form_structure.cc revision ba5b9a6411cb1792fd21f0a078d7a25cd1ceec16 
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"
30#include "third_party/icu/source/i18n/unicode/regex.h"
31#include "third_party/libjingle/source/talk/xmllite/xmlelement.h"
32
33namespace autofill {
34namespace {
35
36const char kFormMethodPost[] = "post";
37
38// XML elements and attributes.
39const char kAttributeAcceptedFeatures[] = "accepts";
40const char kAttributeAutofillUsed[] = "autofillused";
41const char kAttributeAutofillType[] = "autofilltype";
42const char kAttributeClientVersion[] = "clientversion";
43const char kAttributeDataPresent[] = "datapresent";
44const char kAttributeFieldID[] = "fieldid";
45const char kAttributeFieldType[] = "fieldtype";
46const char kAttributeFormSignature[] = "formsignature";
47const char kAttributeName[] = "name";
48const char kAttributeSignature[] = "signature";
49const char kAttributeUrlprefixSignature[] = "urlprefixsignature";
50const char kAcceptedFeaturesExperiment[] = "e"; // e=experiments
51const char kAcceptedFeaturesAutocheckoutExperiment[] = "a,e"; // a=autocheckout
52const char kClientVersion[] = "6.1.1715.1442/en (GGLL)";
53const char kXMLDeclaration[] = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
54const char kXMLElementAutofillQuery[] = "autofillquery";
55const char kXMLElementAutofillUpload[] = "autofillupload";
56const char kXMLElementFieldAssignments[] = "fieldassignments";
57const char kXMLElementField[] = "field";
58const char kXMLElementFields[] = "fields";
59const char kXMLElementForm[] = "form";
60const char kBillingMode[] = "billing";
61const char kShippingMode[] = "shipping";
62
63// Stip away >= 5 consecutive digits.
64const char kIgnorePatternInFieldName[] = "\\d{5,}+";
65
66// Helper for |EncodeUploadRequest()| that creates a bit field corresponding to
67// |available_field_types| and returns the hex representation as a string.
68std::string EncodeFieldTypes(const ServerFieldTypeSet& available_field_types) {
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);
75  for (ServerFieldTypeSet::const_iterator field_type =
76           available_field_types.begin();
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.
105void EncodeFieldForUpload(const AutofillField& field,
106                          buzz::XmlElement* parent) {
107  // Don't upload checkable fields.
108  if (field.is_checkable)
109    return;
110
111  ServerFieldTypeSet types = field.possible_types();
112  // |types| could be empty in unit-tests only.
113  for (ServerFieldTypeSet::iterator field_type = types.begin();
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.
128void 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.
139void EncodeFieldForFieldAssignments(const AutofillField& field,
140                                    buzz::XmlElement* parent) {
141  ServerFieldTypeSet types = field.possible_types();
142  for (ServerFieldTypeSet::iterator field_type = types.begin();
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.
161bool 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
168bool ContactTypeHintMatchesFieldType(const std::string& token,
169                                     HtmlFieldType field_type) {
170  // The "home" and "work" type hints are only appropriate for email and phone
171  // number field types.
172  if (token == "home" || token == "work") {
173    return field_type == HTML_TYPE_EMAIL ||
174        (field_type >= HTML_TYPE_TEL &&
175         field_type <= HTML_TYPE_TEL_LOCAL_SUFFIX);
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") {
182    return field_type >= HTML_TYPE_TEL &&
183        field_type <= HTML_TYPE_TEL_LOCAL_SUFFIX;
184  }
185
186  return false;
187}
188
189// Returns the Chrome Autofill-supported field type corresponding to the given
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
192// http://is.gd/whatwg_autocomplete
193HtmlFieldType FieldTypeFromAutocompleteAttributeValue(
194    const std::string& autocomplete_attribute_value,
195    const AutofillField& field) {
196  if (autocomplete_attribute_value == "name")
197    return HTML_TYPE_NAME;
198
199  if (autocomplete_attribute_value == "given-name")
200    return HTML_TYPE_GIVEN_NAME;
201
202  if (autocomplete_attribute_value == "additional-name") {
203    if (field.max_length == 1)
204      return HTML_TYPE_ADDITIONAL_NAME_INITIAL;
205    else
206      return HTML_TYPE_ADDITIONAL_NAME;
207  }
208
209  if (autocomplete_attribute_value == "family-name")
210    return HTML_TYPE_FAMILY_NAME;
211
212  if (autocomplete_attribute_value == "organization")
213    return HTML_TYPE_ORGANIZATION;
214
215  if (autocomplete_attribute_value == "street-address")
216    return HTML_TYPE_STREET_ADDRESS;
217
218  if (autocomplete_attribute_value == "address-line1")
219    return HTML_TYPE_ADDRESS_LINE1;
220
221  if (autocomplete_attribute_value == "address-line2")
222    return HTML_TYPE_ADDRESS_LINE2;
223
224  if (autocomplete_attribute_value == "locality")
225    return HTML_TYPE_LOCALITY;
226
227  if (autocomplete_attribute_value == "region")
228    return HTML_TYPE_REGION;
229
230  if (autocomplete_attribute_value == "country")
231    return HTML_TYPE_COUNTRY_CODE;
232
233  if (autocomplete_attribute_value == "country-name")
234    return HTML_TYPE_COUNTRY_NAME;
235
236  if (autocomplete_attribute_value == "postal-code")
