| OLD | NEW |
|---|
| (Empty) |
| 1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. | |
| 2 // Use of this source code is governed by a BSD-style license that can be | |
| 3 // found in the LICENSE file. | |
| 4 | |
| 5 #include "chrome/browser/autofill/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/string_util.h" | |
| 15 #include "base/stringprintf.h" | |
| 16 #include "base/strings/string_number_conversions.h" | |
| 17 #include "base/time.h" | |
| 18 #include "base/utf_string_conversions.h" | |
| 19 #include "chrome/browser/autofill/autocheckout_page_meta_data.h" | |
| 20 #include "chrome/browser/autofill/autofill_metrics.h" | |
| 21 #include "chrome/browser/autofill/autofill_type.h" | |
| 22 #include "chrome/browser/autofill/autofill_xml_parser.h" | |
| 23 #include "chrome/browser/autofill/field_types.h" | |
| 24 #include "chrome/browser/autofill/form_field.h" | |
| 25 #include "components/autofill/common/form_data.h" | |
| 26 #include "components/autofill/common/form_data_predictions.h" | |
| 27 #include "components/autofill/common/form_field_data.h" | |
| 28 #include "components/autofill/common/form_field_data_predictions.h" | |
| 29 #include "third_party/libjingle/source/talk/xmllite/xmlelement.h" | |
| 30 | |
| 31 namespace { | |
| 32 | |
| 33 const char kFormMethodPost[] = "post"; | |
| 34 | |
| 35 // XML elements and attributes. | |
| 36 const char kAttributeAcceptedFeatures[] = "accepts"; | |
| 37 const char kAttributeAutofillUsed[] = "autofillused"; | |
| 38 const char kAttributeAutofillType[] = "autofilltype"; | |
| 39 const char kAttributeClientVersion[] = "clientversion"; | |
| 40 const char kAttributeDataPresent[] = "datapresent"; | |
| 41 const char kAttributeFormSignature[] = "formsignature"; | |
| 42 const char kAttributeSignature[] = "signature"; | |
| 43 const char kAttributeUrlprefixSignature[] = "urlprefixsignature"; | |
| 44 const char kAcceptedFeaturesExperiment[] = "e"; // e=experiments | |
| 45 const char kAcceptedFeaturesAutocheckoutExperiment[] = "a,e"; // a=autocheckout | |
| 46 const char kClientVersion[] = "6.1.1715.1442/en (GGLL)"; | |
| 47 const char kXMLDeclaration[] = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"; | |
| 48 const char kXMLElementAutofillQuery[] = "autofillquery"; | |
| 49 const char kXMLElementAutofillUpload[] = "autofillupload"; | |
| 50 const char kXMLElementForm[] = "form"; | |
| 51 const char kXMLElementField[] = "field"; | |
| 52 | |
| 53 // The number of fillable fields necessary for a form to be fillable. | |
| 54 const size_t kRequiredFillableFields = 3; | |
| 55 | |
| 56 // Helper for |EncodeUploadRequest()| that creates a bit field corresponding to | |
| 57 // |available_field_types| and returns the hex representation as a string. | |
| 58 std::string EncodeFieldTypes(const FieldTypeSet& available_field_types) { | |
| 59 // There are |MAX_VALID_FIELD_TYPE| different field types and 8 bits per byte, | |
| 60 // so we need ceil(MAX_VALID_FIELD_TYPE / 8) bytes to encode the bit field. | |
| 61 const size_t kNumBytes = (MAX_VALID_FIELD_TYPE + 0x7) / 8; | |
| 62 | |
| 63 // Pack the types in |available_field_types| into |bit_field|. | |
| 64 std::vector<uint8> bit_field(kNumBytes, 0); | |
| 65 for (FieldTypeSet::const_iterator field_type = available_field_types.begin(); | |
| 66 field_type != available_field_types.end(); | |
| 67 ++field_type) { | |
| 68 // Set the appropriate bit in the field. The bit we set is the one | |
| 69 // |field_type| % 8 from the left of the byte. | |
| 70 const size_t byte = *field_type / 8; | |
| 71 const size_t bit = 0x80 >> (*field_type % 8); | |
| 72 DCHECK(byte < bit_field.size()); | |
| 73 bit_field[byte] |= bit; | |
| 74 } | |
| 75 | |
| 76 // Discard any trailing zeroes. | |
| 77 // If there are no available types, we return the empty string. | |
| 78 size_t data_end = bit_field.size(); | |
| 79 for (; data_end > 0 && !bit_field[data_end - 1]; --data_end) { | |
| 80 } | |
| 81 | |
| 82 // Print all meaningfull bytes into a string. | |
| 83 std::string data_presence; | |
| 84 data_presence.reserve(data_end * 2 + 1); | |
| 85 for (size_t i = 0; i < data_end; ++i) { | |
| 86 base::StringAppendF(&data_presence, "%02x", bit_field[i]); | |
| 87 } | |
| 88 | |
| 89 return data_presence; | |
| 90 } | |
| 91 | |
| 92 // Returns |true| iff the |token| is a type hint for a contact field, as | |
| 93 // specified in the implementation section of http://is.gd/whatwg_autocomplete | |
| 94 // Note that "fax" and "pager" are intentionally ignored, as Chrome does not | |
| 95 // support filling either type of information. | |
| 96 bool IsContactTypeHint(const std::string& token) { | |
| 97 return token == "home" || token == "work" || token == "mobile"; | |
| 98 } | |
| 99 | |
| 100 // Returns |true| iff the |token| is a type hint appropriate for a field of the | |
| 101 // given |field_type|, as specified in the implementation section of | |
| 102 // http://is.gd/whatwg_autocomplete | |
| 103 bool ContactTypeHintMatchesFieldType(const std::string& token, | |
| 104 AutofillFieldType field_type) { | |
| 105 // The "home" and "work" type hints are only appropriate for email and phone | |
| 106 // number field types. | |
| 107 if (token == "home" || token == "work") { | |
| 108 return field_type == EMAIL_ADDRESS || | |
| 109 (field_type >= PHONE_HOME_NUMBER && | |
| 110 field_type <= PHONE_HOME_WHOLE_NUMBER); | |
| 111 } | |
| 112 | |
| 113 // The "mobile" type hint is only appropriate for phone number field types. | |
| 114 // Note that "fax" and "pager" are intentionally ignored, as Chrome does not | |
| 115 // support filling either type of information. | |
| 116 if (token == "mobile") { | |
| 117 return field_type >= PHONE_HOME_NUMBER && | |
| 118 field_type <= PHONE_HOME_WHOLE_NUMBER; | |
| 119 } | |
| 120 | |
| 121 return false; | |
| 122 } | |
| 123 | |
| 124 // Returns the Chrome Autofill-supported field type corresponding to the given | |
| 125 // |autocomplete_type|, if there is one, in the context of the given |field|. | |
| 126 // Chrome Autofill supports a subset of the field types listed at | |
| 127 // http://is.gd/whatwg_autocomplete | |
| 128 AutofillFieldType FieldTypeFromAutocompleteType( | |
| 129 const std::string& autocomplete_type, | |
| 130 const AutofillField& field) { | |
