Consolidate Autofill possible type detection code, and enforce greater match precision.

* Only consider submitted field values to match locally stored data if the strings match exactly
  + This means we will send fewer false positives to the server, at the cost of some false negatives.  Most importantly, we should stop identifying ADDRESS_LINE_2 fields as ADDRESS_LINE_1.
  + There are a few excpetions for structured fields with canonicalized values, like countries (which are canonicalized to country codes)
* Eliminate FormGroup::GetMatchingTypes() -- all of this work is now done in a consistent way within AutofillManager
* Refactor FormGroup::GetNonEmptyTypes() to be shared code.

BUG=81867, 76992
TEST=unit_tests --gtest_filter=AutofillManagerTest.DeterminePossibleFieldTypesForUpload


Review URL: http://codereview.chromium.org/7398019

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@93582 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 25 insertions, 148 deletions

diff --git a/chrome/browser/autofill/address.cc b/chrome/browser/autofill/address.cc
index 0a687a1..6b2d345 100644
--- a/chrome/browser/autofill/address.cc
+++ b/chrome/browser/autofill/address.cc
@@ -28,6 +28,13 @@ const AutofillType::FieldTypeSubGroup kAutofillAddressTypes[] = {
 
 const int kAutofillAddressLength = arraysize(kAutofillAddressTypes);
 
+// Returns the country code corresponding to |country|, which should be a
+// localized country name.
+std::string ToCountryCode(const string16& country) {
+  std::string app_locale = AutofillCountry::ApplicationLocale();
+  return AutofillCountry::GetCountryCode(country, app_locale);
+}
+
 }  // namespace
 
 Address::Address() {}
@@ -42,8 +49,6 @@ Address& Address::operator=(const Address& address) {
   if (this == &address)
     return *this;
 
-  line1_tokens_ = address.line1_tokens_;
-  line2_tokens_= address.line2_tokens_;
   line1_ = address.line1_;
   line2_ = address.line2_;
   city_ = address.city_;
@@ -53,54 +58,13 @@ Address& Address::operator=(const Address& address) {
   return *this;
 }
 
-void Address::GetMatchingTypes(const string16& text,
-                               FieldTypeSet* matching_types) const {
-  DCHECK(matching_types);
-
-  // If the text to match against the field types is empty, then no results will
-  // match.
-  if (text.empty())
-    return;
-
-  if (IsLine1(text))
-    matching_types->insert(ADDRESS_HOME_LINE1);
-
-  if (IsLine2(text))
-    matching_types->insert(ADDRESS_HOME_LINE2);
-
-  if (IsCity(text))
-    matching_types->insert(ADDRESS_HOME_CITY);
-
-  if (IsState(text))
-    matching_types->insert(ADDRESS_HOME_STATE);
-
-  if (IsZipCode(text))
-    matching_types->insert(ADDRESS_HOME_ZIP);
-
-  if (IsCountry(text))
-    matching_types->insert(ADDRESS_HOME_COUNTRY);
-}
-
-void Address::GetNonEmptyTypes(FieldTypeSet* non_empty_types) const {
-  DCHECK(non_empty_types);
-
-  if (!line1_.empty())
-    non_empty_types->insert(ADDRESS_HOME_LINE1);
-
-  if (!line2_.empty())
-    non_empty_types->insert(ADDRESS_HOME_LINE2);
-
-  if (!city_.empty())
-    non_empty_types->insert(ADDRESS_HOME_CITY);
-
-  if (!state_.empty())
-    non_empty_types->insert(ADDRESS_HOME_STATE);
-
-  if (!zip_code_.empty())
-    non_empty_types->insert(ADDRESS_HOME_ZIP);
-
-  if (!country_code_.empty())
-    non_empty_types->insert(ADDRESS_HOME_COUNTRY);
+void Address::GetSupportedTypes(FieldTypeSet* supported_types) const {
+  supported_types->insert(ADDRESS_HOME_LINE1);
+  supported_types->insert(ADDRESS_HOME_LINE2);
+  supported_types->insert(ADDRESS_HOME_CITY);
+  supported_types->insert(ADDRESS_HOME_STATE);
+  supported_types->insert(ADDRESS_HOME_ZIP);
+  supported_types->insert(ADDRESS_HOME_COUNTRY);
 }
 
 string16 Address::GetInfo(AutofillFieldType type) const {
@@ -128,30 +92,29 @@ string16 Address::GetInfo(AutofillFieldType type) const {
 void Address::SetInfo(AutofillFieldType type, const string16& value) {
   FieldTypeSubGroup subgroup = AutofillType(type).subgroup();
   if (subgroup == AutofillType::ADDRESS_LINE1)
-    set_line1(value);
+    line1_ = value;
   else if (subgroup == AutofillType::ADDRESS_LINE2)
-    set_line2(value);
+    line2_ = value;
   else if (subgroup == AutofillType::ADDRESS_CITY)
     city_ = value;
   else if (subgroup == AutofillType::ADDRESS_STATE)
     state_ = value;
   else if (subgroup == AutofillType::ADDRESS_COUNTRY)
-    SetCountry(value);
+    country_code_ = ToCountryCode(value);
   else if (subgroup == AutofillType::ADDRESS_ZIP)
     zip_code_ = value;
   else
     NOTREACHED();
 }
 
