Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1 | // Copyright 2013 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 "components/autofill/core/browser/form_structure.h" |
| 6 | |
| 7 | #include <utility> |
| 8 | |
| 9 | #include "base/basictypes.h" |
| 10 | #include "base/command_line.h" |
| 11 | #include "base/logging.h" |
| 12 | #include "base/memory/scoped_ptr.h" |
| 13 | #include "base/sha1.h" |
| 14 | #include "base/strings/string_number_conversions.h" |
| 15 | #include "base/strings/string_util.h" |
| 16 | #include "base/strings/stringprintf.h" |
| 17 | #include "base/strings/utf_string_conversions.h" |
| 18 | #include "base/time/time.h" |
| 19 | #include "components/autofill/content/browser/autocheckout_page_meta_data.h" |
| 20 | #include "components/autofill/core/browser/autofill_metrics.h" |
| 21 | #include "components/autofill/core/browser/autofill_type.h" |
| 22 | #include "components/autofill/core/browser/autofill_xml_parser.h" |
| 23 | #include "components/autofill/core/browser/field_types.h" |
| 24 | #include "components/autofill/core/browser/form_field.h" |
| 25 | #include "components/autofill/core/common/autofill_constants.h" |
| 26 | #include "components/autofill/core/common/form_data.h" |
| 27 | #include "components/autofill/core/common/form_data_predictions.h" |
| 28 | #include "components/autofill/core/common/form_field_data.h" |
| 29 | #include "components/autofill/core/common/form_field_data_predictions.h" |
Torne (Richard Coles) | a36e592 | 2013-08-05 13:57:33 +0100 | [diff] [blame] | 30 | #include "third_party/icu/source/i18n/unicode/regex.h" |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 31 | #include "third_party/libjingle/source/talk/xmllite/xmlelement.h" |
| 32 | |
| 33 | namespace autofill { |
| 34 | namespace { |
| 35 | |
| 36 | const char kFormMethodPost[] = "post"; |
| 37 | |
| 38 | // XML elements and attributes. |
| 39 | const char kAttributeAcceptedFeatures[] = "accepts"; |
| 40 | const char kAttributeAutofillUsed[] = "autofillused"; |
| 41 | const char kAttributeAutofillType[] = "autofilltype"; |
| 42 | const char kAttributeClientVersion[] = "clientversion"; |
| 43 | const char kAttributeDataPresent[] = "datapresent"; |
| 44 | const char kAttributeFieldID[] = "fieldid"; |
| 45 | const char kAttributeFieldType[] = "fieldtype"; |
| 46 | const char kAttributeFormSignature[] = "formsignature"; |
| 47 | const char kAttributeName[] = "name"; |
| 48 | const char kAttributeSignature[] = "signature"; |
| 49 | const char kAttributeUrlprefixSignature[] = "urlprefixsignature"; |
| 50 | const char kAcceptedFeaturesExperiment[] = "e"; // e=experiments |
| 51 | const char kAcceptedFeaturesAutocheckoutExperiment[] = "a,e"; // a=autocheckout |
| 52 | const char kClientVersion[] = "6.1.1715.1442/en (GGLL)"; |
| 53 | const char kXMLDeclaration[] = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"; |
| 54 | const char kXMLElementAutofillQuery[] = "autofillquery"; |
| 55 | const char kXMLElementAutofillUpload[] = "autofillupload"; |
| 56 | const char kXMLElementFieldAssignments[] = "fieldassignments"; |
| 57 | const char kXMLElementField[] = "field"; |
| 58 | const char kXMLElementFields[] = "fields"; |
| 59 | const char kXMLElementForm[] = "form"; |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 60 | const char kBillingMode[] = "billing"; |
| 61 | const char kShippingMode[] = "shipping"; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 62 | |
Torne (Richard Coles) | a36e592 | 2013-08-05 13:57:33 +0100 | [diff] [blame] | 63 | // Stip away >= 5 consecutive digits. |
| 64 | const char kIgnorePatternInFieldName[] = "\\d{5,}+"; |
| 65 | |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 66 | // Helper for |EncodeUploadRequest()| that creates a bit field corresponding to |
| 67 | // |available_field_types| and returns the hex representation as a string. |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 68 | std::string EncodeFieldTypes(const ServerFieldTypeSet& available_field_types) { |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 69 | // There are |MAX_VALID_FIELD_TYPE| different field types and 8 bits per byte, |
| 70 | // so we need ceil(MAX_VALID_FIELD_TYPE / 8) bytes to encode the bit field. |
| 71 | const size_t kNumBytes = (MAX_VALID_FIELD_TYPE + 0x7) / 8; |
| 72 | |
| 73 | // Pack the types in |available_field_types| into |bit_field|. |
| 74 | std::vector<uint8> bit_field(kNumBytes, 0); |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 75 | for (ServerFieldTypeSet::const_iterator field_type = |
| 76 | available_field_types.begin(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 77 | field_type != available_field_types.end(); |
| 78 | ++field_type) { |
| 79 | // Set the appropriate bit in the field. The bit we set is the one |
| 80 | // |field_type| % 8 from the left of the byte. |
| 81 | const size_t byte = *field_type / 8; |
| 82 | const size_t bit = 0x80 >> (*field_type % 8); |
| 83 | DCHECK(byte < bit_field.size()); |
| 84 | bit_field[byte] |= bit; |
| 85 | } |
| 86 | |
| 87 | // Discard any trailing zeroes. |
| 88 | // If there are no available types, we return the empty string. |
| 89 | size_t data_end = bit_field.size(); |
| 90 | for (; data_end > 0 && !bit_field[data_end - 1]; --data_end) { |
| 91 | } |
| 92 | |
| 93 | // Print all meaningfull bytes into a string. |
| 94 | std::string data_presence; |
| 95 | data_presence.reserve(data_end * 2 + 1); |
| 96 | for (size_t i = 0; i < data_end; ++i) { |
| 97 | base::StringAppendF(&data_presence, "%02x", bit_field[i]); |
| 98 | } |
| 99 | |
| 100 | return data_presence; |
| 101 | } |
| 102 | |
| 103 | // Helper for |EncodeFormRequest()| that creates XmlElements for the given field |
| 104 | // in upload xml, and also add them to the parent XmlElement. |
| 105 | void EncodeFieldForUpload(const AutofillField& field, |
| 106 | buzz::XmlElement* parent) { |
| 107 | // Don't upload checkable fields. |
| 108 | if (field.is_checkable) |
| 109 | return; |
| 110 | |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 111 | ServerFieldTypeSet types = field.possible_types(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 112 | // |types| could be empty in unit-tests only. |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 113 | for (ServerFieldTypeSet::iterator field_type = types.begin(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 114 | field_type != types.end(); ++field_type) { |
| 115 | buzz::XmlElement *field_element = new buzz::XmlElement( |
| 116 | buzz::QName(kXMLElementField)); |
| 117 | |
| 118 | field_element->SetAttr(buzz::QName(kAttributeSignature), |
| 119 | field.FieldSignature()); |
| 120 | field_element->SetAttr(buzz::QName(kAttributeAutofillType), |
| 121 | base::IntToString(*field_type)); |
| 122 | parent->AddElement(field_element); |
| 123 | } |
| 124 | } |
| 125 | |
| 126 | // Helper for |EncodeFormRequest()| that creates XmlElement for the given field |
| 127 | // in query xml, and also add it to the parent XmlElement. |
| 128 | void EncodeFieldForQuery(const AutofillField& field, |
| 129 | buzz::XmlElement* parent) { |
| 130 | buzz::XmlElement *field_element = new buzz::XmlElement( |
| 131 | buzz::QName(kXMLElementField)); |
| 132 | field_element->SetAttr(buzz::QName(kAttributeSignature), |
| 133 | field.FieldSignature()); |
| 134 | parent->AddElement(field_element); |
| 135 | } |
| 136 | |
| 137 | // Helper for |EncodeFormRequest()| that creates XmlElements for the given field |
| 138 | // in field assignments xml, and also add them to the parent XmlElement. |
| 139 | void EncodeFieldForFieldAssignments(const AutofillField& field, |
| 140 | buzz::XmlElement* parent) { |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 141 | ServerFieldTypeSet types = field.possible_types(); |