| 131 if (autocomplete_type == "name") | |
| 132 return NAME_FULL; | |
| 133 | |
| 134 if (autocomplete_type == "given-name") | |
| 135 return NAME_FIRST; | |
| 136 | |
| 137 if (autocomplete_type == "additional-name") { | |
| 138 if (field.max_length == 1) | |
| 139 return NAME_MIDDLE_INITIAL; | |
| 140 else | |
| 141 return NAME_MIDDLE; | |
| 142 } | |
| 143 | |
| 144 if (autocomplete_type == "family-name") | |
| 145 return NAME_LAST; | |
| 146 | |
| 147 if (autocomplete_type == "honorific-suffix") | |
| 148 return NAME_SUFFIX; | |
| 149 | |
| 150 if (autocomplete_type == "organization") | |
| 151 return COMPANY_NAME; | |
| 152 | |
| 153 if (autocomplete_type == "street-address" || | |
| 154 autocomplete_type == "address-line1") | |
| 155 return ADDRESS_HOME_LINE1; | |
| 156 | |
| 157 if (autocomplete_type == "address-line2") | |
| 158 return ADDRESS_HOME_LINE2; | |
| 159 | |
| 160 if (autocomplete_type == "locality") | |
| 161 return ADDRESS_HOME_CITY; | |
| 162 | |
| 163 if (autocomplete_type == "region") | |
| 164 return ADDRESS_HOME_STATE; | |
| 165 | |
| 166 if (autocomplete_type == "country") | |
| 167 return ADDRESS_HOME_COUNTRY; | |
| 168 | |
| 169 if (autocomplete_type == "postal-code") | |
| 170 return ADDRESS_HOME_ZIP; | |
| 171 | |
| 172 if (autocomplete_type == "cc-name") | |
| 173 return CREDIT_CARD_NAME; | |
| 174 | |
| 175 if (autocomplete_type == "cc-number") | |
| 176 return CREDIT_CARD_NUMBER; | |
| 177 | |
| 178 if (autocomplete_type == "cc-exp") { | |
| 179 if (field.max_length == 5) | |
| 180 return CREDIT_CARD_EXP_DATE_2_DIGIT_YEAR; | |
| 181 else | |
| 182 return CREDIT_CARD_EXP_DATE_4_DIGIT_YEAR; | |
| 183 } | |
| 184 | |
| 185 if (autocomplete_type == "cc-exp-month") | |
| 186 return CREDIT_CARD_EXP_MONTH; | |
| 187 | |
| 188 if (autocomplete_type == "cc-exp-year") { | |
| 189 if (field.max_length == 2) | |
| 190 return CREDIT_CARD_EXP_2_DIGIT_YEAR; | |
| 191 else | |
| 192 return CREDIT_CARD_EXP_4_DIGIT_YEAR; | |
| 193 } | |
| 194 | |
| 195 if (autocomplete_type == "cc-csc") | |
| 196 return CREDIT_CARD_VERIFICATION_CODE; | |
| 197 | |
| 198 if (autocomplete_type == "cc-type") | |
| 199 return CREDIT_CARD_TYPE; | |
| 200 | |
| 201 if (autocomplete_type == "tel") | |
| 202 return PHONE_HOME_WHOLE_NUMBER; | |
| 203 | |
| 204 if (autocomplete_type == "tel-country-code") | |
| 205 return PHONE_HOME_COUNTRY_CODE; | |
| 206 | |
| 207 if (autocomplete_type == "tel-national") | |
| 208 return PHONE_HOME_CITY_AND_NUMBER; | |
| 209 | |
| 210 if (autocomplete_type == "tel-area-code") | |
| 211 return PHONE_HOME_CITY_CODE; | |
| 212 | |
| 213 if (autocomplete_type == "tel-local") | |
| 214 return PHONE_HOME_NUMBER; | |
| 215 | |
| 216 if (autocomplete_type == "tel-local-prefix") | |
| 217 return PHONE_HOME_NUMBER; | |
| 218 | |
| 219 if (autocomplete_type == "tel-local-suffix") | |
| 220 return PHONE_HOME_NUMBER; | |
| 221 | |
| 222 if (autocomplete_type == "email") | |
| 223 return EMAIL_ADDRESS; | |
| 224 | |
| 225 return UNKNOWN_TYPE; | |
| 226 } | |
| 227 | |
| 228 } // namespace | |
| 229 | |
| 230 FormStructure::FormStructure(const FormData& form, | |
| 231 const std::string& autocheckout_url_prefix) | |
| 232 : form_name_(form.name), | |
| 233 source_url_(form.origin), | |
| 234 target_url_(form.action), | |
| 235 autofill_count_(0), | |
| 236 checkable_field_count_(0), | |
| 237 upload_required_(USE_UPLOAD_RATES), | |
| 238 server_experiment_id_("no server response"), | |
| 239 has_author_specified_types_(false), | |
| 240 autocheckout_url_prefix_(autocheckout_url_prefix) { | |
| 241 // Copy the form fields. | |
| 242 std::map<string16, size_t> unique_names; | |
| 243 for (std::vector<FormFieldData>::const_iterator field = | |
| 244 form.fields.begin(); | |
| 245 field != form.fields.end(); field++) { | |
| 246 | |
| 247 // Skipping checkable elements when Autocheckout is not enabled, else | |
| 248 // these fields will interfere with existing field signatures with Autofill | |
| 249 // servers. | |
| 250 if (!field->is_checkable || IsAutocheckoutEnabled()) { | |
| 251 // Add all supported form fields (including with empty names) to the | |
| 252 // signature. This is a requirement for Autofill servers. | |
| 253 form_signature_field_names_.append("&"); | |
| 254 form_signature_field_names_.append(UTF16ToUTF8(field->name)); | |
| 255 } | |
| 256 | |
| 257 // Generate a unique name for this field by appending a counter to the name. | |
| 258 // Make sure to prepend the counter with a non-numeric digit so that we are | |
| 259 // guaranteed to avoid collisions. | |
| 260 if (!unique_names.count(field->name)) | |
| 261 unique_names[field->name] = 1; | |
| 262 else | |
| 263 ++unique_names[field->name]; | |
| 264 string16 unique_name = field->name + ASCIIToUTF16("_") + | |
| 265 base::IntToString16(unique_names[field->name]); | |
| 266 fields_.push_back(new AutofillField(*field, unique_name)); | |
| 267 | |
| 268 if (field->is_checkable) | |
| 269 ++checkable_field_count_; | |
| 270 } | |
| 271 | |
| 272 std::string method = UTF16ToUTF8(form.method); | |
| 273 if (StringToLowerASCII(method) == kFormMethodPost) { | |
| 274 method_ = POST; | |
| 275 } else { | |
| 276 // Either the method is 'get', or we don't know. In this case we default | |
| 277 // to GET. | |
| 278 method_ = GET; | |
| 279 } | |
| 280 } | |
| 281 | |
| 282 FormStructure::~FormStructure() {} | |
| 283 | |
| 284 void FormStructure::DetermineHeuristicTypes( | |
| 285 const AutofillMetrics& metric_logger) { | |
| 286 // First, try to detect field types based on each field's |autocomplete| | |
| 287 // attribute value. If there is at least one form field that specifies an | |
| 288 // autocomplete type hint, don't try to apply other heuristics to match fields | |
| 289 // in this form. | |
| 290 bool has_author_specified_sections; | |
| 291 ParseFieldTypesFromAutocompleteAttributes(&has_author_specified_types_, | |
| 292 &has_author_specified_sections); | |
| 293 | |
| 294 if (!has_author_specified_types_) { | |
| 295 FieldTypeMap field_type_map; | |
| 296 FormField::ParseFormFields(fields_.get(), &field_type_map); | |
| 297 for (size_t index = 0; index < field_count(); index++) { | |
| 298 AutofillField* field = fields_[index]; | |
