Make autofill regular expressions unicode again.

chrome/browser/autofill/autofill_regex_constants.cc.utf8

+// Copyright (c) 2011 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// This file contains UTF8 strings that we want as char arrays.  To avoid
+// different compilers, we use a script to convert the UTF8 strings into
+// numeric literals (\x##).
+
+#include "chrome/browser/autofill/autofill_regex_constants.h"
+
+namespace autofill {
+
+/////////////////////////////////////////////////////////////////////////////
+// address_field.cc
+/////////////////////////////////////////////////////////////////////////////
+const char kAttentionIgnoredRe[] = "attention|attn";
+const char kRegionIgnoredRe[] =
+    "province|region|other"
+    "|provincia"  // es
+    "|bairro|suburb";  // pt-BR, pt-PT
+const char kCompanyRe[] =
+    "company|business|organization|organisation"
+    "|firma|firmenname"  // de-DE
+    "|empresa"  // es
+    "|societe|société"  // fr-FR
+    "|ragione.?sociale"  // it-IT
+    "|会社"  // ja-JP
+    "|название.?компании"  // ru
+    "|单位|公司"  // zh-CN
+    "|회사|직장";  // ko-KR
+const char kAddressLine1Re[] =
+    "address.*line|address1|addr1|street"
+    "|strasse|straße|hausnummer|housenumber"  // de-DE
+    "|house.?name"  // en-GB
+    "|direccion|dirección"  // es
+    "|adresse"  // fr-FR
+    "|indirizzo"  // it-IT
+    "|住所1"  // ja-JP
+    "|morada|endereço"  // pt-BR, pt-PT
+    "|Адрес"  // ru
+    "|地址"  // zh-CN
+    "|주소.?1";  // ko-KR
+const char kAddressLine1LabelRe[] =
+    "address"
+    "|adresse"  // fr-FR
+    "|indirizzo"  // it-IT
+    "|住所"  // ja-JP
+    "|地址"  // zh-CN
+    "|주소";  // ko-KR
+const char kAddressLine2Re[] =
+    "address.*line2|address2|addr2|street|suite|unit"
+    "|adresszusatz|ergänzende.?angaben"  // de-DE
+    "|direccion2|colonia|adicional"  // es
+    "|addresssuppl|complementnom|appartement"  // fr-FR
+    "|indirizzo2"  // it-IT
+    "|住所2"  // ja-JP
+    "|complemento|addrcomplement"  // pt-BR, pt-PT
+    "|Улица"  // ru
+    "|地址2"  // zh-CN
+    "|주소.?2";  // ko-KR
+const char kAddressLine2LabelRe[] =
+    "address"
+    "|adresse"  // fr-FR
+    "|indirizzo"  // it-IT
+    "|地址"  // zh-CN
+    "|주소";  // ko-KR
+const char kAddressLine3Re[] =
+    "address.*line3|address3|addr3|street|line3"
+    "|municipio"  // es
+    "|batiment|residence"  // fr-FR
+    "|indirizzo3";  // it-IT
+const char kCountryRe[] =
+    "country|countries|location"
+    "|país|pais"  // es
+    "|国"  // ja-JP
+    "|国家"  // zh-CN
+    "|국가|나라";  // ko-KR
+const char kZipCodeRe[] =
+    "zip|postal|post.*code|pcode|^1z$"
+    "|postleitzahl"  // de-DE
+    "|\\bcp\\b"  // es
+    "|\\bcdp\\b"  // fr-FR
+    "|\\bcap\\b"  // it-IT
+    "|郵便番号"  // ja-JP
+    "|codigo|codpos|\\bcep\\b"  // pt-BR, pt-PT
+    "|Почтовый.?Индекс"  // ru
+    "|邮政编码|邮编"  // zh-CN
+    "|郵遞區號"  // zh-TW
+    "|우편.?번호";  // ko-KR
+const char kZip4Re[] =
+    "zip|^-$|post2"
+    "|codpos2";  // pt-BR, pt-PT
+const char kCityRe[] =
+    "city|town"