| 142 | for (ServerFieldTypeSet::iterator field_type = types.begin(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 143 | field_type != types.end(); ++field_type) { |
| 144 | buzz::XmlElement *field_element = new buzz::XmlElement( |
| 145 | buzz::QName(kXMLElementFields)); |
| 146 | |
| 147 | field_element->SetAttr(buzz::QName(kAttributeFieldID), |
| 148 | field.FieldSignature()); |
| 149 | field_element->SetAttr(buzz::QName(kAttributeFieldType), |
| 150 | base::IntToString(*field_type)); |
| 151 | field_element->SetAttr(buzz::QName(kAttributeName), |
| 152 | UTF16ToUTF8(field.name)); |
| 153 | parent->AddElement(field_element); |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | // Returns |true| iff the |token| is a type hint for a contact field, as |
| 158 | // specified in the implementation section of http://is.gd/whatwg_autocomplete |
| 159 | // Note that "fax" and "pager" are intentionally ignored, as Chrome does not |
| 160 | // support filling either type of information. |
| 161 | bool IsContactTypeHint(const std::string& token) { |
| 162 | return token == "home" || token == "work" || token == "mobile"; |
| 163 | } |
| 164 | |
| 165 | // Returns |true| iff the |token| is a type hint appropriate for a field of the |
| 166 | // given |field_type|, as specified in the implementation section of |
| 167 | // http://is.gd/whatwg_autocomplete |
| 168 | bool ContactTypeHintMatchesFieldType(const std::string& token, |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 169 | HtmlFieldType field_type) { |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 170 | // The "home" and "work" type hints are only appropriate for email and phone |
| 171 | // number field types. |
| 172 | if (token == "home" || token == "work") { |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 173 | return field_type == HTML_TYPE_EMAIL || |
| 174 | (field_type >= HTML_TYPE_TEL && |
| 175 | field_type <= HTML_TYPE_TEL_LOCAL_SUFFIX); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 176 | } |
| 177 | |
| 178 | // The "mobile" type hint is only appropriate for phone number field types. |
| 179 | // Note that "fax" and "pager" are intentionally ignored, as Chrome does not |
| 180 | // support filling either type of information. |
| 181 | if (token == "mobile") { |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 182 | return field_type >= HTML_TYPE_TEL && |
| 183 | field_type <= HTML_TYPE_TEL_LOCAL_SUFFIX; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 184 | } |
| 185 | |
| 186 | return false; |
| 187 | } |
| 188 | |
| 189 | // Returns the Chrome Autofill-supported field type corresponding to the given |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 190 | // |autocomplete_attribute_value|, if there is one, in the context of the given |
| 191 | // |field|. Chrome Autofill supports a subset of the field types listed at |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 192 | // http://is.gd/whatwg_autocomplete |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 193 | HtmlFieldType FieldTypeFromAutocompleteAttributeValue( |
| 194 | const std::string& autocomplete_attribute_value, |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 195 | const AutofillField& field) { |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 196 | if (autocomplete_attribute_value == "name") |
| 197 | return HTML_TYPE_NAME; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 198 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 199 | if (autocomplete_attribute_value == "given-name") |
| 200 | return HTML_TYPE_GIVEN_NAME; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 201 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 202 | if (autocomplete_attribute_value == "additional-name") { |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 203 | if (field.max_length == 1) |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 204 | return HTML_TYPE_ADDITIONAL_NAME_INITIAL; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 205 | else |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 206 | return HTML_TYPE_ADDITIONAL_NAME; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 207 | } |
| 208 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 209 | if (autocomplete_attribute_value == "family-name") |
| 210 | return HTML_TYPE_FAMILY_NAME; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 211 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 212 | if (autocomplete_attribute_value == "organization") |
| 213 | return HTML_TYPE_ORGANIZATION; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 214 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 215 | if (autocomplete_attribute_value == "address-line1") |
| 216 | return HTML_TYPE_ADDRESS_LINE1; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 217 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 218 | if (autocomplete_attribute_value == "address-line2") |
| 219 | return HTML_TYPE_ADDRESS_LINE2; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 220 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 221 | if (autocomplete_attribute_value == "locality") |
| 222 | return HTML_TYPE_LOCALITY; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 223 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 224 | if (autocomplete_attribute_value == "region") |
| 225 | return HTML_TYPE_REGION; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 226 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 227 | if (autocomplete_attribute_value == "country") |
| 228 | return HTML_TYPE_COUNTRY_CODE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 229 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 230 | if (autocomplete_attribute_value == "country-name") |
| 231 | return HTML_TYPE_COUNTRY_NAME; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 232 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 233 | if (autocomplete_attribute_value == "postal-code") |
| 234 | return HTML_TYPE_POSTAL_CODE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 235 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 236 | if (autocomplete_attribute_value == "cc-name") |
| 237 | return HTML_TYPE_CREDIT_CARD_NAME; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 238 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 239 | if (autocomplete_attribute_value == "cc-number") |
| 240 | return HTML_TYPE_CREDIT_CARD_NUMBER; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 241 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 242 | if (autocomplete_attribute_value == "cc-exp") { |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 243 | if (field.max_length == 5) |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 244 | return HTML_TYPE_CREDIT_CARD_EXP_DATE_2_DIGIT_YEAR; |
| 245 | else if (field.max_length == 7) |
| 246 | return HTML_TYPE_CREDIT_CARD_EXP_DATE_4_DIGIT_YEAR; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 247 | else |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 248 | return HTML_TYPE_CREDIT_CARD_EXP; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 249 | } |
| 250 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 251 | if (autocomplete_attribute_value == "cc-exp-month") |
| 252 | return HTML_TYPE_CREDIT_CARD_EXP_MONTH; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 253 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 254 | if (autocomplete_attribute_value == "cc-exp-year") { |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 255 | if (field.max_length == 2) |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 256 | return HTML_TYPE_CREDIT_CARD_EXP_2_DIGIT_YEAR; |
| 257 | else if (field.max_length == 4) |
