[Autofill] Adjust regexes for address components parsing (RU & BR)

- Add Russian keywords for street name and house number. - Add additional Russian/Brazilian keywords for city. - Adjust Brazilian street/addr1 keywords to exclude constructions like identificação do endereço, tipo de endereço, nome do endereço etc. - Adjust Brazilian house number keyword to match labels with * (required) and "número da residência". Change-Id: Ib48a921271cacbf30eae21e95d6c82dabe812218 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2468000Reviewed-by: Matthias Körber <koerber@google.com> Reviewed-by: David Roger <droger@chromium.org> Reviewed-by: Christoph Schwering <schwering@google.com> Commit-Queue: David Roger <droger@chromium.org> Cr-Commit-Position: refs/heads/master@{#821134}

[Autofill] Adjust regexes for address components parsing (RU & BR)
- Add Russian keywords for street name and house number. - Add additional Russian/Brazilian keywords for city. - Adjust Brazilian street/addr1 keywords to exclude constructions like identificação do endereço, tipo de endereço, nome do endereço etc. - Adjust Brazilian house number keyword to match labels with * (required) and "número da residência". Change-Id: Ib48a921271cacbf30eae21e95d6c82dabe812218 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2468000Reviewed-by: Matthias Körber <koerber@google.com> Reviewed-by: David Roger <droger@chromium.org> Reviewed-by: Christoph Schwering <schwering@google.com> Commit-Queue: David Roger <droger@chromium.org> Cr-Commit-Position: refs/heads/master@{#821134}
6c4c70b9 · Elizabeth Popova · Commit Bot · b3383cdc · 6c4c70b9 · 6c4c70b9
Commit 6c4c70b9 authored Oct 27, 2020 by Elizabeth Popova Committed by Commit Bot Oct 27, 2020
11 changed files
--- a/components/autofill/core/common/autofill_regex_constants.cc
+++ b/components/autofill/core/common/autofill_regex_constants.cc
@@ -31,10 +31,16 @@ const char kCompanyRe[] =
    "|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$";
+    "stra(ss|ß)e"               // de
+    "|street"                   // en
+    "|улица|название.?улицы"    // ru
+    "|rua|avenida"              // pt-PT, pt-BR
+    "|((?<!do |de )endereço)";  // pt-BR
+const char kHouseNumberRe[] =
+    "(house.?|street.?|^)number"              // en
+    "|(haus|^)nummer"                         // de
+    "|^\\*?.?número(.?\\*?$| da residência)"  // pt-BR, pt-PT
+    "|дом|номер.?дома";                       // ru
 const char kAddressLine1Re[] =
    "^address$|address[_-]?line(one)?|address1|addr1|street"
    "|(?:shipping|billing)address$"
@@ -44,7 +50,7 @@ const char kAddressLine1Re[] =
    "|adresse"                                         // fr-FR
    "|indirizzo"                                       // it-IT
    "|^住所$|住所1"                                    // ja-JP
-    "|morada|((?<!identificação do )endereço)"         // pt-BR, pt-PT
+    "|morada|((?<!do |de )endereço)"                   // pt-BR, pt-PT
    "|Адрес"                                           // ru
    "|地址"                                            // zh-CN
    "|(\\b|_)adres(?! (başlığı(nız)?|tarifi))(\\b|_)"  // tr
@@ -121,8 +127,8 @@ const char kCityRe[] =
    "|ville|commune"                                    // fr-FR
    "|localita"                                         // it-IT
    "|市区町村"                                         // ja-JP
-    "|cidade"                                           // pt-BR, pt-PT
-    "|Город"                                            // ru
+    "|cidade|município"                                 // pt-BR, pt-PT
+    "|Город|Населённый.?пункт"                          // ru
    "|市"                                               // zh-CN
    "|分區"                                             // zh-TW
    "|شهر"                                              // fa

--- a/components/test/data/autofill/heuristics/input/170_i18n_ru_structured_address_1.html
+++ b/components/test/data/autofill/heuristics/input/170_i18n_ru_structured_address_1.html
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="UTF-8">
+    <title></title>
+  </head>