237    return HTML_TYPE_POSTAL_CODE;
238
239  if (autocomplete_attribute_value == "cc-name")
240    return HTML_TYPE_CREDIT_CARD_NAME;
241
242  if (autocomplete_attribute_value == "cc-number")
243    return HTML_TYPE_CREDIT_CARD_NUMBER;
244
245  if (autocomplete_attribute_value == "cc-exp") {
246    if (field.max_length == 5)
247      return HTML_TYPE_CREDIT_CARD_EXP_DATE_2_DIGIT_YEAR;
248    else if (field.max_length == 7)
249      return HTML_TYPE_CREDIT_CARD_EXP_DATE_4_DIGIT_YEAR;
250    else
251      return HTML_TYPE_CREDIT_CARD_EXP;
252  }
253
254  if (autocomplete_attribute_value == "cc-exp-month")
255    return HTML_TYPE_CREDIT_CARD_EXP_MONTH;
256
257  if (autocomplete_attribute_value == "cc-exp-year") {
258    if (field.max_length == 2)
259      return HTML_TYPE_CREDIT_CARD_EXP_2_DIGIT_YEAR;
260    else if (field.max_length == 4)
261      return HTML_TYPE_CREDIT_CARD_EXP_4_DIGIT_YEAR;
262    else
263      return HTML_TYPE_CREDIT_CARD_EXP_YEAR;
264  }
265
266  if (autocomplete_attribute_value == "cc-csc")
267    return HTML_TYPE_CREDIT_CARD_VERIFICATION_CODE;
268
269  if (autocomplete_attribute_value == "cc-type")
270    return HTML_TYPE_CREDIT_CARD_TYPE;
271
272  if (autocomplete_attribute_value == "tel")
273    return HTML_TYPE_TEL;
274
275  if (autocomplete_attribute_value == "tel-country-code")
276    return HTML_TYPE_TEL_COUNTRY_CODE;
277
278  if (autocomplete_attribute_value == "tel-national")
279    return HTML_TYPE_TEL_NATIONAL;
280
281  if (autocomplete_attribute_value == "tel-area-code")
282    return HTML_TYPE_TEL_AREA_CODE;
283
284  if (autocomplete_attribute_value == "tel-local")
285    return HTML_TYPE_TEL_LOCAL;
286
287  if (autocomplete_attribute_value == "tel-local-prefix")
288    return HTML_TYPE_TEL_LOCAL_PREFIX;
289
290  if (autocomplete_attribute_value == "tel-local-suffix")
291    return HTML_TYPE_TEL_LOCAL_SUFFIX;
292
293  if (autocomplete_attribute_value == "email")
294    return HTML_TYPE_EMAIL;
295
296  return HTML_TYPE_UNKNOWN;
297}
298
299std::string StripDigitsIfRequired(const base::string16& input) {
300  UErrorCode status = U_ZERO_ERROR;
301  CR_DEFINE_STATIC_LOCAL(icu::UnicodeString, icu_pattern,
302                         (kIgnorePatternInFieldName));
303  CR_DEFINE_STATIC_LOCAL(icu::RegexMatcher, matcher,
304                         (icu_pattern, UREGEX_CASE_INSENSITIVE, status));
305  DCHECK_EQ(status, U_ZERO_ERROR);
306
307  icu::UnicodeString icu_input(input.data(), input.length());
308  matcher.reset(icu_input);
309
310  icu::UnicodeString replaced_string = matcher.replaceAll("", status);
311
312  std::string return_string;
313  status = U_ZERO_ERROR;
314  UTF16ToUTF8(replaced_string.getBuffer(),
315              static_cast<size_t>(replaced_string.length()),
316              &return_string);
317  if (status != U_ZERO_ERROR) {
318    DVLOG(1) << "Couldn't strip digits in " << UTF16ToUTF8(input);
319    return UTF16ToUTF8(input);
320  }
321
322  return return_string;
323}
324
325}  // namespace
326
327FormStructure::FormStructure(const FormData& form,
328                             const std::string& autocheckout_url_prefix)
329    : form_name_(form.name),
330      source_url_(form.origin),
331      target_url_(form.action),
332      autofill_count_(0),
333      active_field_count_(0),
334      upload_required_(USE_UPLOAD_RATES),
335      server_experiment_id_("no server response"),
336      has_author_specified_types_(false),
337      autocheckout_url_prefix_(autocheckout_url_prefix),
338      filled_by_autocheckout_(false) {
339  // Copy the form fields.
340  std::map<base::string16, size_t> unique_names;
341  for (std::vector<FormFieldData>::const_iterator field =
342           form.fields.begin();
343       field != form.fields.end(); field++) {
344
345    if (!ShouldSkipField(*field)) {
346      // Add all supported form fields (including with empty names) to the
347      // signature.  This is a requirement for Autofill servers.
348      form_signature_field_names_.append("&");
349      form_signature_field_names_.append(StripDigitsIfRequired(field->name));
350
351      ++active_field_count_;
352    }
353
354    // Generate a unique name for this field by appending a counter to the name.
355    // Make sure to prepend the counter with a non-numeric digit so that we are
356    // guaranteed to avoid collisions.
357    if (!unique_names.count(field->name))
358      unique_names[field->name] = 1;
359    else
360      ++unique_names[field->name];
361    base::string16 unique_name = field->name + ASCIIToUTF16("_") +
362        base::IntToString16(unique_names[field->name]);
363    fields_.push_back(new AutofillField(*field, unique_name));
364  }
365
366  std::string method = UTF16ToUTF8(form.method);
367  if (StringToLowerASCII(method) == kFormMethodPost) {
368    method_ = POST;
369  } else {
370    // Either the method is 'get', or we don't know.  In this case we default
371    // to GET.
372    method_ = GET;
373  }
374}
375
376FormStructure::~FormStructure() {}
377
378void FormStructure::DetermineHeuristicTypes(
379    const AutofillMetrics& metric_logger) {
380  // First, try to detect field types based on each field's |autocomplete|
381  // attribute value.  If there is at least one form field that specifies an
382  // autocomplete type hint, don't try to apply other heuristics to match fields
383  // in this form.