+    "|\\bort\\b|stadt"  // de-DE
+    "|suburb"  // en-AU
+    "|ciudad|provincia|localidad|poblacion"  // es
+    "|ville|commune"  // fr-FR
+    "|localita"  // it-IT
+    "|市区町村"  // ja-JP
+    "|cidade"  // pt-BR, pt-PT
+    "|Город"  // ru
+    "|市"  // zh-CN
+    "|分區"  // zh-TW
+    "|^시[^도·・]|시[·・]?군[·・]?구";  // ko-KR
+const char kStateRe[] =
+    "(?<!united )state|county|region|province"
+    "|land"  // de-DE
+    "|county|principality"  // en-UK
+    "|都道府県"  // ja-JP
+    "|estado|provincia"  // pt-BR, pt-PT
+    "|область"  // ru
+    "|省"  // zh-CN
+    "|地區"  // zh-TW
+    "|^시[·・]?도";  // ko-KR
+const char kAddressTypeSameAsRe[] = "same as";
+const char kAddressTypeUseMyRe[] = "use my";
+const char kBillingDesignatorRe[] = "bill";
+const char kShippingDesignatorRe[] = "ship";
+
+/////////////////////////////////////////////////////////////////////////////
+// credit_card_field.cc
+/////////////////////////////////////////////////////////////////////////////
+const char kNameOnCardRe[] =
+    "card.?holder|name.?on.?card|ccname|ccfullname|owner"
+    "|karteninhaber"  // de-DE
+    "|nombre.*tarjeta"  // es
+    "|nom.*carte"  // fr-FR
+    "|nome.*cart"  // it-IT
+    "|名前"  // ja-JP
+    "|Имя.*карты"  // ru
+    "|信用卡开户名|开户名|持卡人姓名"  // zh-CN
+    "|持卡人姓名";  // zh-TW
+const char kNameOnCardContextualRe[] =
+    "name";
+const char kCardNumberRe[] =
+    "card.?number|card.?#|card.?no|ccnum|acctnum"
+    "|nummer"  // de-DE
+    "|credito|numero|número"  // es
+    "|numéro"  // fr-FR
+    "|カード番号"  // ja-JP
+    "|Номер.*карты"  // ru
+    "|信用卡号|信用卡号码"  // zh-CN
+    "|信用卡卡號"  // zh-TW
+    "|카드";  // ko-KR
+const char kCardCvcRe[] =
+    "verification|card identification|security code|cvn|cvv|cvc|csc";
+
+// "Expiration date" is the most common label here, but some pages have
+// "Expires", "exp. date" or "exp. month" and "exp. year".  We also look
+// for the field names ccmonth and ccyear, which appear on at least 4 of
+// our test pages.
+
+// On at least one page (The China Shop2.html) we find only the labels
+// "month" and "year".  So for now we match these words directly; we'll
+// see if this turns out to be too general.
+
+// Toolbar Bug 51451: indeed, simply matching "month" is too general for
+//   https://rps.fidelity.com/ftgw/rps/RtlCust/CreatePIN/Init.
+// Instead, we match only words beginning with "month".
+const char kExpirationMonthRe[] =
+    "expir|exp.*mo|exp.*date|ccmonth"
+    "|gueltig|gültig|monat"  // de-DE
+    "|fecha"  // es
+    "|date.*exp"  // fr-FR
+    "|scadenza"  // it-IT
+    "|有効期限"  // ja-JP
+    "|validade"  // pt-BR, pt-PT
+    "|Срок действия карты"  // ru
+    "|月";  // zh-CN
+const char kExpirationYearRe[] =
+    "exp|^/|year"
+    "|ablaufdatum|gueltig|gültig|yahr"  // de-DE
+    "|fecha"  // es
+    "|scadenza"  // it-IT
+    "|有効期限"  // ja-JP
+    "|validade"  // pt-BR, pt-PT
+    "|Срок действия карты"  // ru
+    "|年|有效期";  // zh-CN
+
+// This regex is a little bit nasty, but it is simply requiring exactly two
+// adjacent y's.
