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 | |
24 public: | |
25 static GlobalDuplicateRecords* GetInstance(){ | |
26 return Singleton<GlobalDuplicateRecords>::get(); | |
27 } | |
28 | |
29 std::set<uint32>* content_matches() { | |
30 return &content_matches_; | |
31 } | |
32 | |
33 std::set<uint32>* content_and_url_matches(){ | |
34 return &content_and_url_matches_; | |
35 } | |
36 | |
37 private: | |
38 friend class Singleton<GlobalDuplicateRecords>; | |
39 friend struct DefaultSingletonTraits<GlobalDuplicateRecords>; | |
40 | |
41 GlobalDuplicateRecords() {} | |
42 ~GlobalDuplicateRecords() {} | |
43 | |
44 std::set<uint32> content_matches_; | |
45 std::set<uint32> content_and_url_matches_; | |
46 }; | |
47 | |
48 } // namespace | |
49 | |
50 DuplicateResourceHandler::DuplicateResourceHandler( | |
51 scoped_ptr<ResourceHandler> next_handler, | |
52 ResourceType::Type resource_type) | |
53 : LayeredResourceHandler(next_handler.Pass()), | |
54 resource_type_(resource_type), | |
55 pmurhash_ph1_(0), | |
56 pmurhash_pcarry_(0), | |
57 bytes_read_(0) { | |
58 } | |
59 | |
60 DuplicateResourceHandler::~DuplicateResourceHandler() { | |
61 } | |
62 | |
63 bool DuplicateResourceHandler::OnWillRead(int request_id, net::IOBuffer** buf, | |
64 int* buf_size, int min_size) { | |
65 DCHECK_EQ(-1, min_size); | |
66 | |
67 if (!next_handler_->OnWillRead(request_id, buf, buf_size, min_size)) | |
68 return false; | |
69 read_buffer_ = *buf; | |
70 return true; | |
71 } | |
72 | |
73 bool DuplicateResourceHandler::OnReadCompleted(int request_id, int bytes_read, | |
74 bool* defer) { | |
75 | |
76 PMurHash32_Process(&pmurhash_ph1_, &pmurhash_pcarry_, read_buffer_->data(), by tes_read); | |
77 bytes_read_ += bytes_read; | |
78 | |
79 return next_handler_->OnReadCompleted(request_id, bytes_read, defer); | |
80 } | |
81 | |
82 bool DuplicateResourceHandler::OnResponseCompleted( | |
83 int request_id, | |
84 const net::URLRequestStatus& status, | |
85 const std::string& security_info) { | |
86 | |
87 if (status.status() != net::URLRequestStatus::SUCCESS) | |
88 return next_handler_->OnResponseCompleted(request_id, status, security_info) ; | |
89 | |
90 uint32 resource_hash = PMurHash32_Result(pmurhash_ph1_, pmurhash_pcarry_, byte s_read_); | |
91 | |
92 // Hash url into the resource to see whether it is from the same or | |
93 // different url. | |
94 uint32 hashed_with_url; | |
95 const std::string url_spec = request_->url().spec(); | |
frankwang
2012/07/19 16:10:27
The segfault happens here.
gavinp
2012/07/19 16:14:13
It seems you never initialize request_.
| |
96 int url_length = url_spec.size(); | |
97 PMurHash32_Process(&pmurhash_ph1_, &pmurhash_pcarry_, url_spec.data(), url_len gth); | |
98 hashed_with_url = PMurHash32_Result(pmurhash_ph1_, pmurhash_pcarry_, url_lengt h + bytes_read_); | |
99 | |
100 DVLOG(4) << "url: " << url_spec; | |
101 DVLOG(4) << "resource hash: " << resource_hash; | |
102 DVLOG(4) << "hash with url: " << hashed_with_url; | |
103 | |
104 std::set<uint32>* content_hashes = | |
105 GlobalDuplicateRecords::GetInstance()->content_matches(); | |
106 std::set<uint32>* content_and_url_hashes = | |
107 GlobalDuplicateRecords::GetInstance()->content_and_url_matches(); | |
108 | |
109 const bool did_match_contents = content_hashes->count(resource_hash); | |
110 const bool did_match_contents_and_url = | |
111 content_and_url_hashes->count(hashed_with_url); | |
112 | |
113 UMA_HISTOGRAM_BOOLEAN("Duplicate.Hits", did_match_contents); | |
114 UMA_HISTOGRAM_BOOLEAN("Duplicate.HitsSameUrl", did_match_contents && | |
115 did_match_contents_and_url); | |
116 if (did_match_contents && !did_match_contents_and_url) { | |
117 UMA_HISTOGRAM_CUSTOM_COUNTS("Duplicate.Size.HashHitUrlMiss", bytes_read_, | |
118 1, 0x7FFFFFFF, 50); | |
119 UMA_HISTOGRAM_ENUMERATION("Duplicate.ResourceType.HashHitUrlMiss", resourc e_type_, | |
120 ResourceType::LAST_TYPE); | |
121 content_and_url_hashes->insert(hashed_with_url); | |
122 } else { | |
123 content_hashes->insert(resource_hash); | |
124 content_and_url_hashes->insert(hashed_with_url); | |
125 } | |
126 | |
127 bytes_read_ = 0; | |
128 read_buffer_ = NULL; | |
129 return next_handler_->OnResponseCompleted(request_id, status, security_info); | |
130 } | |
131 | |
132 } //namespace content | |
OLD | NEW |