bin/datamasking/pii-knownnames.json

[
    {
        "Name": "First name",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(firstname|fname|first_name)(\\w*)",
            "(?>(fi?rst)_?(na?me?)\\|f_?name)(?!\\w*ID)",
            "(\\w*)(?i)(vorname|vname|vor_name)(\\w*)",
            "(\\w*)(?i)(prenom|pnom)(\\w*)",
            "(\\w*)(?i)(nombredepila|npila|apellido)(\\w*)",
            "(\\w*)(?i)(voornaam|vnaam)(\\w*)"
        ],
        "MaskingType": "Name",
        "MaskingSubType": "FirstName"
    },
    {
        "Name": "Middle name",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(middlename|mname|middle_name)(\\w*)",
            "(?>(mid(dle)?)_?(na?me?)\\|m_?name)(?!\\w*ID)",
            "(\\w*)(?i)(mittelname|mname|mittel_name)(\\w*)"
        ],
        "MaskingType": "Name",
        "MaskingSubType": "FirstName"
    },
    {
        "Name": "Last name",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(lastname|lname|maidenname|second_name|third_name|last_name|maiden_name|nickname|name_suffix)(\\w*)",
            "(\\w*)(?i)(nachname|nname)(\\w*)",
            "(\\w*)(?i)(nomdefamille|nomfamille)(\\w*)",
            "(\\w*)(?i)(nombredepila|npila|apellido)(\\w*)",
            "(\\w*)(?i)(achternaam|anaam)(\\w*)"
        ],
        "MaskingType": "Name",
        "MaskingSubType": "LastName"
    },
    {
        "Name": "Full name",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(fullname|fname)(\\w*)",
            "(\\w*)(?i)(vollstandigername|vollname)(\\w*)",
            "(\\w*)(?i)(nomcomplet|ncomplet)(\\w*)",
            "(\\w*)(?i)(nombrecompleto|ncompleto)(\\w*)",
            "(\\w*)(?i)(volledigenaam|volnaam)(\\w*)"
        ],
        "MaskingType": "Name",
        "MaskingSubType": "FullName"
    },
    {
        "Name": "Company name",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(company|companyname)(\\w*)"
        ],
        "MaskingType": "Company",
        "MaskingSubType": "CompanyName"
    },
    {
        "Name": "Date",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(date_of_birth|dateofbirth|dob|date_of_death|dateofdeath)(\\w*)",
            "(?>dob\\|dtofb\\|(day\\|date?\\|dt)_?(of)?_?(bi?rth))(?!\\w*ID)",
            "(?>(bi?rth)_?(date?\\|day\\|dt))(?!\\w*ID)",
            "(?>b_?(date?\\|day))(?!\\w*ID)",
            "(\\w*)(?i)(Geburtsdatum|Geburtsjahr)(\\w*)",
            "(\\w*)(?i)(datedenaissance|date_de_naissance)(\\w*)",
            "(\\w*)(?i)(geboortedatum)(\\w*)"
        ],
        "MaskingType": "Person",
        "MaskingSubType": "DateOfBirth"
    },
    {
        "Name": "Gender",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(gender)(\\w*)",
            "(\\w*)(?i)(Geschlecht)(\\w*)",
            "(\\w*)(?i)(sexe)(\\w*)",
            "(\\w*)(?i)(sexo)(\\w*)",
            "(\\w*)(?i)(geslacht)(\\w*)"
        ],
        "MaskingType": "Person",
        "MaskingSubType": "Gender"
    },
    {
        "Name": "Resume",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(resume)(\\w*)"
        ],
        "MaskingType": "Lorem",
        "MaskingSubType": "Letter"
    },
    {
        "Name": "E-mail",
        "Category": "Communication",
        "Pattern": [
            "(\\w*)(?i)(email|e-mail|mail)(\\w*)",
            "^(?:(?!invalid).)*email(?!\\w*ID)"
        ],
        "MaskingType": "Internet",
        "MaskingSubType": "Email"
    },
    {
        "Name": "Nationality",
        "Category": "Personal",
        "Pattern": [
            "(\\w*)(?i)(nationality)(\\w*)",
            "(\\w*)(?i)(Staatsangehörigkeit|Staatsangehorigkeit)(\\w*)",
            "(\\w*)(?i)(nationalité|nationalite)(\\w*)",
            "(\\w*)(?i)(nacionalidad)(\\w*)",
            "(\\w*)(?i)(nationaliteit)(\\w*)"
        ],
        "MaskingType": "Random",
        "MaskingSubType": "Int"
    },
    {
        "Name": "Address",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)^(?!ipaddress)(address)(\\w*)",
            "(?>(str(eet)?_?addre?s?s?\\|street))(?!\\w*(ID\\|type))",
            "(\\w*)(?i)adresse(\\w*)",
            "(\\w*)(?i)direccion(\\w*)",
            "(\\w*)(?i)adres(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "StreetAddress"
    },
    {
        "Name": "City",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)city(\\w*)",
            "ci?ty(?!\\w*ID)",
            "(\\w*)(?i)stadt(\\w*)",
            "(\\w*)(?i)ville(\\w*)",
            "(\\w*)(?i)ciudad(\\w*)",
            "(\\w*)(?i)stad(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "City"
    },
    {
        "Name": "State",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)state(\\w*)",