| 258 | return HTML_TYPE_CREDIT_CARD_EXP_4_DIGIT_YEAR; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 259 | else |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 260 | return HTML_TYPE_CREDIT_CARD_EXP_YEAR; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 261 | } |
| 262 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 263 | if (autocomplete_attribute_value == "cc-csc") |
| 264 | return HTML_TYPE_CREDIT_CARD_VERIFICATION_CODE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 265 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 266 | if (autocomplete_attribute_value == "cc-type") |
| 267 | return HTML_TYPE_CREDIT_CARD_TYPE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 268 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 269 | if (autocomplete_attribute_value == "tel") |
| 270 | return HTML_TYPE_TEL; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 271 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 272 | if (autocomplete_attribute_value == "tel-country-code") |
| 273 | return HTML_TYPE_TEL_COUNTRY_CODE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 274 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 275 | if (autocomplete_attribute_value == "tel-national") |
| 276 | return HTML_TYPE_TEL_NATIONAL; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 277 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 278 | if (autocomplete_attribute_value == "tel-area-code") |
| 279 | return HTML_TYPE_TEL_AREA_CODE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 280 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 281 | if (autocomplete_attribute_value == "tel-local") |
| 282 | return HTML_TYPE_TEL_LOCAL; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 283 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 284 | if (autocomplete_attribute_value == "tel-local-prefix") |
| 285 | return HTML_TYPE_TEL_LOCAL_PREFIX; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 286 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 287 | if (autocomplete_attribute_value == "tel-local-suffix") |
| 288 | return HTML_TYPE_TEL_LOCAL_SUFFIX; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 289 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 290 | if (autocomplete_attribute_value == "email") |
| 291 | return HTML_TYPE_EMAIL; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 292 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 293 | return HTML_TYPE_UNKNOWN; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 294 | } |
| 295 | |
Torne (Richard Coles) | a36e592 | 2013-08-05 13:57:33 +0100 | [diff] [blame] | 296 | std::string StripDigitsIfRequired(const base::string16& input) { |
| 297 | UErrorCode status = U_ZERO_ERROR; |
| 298 | CR_DEFINE_STATIC_LOCAL(icu::UnicodeString, icu_pattern, |
| 299 | (kIgnorePatternInFieldName)); |
| 300 | CR_DEFINE_STATIC_LOCAL(icu::RegexMatcher, matcher, |
| 301 | (icu_pattern, UREGEX_CASE_INSENSITIVE, status)); |
| 302 | DCHECK_EQ(status, U_ZERO_ERROR); |
| 303 | |
| 304 | icu::UnicodeString icu_input(input.data(), input.length()); |
| 305 | matcher.reset(icu_input); |
| 306 | |
| 307 | icu::UnicodeString replaced_string = matcher.replaceAll("", status); |
| 308 | |
| 309 | std::string return_string; |
| 310 | status = U_ZERO_ERROR; |
| 311 | UTF16ToUTF8(replaced_string.getBuffer(), |
| 312 | static_cast<size_t>(replaced_string.length()), |
| 313 | &return_string); |
| 314 | if (status != U_ZERO_ERROR) { |
| 315 | DVLOG(1) << "Couldn't strip digits in " << UTF16ToUTF8(input); |
| 316 | return UTF16ToUTF8(input); |
| 317 | } |
| 318 | |
| 319 | return return_string; |
| 320 | } |
| 321 | |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 322 | } // namespace |
| 323 | |
| 324 | FormStructure::FormStructure(const FormData& form, |
| 325 | const std::string& autocheckout_url_prefix) |
| 326 | : form_name_(form.name), |
| 327 | source_url_(form.origin), |
| 328 | target_url_(form.action), |
| 329 | autofill_count_(0), |
| 330 | active_field_count_(0), |
| 331 | upload_required_(USE_UPLOAD_RATES), |
| 332 | server_experiment_id_("no server response"), |
| 333 | has_author_specified_types_(false), |
| 334 | autocheckout_url_prefix_(autocheckout_url_prefix), |
| 335 | filled_by_autocheckout_(false) { |
| 336 | // Copy the form fields. |
| 337 | std::map<base::string16, size_t> unique_names; |
| 338 | for (std::vector<FormFieldData>::const_iterator field = |
| 339 | form.fields.begin(); |
| 340 | field != form.fields.end(); field++) { |
| 341 | |
| 342 | if (!ShouldSkipField(*field)) { |
| 343 | // Add all supported form fields (including with empty names) to the |
| 344 | // signature. This is a requirement for Autofill servers. |
| 345 | form_signature_field_names_.append("&"); |
Torne (Richard Coles) | a36e592 | 2013-08-05 13:57:33 +0100 | [diff] [blame] | 346 | form_signature_field_names_.append(StripDigitsIfRequired(field->name)); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 347 | |
| 348 | ++active_field_count_; |
| 349 | } |
| 350 | |
| 351 | // Generate a unique name for this field by appending a counter to the name. |
| 352 | // Make sure to prepend the counter with a non-numeric digit so that we are |
| 353 | // guaranteed to avoid collisions. |
| 354 | if (!unique_names.count(field->name)) |
| 355 | unique_names[field->name] = 1; |
| 356 | else |
| 357 | ++unique_names[field->name]; |
| 358 | base::string16 unique_name = field->name + ASCIIToUTF16("_") + |
| 359 | base::IntToString16(unique_names[field->name]); |
| 360 | fields_.push_back(new AutofillField(*field, unique_name)); |
| 361 | } |
| 362 | |
| 363 | std::string method = UTF16ToUTF8(form.method); |
| 364 | if (StringToLowerASCII(method) == kFormMethodPost) { |
| 365 | method_ = POST; |
| 366 | } else { |
| 367 | // Either the method is 'get', or we don't know. In this case we default |
| 368 | // to GET. |
| 369 | method_ = GET; |
| 370 | } |
| 371 | } |
| 372 | |
| 373 | FormStructure::~FormStructure() {} |
| 374 | |
| 375 | void FormStructure::DetermineHeuristicTypes( |
| 376 | const AutofillMetrics& metric_logger) { |
| 377 | // First, try to detect field types based on each field's |autocomplete| |
| 378 | // attribute value. If there is at least one form field that specifies an |
| 379 | // autocomplete type hint, don't try to apply other heuristics to match fields |
| 380 | // in this form. |
| 381 | bool has_author_specified_sections; |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 382 | ParseFieldTypesFromAutocompleteAttributes(&has_author_specified_types_, |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 383 | &has_author_specified_sections); |
| 384 | |
| 385 | if (!has_author_specified_types_) { |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 386 | ServerFieldTypeMap field_type_map; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 387 | FormField::ParseFormFields(fields_.get(), &field_type_map); |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 388 | for (size_t i = 0; i < field_count(); ++i) { |
| 389 | AutofillField* field = fields_[i]; |
| 390 | ServerFieldTypeMap::iterator iter = |
| 391 | field_type_map.find(field->unique_name()); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 392 | if (iter != field_type_map.end()) |
| 393 | field->set_heuristic_type(iter->second); |
| 394 | } |
| 395 | } |
| 396 | |
| 397 | UpdateAutofillCount(); |
| 398 | IdentifySections(has_author_specified_sections); |
| 399 | |
| 400 | if (IsAutofillable(true)) { |
| 401 | metric_logger.LogDeveloperEngagementMetric( |
| 402 | AutofillMetrics::FILLABLE_FORM_PARSED); |
| 403 | if (has_author_specified_types_) { |
| 404 | metric_logger.LogDeveloperEngagementMetric( |
| 405 | AutofillMetrics::FILLABLE_FORM_CONTAINS_TYPE_HINTS); |
| 406 | } |
| 407 | } |
| 408 | } |
| 409 | |