| 299 FieldTypeMap::iterator iter = field_type_map.find(field->unique_name()); | |
| 300 if (iter != field_type_map.end()) | |
| 301 field->set_heuristic_type(iter->second); | |
| 302 } | |
| 303 } | |
| 304 | |
| 305 UpdateAutofillCount(); | |
| 306 IdentifySections(has_author_specified_sections); | |
| 307 | |
| 308 if (IsAutofillable(true)) { | |
| 309 metric_logger.LogDeveloperEngagementMetric( | |
| 310 AutofillMetrics::FILLABLE_FORM_PARSED); | |
| 311 if (has_author_specified_types_) { | |
| 312 metric_logger.LogDeveloperEngagementMetric( | |
| 313 AutofillMetrics::FILLABLE_FORM_CONTAINS_TYPE_HINTS); | |
| 314 } | |
| 315 } | |
| 316 } | |
| 317 | |
| 318 bool FormStructure::EncodeUploadRequest( | |
| 319 const FieldTypeSet& available_field_types, | |
| 320 bool form_was_autofilled, | |
| 321 std::string* encoded_xml) const { | |
| 322 if (!ShouldBeCrowdsourced()) { | |
| 323 NOTREACHED(); | |
| 324 return false; | |
| 325 } | |
| 326 | |
| 327 // Verify that |available_field_types| agrees with the possible field types we | |
| 328 // are uploading. | |
| 329 for (std::vector<AutofillField*>::const_iterator field = begin(); | |
| 330 field != end(); | |
| 331 ++field) { | |
| 332 for (FieldTypeSet::const_iterator type = (*field)->possible_types().begin(); | |
| 333 type != (*field)->possible_types().end(); | |
| 334 ++type) { | |
| 335 DCHECK(*type == UNKNOWN_TYPE || | |
| 336 *type == EMPTY_TYPE || | |
| 337 available_field_types.count(*type)); | |
| 338 } | |
| 339 } | |
| 340 | |
| 341 // Set up the <autofillupload> element and its attributes. | |
| 342 buzz::XmlElement autofill_request_xml( | |
| 343 (buzz::QName(kXMLElementAutofillUpload))); | |
| 344 autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion), | |
| 345 kClientVersion); | |
| 346 autofill_request_xml.SetAttr(buzz::QName(kAttributeFormSignature), | |
| 347 FormSignature()); | |
| 348 autofill_request_xml.SetAttr(buzz::QName(kAttributeAutofillUsed), | |
| 349 form_was_autofilled ? "true" : "false"); | |
| 350 autofill_request_xml.SetAttr(buzz::QName(kAttributeDataPresent), | |
| 351 EncodeFieldTypes(available_field_types).c_str()); | |
| 352 | |
| 353 if (!EncodeFormRequest(FormStructure::UPLOAD, &autofill_request_xml)) | |
| 354 return false; // Malformed form, skip it. | |
| 355 | |
| 356 // Obtain the XML structure as a string. | |
| 357 encoded_xml->clear(); | |
| 358 *encoded_xml = kXMLDeclaration; | |
| 359 *encoded_xml += autofill_request_xml.Str().c_str(); | |
| 360 | |
| 361 // To enable this logging, run with the flag --vmodule="form_structure=2". | |
| 362 VLOG(2) << "\n" << *encoded_xml; | |
| 363 | |
| 364 return true; | |
| 365 } | |
| 366 | |
| 367 // static | |
| 368 bool FormStructure::EncodeQueryRequest( | |
| 369 const std::vector<FormStructure*>& forms, | |
| 370 std::vector<std::string>* encoded_signatures, | |
| 371 std::string* encoded_xml) { | |
| 372 DCHECK(encoded_signatures); | |
| 373 DCHECK(encoded_xml); | |
| 374 encoded_xml->clear(); | |
| 375 encoded_signatures->clear(); | |
| 376 encoded_signatures->reserve(forms.size()); | |
| 377 | |
| 378 // Set up the <autofillquery> element and attributes. | |
| 379 buzz::XmlElement autofill_request_xml( | |
| 380 (buzz::QName(kXMLElementAutofillQuery))); | |
| 381 autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion), | |
| 382 kClientVersion); | |
| 383 | |
| 384 // autocheckout_url_prefix tells the Autofill server where the forms in the | |
| 385 // request came from, and the the Autofill server checks internal status and | |
| 386 // decide to enable Autocheckout or not and may return Autocheckout related | |
| 387 // data in the response accordingly. | |
| 388 // There is no page/frame level object associated with FormStructure that | |
| 389 // we could extract URL prefix from. But, all the forms should come from the | |
| 390 // same frame, so they should have the same Autocheckout URL prefix. Thus we | |
| 391 // use URL prefix from the first form with Autocheckout enabled. | |
| 392 std::string autocheckout_url_prefix; | |
| 393 | |
| 394 // Some badly formatted web sites repeat forms - detect that and encode only | |
| 395 // one form as returned data would be the same for all the repeated forms. | |
| 396 std::set<std::string> processed_forms; | |
| 397 for (ScopedVector<FormStructure>::const_iterator it = forms.begin(); | |
| 398 it != forms.end(); | |
| 399 ++it) { | |
| 400 std::string signature((*it)->FormSignature()); | |
| 401 if (processed_forms.find(signature) != processed_forms.end()) | |
| 402 continue; | |
| 403 processed_forms.insert(signature); | |
| 404 scoped_ptr<buzz::XmlElement> encompassing_xml_element( | |
| 405 new buzz::XmlElement(buzz::QName(kXMLElementForm))); | |
| 406 encompassing_xml_element->SetAttr(buzz::QName(kAttributeSignature), | |
| 407 signature); | |
| 408 | |
| 409 if (!(*it)->EncodeFormRequest(FormStructure::QUERY, | |
| 410 encompassing_xml_element.get())) | |
| 411 continue; // Malformed form, skip it. | |
| 412 | |
| 413 if ((*it)->IsAutocheckoutEnabled()) { | |
| 414 if (autocheckout_url_prefix.empty()) { | |
| 415 autocheckout_url_prefix = (*it)->autocheckout_url_prefix_; | |
| 416 } else { | |
| 417 // Making sure all the forms in the request has the same url_prefix. | |
| 418 DCHECK_EQ(autocheckout_url_prefix, (*it)->autocheckout_url_prefix_); | |
| 419 } | |
| 420 } | |
| 421 | |
| 422 autofill_request_xml.AddElement(encompassing_xml_element.release()); | |
| 423 encoded_signatures->push_back(signature); | |
| 424 } | |
| 425 | |
| 426 if (!encoded_signatures->size()) | |
| 427 return false; | |
| 428 | |
| 429 if (autocheckout_url_prefix.empty()) { | |
| 430 autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures), | |
| 431 kAcceptedFeaturesExperiment); | |
| 432 } else { | |
| 433 autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures), | |
| 434 kAcceptedFeaturesAutocheckoutExperiment); | |
| 435 autofill_request_xml.SetAttr(buzz::QName(kAttributeUrlprefixSignature), | |
| 436 Hash64Bit(autocheckout_url_prefix)); | |
| 437 } | |
| 438 | |
| 439 // Obtain the XML structure as a string. | |
| 440 *encoded_xml = kXMLDeclaration; | |
| 441 *encoded_xml += autofill_request_xml.Str().c_str(); | |
| 442 | |
| 443 return true; | |
| 444 } | |
| 445 | |
| 446 // static | |