384  bool has_author_specified_sections;
385  ParseFieldTypesFromAutocompleteAttributes(&has_author_specified_types_,
386                                            &has_author_specified_sections);
387
388  if (!has_author_specified_types_) {
389    ServerFieldTypeMap field_type_map;
390    FormField::ParseFormFields(fields_.get(), &field_type_map);
391    for (size_t i = 0; i < field_count(); ++i) {
392      AutofillField* field = fields_[i];
393      ServerFieldTypeMap::iterator iter =
394          field_type_map.find(field->unique_name());
395      if (iter != field_type_map.end())
396        field->set_heuristic_type(iter->second);
397    }
398  }
399
400  UpdateAutofillCount();
401  IdentifySections(has_author_specified_sections);
402
403  if (IsAutofillable(true)) {
404    metric_logger.LogDeveloperEngagementMetric(
405        AutofillMetrics::FILLABLE_FORM_PARSED);
406    if (has_author_specified_types_) {
407      metric_logger.LogDeveloperEngagementMetric(
408          AutofillMetrics::FILLABLE_FORM_CONTAINS_TYPE_HINTS);
409    }
410  }
411}
412
413bool FormStructure::EncodeUploadRequest(
414    const ServerFieldTypeSet& available_field_types,
415    bool form_was_autofilled,
416    std::string* encoded_xml) const {
417  DCHECK(ShouldBeCrowdsourced());
418
419  // Verify that |available_field_types| agrees with the possible field types we
420  // are uploading.
421  for (std::vector<AutofillField*>::const_iterator field = begin();
422       field != end();
423       ++field) {
424    for (ServerFieldTypeSet::const_iterator type =
425             (*field)->possible_types().begin();
426         type != (*field)->possible_types().end();
427         ++type) {
428      DCHECK(*type == UNKNOWN_TYPE ||
429             *type == EMPTY_TYPE ||
430             available_field_types.count(*type));
431    }
432  }
433
434  // Set up the <autofillupload> element and its attributes.
435  buzz::XmlElement autofill_request_xml(
436      (buzz::QName(kXMLElementAutofillUpload)));
437  autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion),
438                               kClientVersion);
439  autofill_request_xml.SetAttr(buzz::QName(kAttributeFormSignature),
440                               FormSignature());
441  autofill_request_xml.SetAttr(buzz::QName(kAttributeAutofillUsed),
442                               form_was_autofilled ? "true" : "false");
443  autofill_request_xml.SetAttr(buzz::QName(kAttributeDataPresent),
444                               EncodeFieldTypes(available_field_types).c_str());
445
446  if (!EncodeFormRequest(FormStructure::UPLOAD, &autofill_request_xml))
447    return false;  // Malformed form, skip it.
448
449  // Obtain the XML structure as a string.
450  *encoded_xml = kXMLDeclaration;
451  *encoded_xml += autofill_request_xml.Str().c_str();
452
453  // To enable this logging, run with the flag --vmodule="form_structure=2".
454  VLOG(2) << "\n" << *encoded_xml;
455
456  return true;
457}
458
459bool FormStructure::EncodeFieldAssignments(
460    const ServerFieldTypeSet& available_field_types,
461    std::string* encoded_xml) const {
462  DCHECK(ShouldBeCrowdsourced());
463
464  // Set up the <fieldassignments> element and its attributes.
465  buzz::XmlElement autofill_request_xml(
466      (buzz::QName(kXMLElementFieldAssignments)));
467  autofill_request_xml.SetAttr(buzz::QName(kAttributeFormSignature),
468                               FormSignature());
469
470  if (!EncodeFormRequest(FormStructure::FIELD_ASSIGNMENTS,
471                         &autofill_request_xml))
472    return false;  // Malformed form, skip it.
473
474  // Obtain the XML structure as a string.
475  *encoded_xml = kXMLDeclaration;
476  *encoded_xml += autofill_request_xml.Str().c_str();
477
478  return true;
479}
480
481// static
482bool FormStructure::EncodeQueryRequest(
483    const std::vector<FormStructure*>& forms,
484    std::vector<std::string>* encoded_signatures,
485    std::string* encoded_xml) {
486  DCHECK(encoded_signatures);
487  DCHECK(encoded_xml);
488  encoded_xml->clear();
489  encoded_signatures->clear();
490  encoded_signatures->reserve(forms.size());
491
492  // Set up the <autofillquery> element and attributes.
493  buzz::XmlElement autofill_request_xml(
494      (buzz::QName(kXMLElementAutofillQuery)));
495  autofill_request_xml.SetAttr(buzz::QName(kAttributeClientVersion),
496                               kClientVersion);
497
498  // autocheckout_url_prefix tells the Autofill server where the forms in the
499  // request came from, and the the Autofill server checks internal status and
500  // decide to enable Autocheckout or not and may return Autocheckout related
501  // data in the response accordingly.
502  // There is no page/frame level object associated with FormStructure that
503  // we could extract URL prefix from. But, all the forms should come from the
504  // same frame, so they should have the same Autocheckout URL prefix. Thus we
505  // use URL prefix from the first form with Autocheckout enabled.
506  std::string autocheckout_url_prefix;
507
508  // Some badly formatted web sites repeat forms - detect that and encode only
509  // one form as returned data would be the same for all the repeated forms.
510  std::set<std::string> processed_forms;
511  for (ScopedVector<FormStructure>::const_iterator it = forms.begin();
512       it != forms.end();
513       ++it) {
514    std::string signature((*it)->FormSignature());
515    if (processed_forms.find(signature) != processed_forms.end())
516      continue;
517    processed_forms.insert(signature);
518    scoped_ptr<buzz::XmlElement> encompassing_xml_element(
519        new buzz::XmlElement(buzz::QName(kXMLElementForm)));
520    encompassing_xml_element->SetAttr(buzz::QName(kAttributeSignature),
521                                      signature);
522
523    if (!(*it)->EncodeFormRequest(FormStructure::QUERY,
524                                  encompassing_xml_element.get()))
525      continue;  // Malformed form, skip it.