            "(\\w*)(?i)provin(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "State"
    },
    {
        "Name": "County",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)county(\\w*)",
            "c(ou)?nty(?!\\w*ID)",
            "(\\w*)(?i)bezirk(\\w*)",
            "(\\w*)(?i)comte(\\w*)",
            "(\\w*)(?i)condado(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "County"
    },
    {
        "Name": "Country",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)country(\\w*)",
            "c(ou)?ntry(?!\\w*ID)",
            "(\\w*)(?i)land(\\w*)",
            "(\\w*)(?i)pays(\\w*)",
            "(\\w*)(?i)pais(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "Country"
    },
    {
        "Name": "Location",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)location(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "Latitude"
    },
    {
        "Name": "Phone",
        "Category": "Communication",
        "Pattern": [
            "(\\w*)(?i)(phone|phonenumber)(\\w*)",
            "(?>((tele?)?phone)\\|(co?nta?ct\\|tel)_?(num(ber)?\\|nbr\\|no))(?!\\w*(ID\\|type))"
        ],
        "MaskingType": "Phone",
        "MaskingSubType": "PhoneNumber"
    },
    {
        "Name": "Fax",
        "Category": "Communication",
        "Pattern": [
            "(\\w*)(?i)(fax|faxnumber)(\\w*)",
            "(?>fax_?(num(ber)?\\|nbr\\|no)?)(?!\\w*(ID\\|type))"
        ],
        "MaskingType": "Phone",
        "MaskingSubType": "PhoneNumber"
    },
    {
        "Name": "Website",
        "Category": "Communication",
        "Pattern": [
            "(\\w*)(?i)(website|url|link)(\\w*)",
            "(?>(url\\|web_?addre?s?s?))(?!\\w*(ID\\|type))"
        ],
        "MaskingType": "Internet",
        "MaskingSubType": "Url"
    },
    {
        "Name": "IPAddress",
        "Category": "System",
        "Pattern": [
            "(\\w*)(?i)(ip|ipaddress|ipv4|ipv6)(\\w*)",
            "(?>(ip_?addre?s?s?))(?!\\w*(ID\\|type))"
        ],
        "MaskingType": "Internet",
        "MaskingSubType": "Ip"
    },
    {
        "Name": "Username",
        "Category": "Credentials",
        "Pattern": [
            "(\\w*)(?i)(userid|user|username|login)(\\w*)",
            "(\\w*)(?i)(nutzerid|nutzer|nutzername)(\\w*)",
            "(\\w*)(?i)(nomutilisateur|utilisateur)(\\w*)",
            "(\\w*)(?i)(nombredeusuario|usuario)(\\w*)",
            "(\\w*)(?i)(gebruikersid|gebruiker|gebruikersnaam)(\\w*)"
        ],
        "MaskingType": "Internet",
        "MaskingSubType": "Username"
    },
    {
        "Name": "Password",
        "Category": "Credentials",
        "Pattern": [
            "(\\w*)(?i)pass(\\w*)",
            "(\\w*)(?i)contrasena(\\w*)",
            "(\\w*)(?i)wachtwoord(\\w*)"
        ],
        "MaskingType": "Internet",
        "MaskingSubType": "Password"
    },
    {
        "Name": "Zipcode",
        "Category": "Location",
        "Pattern": [
            "(\\w*)(?i)(postalcode|zip)(\\w*)",
            "(?>(zip\\|post(al)?)_?((co?de?)?4?))(?!\\w*ID)",
            "(\\w*)(?i)postal(\\w*)",
            "(\\w*)(?i)postleitzahl(\\w*)",
            "(\\w*)(?i)postcode(\\w*)"
        ],
        "MaskingType": "Address",
        "MaskingSubType": "Zipcode"
    },
    {
        "Name": "Bank Account",
        "Category": "Financial",
        "Pattern": [
            "(\\w*)(?i)(bank|bankaccount)(\\w*)",
            "(?>(acc(oun\\|n)?t)_?(num(ber)?\\|nbrjno)?)(?!\\w\\*(ID\\|type))"
        ],
        "MaskingType": "Finance",
        "MaskingSubType": "Account"
    },
    {
        "Name": "CreditCard",
        "Category": "Financial",
        "Pattern": [
            "(\\w*)(?i)(cvv|creditcardcvv)(\\w*)"
        ],
        "MaskingType": "Finance",
        "MaskingSubType": "CreditcardCvv"
    },
    {
        "Name": "CreditCard",
        "Category": "Financial",
        "Pattern": [
            "(\\w*)(?i)(credit|creditcard)(\\w*)",
            "(?>cre?di?t_?(ca?rd)?_?(num(ber)?\\|nbr\\|no)?)(?!\\w*ID)"
        ],
        "MaskingType": "Finance",
        "MaskingSubType": "CreditcardNumber"
    },
    {
        "Name": "Other",
        "Category": "Financial",
        "Pattern": [
            "(\\w*)(?i)(sales|special|shop|product)(\\w*)"
        ],
        "MaskingType": "Commerce",
        "MaskingSubType": "Categories"
    },
    {
        "Name": "Other",
        "Category": "N/A",
        "Pattern": [
            "(\\w*)(?i)(guid)(\\w*)",
            "(\\w*)(?i)(receipt)(\\w*)"
        ],
        "MaskingType": "Random",
        "MaskingSubType": "String"
    }
]