1dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen// Copyright (c) 2011 The Chromium Authors. All rights reserved. 2c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// Use of this source code is governed by a BSD-style license that can be 3c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// found in the LICENSE file. 4c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 5c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "chrome/browser/autofill/form_structure.h" 6c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 7c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "base/basictypes.h" 8c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "base/logging.h" 9c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "base/sha1.h" 103345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick#include "base/string_number_conversions.h" 11c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "base/utf_string_conversions.h" 12201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch#include "chrome/browser/autofill/autofill_metrics.h" 13c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "chrome/browser/autofill/autofill_xml_parser.h" 14c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "chrome/browser/autofill/field_types.h" 15c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "chrome/browser/autofill/form_field.h" 16c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "third_party/libjingle/source/talk/xmllite/xmlelement.h" 17c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch#include "webkit/glue/form_field.h" 18c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 19c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochusing webkit_glue::FormData; 20c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 21c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochnamespace { 22c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 2372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kFormMethodPost[] = "post"; 2472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 2572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen// XML elements and attributes. 2672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kAttributeAcceptedFeatures[] = "accepts"; 27ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenconst char kAttributeAutofillUsed[] = "autofillused"; 28dc0f95d653279beabeb9817299e2902918ba123eKristian Monsenconst char kAttributeAutofillType[] = "autofilltype"; 2972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kAttributeClientVersion[] = "clientversion"; 3072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kAttributeDataPresent[] = "datapresent"; 3172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kAttributeFormSignature[] = "formsignature"; 3272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kAttributeSignature[] = "signature"; 3372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kAcceptedFeatures[] = "e"; // e=experiments 3472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kClientVersion[] = "6.1.1715.1442/en (GGLL)"; 3572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kXMLDeclaration[] = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"; 36ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenconst char kXMLElementAutofillQuery[] = "autofillquery"; 37ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenconst char kXMLElementAutofillUpload[] = "autofillupload"; 3872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kXMLElementForm[] = "form"; 3972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenconst char kXMLElementField[] = "field"; 40c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 41c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// The number of fillable fields necessary for a form to be fillable. 427d214dfa174224b459660971e5b5cce2e06be02aBen Murdoch#ifdef ANDROID 437d214dfa174224b459660971e5b5cce2e06be02aBen Murdoch// Try and autofill more forms on Android, as filling out forms is 447d214dfa174224b459660971e5b5cce2e06be02aBen Murdoch// more frustrating on a mobile device. 457d214dfa174224b459660971e5b5cce2e06be02aBen Murdochconst size_t kRequiredFillableFields = 2; 467d214dfa174224b459660971e5b5cce2e06be02aBen Murdoch#else 47c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochconst size_t kRequiredFillableFields = 3; 487d214dfa174224b459660971e5b5cce2e06be02aBen Murdoch#endif 49c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 50c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} // namespace 51c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 52c407dc5cd9bdc5668497f21b26b09d988ab439deBen MurdochFormStructure::FormStructure(const FormData& form) 53c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch : form_name_(form.name), 54c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch source_url_(form.origin), 55c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch target_url_(form.action), 563345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick has_credit_card_field_(false), 573345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick has_autofillable_field_(false), 583345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick has_password_fields_(false), 59c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch autofill_count_(0) { 60c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Copy the form fields. 61c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch std::vector<webkit_glue::FormField>::const_iterator field; 62c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch for (field = form.fields.begin(); 63c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch field != form.fields.end(); field++) { 6472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Add all supported form fields (including with empty names) to the 65ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // signature. This is a requirement for Autofill servers. 6672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen form_signature_field_names_.append("&"); 67ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen form_signature_field_names_.append(UTF16ToUTF8(field->name)); 68c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 69c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Generate a unique name for this field by appending a counter to the name. 70ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen string16 unique_name = field->name + 713345a6884c488ff3a535c2c9acdd33d74b37e311Iain Merrick base::IntToString16(fields_.size() + 1); 72dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen fields_.push_back(new AutofillField(*field, unique_name)); 73c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 74c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 75c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Terminate the vector with a NULL item. 76c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch fields_.push_back(NULL); 77c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 78c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch std::string method = UTF16ToUTF8(form.method); 79c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (StringToLowerASCII(method) == kFormMethodPost) { 80c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch method_ = POST; 81c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } else { 82c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Either the method is 'get', or we don't know. In this case we default 83c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // to GET. 84c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch method_ = GET; 85c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 86c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 