| 410 | bool FormStructure::EncodeUploadRequest( |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 411 | const ServerFieldTypeSet& available_field_types, |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 412 | bool form_was_autofilled, |
| 413 | std::string* encoded_xml) const { |
| 414 | DCHECK(ShouldBeCrowdsourced()); |
| 415 | |
| 416 | // Verify that |available_field_types| agrees with the possible field types we |
| 417 | // are uploading. |
| 418 | for (std::vector<AutofillField*>::const_iterator field = begin(); |
| 419 | field != end(); |
| 420 | ++field) { |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 421 | for (ServerFieldTypeSet::const_iterator type = |
| 422 | (*field)->possible_types().begin(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 423 | type != (*field)->possible_types().end(); |
| 424 | ++type) { |
| 425 | DCHECK(*type == UNKNOWN_TYPE || |
| 426 | *type == EMPTY_TYPE || |
| 427 | available_field_types.count(*type)); |
| 428 | } |
| 429 | } |
| 430 | |
| 431 | // Set up the <autofillupload> element and its attributes. |
| 432 | buzz::XmlElement autofill_request_xml( |
| 433 | (buzz::QName(kXMLElementAutofillUpload))); |
| 434 | autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion), |
| 435 | kClientVersion); |
| 436 | autofill_request_xml.SetAttr(buzz::QName(kAttributeFormSignature), |
| 437 | FormSignature()); |
| 438 | autofill_request_xml.SetAttr(buzz::QName(kAttributeAutofillUsed), |
| 439 | form_was_autofilled ? "true" : "false"); |
| 440 | autofill_request_xml.SetAttr(buzz::QName(kAttributeDataPresent), |
| 441 | EncodeFieldTypes(available_field_types).c_str()); |
| 442 | |
| 443 | if (!EncodeFormRequest(FormStructure::UPLOAD, &autofill_request_xml)) |
| 444 | return false; // Malformed form, skip it. |
| 445 | |
| 446 | // Obtain the XML structure as a string. |
| 447 | *encoded_xml = kXMLDeclaration; |
| 448 | *encoded_xml += autofill_request_xml.Str().c_str(); |
| 449 | |
| 450 | // To enable this logging, run with the flag --vmodule="form_structure=2". |
| 451 | VLOG(2) << "\n" << *encoded_xml; |
| 452 | |
| 453 | return true; |
| 454 | } |
| 455 | |
| 456 | bool FormStructure::EncodeFieldAssignments( |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 457 | const ServerFieldTypeSet& available_field_types, |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 458 | std::string* encoded_xml) const { |
| 459 | DCHECK(ShouldBeCrowdsourced()); |
| 460 | |
| 461 | // Set up the <fieldassignments> element and its attributes. |
| 462 | buzz::XmlElement autofill_request_xml( |
| 463 | (buzz::QName(kXMLElementFieldAssignments))); |
| 464 | autofill_request_xml.SetAttr(buzz::QName(kAttributeFormSignature), |
| 465 | FormSignature()); |
| 466 | |
| 467 | if (!EncodeFormRequest(FormStructure::FIELD_ASSIGNMENTS, |
| 468 | &autofill_request_xml)) |
| 469 | return false; // Malformed form, skip it. |
| 470 | |
| 471 | // Obtain the XML structure as a string. |
| 472 | *encoded_xml = kXMLDeclaration; |
| 473 | *encoded_xml += autofill_request_xml.Str().c_str(); |
| 474 | |
| 475 | return true; |
| 476 | } |
| 477 | |
| 478 | // static |
| 479 | bool FormStructure::EncodeQueryRequest( |
| 480 | const std::vector<FormStructure*>& forms, |
| 481 | std::vector<std::string>* encoded_signatures, |
| 482 | std::string* encoded_xml) { |
| 483 | DCHECK(encoded_signatures); |
| 484 | DCHECK(encoded_xml); |
| 485 | encoded_xml->clear(); |
| 486 | encoded_signatures->clear(); |
| 487 | encoded_signatures->reserve(forms.size()); |
| 488 | |
| 489 | // Set up the <autofillquery> element and attributes. |
| 490 | buzz::XmlElement autofill_request_xml( |
| 491 | (buzz::QName(kXMLElementAutofillQuery))); |
| 492 | autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion), |
| 493 | kClientVersion); |
| 494 | |
| 495 | // autocheckout_url_prefix tells the Autofill server where the forms in the |
| 496 | // request came from, and the the Autofill server checks internal status and |
| 497 | // decide to enable Autocheckout or not and may return Autocheckout related |
| 498 | // data in the response accordingly. |
| 499 | // There is no page/frame level object associated with FormStructure that |
| 500 | // we could extract URL prefix from. But, all the forms should come from the |
| 501 | // same frame, so they should have the same Autocheckout URL prefix. Thus we |
| 502 | // use URL prefix from the first form with Autocheckout enabled. |
| 503 | std::string autocheckout_url_prefix; |
| 504 | |
| 505 | // Some badly formatted web sites repeat forms - detect that and encode only |
| 506 | // one form as returned data would be the same for all the repeated forms. |
| 507 | std::set<std::string> processed_forms; |
| 508 | for (ScopedVector<FormStructure>::const_iterator it = forms.begin(); |
| 509 | it != forms.end(); |
| 510 | ++it) { |
| 511 | std::string signature((*it)->FormSignature()); |
| 512 | if (processed_forms.find(signature) != processed_forms.end()) |
| 513 | continue; |
| 514 | processed_forms.insert(signature); |
| 515 | scoped_ptr<buzz::XmlElement> encompassing_xml_element( |
| 516 | new buzz::XmlElement(buzz::QName(kXMLElementForm))); |
| 517 | encompassing_xml_element->SetAttr(buzz::QName(kAttributeSignature), |
| 518 | signature); |
| 519 | |
| 520 | if (!(*it)->EncodeFormRequest(FormStructure::QUERY, |
| 521 | encompassing_xml_element.get())) |
| 522 | continue; // Malformed form, skip it. |
| 523 | |
| 524 | if ((*it)->IsAutocheckoutEnabled()) { |
| 525 | if (autocheckout_url_prefix.empty()) { |
| 526 | autocheckout_url_prefix = (*it)->autocheckout_url_prefix_; |
| 527 | } else { |
| 528 | // Making sure all the forms in the request has the same url_prefix. |
| 529 | DCHECK_EQ(autocheckout_url_prefix, (*it)->autocheckout_url_prefix_); |
| 530 | } |
| 531 | } |
| 532 | |
| 533 | autofill_request_xml.AddElement(encompassing_xml_element.release()); |
| 534 | encoded_signatures->push_back(signature); |
| 535 | } |
| 536 | |
| 537 | if (!encoded_signatures->size()) |
| 538 | return false; |
| 539 | |
| 540 | if (autocheckout_url_prefix.empty()) { |
| 541 | autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures), |
| 542 | kAcceptedFeaturesExperiment); |
| 543 | } else { |
| 544 | autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures), |
| 545 | kAcceptedFeaturesAutocheckoutExperiment); |
| 546 | autofill_request_xml.SetAttr(buzz::QName(kAttributeUrlprefixSignature), |
| 547 | Hash64Bit(autocheckout_url_prefix)); |
| 548 | } |
| 549 | |
| 550 | // Obtain the XML structure as a string. |
| 551 | *encoded_xml = kXMLDeclaration; |
| 552 | *encoded_xml += autofill_request_xml.Str().c_str(); |
| 553 | |
| 554 | return true; |
| 555 | } |
| 556 | |
| 557 | // static |
| 558 | void FormStructure::ParseQueryResponse( |
| 559 | const std::string& response_xml, |
| 560 | const std::vector<FormStructure*>& forms, |
| 561 | autofill::AutocheckoutPageMetaData* page_meta_data, |
| 562 | const AutofillMetrics& metric_logger) { |
| 563 | metric_logger.LogServerQueryMetric(AutofillMetrics::QUERY_RESPONSE_RECEIVED); |
| 564 | |
| 565 | // Parse the field types from the server response to the query. |
| 566 | std::vector<AutofillServerFieldInfo> field_infos; |
| 567 | UploadRequired upload_required; |
| 568 | std::string experiment_id; |
| 569 | AutofillQueryXmlParser parse_handler(&field_infos, |
| 570 | &upload_required, |
| 571 | &experiment_id, |
| 572 | page_meta_data); |
| 573 | buzz::XmlParser parser(&parse_handler); |
| 574 | parser.Parse(response_xml.c_str(), response_xml.length(), true); |
| 575 | if (!parse_handler.succeeded()) |
| 576 | return; |
| 577 | |
| 578 | metric_logger.LogServerQueryMetric(AutofillMetrics::QUERY_RESPONSE_PARSED); |
| 579 | metric_logger.LogServerExperimentIdForQuery(experiment_id); |
| 580 | |
| 581 | bool heuristics_detected_fillable_field = false; |