+const char kExpirationDate2DigitYearRe[] =
+    "exp.*date.*[^y]yy([^y]|$)";
+const char kExpirationDateRe[] =
+    "expir|exp.*date"
+    "|gueltig|gültig"  // de-DE
+    "|fecha"  // es
+    "|date.*exp"  // fr-FR
+    "|scadenza"  // it-IT
+    "|有効期限"  // ja-JP
+    "|validade"  // pt-BR, pt-PT
+    "|Срок действия карты";  // ru
+const char kCardIgnoredRe[] =
+    "^card";
+
+
+/////////////////////////////////////////////////////////////////////////////
+// email_field.cc
+/////////////////////////////////////////////////////////////////////////////
+const char kEmailRe[] =
+    "e.?mail"
+    "|メールアドレス"  // ja-JP
+    "|Электронной.?Почты"  // ru
+    "|邮件|邮箱"  // zh-CN
+    "|電郵地址"  // zh-TW
+    "|(이메일|전자.?우편|[Ee]-?mail)(.?주소)?";  // ko-KR
+
+
+/////////////////////////////////////////////////////////////////////////////
+// name_field.cc
+/////////////////////////////////////////////////////////////////////////////
+const char kNameIgnoredRe[] =
+    "user.?name|user.?id|nickname|maiden name|title|prefix|suffix"
+    "|vollständiger.?name"  // de-DE
+    "|用户名"  // zh-CN
+    "|(사용자.?)?아이디|사용자.?ID";  // ko-KR
+const char kNameRe[] =
+    "^name|full.?name|your.?name|customer.?name|firstandlastname|bill.?name"
+        "|ship.?name"
+    "|nombre.*y.*apellidos"  // es
+    "|^nom"  // fr-FR
+    "|お名前|氏名"  // ja-JP
+    "|^nome"  // pt-BR, pt-PT
+    "|姓名"  // zh-CN
+    "|성명";  // ko-KR
+const char kNameSpecificRe[] =
+    "^name"
+    "|^nom"  // fr-FR
+    "|^nome";  // pt-BR, pt-PT
+const char kFirstNameRe[] =
+    "first.*name|initials|fname|first$"
+    "|vorname"  // de-DE
+    "|nombre"  // es
+    "|forename|prénom|prenom"  // fr-FR
+    "|名"  // ja-JP
+    "|nome"  // pt-BR, pt-PT
+    "|Имя"  // ru
+    "|이름";  // ko-KR
+const char kMiddleInitialRe[] = "middle.*initial|m\\.i\\.|mi$|\\bmi\\b";
+const char kMiddleNameRe[] =
+    "middle.*name|mname|middle$"
+    "|apellido.?materno|lastlastname";  // es
+const char kLastNameRe[] =
+    "last.*name|lname|surname|last$|secondname"
+    "|nachname"  // de-DE
+    "|apellido"  // es
+    "|famille|^nom"  // fr-FR
+    "|cognome"  // it-IT
+    "|姓"  // ja-JP
+    "|morada|apelidos|surename|sobrenome"  // pt-BR, pt-PT
+    "|Фамилия"  // ru
+    "|성[^명]?";  // ko-KR
+
+/////////////////////////////////////////////////////////////////////////////
+// phone_field.cc
+/////////////////////////////////////////////////////////////////////////////
+const char kPhoneRe[] =
+    "phone|mobile"
+    "|telefonnummer"  // de-DE
+    "|telefono|teléfono"  // es
+    "|telfixe"  // fr-FR
+    "|電話"  // ja-JP
+    "|telefone|telemovel"  // pt-BR, pt-PT
+    "|телефон"  // ru
+    "|电话"  // zh-CN
+    "|(전화|핸드폰|휴대폰|휴대전화)(.?번호)?";  // ko-KR
+const char kCountryCodeRe[] =
+    "country.*code|ccode|_cc";
+const char kAreaCodeNotextRe[] =
+    "^\\($";
+const char kAreaCodeRe[] =
+    "area.*code|acode|area"
+    "|지역.?번호";  // ko-KR
+const char kFaxRe[] =
+    "fax"
+    "|télécopie|telecopie"  // fr-FR
+    "|ファックス"  // ja-JP
+    "|факс"  // ru
+    "|传真"  // zh-CN
+    "|傳真"  // zh-TW
+    "|팩스(.?번호)?";  // ko-KR
+const char kPhonePrefixSeparatorRe[] =
+    "^-$|^\\)$";
+const char kPhoneSuffixSeparatorRe[] =
+    "^-$";
+const char kPhonePrefixRe[] =
+    "prefix|exchange"
+    "|preselection"  // fr-FR
+    "|ddd";  // pt-BR, pt-PT
+const char kPhoneSuffixRe[] =
+    "suffix";
+const char kPhoneExtensionRe[] =
+    "\\bext|ext\\b|extension"
+    "|ramal";  // pt-BR, pt-PT
+
+}  // namespace autofill