87c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 88731df977c0511bca2206b5f333555b1205ff1f43Iain MerrickFormStructure::~FormStructure() {} 89731df977c0511bca2206b5f333555b1205ff1f43Iain Merrick 90ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid FormStructure::DetermineHeuristicTypes() { 91ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen has_credit_card_field_ = false; 92ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen has_autofillable_field_ = false; 93ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen autofill_count_ = 0; 94ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 95ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FieldTypeMap field_type_map; 96ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen GetHeuristicFieldInfo(&field_type_map); 97ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 98ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (size_t index = 0; index < field_count(); index++) { 99ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillField* field = fields_[index]; 100ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(field); 101ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FieldTypeMap::iterator iter = field_type_map.find(field->unique_name()); 102ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 103ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillFieldType heuristic_autofill_type; 104ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (iter == field_type_map.end()) { 105ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen heuristic_autofill_type = UNKNOWN_TYPE; 106ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 107ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen heuristic_autofill_type = iter->second; 108ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ++autofill_count_; 109ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 110ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 111ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field->set_heuristic_type(heuristic_autofill_type); 112ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 113ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillType autofill_type(field->type()); 114ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (autofill_type.group() == AutofillType::CREDIT_CARD) 115ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen has_credit_card_field_ = true; 116ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (autofill_type.field_type() != UNKNOWN_TYPE) 117ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen has_autofillable_field_ = true; 118ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 119ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 120ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 121ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenbool FormStructure::EncodeUploadRequest(bool autofill_used, 122c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch std::string* encoded_xml) const { 123513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch DCHECK(encoded_xml); 124513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch encoded_xml->clear(); 125ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen bool autofillable = ShouldBeParsed(true); 126ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(autofillable); // Caller should've checked for search pages. 127ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (!autofillable) 128c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 129c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 13072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Set up the <autofillupload> element and its attributes. 13172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen buzz::XmlElement autofill_request_xml( 132ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen (buzz::QName(kXMLElementAutofillUpload))); 13372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion), 13472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen kClientVersion); 13572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen autofill_request_xml.SetAttr(buzz::QName(kAttributeFormSignature), 13672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen FormSignature()); 137ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen autofill_request_xml.SetAttr(buzz::QName(kAttributeAutofillUsed), 138ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen autofill_used ? "true" : "false"); 13972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen autofill_request_xml.SetAttr(buzz::QName(kAttributeDataPresent), 14072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen ConvertPresenceBitsToString().c_str()); 14172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 14272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen if (!EncodeFormRequest(FormStructure::UPLOAD, &autofill_request_xml)) 143513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return false; // Malformed form, skip it. 144c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 145c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Obtain the XML structure as a string. 14672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen *encoded_xml = kXMLDeclaration; 14772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen *encoded_xml += autofill_request_xml.Str().c_str(); 148c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 149c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 150c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 151c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 152c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// static 153c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool FormStructure::EncodeQueryRequest(const ScopedVector<FormStructure>& forms, 154513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::vector<std::string>* encoded_signatures, 155513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string* encoded_xml) { 156513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch DCHECK(encoded_signatures); 157513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch DCHECK(encoded_xml); 158513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch encoded_xml->clear(); 159513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch encoded_signatures->clear(); 160513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch encoded_signatures->reserve(forms.size()); 16172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 16272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Set up the <autofillquery> element and attributes. 16372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen buzz::XmlElement autofill_request_xml( 164ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen (buzz::QName(kXMLElementAutofillQuery))); 16572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion), 16672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen kClientVersion); 16772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures), 16872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen kAcceptedFeatures); 16972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 170513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // Some badly formatted web sites repeat forms - detect that and encode only 171513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // one form as returned data would be the same for all the repeated forms. 