| 447 void FormStructure::ParseQueryResponse( | |
| 448 const std::string& response_xml, | |
| 449 const std::vector<FormStructure*>& forms, | |
| 450 autofill::AutocheckoutPageMetaData* page_meta_data, | |
| 451 const AutofillMetrics& metric_logger) { | |
| 452 metric_logger.LogServerQueryMetric(AutofillMetrics::QUERY_RESPONSE_RECEIVED); | |
| 453 | |
| 454 // Parse the field types from the server response to the query. | |
| 455 std::vector<AutofillServerFieldInfo> field_infos; | |
| 456 UploadRequired upload_required; | |
| 457 std::string experiment_id; | |
| 458 AutofillQueryXmlParser parse_handler(&field_infos, &upload_required, | |
| 459 &experiment_id); | |
| 460 buzz::XmlParser parser(&parse_handler); | |
| 461 parser.Parse(response_xml.c_str(), response_xml.length(), true); | |
| 462 if (!parse_handler.succeeded()) | |
| 463 return; | |
| 464 | |
| 465 page_meta_data->current_page_number = parse_handler.current_page_number(); | |
| 466 page_meta_data->total_pages = parse_handler.total_pages(); | |
| 467 if (parse_handler.proceed_element_descriptor()) { | |
| 468 page_meta_data->proceed_element_descriptor.reset( | |
| 469 new autofill::WebElementDescriptor( | |
| 470 *parse_handler.proceed_element_descriptor())); | |
| 471 } else { | |
| 472 page_meta_data->proceed_element_descriptor.reset(); | |
| 473 } | |
| 474 | |
| 475 metric_logger.LogServerQueryMetric(AutofillMetrics::QUERY_RESPONSE_PARSED); | |
| 476 metric_logger.LogServerExperimentIdForQuery(experiment_id); | |
| 477 | |
| 478 bool heuristics_detected_fillable_field = false; | |
| 479 bool query_response_overrode_heuristics = false; | |
| 480 | |
| 481 // Copy the field types into the actual form. | |
| 482 std::vector<AutofillServerFieldInfo>::iterator current_info = | |
| 483 field_infos.begin(); | |
| 484 for (std::vector<FormStructure*>::const_iterator iter = forms.begin(); | |
| 485 iter != forms.end(); ++iter) { | |
| 486 FormStructure* form = *iter; | |
| 487 form->upload_required_ = upload_required; | |
| 488 form->server_experiment_id_ = experiment_id; | |
| 489 | |
| 490 for (std::vector<AutofillField*>::iterator field = form->fields_.begin(); | |
| 491 field != form->fields_.end(); ++field, ++current_info) { | |
| 492 // In some cases *successful* response does not return all the fields. | |
| 493 // Quit the update of the types then. | |
| 494 if (current_info == field_infos.end()) | |
| 495 break; | |
| 496 | |
| 497 // UNKNOWN_TYPE is reserved for use by the client. | |
| 498 DCHECK_NE(current_info->field_type, UNKNOWN_TYPE); | |
| 499 | |
| 500 AutofillFieldType heuristic_type = (*field)->type(); | |
| 501 if (heuristic_type != UNKNOWN_TYPE) | |
| 502 heuristics_detected_fillable_field = true; | |
| 503 | |
| 504 (*field)->set_server_type(current_info->field_type); | |
| 505 if (heuristic_type != (*field)->type()) | |
| 506 query_response_overrode_heuristics = true; | |
| 507 | |
| 508 // Copy default value into the field if available. | |
| 509 if (!current_info->default_value.empty()) | |
| 510 (*field)->set_default_value(current_info->default_value); | |
| 511 } | |
| 512 | |
| 513 form->UpdateAutofillCount(); | |
| 514 form->IdentifySections(false); | |
| 515 } | |
| 516 | |
| 517 AutofillMetrics::ServerQueryMetric metric; | |
| 518 if (query_response_overrode_heuristics) { | |
| 519 if (heuristics_detected_fillable_field) { | |
| 520 metric = AutofillMetrics::QUERY_RESPONSE_OVERRODE_LOCAL_HEURISTICS; | |
| 521 } else { | |
| 522 metric = AutofillMetrics::QUERY_RESPONSE_WITH_NO_LOCAL_HEURISTICS; | |
| 523 } | |
| 524 } else { | |
| 525 metric = AutofillMetrics::QUERY_RESPONSE_MATCHED_LOCAL_HEURISTICS; | |
| 526 } | |
| 527 metric_logger.LogServerQueryMetric(metric); | |
| 528 } | |
| 529 | |
| 530 // static | |
| 531 void FormStructure::GetFieldTypePredictions( | |
| 532 const std::vector<FormStructure*>& form_structures, | |
| 533 std::vector<FormDataPredictions>* forms) { | |
| 534 forms->clear(); | |
| 535 forms->reserve(form_structures.size()); | |
| 536 for (size_t i = 0; i < form_structures.size(); ++i) { | |
| 537 FormStructure* form_structure = form_structures[i]; | |
| 538 FormDataPredictions form; | |
| 539 form.data.name = form_structure->form_name_; | |
| 540 form.data.method = | |
| 541 ASCIIToUTF16((form_structure->method_ == POST) ? "POST" : "GET"); | |
| 542 form.data.origin = form_structure->source_url_; | |
| 543 form.data.action = form_structure->target_url_; | |
| 544 form.signature = form_structure->FormSignature(); | |
| 545 form.experiment_id = form_structure->server_experiment_id_; | |
| 546 | |
| 547 for (std::vector<AutofillField*>::const_iterator field = | |
| 548 form_structure->fields_.begin(); | |
| 549 field != form_structure->fields_.end(); ++field) { | |
| 550 form.data.fields.push_back(FormFieldData(**field)); | |
| 551 | |
| 552 FormFieldDataPredictions annotated_field; | |
| 553 annotated_field.signature = (*field)->FieldSignature(); | |
| 554 annotated_field.heuristic_type = | |
| 555 AutofillType::FieldTypeToString((*field)->heuristic_type()); | |
| 556 annotated_field.server_type = | |
| 557 AutofillType::FieldTypeToString((*field)->server_type()); | |
| 558 annotated_field.overall_type = | |
| 559 AutofillType::FieldTypeToString((*field)->type()); | |
| 560 form.fields.push_back(annotated_field); | |
| 561 } | |
| 562 | |
| 563 forms->push_back(form); | |
| 564 } | |
| 565 } | |
| 566 | |
| 567 std::string FormStructure::FormSignature() const { | |
| 568 std::string scheme(target_url_.scheme()); | |
| 569 std::string host(target_url_.host()); | |
| 570 | |
| 571 // If target host or scheme is empty, set scheme and host of source url. | |
| 572 // This is done to match the Toolbar's behavior. | |
| 573 if (scheme.empty() || host.empty()) { | |
| 574 scheme = source_url_.scheme(); | |
| 575 host = source_url_.host(); | |
| 576 } | |
| 577 | |
| 578 std::string form_string = scheme + "://" + host + "&" + | |
| 579 UTF16ToUTF8(form_name_) + | |
| 580 form_signature_field_names_; | |
| 581 | |
| 582 return Hash64Bit(form_string); | |
| 583 } | |
| 584 | |
| 585 bool FormStructure::IsAutocheckoutEnabled() const { | |
| 586 return !autocheckout_url_prefix_.empty(); | |
| 587 } | |
| 588 | |
| 589 size_t FormStructure::RequiredFillableFields() const { | |