+  <body>
+    <form action="http://www.google.com/" method="post">
+      <label for="fn">Имя:</label> <input type="text" id="fn"><br>
+      <label for="ln">Фамилия:</label> <input type="text" id="ln"><br>
+      <label for="cm">Название компании:</label> <input type="text" id="cm"><br>
+      <label for="hn">Дом:</label> <input type="text" id="hn"><br>
+      <label for="sn">Улица:</label> <input type="text" id="sn"><br>
+      <label for="ct">Город:</label> <input type="text" id="ct"><br>
+      <label for="zc">Почтовый индекс:</label> <input type="text" id="zc"><br>
+      <label for="st">Область:</label> <input type="text" id="st"><br>
+      <label for="em">Адрес электронной почты:</label> <input type="text" id="em"><br>
+      <label for="ph">Телефон:</label> <input type="text" id="ph"><br>
+      <label for="c1">Имя владельца карты:</label> <input type="text" id="c1"><br>
+      <label for="c2">Номер кредитной карты:</label> <input type="text" id="c2"><br>
+      <label for="c3">Срок действия карты:</label> <input type="text" id="c3"><br>
+      <label for="c4">Срок действия карты:</label> <input type="text" id="c4"><br>
+    </form>
+  </body>
+</html>
\ No newline at end of file
--- a/components/test/data/autofill/heuristics/input/171_i18n_ru_structured_address_2.html
+++ b/components/test/data/autofill/heuristics/input/171_i18n_ru_structured_address_2.html
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="UTF-8">
+    <title></title>
+  </head>
+  <body>
+    <form action="http://www.google.com/" method="post">
+      <label for="fn">Имя:</label> <input type="text" id="fn"><br>
+      <label for="ln">Фамилия:</label> <input type="text" id="ln"><br>
+      <label for="cm">Название компании:</label> <input type="text" id="cm"><br>
+      <label for="sn">Название улицы:</label> <input type="text" id="sn"><br>
+      <label for="hn">Номер дома:</label> <input type="text" id="hn"><br>
+      <label for="ct">Населённый пункт:</label> <input type="text" id="ct"><br>
+      <label for="zc">Почтовый индекс:</label> <input type="text" id="zc"><br>
+      <label for="st">Область:</label> <input type="text" id="st"><br>
+      <label for="em">Адрес электронной почты:</label> <input type="text" id="em"><br>
+      <label for="ph">Телефон:</label> <input type="text" id="ph"><br>
+      <label for="c1">Имя владельца карты:</label> <input type="text" id="c1"><br>
+      <label for="c2">Номер кредитной карты:</label> <input type="text" id="c2"><br>
+      <label for="c3">Срок действия карты:</label> <input type="text" id="c3"><br>
+      <label for="c4">Срок действия карты:</label> <input type="text" id="c4"><br>
+    </form>
+  </body>
+</html>
\ No newline at end of file
--- a/components/test/data/autofill/heuristics/input/172_i18n_pt_structured_address_1.html
+++ b/components/test/data/autofill/heuristics/input/172_i18n_pt_structured_address_1.html
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="UTF-8">
+    <title></title>
+  </head>
+  <body>
+    <form action="http://www.google.com/" method="post">
+      <label for="fn">Nome:</label> <input type="text" id="fn"><br>
+      <label for="ln">Sobrenome:</label> <input type="text" id="ln"><br>
+      <label for="cm">Empresa:</label> <input type="text" id="cm"><br>
+      <label for="an">Identificação do endereço:</label> <input type="text" id="an"><br>
+      <label for="sn">Endereço:</label> <input type="text" id="sn"><br>
+      <label for="hn">*Número:</label> <input type="text" id="hn"><br>
+      <label for="ct">Cidade:</label> <input type="text" id="ct"><br>
+      <label for="zc">CEP:</label> <input type="text" id="zc"><br>
+      <label for="st">Estado:</label> <input type="text" id="st"><br>
+      <label for="em">Email:</label> <input type="text" id="em"><br>
+      <label for="ph">Telefone:</label> <input type="text" id="ph"><br>
+      <label for="c1">Nome do cartão:</label> <input type="text" id="c1"><br>
+      <label for="c2">Número do cartão:</label> <input type="text" id="c2"><br>
+      <label for="c3">Data de validade:</label> <input type="text" id="c3"><br>