172513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::set<std::string> processed_forms; 173c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch for (ScopedVector<FormStructure>::const_iterator it = forms.begin(); 174c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch it != forms.end(); 175c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ++it) { 176513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch std::string signature((*it)->FormSignature()); 177513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (processed_forms.find(signature) != processed_forms.end()) 178513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch continue; 179513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch processed_forms.insert(signature); 180513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch scoped_ptr<buzz::XmlElement> encompassing_xml_element( 18172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen new buzz::XmlElement(buzz::QName(kXMLElementForm))); 182c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch encompassing_xml_element->SetAttr(buzz::QName(kAttributeSignature), 183513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch signature); 184c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 185513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (!(*it)->EncodeFormRequest(FormStructure::QUERY, 186513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch encompassing_xml_element.get())) 187513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch continue; // Malformed form, skip it. 188c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 18972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen autofill_request_xml.AddElement(encompassing_xml_element.release()); 190513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch encoded_signatures->push_back(signature); 191c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 192c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 193513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch if (!encoded_signatures->size()) 194513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return false; 195513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch 196c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Obtain the XML structure as a string. 19772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen *encoded_xml = kXMLDeclaration; 19872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen *encoded_xml += autofill_request_xml.Str().c_str(); 199c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 200c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 201c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 202c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 203c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch// static 204c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochvoid FormStructure::ParseQueryResponse(const std::string& response_xml, 205c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch const std::vector<FormStructure*>& forms, 20621d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen UploadRequired* upload_required, 207dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen const AutofillMetrics& metric_logger) { 208dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen metric_logger.Log(AutofillMetrics::QUERY_RESPONSE_RECEIVED); 209201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 210c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Parse the field types from the server response to the query. 211dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen std::vector<AutofillFieldType> field_types; 21272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string experiment_id; 213ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillQueryXmlParser parse_handler(&field_types, upload_required, 21472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen &experiment_id); 215c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch buzz::XmlParser parser(&parse_handler); 216c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch parser.Parse(response_xml.c_str(), response_xml.length(), true); 217c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!parse_handler.succeeded()) 218c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return; 219c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 220dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen metric_logger.Log(AutofillMetrics::QUERY_RESPONSE_PARSED); 221201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 222201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch bool heuristics_detected_fillable_field = false; 223201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch bool query_response_overrode_heuristics = false; 224201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 225c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Copy the field types into the actual form. 226dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen std::vector<AutofillFieldType>::iterator current_type = field_types.begin(); 227c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch for (std::vector<FormStructure*>::const_iterator iter = forms.begin(); 228c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch iter != forms.end(); ++iter) { 229c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch FormStructure* form = *iter; 23072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen form->server_experiment_id_ = experiment_id; 231201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 232201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch if (form->has_autofillable_field_) 233201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch heuristics_detected_fillable_field = true; 234201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 235c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch form->has_credit_card_field_ = false; 236c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch form->has_autofillable_field_ = false; 237c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 238dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen for (std::vector<AutofillField*>::iterator field = form->fields_.begin(); 239201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch field != form->fields_.end(); ++field, ++current_type) { 240dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen // The field list is terminated by a NULL AutofillField. 241c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!*field) 242c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch break; 243c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 244c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // In some cases *successful* response does not return all the fields. 245c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Quit the update of the types then. 246c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (current_type == field_types.end()) 247c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch break; 248c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 24921d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen // UNKNOWN_TYPE is reserved for use by the client. 