| OLD | NEW |
| 1 // Copyright 2014 The Chromium Authors. All rights reserved. | 1 // Copyright 2014 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/browser/extensions/activity_log/ad_network_database.h" | 5 #include "chrome/browser/extensions/activity_log/ad_network_database.h" |
| 6 | 6 |
| 7 #include "base/basictypes.h" | |
| 8 #include "base/lazy_instance.h" | 7 #include "base/lazy_instance.h" |
| 9 #include "base/memory/ref_counted_memory.h" | 8 #include "chrome/browser/extensions/activity_log/hashed_ad_network_database.h" |
| 10 #include "crypto/secure_hash.h" | |
| 11 #include "crypto/sha2.h" | |
| 12 #include "grit/browser_resources.h" | 9 #include "grit/browser_resources.h" |
| 13 #include "ui/base/resource/resource_bundle.h" | 10 #include "ui/base/resource/resource_bundle.h" |
| 14 #include "url/gurl.h" | |
| 15 | 11 |
| 16 namespace extensions { | 12 namespace extensions { |
| 17 | 13 |
| 18 namespace { | 14 namespace { |
| 19 | 15 |
| 20 // We use a hash size of 8 for these for three reasons. | |
| 21 // 1. It saves us a bit on space, and, since we have to store these in memory | |
| 22 // (reading from disk would be far too slow because these checks are | |
| 23 // performed synchronously), that space is important. | |
| 24 // 2. Since we don't store full hashes, reconstructing the list is more | |
| 25 // difficult. This may mean we get a few incorrect hits, but the security is | |
| 26 // worth the (very small) amount of noise. | |
| 27 // 3. It fits nicely into a int64. | |
| 28 const size_t kUrlHashSize = 8u; | |
| 29 COMPILE_ASSERT(kUrlHashSize <= sizeof(int64), url_hashes_must_fit_into_a_int64); | |
| 30 | |
| 31 const size_t kChecksumHashSize = 32u; | |
| 32 | |
| 33 class AdNetworkDatabaseImpl : public AdNetworkDatabase { | |
| 34 public: | |
| 35 AdNetworkDatabaseImpl(); | |
| 36 virtual ~AdNetworkDatabaseImpl(); | |
| 37 | |
| 38 private: | |
| 39 virtual bool IsAdNetwork(const GURL& url) const OVERRIDE; | |
| 40 | |
| 41 // Initialize the AdNetworkDatabase. This means initializing the set of | |
| 42 // hashes from the shared memory. | |
| 43 void Init(); | |
| 44 | |
| 45 // The set of partial hashes for known ad networks. | |
| 46 base::hash_set<int64> entries_; | |
| 47 }; | |
| 48 | |
| 49 AdNetworkDatabaseImpl::AdNetworkDatabaseImpl() { | |
| 50 Init(); | |
| 51 } | |
| 52 | |
| 53 AdNetworkDatabaseImpl::~AdNetworkDatabaseImpl() {} | |
| 54 | |
| 55 void AdNetworkDatabaseImpl::Init() { | |
| 56 base::RefCountedStaticMemory* entries_memory = | |
| 57 ResourceBundle::GetSharedInstance().LoadDataResourceBytes( | |
| 58 IDR_AD_NETWORK_HASHES); | |
| 59 | |
| 60 // This can legitimately happen in unit tests. | |
| 61 if (!entries_memory) | |
| 62 return; | |
| 63 | |
| 64 const size_t size = entries_memory->size(); | |
| 65 const unsigned char* const front = entries_memory->front(); | |
| 66 if (size < kChecksumHashSize || | |
| 67 (size - kChecksumHashSize) % kUrlHashSize != 0) { | |
| 68 NOTREACHED(); | |
| 69 return; | |
| 70 } | |
| 71 | |
| 72 // The format of the data resource is fairly straight-forward: | |
| 73 // <32-bit checksum><list of 64-bit hashes of hosts>, with no linebreaks or | |
| 74 // other separations. | |
| 75 scoped_ptr<crypto::SecureHash> hash( | |
| 76 crypto::SecureHash::Create(crypto::SecureHash::SHA256)); | |
| 77 | |
| 78 hash->Update(front + kChecksumHashSize, size - kChecksumHashSize); | |
| 79 char hash_value[kChecksumHashSize]; | |
| 80 hash->Finish(hash_value, kChecksumHashSize); | |
| 81 // If the checksum doesn't match, abort. | |
| 82 if (memcmp(hash_value, front, kChecksumHashSize) != 0) { | |
| 83 NOTREACHED(); | |
| 84 return; | |
| 85 } | |
| 86 | |