| 590 return IsAutocheckoutEnabled() ? 0 : kRequiredFillableFields; | |
| 591 } | |
| 592 | |
| 593 bool FormStructure::IsAutofillable(bool require_method_post) const { | |
| 594 if (autofill_count() < RequiredFillableFields()) | |
| 595 return false; | |
| 596 | |
| 597 return ShouldBeParsed(require_method_post); | |
| 598 } | |
| 599 | |
| 600 void FormStructure::UpdateAutofillCount() { | |
| 601 autofill_count_ = 0; | |
| 602 for (std::vector<AutofillField*>::const_iterator iter = begin(); | |
| 603 iter != end(); ++iter) { | |
| 604 AutofillField* field = *iter; | |
| 605 if (field && field->IsFieldFillable()) | |
| 606 ++autofill_count_; | |
| 607 } | |
| 608 } | |
| 609 | |
| 610 bool FormStructure::ShouldBeParsed(bool require_method_post) const { | |
| 611 // Ignore counting checkable elements towards minimum number of elements | |
| 612 // required to parse. This avoids trying to crowdsource forms with few text | |
| 613 // or select elements. | |
| 614 if ((field_count() - checkable_field_count()) < RequiredFillableFields()) | |
| 615 return false; | |
| 616 | |
| 617 // Rule out http(s)://*/search?... | |
| 618 // e.g. http://www.google.com/search?q=... | |
| 619 // http://search.yahoo.com/search?p=... | |
| 620 if (target_url_.path() == "/search") | |
| 621 return false; | |
| 622 | |
| 623 if (!IsAutocheckoutEnabled()) { | |
| 624 // Make sure there is at least one text field when Autocheckout is | |
| 625 // not enabled. | |
| 626 bool has_text_field = false; | |
| 627 for (std::vector<AutofillField*>::const_iterator it = begin(); | |
| 628 it != end() && !has_text_field; ++it) { | |
| 629 has_text_field |= (*it)->form_control_type != "select-one"; | |
| 630 } | |
| 631 if (!has_text_field) | |
| 632 return false; | |
| 633 } | |
| 634 | |
| 635 return !require_method_post || (method_ == POST); | |
| 636 } | |
| 637 | |
| 638 bool FormStructure::ShouldBeCrowdsourced() const { | |
| 639 return !has_author_specified_types_ && ShouldBeParsed(true); | |
| 640 } | |
| 641 | |
| 642 void FormStructure::UpdateFromCache(const FormStructure& cached_form) { | |
| 643 // Map from field signatures to cached fields. | |
| 644 std::map<std::string, const AutofillField*> cached_fields; | |
| 645 for (size_t i = 0; i < cached_form.field_count(); ++i) { | |
| 646 const AutofillField* field = cached_form.field(i); | |
| 647 cached_fields[field->FieldSignature()] = field; | |
| 648 } | |
| 649 | |
| 650 for (std::vector<AutofillField*>::const_iterator iter = begin(); | |
| 651 iter != end(); ++iter) { | |
| 652 AutofillField* field = *iter; | |
| 653 | |
| 654 std::map<std::string, const AutofillField*>::const_iterator | |
| 655 cached_field = cached_fields.find(field->FieldSignature()); | |
| 656 if (cached_field != cached_fields.end()) { | |
| 657 if (field->form_control_type != "select-one" && | |
| 658 field->value == cached_field->second->value) { | |
| 659 // From the perspective of learning user data, text fields containing | |
| 660 // default values are equivalent to empty fields. | |
| 661 field->value = string16(); | |
| 662 } | |
| 663 | |
| 664 field->set_heuristic_type(cached_field->second->heuristic_type()); | |
| 665 field->set_server_type(cached_field->second->server_type()); | |
| 666 } | |
| 667 } | |
| 668 | |
| 669 UpdateAutofillCount(); | |
| 670 | |
| 671 server_experiment_id_ = cached_form.server_experiment_id(); | |
| 672 | |
| 673 // The form signature should match between query and upload requests to the | |
| 674 // server. On many websites, form elements are dynamically added, removed, or | |
| 675 // rearranged via JavaScript between page load and form submission, so we | |
| 676 // copy over the |form_signature_field_names_| corresponding to the query | |
| 677 // request. | |
| 678 DCHECK_EQ(cached_form.form_name_, form_name_); | |
| 679 DCHECK_EQ(cached_form.source_url_, source_url_); | |
| 680 DCHECK_EQ(cached_form.target_url_, target_url_); | |
| 681 form_signature_field_names_ = cached_form.form_signature_field_names_; | |
| 682 } | |
| 683 | |
| 684 void FormStructure::LogQualityMetrics( | |
| 685 const AutofillMetrics& metric_logger, | |
| 686 const base::TimeTicks& load_time, | |
| 687 const base::TimeTicks& interaction_time, | |
| 688 const base::TimeTicks& submission_time) const { | |
| 689 std::string experiment_id = server_experiment_id(); | |
| 690 metric_logger.LogServerExperimentIdForUpload(experiment_id); | |
| 691 | |
| 692 size_t num_detected_field_types = 0; | |
| 693 bool did_autofill_all_possible_fields = true; | |
| 694 bool did_autofill_some_possible_fields = false; | |
| 695 for (size_t i = 0; i < field_count(); ++i) { | |
| 696 const AutofillField* field = this->field(i); | |
| 697 metric_logger.LogQualityMetric(AutofillMetrics::FIELD_SUBMITTED, | |
| 698 experiment_id); | |
| 699 | |
| 700 // No further logging for empty fields nor for fields where the entered data | |
| 701 // does not appear to already exist in the user's stored Autofill data. | |
| 702 const FieldTypeSet& field_types = field->possible_types(); | |
| 703 DCHECK(!field_types.empty()); | |
| 704 if (field_types.count(EMPTY_TYPE) || field_types.count(UNKNOWN_TYPE)) | |
| 705 continue; | |
| 706 | |
| 707 ++num_detected_field_types; | |
| 708 if (field->is_autofilled) | |
| 709 did_autofill_some_possible_fields = true; | |
| 710 else | |
| 711 did_autofill_all_possible_fields = false; | |
| 712 | |
| 713 // Collapse field types that Chrome treats as identical, e.g. home and | |
| 714 // billing address fields. | |
| 715 FieldTypeSet collapsed_field_types; | |
| 716 for (FieldTypeSet::const_iterator it = field_types.begin(); | |
| 717 it != field_types.end(); | |
| 718 ++it) { | |
| 719 // Since we currently only support US phone numbers, the (city code + main | |
| 720 // digits) number is almost always identical to the whole phone number. | |
| 721 // TODO(isherman): Improve this logic once we add support for | |
| 722 // international numbers. | |
| 723 if (*it == PHONE_HOME_CITY_AND_NUMBER) | |
| 724 collapsed_field_types.insert(PHONE_HOME_WHOLE_NUMBER); | |
| 725 else | |
| 726 collapsed_field_types.insert(AutofillType::GetEquivalentFieldType(*it)); | |
| 727 } | |
| 728 | |
| 729 // Capture the field's type, if it is unambiguous. | |
| 730 AutofillFieldType field_type = UNKNOWN_TYPE; | |
