components/autofill/core/browser/autofill_regex_constants.cc

   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 // This file contains UTF8 strings that we want as char arrays.  To avoid
   6 // different compilers, we use a script to convert the UTF8 strings into
   7 // numeric literals (\x##).
   8
   9 #include "components/autofill/core/browser/autofill_regex_constants.h"
  10
  11 namespace autofill {
  12
  13 /////////////////////////////////////////////////////////////////////////////
  14 // address_field.cc
  15 /////////////////////////////////////////////////////////////////////////////
  16 const char kAttentionIgnoredRe[] = "attention|attn";
  17 const char kRegionIgnoredRe[] =
  18     "province|region|other"
  19     "|provincia"  // es
  20     "|bairro|suburb";  // pt-BR, pt-PT
  21 const char kCompanyRe[] =
  22     "company|business|organization|organisation"
  23     "|firma|firmenname"  // de-DE
  24     "|empresa"  // es
  25     "|societe|société"  // fr-FR
  26     "|ragione.?sociale"  // it-IT
  27     "|会社"  // ja-JP
  28     "|название.?компании"  // ru
  29     "|单位|公司"  // zh-CN
  30     "|회사|직장";  // ko-KR
  31 const char kAddressLine1Re[] =
  32     "address.*line|address1|addr1|street"
  33     "|(shipping|billing)address$"
  34     "|strasse|straße|hausnummer|housenumber"  // de-DE
  35     "|house.?name"  // en-GB
  36     "|direccion|dirección"  // es
  37     "|adresse"  // fr-FR
  38     "|indirizzo"  // it-IT
  39     "|住所1"  // ja-JP
  40     "|morada|endereço"  // pt-BR, pt-PT
  41     "|Адрес"  // ru
  42     "|地址"  // zh-CN
  43     "|주소.?1";  // ko-KR
  44 const char kAddressLine1LabelRe[] =
  45     "address"
  46     "|adresse"  // fr-FR
  47     "|indirizzo"  // it-IT
  48     "|住所"  // ja-JP
  49     "|地址"  // zh-CN
  50     "|주소";  // ko-KR
  51 const char kAddressLine2Re[] =
  52     "address.*line2|address2|addr2|street|suite|unit"
  53     "|adresszusatz|ergänzende.?angaben"  // de-DE
  54     "|direccion2|colonia|adicional"  // es
  55     "|addresssuppl|complementnom|appartement"  // fr-FR
  56     "|indirizzo2"  // it-IT
  57     "|住所2"  // ja-JP
  58     "|complemento|addrcomplement"  // pt-BR, pt-PT
  59     "|Улица"  // ru
  60     "|地址2"  // zh-CN
  61     "|주소.?2";  // ko-KR
  62 const char kAddressLine2LabelRe[] =
  63     "address|line"
  64     "|adresse"  // fr-FR
  65     "|indirizzo"  // it-IT
  66     "|地址"  // zh-CN
  67     "|주소";  // ko-KR
  68 const char kAddressLinesExtraRe[] =
  69     "address.*line[3-9]|address[3-9]|addr[3-9]|street|line[3-9]"
  70     "|municipio"  // es
  71     "|batiment|residence"  // fr-FR
  72     "|indirizzo[3-9]";  // it-IT
  73 const char kAddressLookupRe[] =
  74     "lookup";
  75 const char kCountryRe[] =
  76     "country|countries"
  77     "|país|pais"  // es
  78     "|国"  // ja-JP
  79     "|国家"  // zh-CN
  80     "|국가|나라";  // ko-KR
  81 const char kCountryLocationRe[] =
  82     "location";
  83 const char kZipCodeRe[] =
  84     "zip|postal|post.*code|pcode"
  85     "|pin.?code"  // en-IN
  86     "|postleitzahl"  // de-DE
  87     "|\\bcp\\b"  // es
  88     "|\\bcdp\\b"  // fr-FR
  89     "|\\bcap\\b"  // it-IT
  90     "|郵便番号"  // ja-JP
  91     "|codigo|codpos|\\bcep\\b"  // pt-BR, pt-PT
  92     "|Почтовый.?Индекс"  // ru
  93     "|邮政编码|邮编"  // zh-CN
  94     "|郵遞區號"  // zh-TW
  95     "|우편.?번호";  // ko-KR
