[Autofill][Leizig] Add parsing for house number fields.

Change-Id: Ied46d737413cfb5fcd1eae5995c2bea8c1be50d4
Bug: 978014, 1099202
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2282752
Auto-Submit: Matthias Körber <koerber@google.com>
Commit-Queue: Christoph Schwering <schwering@google.com>
Reviewed-by: Christoph Schwering <schwering@google.com>
Cr-Commit-Position: refs/heads/master@{#804899}

chrome/browser/autofill/form_structure_browsertest.cc

@@ -152,6 +152,8 @@ FormStructureBrowserTest::FormStructureBrowserTest()
       {
           // TODO(crbug.com/1098943): Remove once experiment is over.
           autofill::features::kAutofillEnableSupportForMoreStructureInNames,
+          // TODO(crbug.com/1125978): Remove once launched.
+          autofill::features::kAutofillEnableSupportForHouseNumbers,
       },
       // Disabled
       {autofill::features::kAutofillEnforceMinRequiredFieldsForHeuristics,

components/autofill/core/browser/form_parsing/address_field.cc

@@ -77,7 +77,7 @@ std::unique_ptr<FormField> AddressField::Parse(AutofillScanner* scanner,
                                    MATCH_DEFAULT | MATCH_TEXT_AREA, nullptr,
                                    {log_manager, "kEmailRe"})) {
       continue;
-    } else if (address_field->ParseAddressLines(scanner) ||
+    } else if (address_field->ParseAddress(scanner) ||
                (!is_enabled_merged_city_state_country_zip &&
                 (address_field->ParseCityStateZipCode(scanner) ||
                  address_field->ParseCountry(scanner))) ||
@@ -133,17 +133,7 @@ std::unique_ptr<FormField> AddressField::Parse(AutofillScanner* scanner,
 }
 
 AddressField::AddressField(LogManager* log_manager)
-    : log_manager_(log_manager),
-      company_(nullptr),
-      address1_(nullptr),
-      address2_(nullptr),
-      address3_(nullptr),
-      street_address_(nullptr),
-      city_(nullptr),
-      state_(nullptr),
-      zip_(nullptr),
-      zip4_(nullptr),
-      country_(nullptr) {}
+    : log_manager_(log_manager) {}
 
 void AddressField::AddClassifications(
     FieldCandidatesMap* field_candidates) const {
@@ -172,6 +162,10 @@ void AddressField::AddClassifications(
                     field_candidates);
   AddClassification(country_, ADDRESS_HOME_COUNTRY, kBaseAddressParserScore,
                     field_candidates);
+  AddClassification(house_number_, ADDRESS_HOME_HOUSE_NUMBER,
+                    kBaseAddressParserScore, field_candidates);
+  AddClassification(street_name_, ADDRESS_HOME_STREET_NAME,
+                    kBaseAddressParserScore, field_candidates);
 }
 
 bool AddressField::ParseCompany(AutofillScanner* scanner) {
@@ -182,6 +176,53 @@ bool AddressField::ParseCompany(AutofillScanner* scanner) {
                     {log_manager_, "kCompanyRe"});
 }
 