| 731 if (collapsed_field_types.size() == 1) | |
| 732 field_type = *collapsed_field_types.begin(); | |
| 733 | |
| 734 AutofillFieldType heuristic_type = field->heuristic_type(); | |
| 735 AutofillFieldType server_type = field->server_type(); | |
| 736 AutofillFieldType predicted_type = field->type(); | |
| 737 | |
| 738 // Log heuristic, server, and overall type quality metrics, independently of | |
| 739 // whether the field was autofilled. | |
| 740 if (heuristic_type == UNKNOWN_TYPE) { | |
| 741 metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_UNKNOWN, | |
| 742 field_type, experiment_id); | |
| 743 } else if (field_types.count(heuristic_type)) { | |
| 744 metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_MATCH, | |
| 745 field_type, experiment_id); | |
| 746 } else { | |
| 747 metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_MISMATCH, | |
| 748 field_type, experiment_id); | |
| 749 } | |
| 750 | |
| 751 if (server_type == NO_SERVER_DATA) { | |
| 752 metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_UNKNOWN, | |
| 753 field_type, experiment_id); | |
| 754 } else if (field_types.count(server_type)) { | |
| 755 metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_MATCH, | |
| 756 field_type, experiment_id); | |
| 757 } else { | |
| 758 metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_MISMATCH, | |
| 759 field_type, experiment_id); | |
| 760 } | |
| 761 | |
| 762 if (predicted_type == UNKNOWN_TYPE) { | |
| 763 metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_UNKNOWN, | |
| 764 field_type, experiment_id); | |
| 765 } else if (field_types.count(predicted_type)) { | |
| 766 metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_MATCH, | |
| 767 field_type, experiment_id); | |
| 768 } else { | |
| 769 metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_MISMATCH, | |
| 770 field_type, experiment_id); | |
| 771 } | |
| 772 | |
| 773 // TODO(isherman): <select> fields don't support |is_autofilled()|, so we | |
| 774 // have to skip them for the remaining metrics. | |
| 775 if (field->form_control_type == "select-one") | |
| 776 continue; | |
| 777 | |
| 778 if (field->is_autofilled) { | |
| 779 metric_logger.LogQualityMetric(AutofillMetrics::FIELD_AUTOFILLED, | |
| 780 experiment_id); | |
| 781 } else { | |
| 782 metric_logger.LogQualityMetric(AutofillMetrics::FIELD_NOT_AUTOFILLED, | |
| 783 experiment_id); | |
| 784 | |
| 785 if (heuristic_type == UNKNOWN_TYPE) { | |
| 786 metric_logger.LogQualityMetric( | |
| 787 AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_UNKNOWN, | |
| 788 experiment_id); | |
| 789 } else if (field_types.count(heuristic_type)) { | |
| 790 metric_logger.LogQualityMetric( | |
| 791 AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MATCH, | |
| 792 experiment_id); | |
| 793 } else { | |
| 794 metric_logger.LogQualityMetric( | |
| 795 AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MISMATCH, | |
| 796 experiment_id); | |
| 797 } | |
| 798 | |
| 799 if (server_type == NO_SERVER_DATA) { | |
| 800 metric_logger.LogQualityMetric( | |
| 801 AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_UNKNOWN, | |
| 802 experiment_id); | |
| 803 } else if (field_types.count(server_type)) { | |
| 804 metric_logger.LogQualityMetric( | |
| 805 AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MATCH, | |
| 806 experiment_id); | |
| 807 } else { | |
| 808 metric_logger.LogQualityMetric( | |
| 809 AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MISMATCH, | |
| 810 experiment_id); | |
| 811 } | |
| 812 } | |
| 813 } | |
| 814 | |
| 815 if (num_detected_field_types < RequiredFillableFields()) { | |
| 816 metric_logger.LogUserHappinessMetric( | |
| 817 AutofillMetrics::SUBMITTED_NON_FILLABLE_FORM); | |
| 818 } else { | |
| 819 if (did_autofill_all_possible_fields) { | |
| 820 metric_logger.LogUserHappinessMetric( | |
| 821 AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_ALL); | |
| 822 } else if (did_autofill_some_possible_fields) { | |
| 823 metric_logger.LogUserHappinessMetric( | |
| 824 AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_SOME); | |
| 825 } else { | |
| 826 metric_logger.LogUserHappinessMetric( | |
| 827 AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_NONE); | |
| 828 } | |
| 829 | |
| 830 // Unlike the other times, the |submission_time| should always be available. | |
| 831 DCHECK(!submission_time.is_null()); | |
| 832 | |
| 833 // The |load_time| might be unset, in the case that the form was dynamically | |
| 834 // added to the DOM. | |
| 835 if (!load_time.is_null()) { | |
| 836 // Submission should always chronologically follow form load. | |
| 837 DCHECK(submission_time > load_time); | |
| 838 base::TimeDelta elapsed = submission_time - load_time; | |
| 839 if (did_autofill_some_possible_fields) | |
| 840 metric_logger.LogFormFillDurationFromLoadWithAutofill(elapsed); | |
| 841 else | |
| 842 metric_logger.LogFormFillDurationFromLoadWithoutAutofill(elapsed); | |
| 843 } | |
| 844 | |
| 845 // The |interaction_time| might be unset, in the case that the user | |
| 846 // submitted a blank form. | |
| 847 if (!interaction_time.is_null()) { | |
| 848 // Submission should always chronologically follow interaction. | |
| 849 DCHECK(submission_time > interaction_time); | |
| 850 base::TimeDelta elapsed = submission_time - interaction_time; | |
| 851 if (did_autofill_some_possible_fields) { | |
| 852 metric_logger.LogFormFillDurationFromInteractionWithAutofill(elapsed); | |
| 853 } else { | |
| 854 metric_logger.LogFormFillDurationFromInteractionWithoutAutofill( | |
| 855 elapsed); | |
| 856 } | |
| 857 } | |
| 858 } | |
| 859 } | |
| 860 | |
| 861 const AutofillField* FormStructure::field(size_t index) const { | |
| 862 if (index >= fields_.size()) { | |
| 863 NOTREACHED(); | |
| 864 return NULL; | |
| 865 } | |
| 866 | |
| 867 return fields_[index]; | |
| 868 } | |
| 869 | |
| 870 AutofillField* FormStructure::field(size_t index) { | |
| 871 return const_cast<AutofillField*>( | |
| 872 static_cast<const FormStructure*>(this)->field(index)); | |
| 873 } | |
| 874 | |
| 875 size_t FormStructure::field_count() const { | |
| 876 return fields_.size(); | |
| 877 } | |
| 878 | |
| 879 size_t FormStructure::checkable_field_count() const { | |