  96 const char kZip4Re[] =
  97     "zip|^-$|post2"
  98     "|codpos2";  // pt-BR, pt-PT
  99 const char kCityRe[] =
 100     "city|town"
 101     "|\\bort\\b|stadt"  // de-DE
 102     "|suburb"  // en-AU
 103     "|ciudad|provincia|localidad|poblacion"  // es
 104     "|ville|commune"  // fr-FR
 105     "|localita"  // it-IT
 106     "|市区町村"  // ja-JP
 107     "|cidade"  // pt-BR, pt-PT
 108     "|Город"  // ru
 109     "|市"  // zh-CN
 110     "|分區"  // zh-TW
 111     "|^시[^도·・]|시[·・]?군[·・]?구";  // ko-KR
 112 const char kStateRe[] =
 113     "(?<!united )state|county|region|province"
 114     "|land"  // de-DE
 115     "|county|principality"  // en-UK
 116     "|都道府県"  // ja-JP
 117     "|estado|provincia"  // pt-BR, pt-PT
 118     "|область"  // ru
 119     "|省"  // zh-CN
 120     "|地區"  // zh-TW
 121     "|^시[·・]?도";  // ko-KR
 122
 123 /////////////////////////////////////////////////////////////////////////////
 124 // credit_card_field.cc
 125 /////////////////////////////////////////////////////////////////////////////
 126 const char kNameOnCardRe[] =
 127     "card.?(holder|owner)|name.*\\bon\\b.*card|(card|cc).?name|cc.?full.?name"
 128     "|karteninhaber"  // de-DE
 129     "|nombre.*tarjeta"  // es
 130     "|nom.*carte"  // fr-FR
 131     "|nome.*cart"  // it-IT
 132     "|名前"  // ja-JP
 133     "|Имя.*карты"  // ru
 134     "|信用卡开户名|开户名|持卡人姓名"  // zh-CN
 135     "|持卡人姓名";  // zh-TW
 136 const char kNameOnCardContextualRe[] =
 137     "name";
 138 const char kCardNumberRe[] =
 139     "(card|cc|acct).?(number|#|no|num)"
 140     "|nummer"  // de-DE
 141     "|credito|numero|número"  // es
 142     "|numéro"  // fr-FR
 143     "|カード番号"  // ja-JP
 144     "|Номер.*карты"  // ru
 145     "|信用卡号|信用卡号码"  // zh-CN
 146     "|信用卡卡號"  // zh-TW
 147     "|카드";  // ko-KR
 148 const char kCardCvcRe[] =
 149     "verification|card identification|security code|card code"
 150     "|cvn|cvv|cvc|csc|cvd|cid|ccv"
 151     "|\\bcid\\b";
 152
 153 // "Expiration date" is the most common label here, but some pages have
 154 // "Expires", "exp. date" or "exp. month" and "exp. year".  We also look
 155 // for the field names ccmonth and ccyear, which appear on at least 4 of
 156 // our test pages.
 157
 158 // On at least one page (The China Shop2.html) we find only the labels
 159 // "month" and "year".  So for now we match these words directly; we'll
 160 // see if this turns out to be too general.
 161
 162 // Toolbar Bug 51451: indeed, simply matching "month" is too general for
 163 //   https://rps.fidelity.com/ftgw/rps/RtlCust/CreatePIN/Init.
 164 // Instead, we match only words beginning with "month".
 165 const char kExpirationMonthRe[] =
 166     "expir|exp.*mo|exp.*date|ccmonth|cardmonth"
 167     "|gueltig|gültig|monat"  // de-DE
 168     "|fecha"  // es
 169     "|date.*exp"  // fr-FR
 170     "|scadenza"  // it-IT
 171     "|有効期限"  // ja-JP
 172     "|validade"  // pt-BR, pt-PT
 173     "|Срок действия карты"  // ru
 174     "|月";  // zh-CN
 175 const char kExpirationYearRe[] =
 176     "exp|^/|year"
 177     "|ablaufdatum|gueltig|gültig|yahr"  // de-DE
 178     "|fecha"  // es
 179     "|scadenza"  // it-IT
 180     "|有効期限"  // ja-JP
 181     "|validade"  // pt-BR, pt-PT
 182     "|Срок действия карты"  // ru
 183     "|年|有效期";  // zh-CN
 184
 185 // The "yy" portion of the regex is just looking for two adjacent y's.