+bool AddressField::ParseAddressFieldSequence(AutofillScanner* scanner) {
+  // Search for a sequence of a street name field followed by a house number
+  // field. Only if both are found in an abitrary order, the parsing is
+  // considered successful.
+
+  // TODO(crbug.com/1125978): Remove once launched.
+  if (!base::FeatureList::IsEnabled(
+          features::kAutofillEnableSupportForHouseNumbers)) {
+    return false;
+  }
+
+  const size_t cursor_position = scanner->CursorPosition();
+
+  while (!scanner->IsEnd()) {
+    if (!street_name_ &&
+        ParseFieldSpecifics(scanner, UTF8ToUTF16(kStreetNameRe), MATCH_DEFAULT,
+                            &street_name_, {log_manager_, "kStreetNameRe"})) {
+      continue;
+    }
+    if (!house_number_ &&
+        ParseFieldSpecifics(scanner, UTF8ToUTF16(kHouseNumberRe), MATCH_DEFAULT,
+                            &house_number_, {log_manager_, "kHouseNumberRe"})) {
+      continue;
+    }
+
+    break;
+  }
+
+  if (street_name_ && house_number_)
+    return true;
+
+  // Reset both fields in case one of them was found.
+  if (street_name_ || house_number_) {
+    street_name_ = nullptr;
+    house_number_ = nullptr;
+  }
+  scanner->RewindTo(cursor_position);
+  return false;
+}
+
+bool AddressField::ParseAddress(AutofillScanner* scanner) {
+  if (street_name_ && house_number_) {
+    return false;
+  }
+  return ParseAddressFieldSequence(scanner) || ParseAddressLines(scanner);
+}
+
 bool AddressField::ParseAddressLines(AutofillScanner* scanner) {
   // We only match the string "address" in page text, not in element names,
   // because sometimes every element in a group of address fields will have

components/autofill/core/browser/form_parsing/address_field.h

@@ -53,6 +53,8 @@ class AddressField : public FormField {
   explicit AddressField(LogManager* log_manager);
 
   bool ParseCompany(AutofillScanner* scanner);
+  bool ParseAddress(AutofillScanner* scanner);
+  bool ParseAddressFieldSequence(AutofillScanner* scanner);
   bool ParseAddressLines(AutofillScanner* scanner);
   bool ParseCountry(AutofillScanner* scanner);
   bool ParseZipCode(AutofillScanner* scanner);
@@ -91,16 +93,18 @@ class AddressField : public FormField {
   ParseNameLabelResult ParseNameAndLabelForState(AutofillScanner* scanner);
 
   LogManager* log_manager_;
-  AutofillField* company_;
-  AutofillField* address1_;
-  AutofillField* address2_;
-  AutofillField* address3_;
-  AutofillField* street_address_;
-  AutofillField* city_;
-  AutofillField* state_;
-  AutofillField* zip_;
-  AutofillField* zip4_;  // optional ZIP+4; we don't fill this yet.
-  AutofillField* country_;
+  AutofillField* company_ = nullptr;
+  AutofillField* street_name_ = nullptr;
+  AutofillField* house_number_ = nullptr;
+  AutofillField* address1_ = nullptr;
+  AutofillField* address2_ = nullptr;
+  AutofillField* address3_ = nullptr;
+  AutofillField* street_address_ = nullptr;
+  AutofillField* city_ = nullptr;
+  AutofillField* state_ = nullptr;
+  AutofillField* zip_ = nullptr;
+  AutofillField* zip4_ = nullptr;  // optional ZIP+4; we don't fill this yet.
+  AutofillField* country_ = nullptr;
 
   DISALLOW_COPY_AND_ASSIGN(AddressField);
 };

components/autofill/core/browser/form_parsing/autofill_scanner.cc

@@ -59,6 +59,10 @@ size_t AutofillScanner::SaveCursor() {
   return static_cast<size_t>(cursor_ - begin_);
 }
 
+size_t AutofillScanner::CursorPosition() {
+  return static_cast<size_t>(cursor_ - begin_);
+}
+
 void AutofillScanner::Init(const std::vector<AutofillField*>& fields) {
   cursor_ = fields.begin();
   saved_cursor_ = fields.begin();

components/autofill/core/browser/form_parsing/autofill_scanner.h

@@ -45,6 +45,9 @@ class AutofillScanner {
   // |RewindTo()|.
   size_t SaveCursor();
 
+  // Returns the current cursor position.
+  size_t CursorPosition();
+
  private:
   void Init(const std::vector<AutofillField*>& fields);
 

components/autofill/core/common/autofill_features.cc

@@ -103,6 +103,11 @@ const base::Feature kAutofillEnableSupportForMergingSubsetNames{
     "AutofillEnableSupportForMergingSubsetNames",
     base::FEATURE_DISABLED_BY_DEFAULT};
 