| 880 return checkable_field_count_; | |
| 881 } | |
| 882 | |
| 883 std::string FormStructure::server_experiment_id() const { | |
| 884 return server_experiment_id_; | |
| 885 } | |
| 886 | |
| 887 FormData FormStructure::ToFormData() const { | |
| 888 // |data.user_submitted| will always be false. | |
| 889 FormData data; | |
| 890 data.name = form_name_; | |
| 891 data.origin = source_url_; | |
| 892 data.action = target_url_; | |
| 893 data.method = ASCIIToUTF16(method_ == POST ? "POST" : "GET"); | |
| 894 | |
| 895 for (size_t i = 0; i < fields_.size(); ++i) { | |
| 896 data.fields.push_back(FormFieldData(*fields_[i])); | |
| 897 } | |
| 898 | |
| 899 return data; | |
| 900 } | |
| 901 | |
| 902 bool FormStructure::operator==(const FormData& form) const { | |
| 903 // TODO(jhawkins): Is this enough to differentiate a form? | |
| 904 if (form_name_ == form.name && | |
| 905 source_url_ == form.origin && | |
| 906 target_url_ == form.action) { | |
| 907 return true; | |
| 908 } | |
| 909 | |
| 910 // TODO(jhawkins): Compare field names, IDs and labels once we have labels | |
| 911 // set up. | |
| 912 | |
| 913 return false; | |
| 914 } | |
| 915 | |
| 916 bool FormStructure::operator!=(const FormData& form) const { | |
| 917 return !operator==(form); | |
| 918 } | |
| 919 | |
| 920 std::string FormStructure::Hash64Bit(const std::string& str) { | |
| 921 std::string hash_bin = base::SHA1HashString(str); | |
| 922 DCHECK_EQ(20U, hash_bin.length()); | |
| 923 | |
| 924 uint64 hash64 = (((static_cast<uint64>(hash_bin[0])) & 0xFF) << 56) | | |
| 925 (((static_cast<uint64>(hash_bin[1])) & 0xFF) << 48) | | |
| 926 (((static_cast<uint64>(hash_bin[2])) & 0xFF) << 40) | | |
| 927 (((static_cast<uint64>(hash_bin[3])) & 0xFF) << 32) | | |
| 928 (((static_cast<uint64>(hash_bin[4])) & 0xFF) << 24) | | |
| 929 (((static_cast<uint64>(hash_bin[5])) & 0xFF) << 16) | | |
| 930 (((static_cast<uint64>(hash_bin[6])) & 0xFF) << 8) | | |
| 931 ((static_cast<uint64>(hash_bin[7])) & 0xFF); | |
| 932 | |
| 933 return base::Uint64ToString(hash64); | |
| 934 } | |
| 935 | |
| 936 bool FormStructure::EncodeFormRequest( | |
| 937 FormStructure::EncodeRequestType request_type, | |
| 938 buzz::XmlElement* encompassing_xml_element) const { | |
| 939 if (!field_count()) // Nothing to add. | |
| 940 return false; | |
| 941 | |
| 942 // Some badly formatted web sites repeat fields - limit number of fields to | |
| 943 // 48, which is far larger than any valid form and XML still fits into 2K. | |
| 944 // Do not send requests for forms with more than this many fields, as they are | |
| 945 // near certainly not valid/auto-fillable. | |
| 946 const size_t kMaxFieldsOnTheForm = 48; | |
| 947 if (field_count() > kMaxFieldsOnTheForm) | |
| 948 return false; | |
| 949 | |
| 950 // Add the child nodes for the form fields. | |
| 951 for (size_t index = 0; index < field_count(); ++index) { | |
| 952 const AutofillField* field = fields_[index]; | |
| 953 if (request_type == FormStructure::UPLOAD) { | |
| 954 // Don't upload checkable fields. | |
| 955 if (field->is_checkable) | |
| 956 continue; | |
| 957 | |
| 958 FieldTypeSet types = field->possible_types(); | |
| 959 // |types| could be empty in unit-tests only. | |
| 960 for (FieldTypeSet::iterator field_type = types.begin(); | |
| 961 field_type != types.end(); ++field_type) { | |
| 962 buzz::XmlElement *field_element = new buzz::XmlElement( | |
| 963 buzz::QName(kXMLElementField)); | |
| 964 | |
| 965 field_element->SetAttr(buzz::QName(kAttributeSignature), | |
| 966 field->FieldSignature()); | |
| 967 field_element->SetAttr(buzz::QName(kAttributeAutofillType), | |
| 968 base::IntToString(*field_type)); | |
| 969 encompassing_xml_element->AddElement(field_element); | |
| 970 } | |
| 971 } else { | |
| 972 // Skip putting checkable fields in the request if Autocheckout is not | |
| 973 // enabled. | |
| 974 if (field->is_checkable && !IsAutocheckoutEnabled()) | |
| 975 continue; | |
| 976 | |
| 977 buzz::XmlElement *field_element = new buzz::XmlElement( | |
| 978 buzz::QName(kXMLElementField)); | |
| 979 field_element->SetAttr(buzz::QName(kAttributeSignature), | |
| 980 field->FieldSignature()); | |
| 981 encompassing_xml_element->AddElement(field_element); | |
| 982 } | |
| 983 } | |
| 984 return true; | |
| 985 } | |
| 986 | |
| 987 void FormStructure::ParseFieldTypesFromAutocompleteAttributes( | |
| 988 bool* found_types, | |
| 989 bool* found_sections) { | |
| 990 const std::string kDefaultSection = "-default"; | |
| 991 | |
| 992 *found_types = false; | |
| 993 *found_sections = false; | |
| 994 for (std::vector<AutofillField*>::iterator it = fields_.begin(); | |
| 995 it != fields_.end(); ++it) { | |
| 996 AutofillField* field = *it; | |
| 997 | |
| 998 // To prevent potential section name collisions, add a default suffix for | |
| 999 // other fields. Without this, 'autocomplete' attribute values | |
| 1000 // "section--shipping street-address" and "shipping street-address" would be | |
| 1001 // parsed identically, given the section handling code below. We do this | |
| 1002 // before any validation so that fields with invalid attributes still end up | |
| 1003 // in the default section. These default section names will be overridden | |
| 1004 // by subsequent heuristic parsing steps if there are no author-specified | |
| 1005 // section names. | |
| 1006 field->set_section(kDefaultSection); | |
| 1007 | |
| 1008 // Canonicalize the attribute value by trimming whitespace, collapsing | |
| 1009 // non-space characters (e.g. tab) to spaces, and converting to lowercase. | |
| 1010 std::string autocomplete_attribute = | |
| 1011 CollapseWhitespaceASCII(field->autocomplete_attribute, false); | |
| 1012 autocomplete_attribute = StringToLowerASCII(autocomplete_attribute); | |
| 1013 | |
| 1014 // The autocomplete attribute is overloaded: it can specify either a field | |
| 1015 // type hint or whether autocomplete should be enabled at all. Ignore the | |
| 1016 // latter type of attribute value. | |
| 1017 if (autocomplete_attribute.empty() || | |
| 1018 autocomplete_attribute == "on" || | |
| 1019 autocomplete_attribute == "off") { | |
| 1020 continue; | |
| 1021 } | |
| 1022 | |
| 1023 // Any other value, even it is invalid, is considered to be a type hint. | |