| 582 | bool query_response_overrode_heuristics = false; |
| 583 | |
| 584 | // Copy the field types into the actual form. |
| 585 | std::vector<AutofillServerFieldInfo>::iterator current_info = |
| 586 | field_infos.begin(); |
| 587 | for (std::vector<FormStructure*>::const_iterator iter = forms.begin(); |
| 588 | iter != forms.end(); ++iter) { |
| 589 | FormStructure* form = *iter; |
| 590 | form->upload_required_ = upload_required; |
| 591 | form->server_experiment_id_ = experiment_id; |
| 592 | |
| 593 | for (std::vector<AutofillField*>::iterator field = form->fields_.begin(); |
| 594 | field != form->fields_.end(); ++field) { |
| 595 | if (form->ShouldSkipField(**field)) |
| 596 | continue; |
| 597 | |
| 598 | // In some cases *successful* response does not return all the fields. |
| 599 | // Quit the update of the types then. |
| 600 | if (current_info == field_infos.end()) |
| 601 | break; |
| 602 | |
| 603 | // UNKNOWN_TYPE is reserved for use by the client. |
| 604 | DCHECK_NE(current_info->field_type, UNKNOWN_TYPE); |
| 605 | |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 606 | ServerFieldType heuristic_type = (*field)->heuristic_type(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 607 | if (heuristic_type != UNKNOWN_TYPE) |
| 608 | heuristics_detected_fillable_field = true; |
| 609 | |
| 610 | (*field)->set_server_type(current_info->field_type); |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 611 | if (heuristic_type != (*field)->Type().GetStorableType()) |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 612 | query_response_overrode_heuristics = true; |
| 613 | |
| 614 | // Copy default value into the field if available. |
| 615 | if (!current_info->default_value.empty()) |
| 616 | (*field)->set_default_value(current_info->default_value); |
| 617 | |
| 618 | ++current_info; |
| 619 | } |
| 620 | |
| 621 | form->UpdateAutofillCount(); |
| 622 | form->IdentifySections(false); |
| 623 | } |
| 624 | |
| 625 | AutofillMetrics::ServerQueryMetric metric; |
| 626 | if (query_response_overrode_heuristics) { |
| 627 | if (heuristics_detected_fillable_field) { |
| 628 | metric = AutofillMetrics::QUERY_RESPONSE_OVERRODE_LOCAL_HEURISTICS; |
| 629 | } else { |
| 630 | metric = AutofillMetrics::QUERY_RESPONSE_WITH_NO_LOCAL_HEURISTICS; |
| 631 | } |
| 632 | } else { |
| 633 | metric = AutofillMetrics::QUERY_RESPONSE_MATCHED_LOCAL_HEURISTICS; |
| 634 | } |
| 635 | metric_logger.LogServerQueryMetric(metric); |
| 636 | } |
| 637 | |
| 638 | // static |
| 639 | void FormStructure::GetFieldTypePredictions( |
| 640 | const std::vector<FormStructure*>& form_structures, |
| 641 | std::vector<FormDataPredictions>* forms) { |
| 642 | forms->clear(); |
| 643 | forms->reserve(form_structures.size()); |
| 644 | for (size_t i = 0; i < form_structures.size(); ++i) { |
| 645 | FormStructure* form_structure = form_structures[i]; |
| 646 | FormDataPredictions form; |
| 647 | form.data.name = form_structure->form_name_; |
| 648 | form.data.method = |
| 649 | ASCIIToUTF16((form_structure->method_ == POST) ? "POST" : "GET"); |
| 650 | form.data.origin = form_structure->source_url_; |
| 651 | form.data.action = form_structure->target_url_; |
| 652 | form.signature = form_structure->FormSignature(); |
| 653 | form.experiment_id = form_structure->server_experiment_id_; |
| 654 | |
| 655 | for (std::vector<AutofillField*>::const_iterator field = |
| 656 | form_structure->fields_.begin(); |
| 657 | field != form_structure->fields_.end(); ++field) { |
| 658 | form.data.fields.push_back(FormFieldData(**field)); |
| 659 | |
| 660 | FormFieldDataPredictions annotated_field; |
| 661 | annotated_field.signature = (*field)->FieldSignature(); |
| 662 | annotated_field.heuristic_type = |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 663 | AutofillType((*field)->heuristic_type()).ToString(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 664 | annotated_field.server_type = |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 665 | AutofillType((*field)->server_type()).ToString(); |
| 666 | annotated_field.overall_type = (*field)->Type().ToString(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 667 | form.fields.push_back(annotated_field); |
| 668 | } |
| 669 | |
| 670 | forms->push_back(form); |
| 671 | } |
| 672 | } |
| 673 | |
| 674 | std::string FormStructure::FormSignature() const { |
| 675 | std::string scheme(target_url_.scheme()); |
| 676 | std::string host(target_url_.host()); |
| 677 | |
| 678 | // If target host or scheme is empty, set scheme and host of source url. |
| 679 | // This is done to match the Toolbar's behavior. |
| 680 | if (scheme.empty() || host.empty()) { |
| 681 | scheme = source_url_.scheme(); |
| 682 | host = source_url_.host(); |
| 683 | } |
| 684 | |
| 685 | std::string form_string = scheme + "://" + host + "&" + |
| 686 | UTF16ToUTF8(form_name_) + |
| 687 | form_signature_field_names_; |
| 688 | |
| 689 | return Hash64Bit(form_string); |
| 690 | } |
| 691 | |
| 692 | bool FormStructure::IsAutocheckoutEnabled() const { |
| 693 | return !autocheckout_url_prefix_.empty(); |
| 694 | } |
| 695 | |
| 696 | bool FormStructure::ShouldSkipField(const FormFieldData& field) const { |
| 697 | return (field.is_checkable || field.form_control_type == "password") && |
| 698 | !IsAutocheckoutEnabled(); |
| 699 | } |
| 700 | |
| 701 | size_t FormStructure::RequiredFillableFields() const { |
| 702 | return IsAutocheckoutEnabled() ? 0 : kRequiredAutofillFields; |
| 703 | } |
| 704 | |
| 705 | bool FormStructure::IsAutofillable(bool require_method_post) const { |
| 706 | if (autofill_count() < RequiredFillableFields()) |
| 707 | return false; |
| 708 | |
| 709 | return ShouldBeParsed(require_method_post); |
| 710 | } |
| 711 | |
| 712 | void FormStructure::UpdateAutofillCount() { |
| 713 | autofill_count_ = 0; |
| 714 | for (std::vector<AutofillField*>::const_iterator iter = begin(); |
| 715 | iter != end(); ++iter) { |
| 716 | AutofillField* field = *iter; |
| 717 | if (field && field->IsFieldFillable()) |
| 718 | ++autofill_count_; |
| 719 | } |
| 720 | } |
| 721 | |
| 722 | bool FormStructure::ShouldBeParsed(bool require_method_post) const { |
| 723 | if (active_field_count() < RequiredFillableFields()) |
| 724 | return false; |
| 725 | |
| 726 | // Rule out http(s)://*/search?... |
| 727 | // e.g. http://www.google.com/search?q=... |
| 728 | // http://search.yahoo.com/search?p=... |
| 729 | if (target_url_.path() == "/search") |
| 730 | return false; |
| 731 | |
| 732 | if (!IsAutocheckoutEnabled()) { |
| 733 | // Make sure there is at least one text field when Autocheckout is |
| 734 | // not enabled. |
| 735 | bool has_text_field = false; |
| 736 | for (std::vector<AutofillField*>::const_iterator it = begin(); |
| 737 | it != end() && !has_text_field; ++it) { |
| 738 | has_text_field |= (*it)->form_control_type != "select-one"; |
| 739 | } |
| 740 | if (!has_text_field) |
| 741 | return false; |
| 742 | } |
| 743 | |
| 744 | return !require_method_post || (method_ == POST); |
| 745 | } |
| 746 | |
| 747 | bool FormStructure::ShouldBeCrowdsourced() const { |
| 748 | // Allow all forms in Autocheckout flow to be crowdsourced. |
| 749 | return (!has_author_specified_types_ && ShouldBeParsed(true)) || |
| 750 | IsAutocheckoutEnabled(); |
| 751 | } |
| 752 | |
| 753 | void FormStructure::UpdateFromCache(const FormStructure& cached_form) { |
| 754 | // Map from field signatures to cached fields. |
| 755 | std::map<std::string, const AutofillField*> cached_fields; |
| 756 | for (size_t i = 0; i < cached_form.field_count(); ++i) { |
| 757 | const AutofillField* field = cached_form.field(i); |
| 758 | cached_fields[field->FieldSignature()] = field; |
| 759 | } |
| 760 | |
| 761 | for (std::vector<AutofillField*>::const_iterator iter = begin(); |