526
527    if ((*it)->IsAutocheckoutEnabled()) {
528      if (autocheckout_url_prefix.empty()) {
529        autocheckout_url_prefix = (*it)->autocheckout_url_prefix_;
530      } else {
531        // Making sure all the forms in the request has the same url_prefix.
532        DCHECK_EQ(autocheckout_url_prefix, (*it)->autocheckout_url_prefix_);
533      }
534    }
535
536    autofill_request_xml.AddElement(encompassing_xml_element.release());
537    encoded_signatures->push_back(signature);
538  }
539
540  if (!encoded_signatures->size())
541    return false;
542
543  if (autocheckout_url_prefix.empty()) {
544    autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures),
545                                 kAcceptedFeaturesExperiment);
546  } else {
547    autofill_request_xml.SetAttr(buzz::QName(kAttributeAcceptedFeatures),
548                                 kAcceptedFeaturesAutocheckoutExperiment);
549    autofill_request_xml.SetAttr(buzz::QName(kAttributeUrlprefixSignature),
550                                 Hash64Bit(autocheckout_url_prefix));
551  }
552
553  // Obtain the XML structure as a string.
554  *encoded_xml = kXMLDeclaration;
555  *encoded_xml += autofill_request_xml.Str().c_str();
556
557  return true;
558}
559
560// static
561void FormStructure::ParseQueryResponse(
562    const std::string& response_xml,
563    const std::vector<FormStructure*>& forms,
564    autofill::AutocheckoutPageMetaData* page_meta_data,
565    const AutofillMetrics& metric_logger) {
566  metric_logger.LogServerQueryMetric(AutofillMetrics::QUERY_RESPONSE_RECEIVED);
567
568  // Parse the field types from the server response to the query.
569  std::vector<AutofillServerFieldInfo> field_infos;
570  UploadRequired upload_required;
571  std::string experiment_id;
572  AutofillQueryXmlParser parse_handler(&field_infos,
573                                       &upload_required,
574                                       &experiment_id,
575                                       page_meta_data);
576  buzz::XmlParser parser(&parse_handler);
577  parser.Parse(response_xml.c_str(), response_xml.length(), true);
578  if (!parse_handler.succeeded())
579    return;
580
581  metric_logger.LogServerQueryMetric(AutofillMetrics::QUERY_RESPONSE_PARSED);
582  metric_logger.LogServerExperimentIdForQuery(experiment_id);
583
584  bool heuristics_detected_fillable_field = false;
585  bool query_response_overrode_heuristics = false;
586
587  // Copy the field types into the actual form.
588  std::vector<AutofillServerFieldInfo>::iterator current_info =
589      field_infos.begin();
590  for (std::vector<FormStructure*>::const_iterator iter = forms.begin();
591       iter != forms.end(); ++iter) {
592    FormStructure* form = *iter;
593    form->upload_required_ = upload_required;
594    form->server_experiment_id_ = experiment_id;
595
596    for (std::vector<AutofillField*>::iterator field = form->fields_.begin();
597         field != form->fields_.end(); ++field) {
598      if (form->ShouldSkipField(**field))
599        continue;
600
601      // In some cases *successful* response does not return all the fields.
602      // Quit the update of the types then.
603      if (current_info == field_infos.end())
604        break;
605
606      // UNKNOWN_TYPE is reserved for use by the client.
607      DCHECK_NE(current_info->field_type, UNKNOWN_TYPE);
608
609      ServerFieldType heuristic_type = (*field)->heuristic_type();
610      if (heuristic_type != UNKNOWN_TYPE)
611        heuristics_detected_fillable_field = true;
612
613      (*field)->set_server_type(current_info->field_type);
614      if (heuristic_type != (*field)->Type().GetStorableType())
615        query_response_overrode_heuristics = true;
616
617      // Copy default value into the field if available.
618      if (!current_info->default_value.empty())
619        (*field)->set_default_value(current_info->default_value);
620
621      ++current_info;
622    }
623
624    form->UpdateAutofillCount();
625    form->IdentifySections(false);
626  }
627
628  AutofillMetrics::ServerQueryMetric metric;
629  if (query_response_overrode_heuristics) {
630    if (heuristics_detected_fillable_field) {
631      metric = AutofillMetrics::QUERY_RESPONSE_OVERRODE_LOCAL_HEURISTICS;
632    } else {
633      metric = AutofillMetrics::QUERY_RESPONSE_WITH_NO_LOCAL_HEURISTICS;
634    }
635  } else {
636    metric = AutofillMetrics::QUERY_RESPONSE_MATCHED_LOCAL_HEURISTICS;
637  }
638  metric_logger.LogServerQueryMetric(metric);
639}
640
641// static
642void FormStructure::GetFieldTypePredictions(
643    const std::vector<FormStructure*>& form_structures,
644    std::vector<FormDataPredictions>* forms) {
645  forms->clear();
646  forms->reserve(form_structures.size());
647  for (size_t i = 0; i < form_structures.size(); ++i) {
648    FormStructure* form_structure = form_structures[i];
649    FormDataPredictions form;
650    form.data.name = form_structure->form_name_;
651    form.data.method =
652        ASCIIToUTF16((form_structure->method_ == POST) ? "POST" : "GET");
653    form.data.origin = form_structure->source_url_;
654    form.data.action = form_structure->target_url_;
655    form.signature = form_structure->FormSignature();
656    form.experiment_id = form_structure->server_experiment_id_;
657
658    for (std::vector<AutofillField*>::const_iterator field =
659             form_structure->fields_.begin();
660         field != form_structure->fields_.end(); ++field) {
661      form.data.fields.push_back(FormFieldData(**field));
662
663      FormFieldDataPredictions annotated_field;
664      annotated_field.signature = (*field)->FieldSignature();
665      annotated_field.heuristic_type =
666          AutofillType((*field)->heuristic_type()).ToString();
667      annotated_field.server_type =
668          AutofillType((*field)->server_type()).ToString();
669      annotated_field.overall_type = (*field)->Type().ToString();
670      form.fields.push_back(annotated_field);
671    }
672
673    forms->push_back(form);
674  }
675}
676
677std::string FormStructure::FormSignature() const {
678  std::string scheme(target_url_.scheme());
679  std::string host(target_url_.host());
680
681  // If target host or scheme is empty, set scheme and host of source url.