| 87 // Construct and insert all hashes. | |
| 88 for (const unsigned char* index = front + kChecksumHashSize; | |
| 89 index < front + size; | |
| 90 index += kUrlHashSize) { | |
| 91 int64 value = 0; | |
| 92 memcpy(&value, index, kUrlHashSize); | |
| 93 entries_.insert(value); | |
| 94 } | |
| 95 } | |
| 96 | |
| 97 bool AdNetworkDatabaseImpl::IsAdNetwork(const GURL& url) const { | |
| 98 int64 hash = 0; | |
| 99 crypto::SHA256HashString(url.host(), &hash, sizeof(hash)); | |
| 100 // If initialization failed (most likely because this is a unittest), then | |
| 101 // |entries_| is never populated and we are guaranteed to return false - which | |
| 102 // is desired default behavior. | |
| 103 return entries_.count(hash) != 0; | |
| 104 } | |
| 105 | |
| 106 class AdNetworkDatabaseFactory { | 16 class AdNetworkDatabaseFactory { |
| 107 public: | 17 public: |
| 108 AdNetworkDatabaseFactory(); | 18 AdNetworkDatabaseFactory(); |
| 109 ~AdNetworkDatabaseFactory(); | 19 ~AdNetworkDatabaseFactory(); |
| 110 | 20 |
| 111 const AdNetworkDatabase* GetDatabase(); | 21 const AdNetworkDatabase* GetDatabase(); |
| 112 void SetDatabase(scoped_ptr<AdNetworkDatabase> database); | 22 void SetDatabase(scoped_ptr<AdNetworkDatabase> database); |
| 113 | 23 |
| 114 private: | 24 private: |
| 115 scoped_ptr<AdNetworkDatabase> database_; | 25 scoped_ptr<AdNetworkDatabase> database_; |
| 116 }; | 26 }; |
| 117 | 27 |
| 118 AdNetworkDatabaseFactory::AdNetworkDatabaseFactory() {} | 28 AdNetworkDatabaseFactory::AdNetworkDatabaseFactory() {} |
| 119 AdNetworkDatabaseFactory::~AdNetworkDatabaseFactory() {} | 29 AdNetworkDatabaseFactory::~AdNetworkDatabaseFactory() {} |
| 120 | 30 |
| 121 const AdNetworkDatabase* AdNetworkDatabaseFactory::GetDatabase() { | 31 const AdNetworkDatabase* AdNetworkDatabaseFactory::GetDatabase() { |
| 122 // Construct a new database, if we don't have one. | 32 // Construct a new database, if we don't have one. |
| 123 if (!database_.get()) | 33 if (!database_.get()) { |
| 124 database_.reset(new AdNetworkDatabaseImpl()); | 34 database_.reset(new HashedAdNetworkDatabase( |
| 35 ResourceBundle::GetSharedInstance().LoadDataResourceBytes( |
| 36 IDR_AD_NETWORK_HASHES))); |
| 37 } |
| 125 | 38 |
| 126 return database_.get(); | 39 return database_.get(); |
| 127 } | 40 } |
| 128 | 41 |
| 129 void AdNetworkDatabaseFactory::SetDatabase( | 42 void AdNetworkDatabaseFactory::SetDatabase( |
| 130 scoped_ptr<AdNetworkDatabase> database) { | 43 scoped_ptr<AdNetworkDatabase> database) { |
| 131 database_.reset(database.release()); | 44 database_.reset(database.release()); |
| 132 } | 45 } |
| 133 | 46 |
| 134 base::LazyInstance<AdNetworkDatabaseFactory> g_factory = | 47 base::LazyInstance<AdNetworkDatabaseFactory> g_factory = |
| 135 LAZY_INSTANCE_INITIALIZER; | 48 LAZY_INSTANCE_INITIALIZER; |
| 136 | 49 |
| 137 } // namespace | 50 } // namespace |
| 138 | 51 |
| 139 AdNetworkDatabase::~AdNetworkDatabase() {} | 52 AdNetworkDatabase::~AdNetworkDatabase() {} |
| 140 | 53 |
| 141 // static | 54 // static |
| 142 const AdNetworkDatabase* AdNetworkDatabase::Get() { | 55 const AdNetworkDatabase* AdNetworkDatabase::Get() { |
| 143 return g_factory.Get().GetDatabase(); | 56 return g_factory.Get().GetDatabase(); |
| 144 } | 57 } |
| 145 | 58 |
| 146 // static | 59 // static |
| 147 void AdNetworkDatabase::SetForTesting(scoped_ptr<AdNetworkDatabase> database) { | 60 void AdNetworkDatabase::SetForTesting(scoped_ptr<AdNetworkDatabase> database) { |
| 148 g_factory.Get().SetDatabase(database.Pass()); | 61 g_factory.Get().SetDatabase(database.Pass()); |
| 149 } | 62 } |
| 150 | 63 |
| 151 } // namespace extensions | 64 } // namespace extensions |
| OLD | NEW |