| 762 | iter != end(); ++iter) { |
| 763 | AutofillField* field = *iter; |
| 764 | |
| 765 | std::map<std::string, const AutofillField*>::const_iterator |
| 766 | cached_field = cached_fields.find(field->FieldSignature()); |
| 767 | if (cached_field != cached_fields.end()) { |
| 768 | if (field->form_control_type != "select-one" && |
| 769 | field->value == cached_field->second->value) { |
| 770 | // From the perspective of learning user data, text fields containing |
| 771 | // default values are equivalent to empty fields. |
| 772 | field->value = base::string16(); |
| 773 | } |
| 774 | |
| 775 | field->set_heuristic_type(cached_field->second->heuristic_type()); |
| 776 | field->set_server_type(cached_field->second->server_type()); |
| 777 | } |
| 778 | } |
| 779 | |
| 780 | UpdateAutofillCount(); |
| 781 | |
| 782 | filled_by_autocheckout_ = cached_form.filled_by_autocheckout(); |
| 783 | server_experiment_id_ = cached_form.server_experiment_id(); |
| 784 | |
| 785 | // The form signature should match between query and upload requests to the |
| 786 | // server. On many websites, form elements are dynamically added, removed, or |
| 787 | // rearranged via JavaScript between page load and form submission, so we |
| 788 | // copy over the |form_signature_field_names_| corresponding to the query |
| 789 | // request. |
| 790 | DCHECK_EQ(cached_form.form_name_, form_name_); |
| 791 | DCHECK_EQ(cached_form.source_url_, source_url_); |
| 792 | DCHECK_EQ(cached_form.target_url_, target_url_); |
| 793 | form_signature_field_names_ = cached_form.form_signature_field_names_; |
| 794 | } |
| 795 | |
| 796 | void FormStructure::LogQualityMetrics( |
| 797 | const AutofillMetrics& metric_logger, |
| 798 | const base::TimeTicks& load_time, |
| 799 | const base::TimeTicks& interaction_time, |
| 800 | const base::TimeTicks& submission_time) const { |
| 801 | std::string experiment_id = server_experiment_id(); |
| 802 | metric_logger.LogServerExperimentIdForUpload(experiment_id); |
| 803 | |
| 804 | size_t num_detected_field_types = 0; |
| 805 | bool did_autofill_all_possible_fields = true; |
| 806 | bool did_autofill_some_possible_fields = false; |
| 807 | for (size_t i = 0; i < field_count(); ++i) { |
| 808 | const AutofillField* field = this->field(i); |
| 809 | metric_logger.LogQualityMetric(AutofillMetrics::FIELD_SUBMITTED, |
| 810 | experiment_id); |
| 811 | |
| 812 | // No further logging for empty fields nor for fields where the entered data |
| 813 | // does not appear to already exist in the user's stored Autofill data. |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 814 | const ServerFieldTypeSet& field_types = field->possible_types(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 815 | DCHECK(!field_types.empty()); |
| 816 | if (field_types.count(EMPTY_TYPE) || field_types.count(UNKNOWN_TYPE)) |
| 817 | continue; |
| 818 | |
| 819 | ++num_detected_field_types; |
| 820 | if (field->is_autofilled) |
| 821 | did_autofill_some_possible_fields = true; |
| 822 | else |
| 823 | did_autofill_all_possible_fields = false; |
| 824 | |
| 825 | // Collapse field types that Chrome treats as identical, e.g. home and |
| 826 | // billing address fields. |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 827 | ServerFieldTypeSet collapsed_field_types; |
| 828 | for (ServerFieldTypeSet::const_iterator it = field_types.begin(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 829 | it != field_types.end(); |
| 830 | ++it) { |
| 831 | // Since we currently only support US phone numbers, the (city code + main |
| 832 | // digits) number is almost always identical to the whole phone number. |
| 833 | // TODO(isherman): Improve this logic once we add support for |
| 834 | // international numbers. |
| 835 | if (*it == PHONE_HOME_CITY_AND_NUMBER) |
| 836 | collapsed_field_types.insert(PHONE_HOME_WHOLE_NUMBER); |
| 837 | else |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 838 | collapsed_field_types.insert(AutofillType(*it).GetStorableType()); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 839 | } |
| 840 | |
| 841 | // Capture the field's type, if it is unambiguous. |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 842 | ServerFieldType field_type = UNKNOWN_TYPE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 843 | if (collapsed_field_types.size() == 1) |
| 844 | field_type = *collapsed_field_types.begin(); |
| 845 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 846 | ServerFieldType heuristic_type = |
| 847 | AutofillType(field->heuristic_type()).GetStorableType(); |
| 848 | ServerFieldType server_type = |
| 849 | AutofillType(field->server_type()).GetStorableType(); |
| 850 | ServerFieldType predicted_type = field->Type().GetStorableType(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 851 | |
| 852 | // Log heuristic, server, and overall type quality metrics, independently of |
| 853 | // whether the field was autofilled. |
| 854 | if (heuristic_type == UNKNOWN_TYPE) { |
| 855 | metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_UNKNOWN, |
| 856 | field_type, experiment_id); |
| 857 | } else if (field_types.count(heuristic_type)) { |
| 858 | metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_MATCH, |
| 859 | field_type, experiment_id); |
| 860 | } else { |
| 861 | metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_MISMATCH, |
| 862 | field_type, experiment_id); |
| 863 | } |
| 864 | |
| 865 | if (server_type == NO_SERVER_DATA) { |
| 866 | metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_UNKNOWN, |
| 867 | field_type, experiment_id); |
| 868 | } else if (field_types.count(server_type)) { |
| 869 | metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_MATCH, |
| 870 | field_type, experiment_id); |
| 871 | } else { |
| 872 | metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_MISMATCH, |
| 873 | field_type, experiment_id); |
| 874 | } |
| 875 | |
| 876 | if (predicted_type == UNKNOWN_TYPE) { |
| 877 | metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_UNKNOWN, |
| 878 | field_type, experiment_id); |
| 879 | } else if (field_types.count(predicted_type)) { |
| 880 | metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_MATCH, |
| 881 | field_type, experiment_id); |
| 882 | } else { |
| 883 | metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_MISMATCH, |
| 884 | field_type, experiment_id); |
| 885 | } |
| 886 | |
| 887 | // TODO(isherman): <select> fields don't support |is_autofilled()|, so we |
| 888 | // have to skip them for the remaining metrics. |
| 889 | if (field->form_control_type == "select-one") |
| 890 | continue; |
| 891 | |
| 892 | if (field->is_autofilled) { |
| 893 | metric_logger.LogQualityMetric(AutofillMetrics::FIELD_AUTOFILLED, |
| 894 | experiment_id); |
| 895 | } else { |
| 896 | metric_logger.LogQualityMetric(AutofillMetrics::FIELD_NOT_AUTOFILLED, |
| 897 | experiment_id); |
| 898 | |
| 899 | if (heuristic_type == UNKNOWN_TYPE) { |
| 900 | metric_logger.LogQualityMetric( |
| 901 | AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_UNKNOWN, |
| 902 | experiment_id); |
| 903 | } else if (field_types.count(heuristic_type)) { |
| 904 | metric_logger.LogQualityMetric( |
| 905 | AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MATCH, |
| 906 | experiment_id); |
| 907 | } else { |
| 908 | metric_logger.LogQualityMetric( |
| 909 | AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MISMATCH, |
| 910 | experiment_id); |
| 911 | } |
| 912 | |
| 913 | if (server_type == NO_SERVER_DATA) { |
| 914 | metric_logger.LogQualityMetric( |
| 915 | AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_UNKNOWN, |
| 916 | experiment_id); |
| 917 | } else if (field_types.count(server_type)) { |
| 918 | metric_logger.LogQualityMetric( |