25021d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen DCHECK_NE(*current_type, UNKNOWN_TYPE); 25121d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen 252dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen AutofillFieldType heuristic_type = (*field)->type(); 253c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch (*field)->set_server_type(*current_type); 254201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch if (heuristic_type != (*field)->type()) 255201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch query_response_overrode_heuristics = true; 256201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 257dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen AutofillType autofill_type((*field)->type()); 258dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen if (autofill_type.group() == AutofillType::CREDIT_CARD) 259c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch form->has_credit_card_field_ = true; 260c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (autofill_type.field_type() != UNKNOWN_TYPE) 261c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch form->has_autofillable_field_ = true; 262c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 263c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 264ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen form->UpdateAutofillCount(); 265c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 266201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch 267dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen AutofillMetrics::ServerQueryMetric metric; 268201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch if (query_response_overrode_heuristics) { 269201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch if (heuristics_detected_fillable_field) { 270dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen metric = AutofillMetrics::QUERY_RESPONSE_OVERRODE_LOCAL_HEURISTICS; 271201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch } else { 272dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen metric = AutofillMetrics::QUERY_RESPONSE_WITH_NO_LOCAL_HEURISTICS; 273201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch } 274201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch } else { 275dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen metric = AutofillMetrics::QUERY_RESPONSE_MATCHED_LOCAL_HEURISTICS; 276201ade2fbba22bfb27ae029f4d23fca6ded109a0Ben Murdoch } 27721d179b334e59e9a3bfcaed4c4430bef1bc5759dKristian Monsen metric_logger.Log(metric); 278c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 279c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 280c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochstd::string FormStructure::FormSignature() const { 28172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string scheme(target_url_.scheme()); 28272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string host(target_url_.host()); 28372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 28472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // If target host or scheme is empty, set scheme and host of source url. 28572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // This is done to match the Toolbar's behavior. 28672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen if (scheme.empty() || host.empty()) { 28772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen scheme = source_url_.scheme(); 28872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen host = source_url_.host(); 28972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen } 29072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 29172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string form_string = scheme + "://" + host + "&" + 292c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch UTF16ToUTF8(form_name_) + 293c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch form_signature_field_names_; 294c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 295c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return Hash64Bit(form_string); 296c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 297c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 298ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenbool FormStructure::IsAutofillable(bool require_method_post) const { 299c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (autofill_count() < kRequiredFillableFields) 300c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 301c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 3024a5e2dc747d50c653511c68ccb2cfbfb740bd5a7Ben Murdoch return ShouldBeParsed(require_method_post); 303c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 304c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 305ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid FormStructure::UpdateAutofillCount() { 306c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch autofill_count_ = 0; 307dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen for (std::vector<AutofillField*>::const_iterator iter = begin(); 308c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch iter != end(); ++iter) { 309dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen AutofillField* field = *iter; 310c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (field && field->IsFieldFillable()) 311c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch ++autofill_count_; 312c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 313c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 314c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 3154a5e2dc747d50c653511c68ccb2cfbfb740bd5a7Ben Murdochbool FormStructure::ShouldBeParsed(bool require_method_post) const { 316c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (field_count() < kRequiredFillableFields) 317c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 318c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 319c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Rule out http(s)://*/search?... 320c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // e.g. http://www.google.com/search?q=... 321c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // http://search.yahoo.com/search?p=... 322c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (target_url_.path() == "/search") 323c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 324c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 3254a5e2dc747d50c653511c68ccb2cfbfb740bd5a7Ben Murdoch return !require_method_post || (method_ == POST); 326c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 327c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 328ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid FormStructure::UpdateFromCache(const FormStructure& cached_form) { 329ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Map from field signatures to cached fields. 330ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::map<std::string, const AutofillField*> cached_fields; 331ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (size_t i = 0; i < cached_form.field_count(); ++i) { 332ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const AutofillField* field = cached_form.field(i); 333ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen cached_fields[field->FieldSignature()] = field; 334ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 335ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 336ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (std::vector<AutofillField*>::const_iterator iter = begin(); 337ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen iter != end(); ++iter) { 338ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillField* field = *iter; 339ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (!field) 340ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen continue; 341ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 342ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::map<std::string, const AutofillField*>::const_iterator 343ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen cached_field = cached_fields.find(field->FieldSignature()); 344ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (cached_field != cached_fields.end()) { 345ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field->set_heuristic_type(cached_field->second->heuristic_type()); 346ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field->set_server_type(cached_field->second->server_type()); 347ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 348ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 349ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 350ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen UpdateAutofillCount(); 351ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 352ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen server_experiment_id_ = cached_form.server_experiment_id(); 353ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 354ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 355ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsenvoid FormStructure::LogQualityMetrics( 356ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const AutofillMetrics& metric_logger) const { 357ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen std::string experiment_id = server_experiment_id(); 358ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (size_t i = 0; i < field_count(); ++i) { 359ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const AutofillField* field = this->field(i); 360ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::FIELD_SUBMITTED, experiment_id); 361ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 362ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // No further logging for empty fields nor for fields where the entered data 363ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // does not appear to already exist in the user's stored Autofill data. 364ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen const FieldTypeSet& field_types = field->possible_types(); 365ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen DCHECK(!field_types.empty()); 366ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (field_types.count(EMPTY_TYPE) || field_types.count(UNKNOWN_TYPE)) 367ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen continue; 368ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 369ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Collapse field types that Chrome treats as identical, e.g. home and 370ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // billing address fields. 371ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen FieldTypeSet collapsed_field_types; 372ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen for (FieldTypeSet::const_iterator it = field_types.begin(); 373ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen it != field_types.end(); 374ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen ++it) { 375ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Since we currently only support US phone numbers, the (city code + main 376ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // digits) number is almost always identical to the whole phone number. 377ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // TODO(isherman): Improve this logic once we add support for 378ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // international numbers. 379ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (*it == PHONE_HOME_CITY_AND_NUMBER) 380ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen collapsed_field_types.insert(PHONE_HOME_WHOLE_NUMBER); 381ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen else if (*it == PHONE_FAX_CITY_AND_NUMBER) 382ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen collapsed_field_types.insert(PHONE_FAX_WHOLE_NUMBER); 383ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen else 384ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen collapsed_field_types.insert(AutofillType::GetEquivalentFieldType(*it)); 385ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 386ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 387ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Capture the field's type, if it is unambiguous. 388ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillFieldType field_type = UNKNOWN_TYPE; 389ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (collapsed_field_types.size() == 1) 390ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type = *collapsed_field_types.begin(); 391ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 392ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillFieldType heuristic_type = field->heuristic_type(); 393ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillFieldType server_type = field->server_type(); 394ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillFieldType predicted_type = field->type(); 395ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 396ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // Log heuristic, server, and overall type quality metrics, independently of 397ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // whether the field was autofilled. 398ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (heuristic_type == UNKNOWN_TYPE) { 399ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::HEURISTIC_TYPE_UNKNOWN, 400ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 401ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else if (field_types.count(heuristic_type)) { 402ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::HEURISTIC_TYPE_MATCH, 403ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 404ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 405ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::HEURISTIC_TYPE_MISMATCH, 406ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 407ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 408ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 409ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (server_type == NO_SERVER_DATA) { 410ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::SERVER_TYPE_UNKNOWN, 411ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 412ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else if (field_types.count(server_type)) { 413ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::SERVER_TYPE_MATCH, 414ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 415ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 416ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::SERVER_TYPE_MISMATCH, 417ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 418ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 419ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 420ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (predicted_type == UNKNOWN_TYPE) { 421ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::PREDICTED_TYPE_UNKNOWN, 422ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 423ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else if (field_types.count(predicted_type)) { 424ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::PREDICTED_TYPE_MATCH, 425ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 426ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 427ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::PREDICTED_TYPE_MISMATCH, 428ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen field_type, experiment_id); 429ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 430ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 431ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // TODO(isherman): <select> fields don't support |is_autofilled()|, so we 432ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen // have to skip them for the remaining metrics. 433ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (field->form_control_type == ASCIIToUTF16("select-one")) 434ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen continue; 435ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 436ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (field->is_autofilled) { 437ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::FIELD_AUTOFILLED, experiment_id); 438ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 439ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::FIELD_NOT_AUTOFILLED, 440ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 441ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 442ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (heuristic_type == UNKNOWN_TYPE) { 443ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log( 444ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_UNKNOWN, 445ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 446ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else if (field_types.count(heuristic_type)) { 447ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MATCH, 448ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 449ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 450ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log( 451ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MISMATCH, 452ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 453ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 454ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 455ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen if (server_type == NO_SERVER_DATA) { 456ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_UNKNOWN, 457ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 458ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else if (field_types.count(server_type)) { 459ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MATCH, 460ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 461ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } else { 462ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen metric_logger.Log(AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MISMATCH, 463ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen experiment_id); 464ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 465ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 466ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen } 467ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen} 468ddb351dbec246cf1fab5ec20d2d5520909041de1Kristian Monsen 469c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochvoid FormStructure::set_possible_types(int index, const FieldTypeSet& types) { 470c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch int num_fields = static_cast<int>(field_count()); 471c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(index >= 0 && index < num_fields); 472c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (index >= 0 && index < num_fields) 473c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch fields_[index]->set_possible_types(types); 474c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 475c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 476dc0f95d653279beabeb9817299e2902918ba123eKristian Monsenconst AutofillField* FormStructure::field(int index) const { 477c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return fields_[index]; 478c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 479c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 480c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochsize_t FormStructure::field_count() const { 481c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Don't count the NULL terminator. 482c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch size_t field_size = fields_.size(); 483c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return (field_size == 0) ? 0 : field_size - 1; 484c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 485c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 48672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstd::string FormStructure::server_experiment_id() const { 48772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return server_experiment_id_; 48872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 48972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 490c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool FormStructure::operator==(const FormData& form) const { 491c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // TODO(jhawkins): Is this enough to differentiate a form? 492c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (form_name_ == form.name && 493c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch source_url_ == form.origin && 494c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch target_url_ == form.action) { 495c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 496c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 497c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 498c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // TODO(jhawkins): Compare field names, IDs and labels once we have labels 499c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // set up. 500c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 501c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 502c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 503c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 504c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool FormStructure::operator!=(const FormData& form) const { 505c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return !operator==(form); 506c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 507c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 50872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstd::string FormStructure::Hash64Bit(const std::string& str) { 50972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string hash_bin = base::SHA1HashString(str); 51072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DCHECK_EQ(20U, hash_bin.length()); 51172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 51272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen uint64 hash64 = (((static_cast<uint64>(hash_bin[0])) & 0xFF) << 56) | 51372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen (((static_cast<uint64>(hash_bin[1])) & 0xFF) << 48) | 51472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen (((static_cast<uint64>(hash_bin[2])) & 0xFF) << 40) | 51572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen (((static_cast<uint64>(hash_bin[3])) & 0xFF) << 32) | 51672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen (((static_cast<uint64>(hash_bin[4])) & 0xFF) << 24) | 51772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen (((static_cast<uint64>(hash_bin[5])) & 0xFF) << 16) | 51872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen (((static_cast<uint64>(hash_bin[6])) & 0xFF) << 8) | 51972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen ((static_cast<uint64>(hash_bin[7])) & 0xFF); 52072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 52172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return base::Uint64ToString(hash64); 52272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 52372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 524c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochvoid FormStructure::GetHeuristicFieldInfo(FieldTypeMap* field_type_map) { 525c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch FormFieldSet fields(this); 526c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 527c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch FormFieldSet::const_iterator field; 528c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch for (field = fields.begin(); field != fields.end(); field++) { 529c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch bool ok = (*field)->GetFieldInfo(field_type_map); 530c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch DCHECK(ok); 531c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 532c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 533c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 534c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdochbool FormStructure::EncodeFormRequest( 535c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch FormStructure::EncodeRequestType request_type, 536c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch buzz::XmlElement* encompassing_xml_element) const { 537c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (!field_count()) // Nothing to add. 538c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return false; 53972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 540513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // Some badly formatted web sites repeat fields - limit number of fields to 541513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch // 48, which is far larger than any valid form and XML still fits into 2K. 54272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Do not send requests for forms with more than this many fields, as they are 54372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // near certainly not valid/auto-fillable. 