+      <label for="c4">Data de validade:</label> <input type="text" id="c4"><br>
+    </form>
+  </body>
+</html>
\ No newline at end of file
--- a/components/test/data/autofill/heuristics/input/173_i18n_pt_structured_address_2.html
+++ b/components/test/data/autofill/heuristics/input/173_i18n_pt_structured_address_2.html
+<!DOCTYPE html>
+<html>
+  <head>
+    <meta charset="UTF-8">
+    <title></title>
+  </head>
+  <body>
+    <form action="http://www.google.com/" method="post">
+      <label for="fn">Nome:</label> <input type="text" id="fn"><br>
+      <label for="ln">Sobrenome:</label> <input type="text" id="ln"><br>
+      <label for="cm">Empresa:</label> <input type="text" id="cm"><br>
+      <label for="hn">Número da residência:</label> <input type="text" id="hn"><br>
+      <label for="sn">Endereço:</label> <input type="text" id="sn"><br>
+      <label for="ct">Município:</label> <input type="text" id="ct"><br>
+      <label for="zc">CEP:</label> <input type="text" id="zc"><br>
+      <label for="st">Estado:</label> <input type="text" id="st"><br>
+      <label for="em">Email:</label> <input type="text" id="em"><br>
+      <label for="ph">Telefone:</label> <input type="text" id="ph"><br>
+      <label for="c1">Nome do cartão:</label> <input type="text" id="c1"><br>
+      <label for="c2">Número do cartão:</label> <input type="text" id="c2"><br>
+      <label for="c3">Data de validade:</label> <input type="text" id="c3"><br>
+      <label for="c4">Data de validade:</label> <input type="text" id="c4"><br>
+    </form>
+  </body>
+</html>
\ No newline at end of file
--- a/components/test/data/autofill/heuristics/output/170_i18n_ru_structured_address_1.out
+++ b/components/test/data/autofill/heuristics/output/170_i18n_ru_structured_address_1.out
+NAME_FIRST | fn | Имя: |  | fn_1-default
+NAME_LAST | ln | Фамилия: |  | fn_1-default
+COMPANY_NAME | cm | Название компании: |  | fn_1-default
+ADDRESS_HOME_HOUSE_NUMBER | hn | Дом: |  | fn_1-default
+ADDRESS_HOME_STREET_NAME | sn | Улица: |  | fn_1-default
+ADDRESS_HOME_CITY | ct | Город: |  | fn_1-default
+ADDRESS_HOME_ZIP | zc | Почтовый индекс: |  | fn_1-default
+ADDRESS_HOME_STATE | st | Область: |  | fn_1-default
+EMAIL_ADDRESS | em | Адрес электронной почты: |  | fn_1-default
+PHONE_HOME_WHOLE_NUMBER | ph | Телефон: |  | fn_1-default
+CREDIT_CARD_NAME_FULL | c1 | Имя владельца карты: |  | credit-card-cc
+CREDIT_CARD_NUMBER | c2 | Номер кредитной карты: |  | credit-card-cc
+CREDIT_CARD_EXP_MONTH | c3 | Срок действия карты: |  | credit-card-cc
+CREDIT_CARD_EXP_4_DIGIT_YEAR | c4 | Срок действия карты: |  | credit-card-cc
--- a/components/test/data/autofill/heuristics/output/171_i18n_ru_structured_address_2.out
+++ b/components/test/data/autofill/heuristics/output/171_i18n_ru_structured_address_2.out
+NAME_FIRST | fn | Имя: |  | fn_1-default
+NAME_LAST | ln | Фамилия: |  | fn_1-default
+COMPANY_NAME | cm | Название компании: |  | fn_1-default
+ADDRESS_HOME_STREET_NAME | sn | Название улицы: |  | fn_1-default
+ADDRESS_HOME_HOUSE_NUMBER | hn | Номер дома: |  | fn_1-default
+ADDRESS_HOME_CITY | ct | Населённый пункт: |  | fn_1-default
+ADDRESS_HOME_ZIP | zc | Почтовый индекс: |  | fn_1-default
+ADDRESS_HOME_STATE | st | Область: |  | fn_1-default
+EMAIL_ADDRESS | em | Адрес электронной почты: |  | fn_1-default
+PHONE_HOME_WHOLE_NUMBER | ph | Телефон: |  | fn_1-default
+CREDIT_CARD_NAME_FULL | c1 | Имя владельца карты: |  | credit-card-cc
+CREDIT_CARD_NUMBER | c2 | Номер кредитной карты: |  | credit-card-cc
+CREDIT_CARD_EXP_MONTH | c3 | Срок действия карты: |  | credit-card-cc
+CREDIT_CARD_EXP_4_DIGIT_YEAR | c4 | Срок действия карты: |  | credit-card-cc
--- a/components/test/data/autofill/heuristics/output/172_i18n_pt_structured_address_1.out
+++ b/components/test/data/autofill/heuristics/output/172_i18n_pt_structured_address_1.out
+NAME_FIRST | fn | Nome: |  | fn_1-default
+NAME_LAST | ln | Sobrenome: |  | fn_1-default
+COMPANY_NAME | cm | Empresa: |  | fn_1-default
+UNKNOWN_TYPE | an | Identificação do endereço: |  | fn_1-default
+ADDRESS_HOME_STREET_NAME | sn | Endereço: |  | fn_1-default