| 919 | AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MATCH, |
| 920 | experiment_id); |
| 921 | } else { |
| 922 | metric_logger.LogQualityMetric( |
| 923 | AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MISMATCH, |
| 924 | experiment_id); |
| 925 | } |
| 926 | } |
| 927 | } |
| 928 | |
| 929 | if (num_detected_field_types < RequiredFillableFields()) { |
| 930 | metric_logger.LogUserHappinessMetric( |
| 931 | AutofillMetrics::SUBMITTED_NON_FILLABLE_FORM); |
| 932 | } else { |
| 933 | if (did_autofill_all_possible_fields) { |
| 934 | metric_logger.LogUserHappinessMetric( |
| 935 | AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_ALL); |
| 936 | } else if (did_autofill_some_possible_fields) { |
| 937 | metric_logger.LogUserHappinessMetric( |
| 938 | AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_SOME); |
| 939 | } else { |
| 940 | metric_logger.LogUserHappinessMetric( |
| 941 | AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_NONE); |
| 942 | } |
| 943 | |
| 944 | // Unlike the other times, the |submission_time| should always be available. |
| 945 | DCHECK(!submission_time.is_null()); |
| 946 | |
| 947 | // The |load_time| might be unset, in the case that the form was dynamically |
| 948 | // added to the DOM. |
| 949 | if (!load_time.is_null()) { |
| 950 | // Submission should always chronologically follow form load. |
| 951 | DCHECK(submission_time > load_time); |
| 952 | base::TimeDelta elapsed = submission_time - load_time; |
| 953 | if (did_autofill_some_possible_fields) |
| 954 | metric_logger.LogFormFillDurationFromLoadWithAutofill(elapsed); |
| 955 | else |
| 956 | metric_logger.LogFormFillDurationFromLoadWithoutAutofill(elapsed); |
| 957 | } |
| 958 | |
| 959 | // The |interaction_time| might be unset, in the case that the user |
| 960 | // submitted a blank form. |
| 961 | if (!interaction_time.is_null()) { |
| 962 | // Submission should always chronologically follow interaction. |
| 963 | DCHECK(submission_time > interaction_time); |
| 964 | base::TimeDelta elapsed = submission_time - interaction_time; |
| 965 | if (did_autofill_some_possible_fields) { |
| 966 | metric_logger.LogFormFillDurationFromInteractionWithAutofill(elapsed); |
| 967 | } else { |
| 968 | metric_logger.LogFormFillDurationFromInteractionWithoutAutofill( |
| 969 | elapsed); |
| 970 | } |
| 971 | } |
| 972 | } |
| 973 | } |
| 974 | |
| 975 | const AutofillField* FormStructure::field(size_t index) const { |
| 976 | if (index >= fields_.size()) { |
| 977 | NOTREACHED(); |
| 978 | return NULL; |
| 979 | } |
| 980 | |
| 981 | return fields_[index]; |
| 982 | } |
| 983 | |
| 984 | AutofillField* FormStructure::field(size_t index) { |
| 985 | return const_cast<AutofillField*>( |
| 986 | static_cast<const FormStructure*>(this)->field(index)); |
| 987 | } |
| 988 | |
| 989 | size_t FormStructure::field_count() const { |
| 990 | return fields_.size(); |
| 991 | } |
| 992 | |
| 993 | size_t FormStructure::active_field_count() const { |
| 994 | return active_field_count_; |
| 995 | } |
| 996 | |
| 997 | std::string FormStructure::server_experiment_id() const { |
| 998 | return server_experiment_id_; |
| 999 | } |
| 1000 | |
| 1001 | FormData FormStructure::ToFormData() const { |
| 1002 | // |data.user_submitted| will always be false. |
| 1003 | FormData data; |
| 1004 | data.name = form_name_; |
| 1005 | data.origin = source_url_; |
| 1006 | data.action = target_url_; |
| 1007 | data.method = ASCIIToUTF16(method_ == POST ? "POST" : "GET"); |
| 1008 | |
| 1009 | for (size_t i = 0; i < fields_.size(); ++i) { |
| 1010 | data.fields.push_back(FormFieldData(*fields_[i])); |
| 1011 | } |
| 1012 | |
| 1013 | return data; |
| 1014 | } |
| 1015 | |
| 1016 | bool FormStructure::operator==(const FormData& form) const { |
| 1017 | // TODO(jhawkins): Is this enough to differentiate a form? |
| 1018 | if (form_name_ == form.name && |
| 1019 | source_url_ == form.origin && |
| 1020 | target_url_ == form.action) { |
| 1021 | return true; |
| 1022 | } |
| 1023 | |
| 1024 | // TODO(jhawkins): Compare field names, IDs and labels once we have labels |
| 1025 | // set up. |
| 1026 | |
| 1027 | return false; |
| 1028 | } |
| 1029 | |
| 1030 | bool FormStructure::operator!=(const FormData& form) const { |
| 1031 | return !operator==(form); |
| 1032 | } |
| 1033 | |
| 1034 | std::string FormStructure::Hash64Bit(const std::string& str) { |
| 1035 | std::string hash_bin = base::SHA1HashString(str); |
| 1036 | DCHECK_EQ(20U, hash_bin.length()); |
| 1037 | |
| 1038 | uint64 hash64 = (((static_cast<uint64>(hash_bin[0])) & 0xFF) << 56) | |
| 1039 | (((static_cast<uint64>(hash_bin[1])) & 0xFF) << 48) | |
| 1040 | (((static_cast<uint64>(hash_bin[2])) & 0xFF) << 40) | |
| 1041 | (((static_cast<uint64>(hash_bin[3])) & 0xFF) << 32) | |
| 1042 | (((static_cast<uint64>(hash_bin[4])) & 0xFF) << 24) | |
| 1043 | (((static_cast<uint64>(hash_bin[5])) & 0xFF) << 16) | |
| 1044 | (((static_cast<uint64>(hash_bin[6])) & 0xFF) << 8) | |
| 1045 | ((static_cast<uint64>(hash_bin[7])) & 0xFF); |
| 1046 | |
| 1047 | return base::Uint64ToString(hash64); |
| 1048 | } |
| 1049 | |
| 1050 | bool FormStructure::EncodeFormRequest( |
| 1051 | FormStructure::EncodeRequestType request_type, |
| 1052 | buzz::XmlElement* encompassing_xml_element) const { |
| 1053 | if (!field_count()) // Nothing to add. |
| 1054 | return false; |
| 1055 | |
| 1056 | // Some badly formatted web sites repeat fields - limit number of fields to |
| 1057 | // 48, which is far larger than any valid form and XML still fits into 2K. |
| 1058 | // Do not send requests for forms with more than this many fields, as they are |
| 1059 | // near certainly not valid/auto-fillable. |
| 1060 | const size_t kMaxFieldsOnTheForm = 48; |
| 1061 | if (field_count() > kMaxFieldsOnTheForm) |
| 1062 | return false; |
| 1063 | |
| 1064 | // Add the child nodes for the form fields. |
| 1065 | for (size_t index = 0; index < field_count(); ++index) { |
| 1066 | const AutofillField* field = fields_[index]; |
| 1067 | switch (request_type) { |
| 1068 | case FormStructure::UPLOAD: |
| 1069 | EncodeFieldForUpload(*field, encompassing_xml_element); |
| 1070 | break; |
| 1071 | case FormStructure::QUERY: |
| 1072 | if (ShouldSkipField(*field)) |
| 1073 | continue; |
| 1074 | EncodeFieldForQuery(*field, encompassing_xml_element); |
| 1075 | break; |
| 1076 | case FormStructure::FIELD_ASSIGNMENTS: |
| 1077 | EncodeFieldForFieldAssignments(*field, encompassing_xml_element); |
| 1078 | break; |
| 1079 | } |
| 1080 | } |
| 1081 | return true; |
| 1082 | } |
| 1083 | |
| 1084 | void FormStructure::ParseFieldTypesFromAutocompleteAttributes( |
| 1085 | bool* found_types, |
| 1086 | bool* found_sections) { |
| 1087 | const std::string kDefaultSection = "-default"; |
| 1088 | |
| 1089 | *found_types = false; |
| 1090 | *found_sections = false; |
| 1091 | for (std::vector<AutofillField*>::iterator it = fields_.begin(); |
| 1092 | it != fields_.end(); ++it) { |
| 1093 | AutofillField* field = *it; |
| 1094 | |
| 1095 | // To prevent potential section name collisions, add a default suffix for |
| 1096 | // other fields. Without this, 'autocomplete' attribute values |
| 1097 | // "section--shipping street-address" and "shipping street-address" would be |
| 1098 | // parsed identically, given the section handling code below. We do this |
| 1099 | // before any validation so that fields with invalid attributes still end up |
| 1100 | // in the default section. These default section names will be overridden |
| 1101 | // by subsequent heuristic parsing steps if there are no author-specified |
| 1102 | // section names. |
| 1103 | field->set_section(kDefaultSection); |
| 1104 | |
| 1105 | // Canonicalize the attribute value by trimming whitespace, collapsing |
