OLD | NEW |
---|---|
(Empty) | |
1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. | |
2 // Use of this source code is governed by a BSD-style license that can be | |
3 // found in the LICENSE file. | |
4 | |
5 #include "content/browser/renderer_host/duplicate_resource_handler.h" | |
6 | |
7 #include <set> | |
8 | |
9 #include "base/logging.h" | |
10 #include "base/memory/singleton.h" | |
11 #include "base/metrics/histogram.h" | |
12 #include "content/browser/renderer_host/resource_request_info_impl.h" | |
13 #include "net/base/io_buffer.h" | |
14 #include "net/url_request/url_request.h" | |
15 #include "third_party/smhasher/src/PMurHash.h" | |
16 | |
17 | |
18 namespace content { | |
19 | |
20 namespace { | |
21 | |
22 class GlobalDuplicateRecords { | |
23 public: | |
24 static GlobalDuplicateRecords* GetInstance() { | |
25 return Singleton<GlobalDuplicateRecords>::get(); | |
26 } | |
27 | |
28 std::set<uint32>* content_matches() { | |
29 return &content_matches_; | |
30 } | |
31 | |
32 std::set<uint32>* content_and_url_matches() { | |
33 return &content_and_url_matches_; | |
34 } | |
35 | |
36 int* total_bytes_seen() { | |
37 return &total_bytes_seen_; | |
38 } | |
39 | |
40 private: | |
41 friend class Singleton<GlobalDuplicateRecords>; | |
42 friend struct DefaultSingletonTraits<GlobalDuplicateRecords>; | |
43 | |
44 GlobalDuplicateRecords() {} | |
45 ~GlobalDuplicateRecords() {} | |
46 | |
47 std::set<uint32> content_matches_; | |
48 std::set<uint32> content_and_url_matches_; | |
49 int total_bytes_seen_; | |
gavinp
2012/07/20 11:38:47
I think you can't use an int here, it's very likel
frankwang
2012/07/20 17:51:38
I took this out and went with session length.
| |
50 }; | |
51 | |
52 } // namespace | |
53 | |
54 DuplicateResourceHandler::DuplicateResourceHandler( | |
55 scoped_ptr<ResourceHandler> next_handler, | |
56 ResourceType::Type resource_type, | |
57 net::URLRequest* request) | |
58 : LayeredResourceHandler(next_handler.Pass()), | |
59 resource_type_(resource_type), | |
60 bytes_read_(0), | |
61 request_(request), | |
62 pmurhash_ph1_(0), | |
63 pmurhash_pcarry_(0) { | |
64 } | |
65 | |
66 DuplicateResourceHandler::~DuplicateResourceHandler() { | |
67 } | |
68 | |
69 bool DuplicateResourceHandler::OnWillRead(int request_id, net::IOBuffer** buf, | |
70 int* buf_size, int min_size) { | |
71 DCHECK_EQ(-1, min_size); | |
72 | |
73 if (!next_handler_->OnWillRead(request_id, buf, buf_size, min_size)) | |
74 return false; | |
75 read_buffer_ = *buf; | |
76 return true; | |
77 } | |
78 | |
79 bool DuplicateResourceHandler::OnReadCompleted(int request_id, int bytes_read, | |
80 bool* defer) { | |
81 PMurHash32_Process(&pmurhash_ph1_, &pmurhash_pcarry_, | |
82 read_buffer_->data(), bytes_read); | |
83 bytes_read_ += bytes_read; | |
84 return next_handler_->OnReadCompleted(request_id, bytes_read, defer); | |
85 } | |
86 | |
87 bool DuplicateResourceHandler::OnResponseCompleted( | |
88 int request_id, | |
89 const net::URLRequestStatus& status, | |
90 const std::string& security_info) { | |
91 | |
92 if (status.status() != net::URLRequestStatus::SUCCESS) | |
93 return next_handler_->OnResponseCompleted(request_id, | |
94 status, security_info); | |
95 | |
96 int* total_bytes_seen = | |
97 GlobalDuplicateRecords::GetInstance()->total_bytes_seen(); | |
98 *total_bytes_seen += bytes_read_; | |
99 uint32 contents_hash = PMurHash32_Result(pmurhash_ph1_, | |
100 pmurhash_pcarry_, bytes_read_); | |
101 | |
102 // Combine the contents_hash with the url, so we can test if future content | |
103 // identical resources have the same original url or not. | |
104 uint32 hashed_with_url; | |
105 const std::string url_spec = request_->url().spec(); | |
106 PMurHash32_Process(&pmurhash_ph1_, &pmurhash_pcarry_, | |
107 url_spec.data(), url_spec.length()); | |
108 hashed_with_url = PMurHash32_Result(pmurhash_ph1_, pmurhash_pcarry_, | |
109 url_spec.length() + bytes_read_); | |
110 | |
111 DVLOG(4) << "url: " << url_spec; | |
112 DVLOG(4) << "contents hash: " << contents_hash; | |
113 DVLOG(4) << "hash with url: " << hashed_with_url; | |
114 | |
115 std::set<uint32>* content_matches = | |
116 GlobalDuplicateRecords::GetInstance()->content_matches(); | |
117 std::set<uint32>* content_and_url_matches = | |
118 GlobalDuplicateRecords::GetInstance()->content_and_url_matches(); | |
119 | |
120 const bool did_match_contents = content_matches->count(contents_hash); | |
121 const bool did_match_contents_and_url = | |
122 content_and_url_matches->count(hashed_with_url); | |
123 | |
124 UMA_HISTOGRAM_BOOLEAN("Duplicate.Hits", did_match_contents); | |
125 UMA_HISTOGRAM_BOOLEAN("Duplicate.HitsSameUrl", did_match_contents && | |
126 did_match_contents_and_url); | |
127 if (did_match_contents && !did_match_contents_and_url) { | |
128 content_and_url_matches->insert(hashed_with_url); | |
129 UMA_HISTOGRAM_CUSTOM_COUNTS("Duplicate.Size.HashHitUrlMiss", bytes_read_, | |
130 1, 0x7FFFFFFF, 50); | |
131 UMA_HISTOGRAM_ENUMERATION("Duplicate.ResourceType.HashHitUrlMiss", | |
132 resource_type_, ResourceType::LAST_TYPE); | |
133 UMA_HISTOGRAM_CUSTOM_COUNTS("Duplicate.TotalBytesSeen", *total_bytes_seen, | |
134 1, 0x7FFFFFFF, 50); | |
135 } else { | |
gavinp
2012/07/20 11:38:47
I think you don't need this else (insert does noth
frankwang
2012/07/20 17:51:38
Done.
| |
136 content_matches->insert(contents_hash); | |
137 content_and_url_matches->insert(hashed_with_url); | |
138 } | |
139 | |
140 bytes_read_ = 0; | |
141 read_buffer_ = NULL; | |
142 return next_handler_->OnResponseCompleted(request_id, status, security_info); | |
143 } | |
144 | |
145 } // namespace content | |
146 | |
OLD | NEW |