+ADDRESS_HOME_HOUSE_NUMBER | hn | *Número: |  | fn_1-default
+ADDRESS_HOME_CITY | ct | Cidade: |  | fn_1-default
+ADDRESS_HOME_ZIP | zc | CEP: |  | fn_1-default
+ADDRESS_HOME_STATE | st | Estado: |  | fn_1-default
+EMAIL_ADDRESS | em | Email: |  | fn_1-default
+PHONE_HOME_WHOLE_NUMBER | ph | Telefone: |  | fn_1-default
+CREDIT_CARD_NAME_FULL | c1 | Nome do cartão: |  | credit-card-cc
+CREDIT_CARD_NUMBER | c2 | Número do cartão: |  | credit-card-cc
+CREDIT_CARD_EXP_MONTH | c3 | Data de validade: |  | credit-card-cc
+CREDIT_CARD_EXP_4_DIGIT_YEAR | c4 | Data de validade: |  | credit-card-cc
--- a/components/test/data/autofill/heuristics/output/173_i18n_pt_structured_address_2.out
+++ b/components/test/data/autofill/heuristics/output/173_i18n_pt_structured_address_2.out
+NAME_FIRST | fn | Nome: |  | fn_1-default
+NAME_LAST | ln | Sobrenome: |  | fn_1-default
+COMPANY_NAME | cm | Empresa: |  | fn_1-default
+ADDRESS_HOME_HOUSE_NUMBER | hn | Número da residência: |  | fn_1-default
+ADDRESS_HOME_STREET_NAME | sn | Endereço: |  | fn_1-default
+ADDRESS_HOME_CITY | ct | Município: |  | fn_1-default
+ADDRESS_HOME_ZIP | zc | CEP: |  | fn_1-default
+ADDRESS_HOME_STATE | st | Estado: |  | fn_1-default
+EMAIL_ADDRESS | em | Email: |  | fn_1-default
+PHONE_HOME_WHOLE_NUMBER | ph | Telefone: |  | fn_1-default
+CREDIT_CARD_NAME_FULL | c1 | Nome do cartão: |  | credit-card-cc
+CREDIT_CARD_NUMBER | c2 | Número do cartão: |  | credit-card-cc
+CREDIT_CARD_EXP_MONTH | c3 | Data de validade: |  | credit-card-cc
+CREDIT_CARD_EXP_4_DIGIT_YEAR | c4 | Data de validade: |  | credit-card-cc
--- a/ios/chrome/browser/autofill/BUILD.gn
+++ b/ios/chrome/browser/autofill/BUILD.gn
@@ -359,6 +359,11 @@ bundle_data("unit_tests_bundle_data") {
    "//components/test/data/autofill/heuristics/input/166_i18n_en_honorific_prefix.html",
    "//components/test/data/autofill/heuristics/input/167_helity.es.html",
    "//components/test/data/autofill/heuristics/input/168_i18n_hi.html",
+    "//components/test/data/autofill/heuristics/input/169_dominos_de.html",
+    "//components/test/data/autofill/heuristics/input/170_i18n_ru_structured_address_1.html",
+    "//components/test/data/autofill/heuristics/input/171_i18n_ru_structured_address_2.html",
+    "//components/test/data/autofill/heuristics/input/172_i18n_pt_structured_address_1.html",
+    "//components/test/data/autofill/heuristics/input/173_i18n_pt_structured_address_2.html",
    "//components/test/data/autofill/heuristics/output/000_i18n_de.out",
    "//components/test/data/autofill/heuristics/output/001_i18n_de2.out",
    "//components/test/data/autofill/heuristics/output/002_i18n_en.out",
@@ -527,6 +532,11 @@ bundle_data("unit_tests_bundle_data") {
    "//components/test/data/autofill/heuristics/output/166_i18n_en_honorific_prefix.out",
    "//components/test/data/autofill/heuristics/output/167_helity.es.out",
    "//components/test/data/autofill/heuristics/output/168_i18n_hi.out",
+    "//components/test/data/autofill/heuristics/output/169_dominos_de.out",
+    "//components/test/data/autofill/heuristics/output/170_i18n_ru_structured_address_1.out",
+    "//components/test/data/autofill/heuristics/output/171_i18n_ru_structured_address_2.out",
+    "//components/test/data/autofill/heuristics/output/172_i18n_pt_structured_address_1.out",
+    "//components/test/data/autofill/heuristics/output/173_i18n_pt_structured_address_2.out",
  ]
  outputs = [ "{{bundle_resources_dir}}/" +
              "{{source_root_relative_dir}}/{{source_file_part}}" ]

--- a/ios/chrome/browser/autofill/form_structure_browsertest.mm
+++ b/ios/chrome/browser/autofill/form_structure_browsertest.mm
@@ -136,6 +136,8 @@ FormStructureBrowserTest::FormStructureBrowserTest()
      {
          // TODO(crbug.com/1098943): Remove once experiment is over.
          autofill::features::kAutofillEnableSupportForMoreStructureInNames,
+          // TODO(crbug.com/1125978): Remove once launched.
+          autofill::features::kAutofillEnableSupportForMoreStructureInAddresses,
      },
      // Disabled
      {autofill::features::kAutofillEnforceMinRequiredFieldsForHeuristics,