-void Address::Clear() {
-  line1_tokens_.clear();
-  line1_.clear();
-  line2_tokens_.clear();
-  line2_.clear();
-  city_.clear();
-  state_.clear();
-  country_code_.clear();
-  zip_code_.clear();
+void Address::GetMatchingTypes(const string16& text,
+                               FieldTypeSet* matching_types) const {
+  FormGroup::GetMatchingTypes(text, matching_types);
+
+  // Check to see if the |text| canonicalized as a country name is a match.
+  std::string country_code = ToCountryCode(text);
+  if (!country_code.empty() && country_code_ == country_code)
+    matching_types->insert(ADDRESS_HOME_COUNTRY);
 }
 
 string16 Address::Country() const {
@@ -161,89 +124,3 @@ string16 Address::Country() const {
   std::string app_locale = AutofillCountry::ApplicationLocale();
   return AutofillCountry(country_code(), app_locale).name();
 }
-
-void Address::set_line1(const string16& line1) {
-  line1_ = line1;
-  line1_tokens_.clear();
-  Tokenize(line1, kAddressSplitChars, &line1_tokens_);
-  LineTokens::iterator iter;
-  for (iter = line1_tokens_.begin(); iter != line1_tokens_.end(); ++iter)
-    *iter = StringToLowerASCII(*iter);
-}
-
-void Address::set_line2(const string16& line2) {
-  line2_ = line2;
-  line2_tokens_.clear();
-  Tokenize(line2, kAddressSplitChars, &line2_tokens_);
-  LineTokens::iterator iter;
-  for (iter = line2_tokens_.begin(); iter != line2_tokens_.end(); ++iter)
-    *iter = StringToLowerASCII(*iter);
-}
-
-void Address::SetCountry(const string16& country) {
-  std::string app_locale = AutofillCountry::ApplicationLocale();
-  country_code_ = AutofillCountry::GetCountryCode(country, app_locale);
-}
-
-bool Address::IsLine1(const string16& text) const {
-  return IsLineMatch(text, line1_tokens_);
-}
-
-bool Address::IsLine2(const string16& text) const {
-  return IsLineMatch(text, line2_tokens_);
-}
-
-bool Address::IsCity(const string16& text) const {
-  return (StringToLowerASCII(city_) == StringToLowerASCII(text));
-}
-
-bool Address::IsState(const string16& text) const {
-  return (StringToLowerASCII(state_) == StringToLowerASCII(text));
-}
-
-bool Address::IsCountry(const string16& text) const {
-  std::string app_locale = AutofillCountry::ApplicationLocale();
-  std::string country_code = AutofillCountry::GetCountryCode(text, app_locale);
-  return (!country_code.empty() && country_code_ == country_code);
-}
-
-bool Address::IsZipCode(const string16& text) const {
-  return zip_code_ == text;
-}
-
-bool Address::IsLineMatch(const string16& text,
-                          const LineTokens& line_tokens) const {
-  size_t line_tokens_size = line_tokens.size();
-  if (line_tokens_size == 0)
-    return false;
-
-  LineTokens text_tokens;
-  Tokenize(text, kAddressSplitChars, &text_tokens);
-  size_t text_tokens_size = text_tokens.size();
-  if (text_tokens_size == 0)
-    return false;
-
-  if (text_tokens_size > line_tokens_size)
-    return false;
-
-  // If each of the 'words' contained in the text are also present in the line,
-  // then we will consider the text to match the line.
-  LineTokens::iterator iter;
-  for (iter = text_tokens.begin(); iter != text_tokens.end(); ++iter) {
-    if (!IsWordInLine(*iter, line_tokens))
-      return false;
-  }
-
-  return true;
-}
-
-bool Address::IsWordInLine(const string16& word,
-                           const LineTokens& line_tokens) const {
-  LineTokens::const_iterator iter;
-  for (iter = line_tokens.begin(); iter != line_tokens.end(); ++iter) {
-    if (StringToLowerASCII(word) == *iter)
-      return true;
-  }
-
-  return false;
-}