682  // This is done to match the Toolbar's behavior.
683  if (scheme.empty() || host.empty()) {
684    scheme = source_url_.scheme();
685    host = source_url_.host();
686  }
687
688  std::string form_string = scheme + "://" + host + "&" +
689                            UTF16ToUTF8(form_name_) +
690                            form_signature_field_names_;
691
692  return Hash64Bit(form_string);
693}
694
695bool FormStructure::IsAutocheckoutEnabled() const {
696  return !autocheckout_url_prefix_.empty();
697}
698
699bool FormStructure::ShouldSkipField(const FormFieldData& field) const {
700  return (field.is_checkable || field.form_control_type == "password") &&
701      !IsAutocheckoutEnabled();
702}
703
704size_t FormStructure::RequiredFillableFields() const {
705  return IsAutocheckoutEnabled() ? 0 : kRequiredAutofillFields;
706}
707
708bool FormStructure::IsAutofillable(bool require_method_post) const {
709  if (autofill_count() < RequiredFillableFields())
710    return false;
711
712  return ShouldBeParsed(require_method_post);
713}
714
715void FormStructure::UpdateAutofillCount() {
716  autofill_count_ = 0;
717  for (std::vector<AutofillField*>::const_iterator iter = begin();
718       iter != end(); ++iter) {
719    AutofillField* field = *iter;
720    if (field && field->IsFieldFillable())
721      ++autofill_count_;
722  }
723}
724
725bool FormStructure::ShouldBeParsed(bool require_method_post) const {
726  if (active_field_count() < RequiredFillableFields())
727    return false;
728
729  // Rule out http(s)://*/search?...
730  //  e.g. http://www.google.com/search?q=...
731  //       http://search.yahoo.com/search?p=...
732  if (target_url_.path() == "/search")
733    return false;
734
735  if (!IsAutocheckoutEnabled()) {
736    // Make sure there is at least one text field when Autocheckout is
737    // not enabled.
738    bool has_text_field = false;
739    for (std::vector<AutofillField*>::const_iterator it = begin();
740         it != end() && !has_text_field; ++it) {
741      has_text_field |= (*it)->form_control_type != "select-one";
742    }
743    if (!has_text_field)
744      return false;
745  }
746
747  return !require_method_post || (method_ == POST);
748}
749
750bool FormStructure::ShouldBeCrowdsourced() const {
751  // Allow all forms in Autocheckout flow to be crowdsourced.
752  return (!has_author_specified_types_ && ShouldBeParsed(true)) ||
753      IsAutocheckoutEnabled();
754}
755
756void FormStructure::UpdateFromCache(const FormStructure& cached_form) {
757  // Map from field signatures to cached fields.
758  std::map<std::string, const AutofillField*> cached_fields;
759  for (size_t i = 0; i < cached_form.field_count(); ++i) {
760    const AutofillField* field = cached_form.field(i);
761    cached_fields[field->FieldSignature()] = field;
762  }
763
764  for (std::vector<AutofillField*>::const_iterator iter = begin();
765       iter != end(); ++iter) {
766    AutofillField* field = *iter;
767
768    std::map<std::string, const AutofillField*>::const_iterator
769        cached_field = cached_fields.find(field->FieldSignature());
770    if (cached_field != cached_fields.end()) {
771      if (field->form_control_type != "select-one" &&
772          field->value == cached_field->second->value) {
773        // From the perspective of learning user data, text fields containing
774        // default values are equivalent to empty fields.
775        field->value = base::string16();
776      }
777
778      field->set_heuristic_type(cached_field->second->heuristic_type());
779      field->set_server_type(cached_field->second->server_type());
780    }
781  }
782
783  UpdateAutofillCount();
784
785  filled_by_autocheckout_ = cached_form.filled_by_autocheckout();
786  server_experiment_id_ = cached_form.server_experiment_id();
787
788  // The form signature should match between query and upload requests to the
789  // server. On many websites, form elements are dynamically added, removed, or
790  // rearranged via JavaScript between page load and form submission, so we
791  // copy over the |form_signature_field_names_| corresponding to the query
792  // request.
793  DCHECK_EQ(cached_form.form_name_, form_name_);
794  DCHECK_EQ(cached_form.source_url_, source_url_);
795  DCHECK_EQ(cached_form.target_url_, target_url_);
796  form_signature_field_names_ = cached_form.form_signature_field_names_;
797}
798
799void FormStructure::LogQualityMetrics(
800    const AutofillMetrics& metric_logger,
801    const base::TimeTicks& load_time,
802    const base::TimeTicks& interaction_time,
803    const base::TimeTicks& submission_time) const {
804  std::string experiment_id = server_experiment_id();
805  metric_logger.LogServerExperimentIdForUpload(experiment_id);
806
807  size_t num_detected_field_types = 0;
808  bool did_autofill_all_possible_fields = true;
809  bool did_autofill_some_possible_fields = false;
810  for (size_t i = 0; i < field_count(); ++i) {
811    const AutofillField* field = this->field(i);
812    metric_logger.LogQualityMetric(AutofillMetrics::FIELD_SUBMITTED,
813                                   experiment_id);
814
815    // No further logging for empty fields nor for fields where the entered data
816    // does not appear to already exist in the user's stored Autofill data.
817    const ServerFieldTypeSet& field_types = field->possible_types();
818    DCHECK(!field_types.empty());
819    if (field_types.count(EMPTY_TYPE) || field_types.count(UNKNOWN_TYPE))
820      continue;
821
822    ++num_detected_field_types;
823    if (field->is_autofilled)
824      did_autofill_some_possible_fields = true;
825    else
826      did_autofill_all_possible_fields = false;
827
828    // Collapse field types that Chrome treats as identical, e.g. home and
829    // billing address fields.