 186 const char kExpirationDate2DigitYearRe[] =
 187     "(exp.*date.*|mm\\s*[-/]\\s*)[^y]yy([^y]|$)";
 188 const char kExpirationDate4DigitYearRe[] =
 189     "^mm\\s*[-/]\\syyyy$";
 190 const char kExpirationDateRe[] =
 191     "expir|exp.*date"
 192     "|gueltig|gültig"  // de-DE
 193     "|fecha"  // es
 194     "|date.*exp"  // fr-FR
 195     "|scadenza"  // it-IT
 196     "|有効期限"  // ja-JP
 197     "|validade"  // pt-BR, pt-PT
 198     "|Срок действия карты";  // ru
 199 const char kGiftCardRe[] =
 200     "gift.?card";
 201 const char kDebitGiftCardRe[] =
 202     "(visa|mastercard|discover|amex|american express).*gift.?card";
 203 const char kDebitCardRe[] =
 204     "debit.*card";
 205
 206
 207 /////////////////////////////////////////////////////////////////////////////
 208 // email_field.cc
 209 /////////////////////////////////////////////////////////////////////////////
 210 const char kEmailRe[] =
 211     "e.?mail"
 212     "|courriel"  // fr
 213     "|メールアドレス"  // ja-JP
 214     "|Электронной.?Почты"  // ru
 215     "|邮件|邮箱"  // zh-CN
 216     "|電郵地址"  // zh-TW
 217     "|(이메일|전자.?우편|[Ee]-?mail)(.?주소)?";  // ko-KR
 218
 219
 220 /////////////////////////////////////////////////////////////////////////////
 221 // name_field.cc
 222 /////////////////////////////////////////////////////////////////////////////
 223 const char kNameIgnoredRe[] =
 224     "user.?name|user.?id|nickname|maiden name|title|prefix|suffix"
 225     "|vollständiger.?name"  // de-DE
 226     "|用户名"  // zh-CN
 227     "|(사용자.?)?아이디|사용자.?ID";  // ko-KR
 228 const char kNameRe[] =
 229     "^name|full.?name|your.?name|customer.?name|bill.?name|ship.?name"
 230     "|name.*first.*last|firstandlastname"
 231     "|nombre.*y.*apellidos"  // es
 232     "|^nom"  // fr-FR
 233     "|お名前|氏名"  // ja-JP
 234     "|^nome"  // pt-BR, pt-PT
 235     "|姓名"  // zh-CN
 236     "|성명";  // ko-KR
 237 const char kNameSpecificRe[] =
 238     "^name"
 239     "|^nom"  // fr-FR
 240     "|^nome";  // pt-BR, pt-PT
 241 const char kFirstNameRe[] =
 242     "first.*name|initials|fname|first$|given.*name"
 243     "|vorname"  // de-DE
 244     "|nombre"  // es
 245     "|forename|prénom|prenom"  // fr-FR
 246     "|名"  // ja-JP
 247     "|nome"  // pt-BR, pt-PT
 248     "|Имя"  // ru
 249     "|이름";  // ko-KR
 250 const char kMiddleInitialRe[] = "middle.*initial|m\\.i\\.|mi$|\\bmi\\b";
 251 const char kMiddleNameRe[] =
 252     "middle.*name|mname|middle$"
 253     "|apellido.?materno|lastlastname";  // es
 254 const char kLastNameRe[] =
 255     "last.*name|lname|surname|last$|secondname|family.*name"
 256     "|nachname"  // de-DE
 257     "|apellido"  // es
 258     "|famille|^nom"  // fr-FR
 259     "|cognome"  // it-IT
 260     "|姓"  // ja-JP
 261     "|morada|apelidos|surename|sobrenome"  // pt-BR, pt-PT
 262     "|Фамилия"  // ru
 263     "|성[^명]?";  // ko-KR
 264
 265 /////////////////////////////////////////////////////////////////////////////
 266 // phone_field.cc
 267 /////////////////////////////////////////////////////////////////////////////
 268 const char kPhoneRe[] =
 269     "phone|mobile"
 270     "|telefonnummer"  // de-DE
 271     "|telefono|teléfono"  // es
 272     "|telfixe"  // fr-FR
 273     "|電話"  // ja-JP
 274     "|telefone|telemovel"  // pt-BR, pt-PT
 275     "|телефон"  // ru
 276     "|电话"  // zh-CN
 277     "|(전화|핸드폰|휴대폰|휴대전화)(.?번호)?";  // ko-KR
 278 const char kCountryCodeRe[] =
 279     "country.*code|ccode|_cc";
 280 const char kAreaCodeNotextRe[] =
 281     "^\\($";
 282 const char kAreaCodeRe[] =
 283     "area.*code|acode|area"
 284     "|지역.?번호";  // ko-KR
 285 const char kPhonePrefixSeparatorRe[] =
 286     "^-$|^\\)$";
 287 const char kPhoneSuffixSeparatorRe[] =
 288     "^-$";
 289 const char kPhonePrefixRe[] =
 290     "prefix|exchange"
 291     "|preselection"  // fr-FR
 292     "|ddd";  // pt-BR, pt-PT
 293 const char kPhoneSuffixRe[] =
 294     "suffix";
 295 const char kPhoneExtensionRe[] =
 296     "\\bext|ext\\b|extension"
 297     "|ramal";  // pt-BR, pt-PT
 298
 299 }  // namespace autofill