544513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch const size_t kMaxFieldsOnTheForm = 48; 54572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen if (field_count() > kMaxFieldsOnTheForm) 546513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch return false; 54772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 548c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch // Add the child nodes for the form fields. 549513209b27ff55e2841eac0e4120199c23acce758Ben Murdoch for (size_t index = 0; index < field_count(); ++index) { 550dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen const AutofillField* field = fields_[index]; 551c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch if (request_type == FormStructure::UPLOAD) { 552c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch FieldTypeSet types = field->possible_types(); 55372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // |types| could be empty in unit-tests only. 55472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen for (FieldTypeSet::iterator field_type = types.begin(); 55572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen field_type != types.end(); ++field_type) { 556c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch buzz::XmlElement *field_element = new buzz::XmlElement( 557c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch buzz::QName(kXMLElementField)); 558c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch 559c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch field_element->SetAttr(buzz::QName(kAttributeSignature), 560c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch field->FieldSignature()); 561dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen field_element->SetAttr(buzz::QName(kAttributeAutofillType), 56272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen base::IntToString(*field_type)); 563c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch encompassing_xml_element->AddElement(field_element); 564c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 565c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } else { 566c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch buzz::XmlElement *field_element = new buzz::XmlElement( 567c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch buzz::QName(kXMLElementField)); 568c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch field_element->SetAttr(buzz::QName(kAttributeSignature), 569c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch field->FieldSignature()); 570c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch encompassing_xml_element->AddElement(field_element); 571c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 572c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch } 573c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch return true; 574c407dc5cd9bdc5668497f21b26b09d988ab439deBen Murdoch} 57572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 57672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsenstd::string FormStructure::ConvertPresenceBitsToString() const { 57772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::vector<uint8> presence_bitfield; 57872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Determine all of the field types that were autofilled. Pack bits into 57972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // |presence_bitfield|. The necessary size for |presence_bitfield| is 58072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // ceil((MAX_VALID_FIELD_TYPE + 7) / 8) bytes (uint8). 58172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen presence_bitfield.resize((MAX_VALID_FIELD_TYPE + 0x7) / 8); 58272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen for (size_t i = 0; i < presence_bitfield.size(); ++i) 58372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen presence_bitfield[i] = 0; 58472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 58572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen for (size_t i = 0; i < field_count(); ++i) { 586dc0f95d653279beabeb9817299e2902918ba123eKristian Monsen const AutofillField* field = fields_[i]; 58772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen FieldTypeSet types = field->possible_types(); 58872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // |types| could be empty in unit-tests only. 58972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen for (FieldTypeSet::iterator field_type = types.begin(); 59072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen field_type != types.end(); ++field_type) { 59172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen DCHECK(presence_bitfield.size() > (static_cast<size_t>(*field_type) / 8)); 59272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Set bit in the bitfield: byte |field_type| / 8, bit in byte 59372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // |field_type| % 8 from the left. 59472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen presence_bitfield[*field_type / 8] |= (0x80 >> (*field_type % 8)); 59572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen } 59672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen } 59772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 59872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen std::string data_presence; 59972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen data_presence.reserve(presence_bitfield.size() * 2 + 1); 60072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 60172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Skip trailing zeroes. If all mask is 0 - return empty string. 60272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen size_t data_end = presence_bitfield.size(); 60372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen for (; data_end > 0 && !presence_bitfield[data_end - 1]; --data_end) { 60472a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen } 60572a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 60672a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen // Print all meaningfull bytes into the string. 60772a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen for (size_t i = 0; i < data_end; ++i) { 60872a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen base::StringAppendF(&data_presence, "%02x", presence_bitfield[i]); 60972a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen } 61072a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 61172a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen return data_presence; 61272a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen} 61372a454cd3513ac24fbdd0e0cb9ad70b86a99b801Kristian Monsen 614