830    ServerFieldTypeSet collapsed_field_types;
831    for (ServerFieldTypeSet::const_iterator it = field_types.begin();
832         it != field_types.end();
833         ++it) {
834      // Since we currently only support US phone numbers, the (city code + main
835      // digits) number is almost always identical to the whole phone number.
836      // TODO(isherman): Improve this logic once we add support for
837      // international numbers.
838      if (*it == PHONE_HOME_CITY_AND_NUMBER)
839        collapsed_field_types.insert(PHONE_HOME_WHOLE_NUMBER);
840      else
841        collapsed_field_types.insert(AutofillType(*it).GetStorableType());
842    }
843
844    // Capture the field's type, if it is unambiguous.
845    ServerFieldType field_type = UNKNOWN_TYPE;
846    if (collapsed_field_types.size() == 1)
847      field_type = *collapsed_field_types.begin();
848
849    ServerFieldType heuristic_type =
850        AutofillType(field->heuristic_type()).GetStorableType();
851    ServerFieldType server_type =
852        AutofillType(field->server_type()).GetStorableType();
853    ServerFieldType predicted_type = field->Type().GetStorableType();
854
855    // Log heuristic, server, and overall type quality metrics, independently of
856    // whether the field was autofilled.
857    if (heuristic_type == UNKNOWN_TYPE) {
858      metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_UNKNOWN,
859                                               field_type, experiment_id);
860    } else if (field_types.count(heuristic_type)) {
861      metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_MATCH,
862                                               field_type, experiment_id);
863    } else {
864      metric_logger.LogHeuristicTypePrediction(AutofillMetrics::TYPE_MISMATCH,
865                                               field_type, experiment_id);
866    }
867
868    if (server_type == NO_SERVER_DATA) {
869      metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_UNKNOWN,
870                                            field_type, experiment_id);
871    } else if (field_types.count(server_type)) {
872      metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_MATCH,
873                                            field_type, experiment_id);
874    } else {
875      metric_logger.LogServerTypePrediction(AutofillMetrics::TYPE_MISMATCH,
876                                            field_type, experiment_id);
877    }
878
879    if (predicted_type == UNKNOWN_TYPE) {
880      metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_UNKNOWN,
881                                             field_type, experiment_id);
882    } else if (field_types.count(predicted_type)) {
883      metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_MATCH,
884                                             field_type, experiment_id);
885    } else {
886      metric_logger.LogOverallTypePrediction(AutofillMetrics::TYPE_MISMATCH,
887                                             field_type, experiment_id);
888    }
889
890    // TODO(isherman): <select> fields don't support |is_autofilled()|, so we
891    // have to skip them for the remaining metrics.
892    if (field->form_control_type == "select-one")
893      continue;
894
895    if (field->is_autofilled) {
896      metric_logger.LogQualityMetric(AutofillMetrics::FIELD_AUTOFILLED,
897                                     experiment_id);
898    } else {
899      metric_logger.LogQualityMetric(AutofillMetrics::FIELD_NOT_AUTOFILLED,
900                                     experiment_id);
901
902      if (heuristic_type == UNKNOWN_TYPE) {
903        metric_logger.LogQualityMetric(
904            AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_UNKNOWN,
905            experiment_id);
906      } else if (field_types.count(heuristic_type)) {
907        metric_logger.LogQualityMetric(
908            AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MATCH,
909            experiment_id);
910      } else {
911        metric_logger.LogQualityMetric(
912            AutofillMetrics::NOT_AUTOFILLED_HEURISTIC_TYPE_MISMATCH,
913            experiment_id);
914      }
915
916      if (server_type == NO_SERVER_DATA) {
917        metric_logger.LogQualityMetric(
918            AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_UNKNOWN,
919            experiment_id);
920      } else if (field_types.count(server_type)) {
921        metric_logger.LogQualityMetric(
922            AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MATCH,
923            experiment_id);
924      } else {
925        metric_logger.LogQualityMetric(
926            AutofillMetrics::NOT_AUTOFILLED_SERVER_TYPE_MISMATCH,
927            experiment_id);
928      }
929    }
930  }
931
932  if (num_detected_field_types < RequiredFillableFields()) {
933    metric_logger.LogUserHappinessMetric(
934        AutofillMetrics::SUBMITTED_NON_FILLABLE_FORM);
935  } else {
936    if (did_autofill_all_possible_fields) {
937      metric_logger.LogUserHappinessMetric(
938          AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_ALL);
939    } else if (did_autofill_some_possible_fields) {
940      metric_logger.LogUserHappinessMetric(
941          AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_SOME);
942    } else {
943      metric_logger.LogUserHappinessMetric(
944          AutofillMetrics::SUBMITTED_FILLABLE_FORM_AUTOFILLED_NONE);
945    }
946
947    // Unlike the other times, the |submission_time| should always be available.
948    DCHECK(!submission_time.is_null());
949
950    // The |load_time| might be unset, in the case that the form was dynamically
951    // added to the DOM.
952    if (!load_time.is_null()) {
953      // Submission should always chronologically follow form load.
954      DCHECK(submission_time > load_time);
955      base::TimeDelta elapsed = submission_time - load_time;
956      if (did_autofill_some_possible_fields)
957        metric_logger.LogFormFillDurationFromLoadWithAutofill(elapsed);
958      else
959        metric_logger.LogFormFillDurationFromLoadWithoutAutofill(elapsed);
960    }
961
962    // The |interaction_time| might be unset, in the case that the user
963    // submitted a blank form.
964    if (!interaction_time.is_null()) {
965      // Submission should always chronologically follow interaction.