+// Enables filling support for separate house number fields.
+// TODO(crbug.com/1125978): Remove once launched.
+const base::Feature kAutofillEnableSupportForHouseNumbers{
+    "AutofillEnableSupportForHouseNumbers", base::FEATURE_DISABLED_BY_DEFAULT};
+
 // Controls whether or not a minimum number of fields is required before
 // heuristic field type prediction is run for a form.
 const base::Feature kAutofillEnforceMinRequiredFieldsForHeuristics{

components/autofill/core/common/autofill_features.h

@@ -36,6 +36,7 @@ extern const base::Feature kAutofillEnableCompanyName;
 extern const base::Feature kAutofillEnableHideSuggestionsUI;
 extern const base::Feature kAutofillEnableSupportForMoreStructureInNames;
 extern const base::Feature kAutofillEnableSupportForMergingSubsetNames;
+extern const base::Feature kAutofillEnableSupportForHouseNumbers;
 extern const base::Feature kAutofillEnforceMinRequiredFieldsForHeuristics;
 extern const base::Feature kAutofillEnforceMinRequiredFieldsForQuery;
 extern const base::Feature kAutofillEnforceMinRequiredFieldsForUpload;

components/autofill/core/common/autofill_regex_constants.cc

@@ -30,6 +30,11 @@ const char kCompanyRe[] =
     "|单位|公司"                 // zh-CN
     "|شرکت"                      // fa
     "|회사|직장";                // ko-KR
+const char kStreetNameRe[] =
+    "stra(ss|ß)e"    // de
+    "|street"        // en
+    "|rua|avenida";  // br
+const char kHouseNumberRe[] = "(house |^)number|(haus|^)nummer|^número$";
 const char kAddressLine1Re[] =
     "^address$|address[_-]?line(one)?|address1|addr1|street"
     "|(?:shipping|billing)address$"

components/autofill/core/common/autofill_regex_constants.h

@@ -11,6 +11,8 @@ extern const char kAttentionIgnoredRe[];
 extern const char kRegionIgnoredRe[];
 extern const char kAddressNameIgnoredRe[];
 extern const char kCompanyRe[];
+extern const char kHouseNumberRe[];
+extern const char kStreetNameRe[];
 extern const char kAddressLine1Re[];
 extern const char kAddressLine1LabelRe[];
 extern const char kAddressLine2Re[];

components/test/data/autofill/heuristics/input/169_dominos_de.html

+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="UTF-8">
+    <title></title>
+  </head>
+  <body>
+    <form id="delivery-details" method="POST">
+        <label id="customer-postcode-label" class="form-label" for="customer-postcode">Postleitzahl</label>
+        <input value="" placeholder="" id="customer-postcode" type="text" class="form-control" name="Customer.Postcode">
+        <label id="customer-street-name-label" for="customer-street-name" class="form-label">
+            Straßenname
+        </label>
+        <input class="form-control" value="" id="customer-street-name" type="text" name="Customer.Street" >
+        <label id="customer-street-no-label" for="customer-street-no" class="form-label">
+            Hausnummer
+        </label>
+        <input value="" id="customer-street-no" class="form-control" type="text" name="Customer.StreetNo" >
+        <input type="checkbox" id="at_remember-my-details" name="Customer.RememberMyDetails" value="true">
+        <label class="form-label checkbox" id="remember-my-details-label" for="at_remember-my-details">Filialauswahl merken</label>
+    </form>
+  </body>
+</html>

components/test/data/autofill/heuristics/output/169_dominos_de.out

+ADDRESS_HOME_ZIP | Customer.Postcode | Postleitzahl |  | Customer.Postcode_1-default
+ADDRESS_HOME_STREET_NAME | Customer.Street | Straßenname |  | Customer.Postcode_1-default
+ADDRESS_HOME_HOUSE_NUMBER | Customer.StreetNo | Hausnummer |  | Customer.Postcode_1-default
+UNKNOWN_TYPE | Customer.RememberMyDetails | Filialauswahl merken | true | Customer.Postcode_1-default