| 1024 // This allows a website's author to specify an attribute like | |
| 1025 // autocomplete="other" on a field to disable all Autofill heuristics for | |
| 1026 // the form. | |
| 1027 *found_types = true; | |
| 1028 | |
| 1029 // Tokenize the attribute value. Per the spec, the tokens are parsed in | |
| 1030 // reverse order. | |
| 1031 std::vector<std::string> tokens; | |
| 1032 Tokenize(autocomplete_attribute, " ", &tokens); | |
| 1033 | |
| 1034 // The final token must be the field type. | |
| 1035 // If it is not one of the known types, abort. | |
| 1036 DCHECK(!tokens.empty()); | |
| 1037 std::string field_type_token = tokens.back(); | |
| 1038 tokens.pop_back(); | |
| 1039 AutofillFieldType field_type = | |
| 1040 FieldTypeFromAutocompleteType(field_type_token, *field); | |
| 1041 if (field_type == UNKNOWN_TYPE) | |
| 1042 continue; | |
| 1043 | |
| 1044 // The preceding token, if any, may be a type hint. | |
| 1045 if (!tokens.empty() && IsContactTypeHint(tokens.back())) { | |
| 1046 // If it is, it must match the field type; otherwise, abort. | |
| 1047 // Note that an invalid token invalidates the entire attribute value, even | |
| 1048 // if the other tokens are valid. | |
| 1049 if (!ContactTypeHintMatchesFieldType(tokens.back(), field_type)) | |
| 1050 continue; | |
| 1051 | |
| 1052 // Chrome Autofill ignores these type hints. | |
| 1053 tokens.pop_back(); | |
| 1054 } | |
| 1055 | |
| 1056 // The preceding token, if any, may be a fixed string that is either | |
| 1057 // "shipping" or "billing". Chrome Autofill treats these as implicit | |
| 1058 // section name suffixes. | |
| 1059 DCHECK_EQ(kDefaultSection, field->section()); | |
| 1060 std::string section = field->section(); | |
| 1061 if (!tokens.empty() && | |
| 1062 (tokens.back() == "shipping" || tokens.back() == "billing")) { | |
| 1063 section = "-" + tokens.back(); | |
| 1064 tokens.pop_back(); | |
| 1065 } | |
| 1066 | |
| 1067 // The preceding token, if any, may be a named section. | |
| 1068 const std::string kSectionPrefix = "section-"; | |
| 1069 if (!tokens.empty() && | |
| 1070 StartsWithASCII(tokens.back(), kSectionPrefix, true)) { | |
| 1071 // Prepend this section name to the suffix set in the preceding block. | |
| 1072 section = tokens.back().substr(kSectionPrefix.size()) + section; | |
| 1073 tokens.pop_back(); | |
| 1074 } | |
| 1075 | |
| 1076 // No other tokens are allowed. If there are any remaining, abort. | |
| 1077 if (!tokens.empty()) | |
| 1078 continue; | |
| 1079 | |
| 1080 if (section != kDefaultSection) { | |
| 1081 *found_sections = true; | |
| 1082 field->set_section(section); | |
| 1083 } | |
| 1084 | |
| 1085 // No errors encountered while parsing! | |
| 1086 // Update the |field|'s type based on what was parsed from the attribute. | |
| 1087 field->set_heuristic_type(field_type); | |
| 1088 if (field_type_token == "tel-local-prefix") | |
| 1089 field->set_phone_part(AutofillField::PHONE_PREFIX); | |
| 1090 else if (field_type_token == "tel-local-suffix") | |
| 1091 field->set_phone_part(AutofillField::PHONE_SUFFIX); | |
| 1092 } | |
| 1093 } | |
| 1094 | |
| 1095 void FormStructure::IdentifySections(bool has_author_specified_sections) { | |
| 1096 if (fields_.empty()) | |
| 1097 return; | |
| 1098 | |
| 1099 if (!has_author_specified_sections) { | |
| 1100 // Name sections after the first field in the section. | |
| 1101 string16 current_section = fields_.front()->unique_name(); | |
| 1102 | |
| 1103 // Keep track of the types we've seen in this section. | |
| 1104 std::set<AutofillFieldType> seen_types; | |
| 1105 AutofillFieldType previous_type = UNKNOWN_TYPE; | |
| 1106 | |
| 1107 for (std::vector<AutofillField*>::iterator field = fields_.begin(); | |
| 1108 field != fields_.end(); ++field) { | |
| 1109 const AutofillFieldType current_type = | |
| 1110 AutofillType::GetEquivalentFieldType((*field)->type()); | |
| 1111 | |
| 1112 bool already_saw_current_type = seen_types.count(current_type) > 0; | |
| 1113 | |
| 1114 // Forms often ask for multiple phone numbers -- e.g. both a daytime and | |
| 1115 // evening phone number. Our phone number detection is also generally a | |
| 1116 // little off. Hence, ignore this field type as a signal here. | |
| 1117 if (AutofillType(current_type).group() == AutofillType::PHONE) | |
| 1118 already_saw_current_type = false; | |
| 1119 | |
| 1120 // Some forms have adjacent fields of the same type. Two common examples: | |
| 1121 // * Forms with two email fields, where the second is meant to "confirm" | |
| 1122 // the first. | |
| 1123 // * Forms with a <select> menu for states in some countries, and a | |
| 1124 // freeform <input> field for states in other countries. (Usually, | |
| 1125 // only one of these two will be visible for any given choice of | |
| 1126 // country.) | |
| 1127 // Generally, adjacent fields of the same type belong in the same logical | |
| 1128 // section. | |
| 1129 if (current_type == previous_type) | |
| 1130 already_saw_current_type = false; | |
| 1131 | |
| 1132 previous_type = current_type; | |
| 1133 | |
| 1134 if (current_type != UNKNOWN_TYPE && already_saw_current_type) { | |
| 1135 // We reached the end of a section, so start a new section. | |
| 1136 seen_types.clear(); | |
| 1137 current_section = (*field)->unique_name(); | |
| 1138 } | |
| 1139 | |
| 1140 seen_types.insert(current_type); | |
| 1141 (*field)->set_section(UTF16ToUTF8(current_section)); | |
| 1142 } | |
| 1143 } | |
| 1144 | |
| 1145 // Ensure that credit card and address fields are in separate sections. | |
| 1146 // This simplifies the section-aware logic in autofill_manager.cc. | |
| 1147 for (std::vector<AutofillField*>::iterator field = fields_.begin(); | |
| 1148 field != fields_.end(); ++field) { | |
| 1149 AutofillType::FieldTypeGroup field_type_group = | |
| 1150 AutofillType((*field)->type()).group(); | |
| 1151 if (field_type_group == AutofillType::CREDIT_CARD) | |
| 1152 (*field)->set_section((*field)->section() + "-cc"); | |
| 1153 else | |
| 1154 (*field)->set_section((*field)->section() + "-default"); | |
| 1155 } | |
| 1156 } | |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 12434004:
Move remaining Autofill code to //components/autofill. (Closed)
Base URL: svn://svn.chromium.org/chrome/trunk/src