966      DCHECK(submission_time > interaction_time);
967      base::TimeDelta elapsed = submission_time - interaction_time;
968      if (did_autofill_some_possible_fields) {
969        metric_logger.LogFormFillDurationFromInteractionWithAutofill(elapsed);
970      } else {
971        metric_logger.LogFormFillDurationFromInteractionWithoutAutofill(
972            elapsed);
973      }
974    }
975  }
976}
977
978const AutofillField* FormStructure::field(size_t index) const {
979  if (index >= fields_.size()) {
980    NOTREACHED();
981    return NULL;
982  }
983
984  return fields_[index];
985}
986
987AutofillField* FormStructure::field(size_t index) {
988  return const_cast<AutofillField*>(
989      static_cast<const FormStructure*>(this)->field(index));
990}
991
992size_t FormStructure::field_count() const {
993  return fields_.size();
994}
995
996size_t FormStructure::active_field_count() const {
997  return active_field_count_;
998}
999
1000std::string FormStructure::server_experiment_id() const {
1001  return server_experiment_id_;
1002}
1003
1004FormData FormStructure::ToFormData() const {
1005  // |data.user_submitted| will always be false.
1006  FormData data;
1007  data.name = form_name_;
1008  data.origin = source_url_;
1009  data.action = target_url_;
1010  data.method = ASCIIToUTF16(method_ == POST ? "POST" : "GET");
1011
1012  for (size_t i = 0; i < fields_.size(); ++i) {
1013    data.fields.push_back(FormFieldData(*fields_[i]));
1014  }
1015
1016  return data;
1017}
1018
1019bool FormStructure::operator==(const FormData& form) const {
1020  // TODO(jhawkins): Is this enough to differentiate a form?
1021  if (form_name_ == form.name &&
1022      source_url_ == form.origin &&
1023      target_url_ == form.action) {
1024    return true;
1025  }
1026
1027  // TODO(jhawkins): Compare field names, IDs and labels once we have labels
1028  // set up.
1029
1030  return false;
1031}
1032
1033bool FormStructure::operator!=(const FormData& form) const {
1034  return !operator==(form);
1035}
1036
1037std::string FormStructure::Hash64Bit(const std::string& str) {
1038  std::string hash_bin = base::SHA1HashString(str);
1039  DCHECK_EQ(20U, hash_bin.length());
1040
1041  uint64 hash64 = (((static_cast<uint64>(hash_bin[0])) & 0xFF) << 56) |
1042                  (((static_cast<uint64>(hash_bin[1])) & 0xFF) << 48) |
1043                  (((static_cast<uint64>(hash_bin[2])) & 0xFF) << 40) |
1044                  (((static_cast<uint64>(hash_bin[3])) & 0xFF) << 32) |
1045                  (((static_cast<uint64>(hash_bin[4])) & 0xFF) << 24) |
1046                  (((static_cast<uint64>(hash_bin[5])) & 0xFF) << 16) |
1047                  (((static_cast<uint64>(hash_bin[6])) & 0xFF) << 8) |
1048                   ((static_cast<uint64>(hash_bin[7])) & 0xFF);
1049
1050  return base::Uint64ToString(hash64);
1051}
1052
1053bool FormStructure::EncodeFormRequest(
1054    FormStructure::EncodeRequestType request_type,
1055    buzz::XmlElement* encompassing_xml_element) const {
1056  if (!field_count())  // Nothing to add.
1057    return false;
1058
1059  // Some badly formatted web sites repeat fields - limit number of fields to
1060  // 48, which is far larger than any valid form and XML still fits into 2K.
1061  // Do not send requests for forms with more than this many fields, as they are
1062  // near certainly not valid/auto-fillable.
1063  const size_t kMaxFieldsOnTheForm = 48;
1064  if (field_count() > kMaxFieldsOnTheForm)
1065    return false;
1066
1067  // Add the child nodes for the form fields.
1068  for (size_t index = 0; index < field_count(); ++index) {
1069    const AutofillField* field = fields_[index];
1070    switch (request_type) {
1071      case FormStructure::UPLOAD:
1072        EncodeFieldForUpload(*field, encompassing_xml_element);
1073        break;
1074      case FormStructure::QUERY:
1075        if (ShouldSkipField(*field))
1076          continue;
1077        EncodeFieldForQuery(*field, encompassing_xml_element);
1078        break;
1079      case FormStructure::FIELD_ASSIGNMENTS:
1080        EncodeFieldForFieldAssignments(*field, encompassing_xml_element);
1081        break;
1082    }
1083  }
1084  return true;
1085}
1086
1087void FormStructure::ParseFieldTypesFromAutocompleteAttributes(
1088    bool* found_types,
1089    bool* found_sections) {
1090  const std::string kDefaultSection = "-default";
1091
1092  *found_types = false;
1093  *found_sections = false;
1094  for (std::vector<AutofillField*>::iterator it = fields_.begin();
1095       it != fields_.end(); ++it) {
1096    AutofillField* field = *it;
1097
1098    // To prevent potential section name collisions, add a default suffix for
1099    // other fields.  Without this, 'autocomplete' attribute values
1100    // "section--shipping street-address" and "shipping street-address" would be
1101    // parsed identically, given the section handling code below.  We do this
1102    // before any validation so that fields with invalid attributes still end up
1103    // in the default section.  These default section names will be overridden
1104    // by subsequent heuristic parsing steps if there are no author-specified
1105    // section names.
1106    field->set_section(kDefaultSection);
1107
1108    // Canonicalize the attribute value by trimming whitespace, collapsing
1109    // non-space characters (e.g. tab) to spaces, and converting to lowercase.
1110    std::string autocomplete_attribute =
1111        CollapseWhitespaceASCII(field->autocomplete_attribute, false);
1112    autocomplete_attribute = StringToLowerASCII(autocomplete_attribute);
1113
1114    // The autocomplete attribute is overloaded: it can specify either a field
1115    // type hint or whether autocomplete should be enabled at all.  Ignore the
1116    // latter type of attribute value.
