe_browsing/phishing_term_feature_extractor.h" | 5 #include "chrome/renderer/safe_browsing/phishing_term_feature_extractor.h" |
6 | 6 |
7 #include <string> | 7 #include <string> |
8 | 8 |
9 #include "base/callback.h" | 9 #include "base/callback.h" |
10 #include "base/hash_tables.h" | 10 #include "base/hash_tables.h" |
11 #include "base/memory/scoped_ptr.h" | 11 #include "base/memory/scoped_ptr.h" |
12 #include "base/message_loop.h" | 12 #include "base/message_loop.h" |
13 #include "base/sha2.h" | |
14 #include "base/string16.h" | 13 #include "base/string16.h" |
15 #include "base/stringprintf.h" | 14 #include "base/stringprintf.h" |
16 #include "base/time.h" | 15 #include "base/time.h" |
17 #include "base/utf_string_conversions.h" | 16 #include "base/utf_string_conversions.h" |
| 17 #include "crypto/sha2.h" |
18 #include "chrome/renderer/safe_browsing/features.h" | 18 #include "chrome/renderer/safe_browsing/features.h" |
19 #include "chrome/renderer/safe_browsing/mock_feature_extractor_clock.h" | 19 #include "chrome/renderer/safe_browsing/mock_feature_extractor_clock.h" |
20 #include "testing/gmock/include/gmock/gmock.h" | 20 #include "testing/gmock/include/gmock/gmock.h" |
21 #include "testing/gtest/include/gtest/gtest.h" | 21 #include "testing/gtest/include/gtest/gtest.h" |
22 | 22 |
23 using ::testing::ContainerEq; | 23 using ::testing::ContainerEq; |
24 using ::testing::Return; | 24 using ::testing::Return; |
25 | 25 |
26 namespace safe_browsing { | 26 namespace safe_browsing { |
27 | 27 |
28 class PhishingTermFeatureExtractorTest : public ::testing::Test { | 28 class PhishingTermFeatureExtractorTest : public ::testing::Test { |
29 protected: | 29 protected: |
30 virtual void SetUp() { | 30 virtual void SetUp() { |
31 base::hash_set<std::string> terms; | 31 base::hash_set<std::string> terms; |
32 terms.insert("one"); | 32 terms.insert("one"); |
33 terms.insert("one one"); | 33 terms.insert("one one"); |
34 terms.insert("two"); | 34 terms.insert("two"); |
35 terms.insert("multi word test"); | 35 terms.insert("multi word test"); |
36 terms.insert("capitalization"); | 36 terms.insert("capitalization"); |
37 terms.insert("space"); | 37 terms.insert("space"); |
38 terms.insert("separator"); | 38 terms.insert("separator"); |
39 terms.insert("punctuation"); | 39 terms.insert("punctuation"); |
40 // Chinese (translation of "hello") | 40 // Chinese (translation of "hello") |
41 terms.insert("\xe4\xbd\xa0\xe5\xa5\xbd"); | 41 terms.insert("\xe4\xbd\xa0\xe5\xa5\xbd"); |
42 // Chinese (translation of "goodbye") | 42 // Chinese (translation of "goodbye") |
43 terms.insert("\xe5\x86\x8d\xe8\xa7\x81"); | 43 terms.insert("\xe5\x86\x8d\xe8\xa7\x81"); |
44 | 44 |
45 for (base::hash_set<std::string>::iterator it = terms.begin(); | 45 for (base::hash_set<std::string>::iterator it = terms.begin(); |
46 it != terms.end(); ++it) { | 46 it != terms.end(); ++it) { |
47 term_hashes_.insert(base::SHA256HashString(*it)); | 47 term_hashes_.insert(crypto::SHA256HashString(*it)); |
48 } | 48 } |
49 | 49 |
50 base::hash_set<std::string> words; | 50 base::hash_set<std::string> words; |
51 words.insert("one"); | 51 words.insert("one"); |
52 words.insert("two"); | 52 words.insert("two"); |
53 words.insert("multi"); | 53 words.insert("multi"); |
54 words.insert("word"); | 54 words.insert("word"); |
55 words.insert("test"); | 55 words.insert("test"); |
56 words.insert("capitalization"); | 56 words.insert("capitalization"); |
57 words.insert("space"); | 57 words.insert("space"); |
58 words.insert("separator"); | 58 words.insert("separator"); |
59 words.insert("punctuation"); | 59 words.insert("punctuation"); |
60 words.insert("\xe4\xbd\xa0\xe5\xa5\xbd"); | 60 words.insert("\xe4\xbd\xa0\xe5\xa5\xbd"); |
61 words.insert("\xe5\x86\x8d\xe8\xa7\x81"); | 61 words.insert("\xe5\x86\x8d\xe8\xa7\x81"); |
62 | 62 |
63 for (base::hash_set<std::string>::iterator it = words.begin(); | 63 for (base::hash_set<std::string>::iterator it = words.begin(); |
64 it != words.end(); ++it) { | 64 it != words.end(); ++it) { |
65 word_hashes_.insert(base::SHA256HashString(*it)); | 65 word_hashes_.insert(crypto::SHA256HashString(*it)); |
66 } | 66 } |
67 | 67 |
68 extractor_.reset(new PhishingTermFeatureExtractor( | 68 extractor_.reset(new PhishingTermFeatureExtractor( |
69 &term_hashes_, | 69 &term_hashes_, |
70 &word_hashes_, | 70 &word_hashes_, |
71 3 /* max_words_per_term */, | 71 3 /* max_words_per_term */, |
72 &clock_)); | 72 &clock_)); |
73 } | 73 } |
74 | 74 |
75 // Runs the TermFeatureExtractor on |page_text|, waiting for the | 75 // Runs the TermFeatureExtractor on |page_text|, waiting for the |
(...skipping 160 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
236 // Time check after the next 10 words. This is over the limit. | 236 // Time check after the next 10 words. This is over the limit. |
237 .WillOnce(Return(now + base::TimeDelta::FromMilliseconds(600))) | 237 .WillOnce(Return(now + base::TimeDelta::FromMilliseconds(600))) |
238 // A final time check for the histograms. | 238 // A final time check for the histograms. |
239 .WillOnce(Return(now + base::TimeDelta::FromMilliseconds(620))); | 239 .WillOnce(Return(now + base::TimeDelta::FromMilliseconds(620))); |
240 | 240 |
241 features.Clear(); | 241 features.Clear(); |
242 EXPECT_FALSE(ExtractFeatures(&page_text, &features)); | 242 EXPECT_FALSE(ExtractFeatures(&page_text, &features)); |
243 } | 243 } |
244 | 244 |
245 } // namespace safe_browsing | 245 } // namespace safe_browsing |
OLD | NEW |