| 1106 | // non-space characters (e.g. tab) to spaces, and converting to lowercase. |
| 1107 | std::string autocomplete_attribute = |
| 1108 | CollapseWhitespaceASCII(field->autocomplete_attribute, false); |
| 1109 | autocomplete_attribute = StringToLowerASCII(autocomplete_attribute); |
| 1110 | |
| 1111 | // The autocomplete attribute is overloaded: it can specify either a field |
| 1112 | // type hint or whether autocomplete should be enabled at all. Ignore the |
| 1113 | // latter type of attribute value. |
| 1114 | if (autocomplete_attribute.empty() || |
| 1115 | autocomplete_attribute == "on" || |
| 1116 | autocomplete_attribute == "off") { |
| 1117 | continue; |
| 1118 | } |
| 1119 | |
| 1120 | // Any other value, even it is invalid, is considered to be a type hint. |
| 1121 | // This allows a website's author to specify an attribute like |
| 1122 | // autocomplete="other" on a field to disable all Autofill heuristics for |
| 1123 | // the form. |
| 1124 | *found_types = true; |
| 1125 | |
| 1126 | // Tokenize the attribute value. Per the spec, the tokens are parsed in |
| 1127 | // reverse order. |
| 1128 | std::vector<std::string> tokens; |
| 1129 | Tokenize(autocomplete_attribute, " ", &tokens); |
| 1130 | |
| 1131 | // The final token must be the field type. |
| 1132 | // If it is not one of the known types, abort. |
| 1133 | DCHECK(!tokens.empty()); |
| 1134 | std::string field_type_token = tokens.back(); |
| 1135 | tokens.pop_back(); |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 1136 | HtmlFieldType field_type = |
| 1137 | FieldTypeFromAutocompleteAttributeValue(field_type_token, *field); |
| 1138 | if (field_type == HTML_TYPE_UNKNOWN) |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1139 | continue; |
| 1140 | |
| 1141 | // The preceding token, if any, may be a type hint. |
| 1142 | if (!tokens.empty() && IsContactTypeHint(tokens.back())) { |
| 1143 | // If it is, it must match the field type; otherwise, abort. |
| 1144 | // Note that an invalid token invalidates the entire attribute value, even |
| 1145 | // if the other tokens are valid. |
| 1146 | if (!ContactTypeHintMatchesFieldType(tokens.back(), field_type)) |
| 1147 | continue; |
| 1148 | |
| 1149 | // Chrome Autofill ignores these type hints. |
| 1150 | tokens.pop_back(); |
| 1151 | } |
| 1152 | |
| 1153 | // The preceding token, if any, may be a fixed string that is either |
| 1154 | // "shipping" or "billing". Chrome Autofill treats these as implicit |
| 1155 | // section name suffixes. |
| 1156 | DCHECK_EQ(kDefaultSection, field->section()); |
| 1157 | std::string section = field->section(); |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 1158 | HtmlFieldMode mode = HTML_MODE_NONE; |
| 1159 | if (!tokens.empty()) { |
| 1160 | if (tokens.back() == kShippingMode) |
| 1161 | mode = HTML_MODE_SHIPPING; |
| 1162 | else if (tokens.back() == kBillingMode) |
| 1163 | mode = HTML_MODE_BILLING; |
| 1164 | } |
Ben Murdoch | 7dbb3d5 | 2013-07-17 14:55:54 +0100 | [diff] [blame] | 1165 | |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 1166 | if (mode != HTML_MODE_NONE) { |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1167 | section = "-" + tokens.back(); |
| 1168 | tokens.pop_back(); |
| 1169 | } |
| 1170 | |
| 1171 | // The preceding token, if any, may be a named section. |
| 1172 | const std::string kSectionPrefix = "section-"; |
| 1173 | if (!tokens.empty() && |
| 1174 | StartsWithASCII(tokens.back(), kSectionPrefix, true)) { |
| 1175 | // Prepend this section name to the suffix set in the preceding block. |
| 1176 | section = tokens.back().substr(kSectionPrefix.size()) + section; |
| 1177 | tokens.pop_back(); |
| 1178 | } |
| 1179 | |
| 1180 | // No other tokens are allowed. If there are any remaining, abort. |
| 1181 | if (!tokens.empty()) |
| 1182 | continue; |
| 1183 | |
| 1184 | if (section != kDefaultSection) { |
| 1185 | *found_sections = true; |
| 1186 | field->set_section(section); |
| 1187 | } |
| 1188 | |
| 1189 | // No errors encountered while parsing! |
| 1190 | // Update the |field|'s type based on what was parsed from the attribute. |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 1191 | field->SetHtmlType(field_type, mode); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1192 | } |
| 1193 | } |
| 1194 | |
| 1195 | void FormStructure::IdentifySections(bool has_author_specified_sections) { |
| 1196 | if (fields_.empty()) |
| 1197 | return; |
| 1198 | |
| 1199 | if (!has_author_specified_sections) { |
| 1200 | // Name sections after the first field in the section. |
| 1201 | base::string16 current_section = fields_.front()->unique_name(); |
| 1202 | |
| 1203 | // Keep track of the types we've seen in this section. |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 1204 | std::set<ServerFieldType> seen_types; |
| 1205 | ServerFieldType previous_type = UNKNOWN_TYPE; |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1206 | |
| 1207 | for (std::vector<AutofillField*>::iterator field = fields_.begin(); |
| 1208 | field != fields_.end(); ++field) { |
Ben Murdoch | bb1529c | 2013-08-08 10:24:53 +0100 | [diff] [blame^] | 1209 | const ServerFieldType current_type = (*field)->Type().GetStorableType(); |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1210 | |
| 1211 | bool already_saw_current_type = seen_types.count(current_type) > 0; |
| 1212 | |
| 1213 | // Forms often ask for multiple phone numbers -- e.g. both a daytime and |
| 1214 | // evening phone number. Our phone number detection is also generally a |
| 1215 | // little off. Hence, ignore this field type as a signal here. |
Ben Murdoch | 2385ea3 | 2013-08-06 11:01:04 +0100 | [diff] [blame] | 1216 | if (AutofillType(current_type).group() == PHONE_HOME) |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1217 | already_saw_current_type = false; |
| 1218 | |
| 1219 | // Some forms have adjacent fields of the same type. Two common examples: |
| 1220 | // * Forms with two email fields, where the second is meant to "confirm" |
| 1221 | // the first. |
| 1222 | // * Forms with a <select> menu for states in some countries, and a |
| 1223 | // freeform <input> field for states in other countries. (Usually, |
| 1224 | // only one of these two will be visible for any given choice of |
| 1225 | // country.) |
| 1226 | // Generally, adjacent fields of the same type belong in the same logical |
| 1227 | // section. |
| 1228 | if (current_type == previous_type) |
| 1229 | already_saw_current_type = false; |
| 1230 | |
| 1231 | previous_type = current_type; |
| 1232 | |
| 1233 | if (current_type != UNKNOWN_TYPE && already_saw_current_type) { |
| 1234 | // We reached the end of a section, so start a new section. |
| 1235 | seen_types.clear(); |
| 1236 | current_section = (*field)->unique_name(); |
| 1237 | } |
| 1238 | |
| 1239 | seen_types.insert(current_type); |
| 1240 | (*field)->set_section(UTF16ToUTF8(current_section)); |
| 1241 | } |
| 1242 | } |
| 1243 | |
| 1244 | // Ensure that credit card and address fields are in separate sections. |
| 1245 | // This simplifies the section-aware logic in autofill_manager.cc. |
| 1246 | for (std::vector<AutofillField*>::iterator field = fields_.begin(); |
| 1247 | field != fields_.end(); ++field) { |
Ben Murdoch | 3240926 | 2013-08-07 11:04:47 +0100 | [diff] [blame] | 1248 | FieldTypeGroup field_type_group = (*field)->Type().group(); |
Ben Murdoch | 2385ea3 | 2013-08-06 11:01:04 +0100 | [diff] [blame] | 1249 | if (field_type_group == CREDIT_CARD) |
Ben Murdoch | eb525c5 | 2013-07-10 11:40:50 +0100 | [diff] [blame] | 1250 | (*field)->set_section((*field)->section() + "-cc"); |
| 1251 | else |
| 1252 | (*field)->set_section((*field)->section() + "-default"); |
| 1253 | } |
| 1254 | } |
| 1255 | |
| 1256 | } // namespace autofill |