1117    if (autocomplete_attribute.empty() ||
1118        autocomplete_attribute == "on" ||
1119        autocomplete_attribute == "off") {
1120      continue;
1121    }
1122
1123    // Any other value, even it is invalid, is considered to be a type hint.
1124    // This allows a website's author to specify an attribute like
1125    // autocomplete="other" on a field to disable all Autofill heuristics for
1126    // the form.
1127    *found_types = true;
1128
1129    // Tokenize the attribute value.  Per the spec, the tokens are parsed in
1130    // reverse order.
1131    std::vector<std::string> tokens;
1132    Tokenize(autocomplete_attribute, " ", &tokens);
1133
1134    // The final token must be the field type.
1135    // If it is not one of the known types, abort.
1136    DCHECK(!tokens.empty());
1137    std::string field_type_token = tokens.back();
1138    tokens.pop_back();
1139    HtmlFieldType field_type =
1140        FieldTypeFromAutocompleteAttributeValue(field_type_token, *field);
1141    if (field_type == HTML_TYPE_UNKNOWN)
1142      continue;
1143
1144    // The preceding token, if any, may be a type hint.
1145    if (!tokens.empty() && IsContactTypeHint(tokens.back())) {
1146      // If it is, it must match the field type; otherwise, abort.
1147      // Note that an invalid token invalidates the entire attribute value, even
1148      // if the other tokens are valid.
1149      if (!ContactTypeHintMatchesFieldType(tokens.back(), field_type))
1150        continue;
1151
1152      // Chrome Autofill ignores these type hints.
1153      tokens.pop_back();
1154    }
1155
1156    // The preceding token, if any, may be a fixed string that is either
1157    // "shipping" or "billing".  Chrome Autofill treats these as implicit
1158    // section name suffixes.
1159    DCHECK_EQ(kDefaultSection, field->section());
1160    std::string section = field->section();
1161    HtmlFieldMode mode = HTML_MODE_NONE;
1162    if (!tokens.empty()) {
1163      if (tokens.back() == kShippingMode)
1164        mode = HTML_MODE_SHIPPING;
1165      else if (tokens.back() == kBillingMode)
1166        mode = HTML_MODE_BILLING;
1167    }
1168
1169    if (mode != HTML_MODE_NONE) {
1170      section = "-" + tokens.back();
1171      tokens.pop_back();
1172    }
1173
1174    // The preceding token, if any, may be a named section.
1175    const std::string kSectionPrefix = "section-";
1176    if (!tokens.empty() &&
1177        StartsWithASCII(tokens.back(), kSectionPrefix, true)) {
1178      // Prepend this section name to the suffix set in the preceding block.
1179      section = tokens.back().substr(kSectionPrefix.size()) + section;
1180      tokens.pop_back();
1181    }
1182
1183    // No other tokens are allowed.  If there are any remaining, abort.
1184    if (!tokens.empty())
1185      continue;
1186
1187    if (section != kDefaultSection) {
1188      *found_sections = true;
1189      field->set_section(section);
1190    }
1191
1192    // No errors encountered while parsing!
1193    // Update the |field|'s type based on what was parsed from the attribute.
1194    field->SetHtmlType(field_type, mode);
1195  }
1196}
1197
1198void FormStructure::IdentifySections(bool has_author_specified_sections) {
1199  if (fields_.empty())
1200    return;
1201
1202  if (!has_author_specified_sections) {
1203    // Name sections after the first field in the section.
1204    base::string16 current_section = fields_.front()->unique_name();
1205
1206    // Keep track of the types we've seen in this section.
1207    std::set<ServerFieldType> seen_types;
1208    ServerFieldType previous_type = UNKNOWN_TYPE;
1209
1210    for (std::vector<AutofillField*>::iterator field = fields_.begin();
1211         field != fields_.end(); ++field) {
1212      const ServerFieldType current_type = (*field)->Type().GetStorableType();
1213
1214      bool already_saw_current_type = seen_types.count(current_type) > 0;
1215
1216      // Forms often ask for multiple phone numbers -- e.g. both a daytime and
1217      // evening phone number.  Our phone number detection is also generally a
1218      // little off.  Hence, ignore this field type as a signal here.
1219      if (AutofillType(current_type).group() == PHONE_HOME)
1220        already_saw_current_type = false;
1221
1222      // Some forms have adjacent fields of the same type.  Two common examples:
1223      //  * Forms with two email fields, where the second is meant to "confirm"
1224      //    the first.
1225      //  * Forms with a <select> menu for states in some countries, and a
1226      //    freeform <input> field for states in other countries.  (Usually,
1227      //    only one of these two will be visible for any given choice of
1228      //    country.)
1229      // Generally, adjacent fields of the same type belong in the same logical
1230      // section.
1231      if (current_type == previous_type)
1232        already_saw_current_type = false;
1233
1234      previous_type = current_type;
1235
1236      if (current_type != UNKNOWN_TYPE && already_saw_current_type) {
1237        // We reached the end of a section, so start a new section.
1238        seen_types.clear();
1239        current_section = (*field)->unique_name();
1240      }
1241
1242      seen_types.insert(current_type);
1243      (*field)->set_section(UTF16ToUTF8(current_section));
1244    }
1245  }
1246
1247  // Ensure that credit card and address fields are in separate sections.
1248  // This simplifies the section-aware logic in autofill_manager.cc.
1249  for (std::vector<AutofillField*>::iterator field = fields_.begin();
1250       field != fields_.end(); ++field) {
1251    FieldTypeGroup field_type_group = (*field)->Type().group();
1252    if (field_type_group == CREDIT_CARD)
1253      (*field)->set_section((*field)->section() + "-cc");
1254    else
1255      (*field)->set_section((*field)->section() + "-default");
1256  }
1257}
1258
1259}  // namespace autofill
1260