Changeset 9853 in josm


Ignore:
Timestamp:
2016-02-23T01:09:36+01:00 (4 years ago)
Author:
Don-vip
Message:

fix #12557 : update to Apache Commons Validator 1.5.0 + updates from trunk + unit/integration tests + anticipated patches to add missing TLDs and Country Codes from IANA

Location:
trunk
Files:
8 added
8 edited

Legend:

Unmodified
Added
Removed
  • trunk/src/org/openstreetmap/josm/data/validation/routines/DomainValidator.java

    r9239 r9853  
    1717package org.openstreetmap.josm.data.validation.routines;
    1818
     19import java.net.IDN;
    1920import java.util.Arrays;
    2021import java.util.Locale;
     
    3132 * to the standards <a href="http://www.ietf.org/rfc/rfc1034.txt">RFC1034</a>,
    3233 * section 3, and <a href="http://www.ietf.org/rfc/rfc1123.txt">RFC1123</a>,
    33  * section 2.1. No accomodation is provided for the specialized needs of
     34 * section 2.1. No accommodation is provided for the specialized needs of
    3435 * other applications; if the domain name has been URL-encoded, for example,
    3536 * validation will fail even though the equivalent plaintext version of the
     
    4748 *     <li>{@link #isValidGenericTld} - validates generic TLDs
    4849 *         (<code>.com, .org</code>, etc.)</li>
    49  *     <li>{@link #isValidIdnTld} - validates IDN TLDs
    50  *         (<code>.xn--*</code>, etc.)</li>
    5150 *     <li>{@link #isValidCountryCodeTld} - validates country code TLDs
    5251 *         (<code>.us, .uk, .cn</code>, etc.)</li>
     
    5958 * </p>
    6059 *
    61  * @version $Revision: 1640271 $ $Date: 2014-11-18 02:32:15 2014 UTC (Tue, 18 Nov 2014) $
     60 * @version $Revision: 1725571 $
    6261 * @since Validator 1.4
    6362 */
    6463public final class DomainValidator extends AbstractValidator {
    6564
     65    private static final int MAX_DOMAIN_LENGTH = 253;
     66
     67    private static final String[] EMPTY_STRING_ARRAY = new String[0];
     68
    6669    // Regular expression strings for hostnames (derived from RFC2396 and RFC 1123)
    67     private static final String DOMAIN_LABEL_REGEX = "\\p{Alnum}(?>[\\p{Alnum}-]*\\p{Alnum})*";
    68     private static final String TOP_LABEL_REGEX = "\\p{Alpha}{2,}";
    69     // JOSM PATCH BEGIN
    70     // See #10862 - IDN TLDs in ASCII form
    71     private static final String TOP_LABEL_IDN_REGEX = "(?:xn|XN)--\\p{Alnum}{2,}(?:-\\p{Alpha}{2,})?";
     70
     71    // RFC2396: domainlabel   = alphanum | alphanum *( alphanum | "-" ) alphanum
     72    // Max 63 characters
     73    private static final String DOMAIN_LABEL_REGEX = "\\p{Alnum}(?>[\\p{Alnum}-]{0,61}\\p{Alnum})?";
     74
     75    // RFC2396 toplabel = alpha | alpha *( alphanum | "-" ) alphanum
     76    // Max 63 characters
     77    private static final String TOP_LABEL_REGEX = "\\p{Alpha}(?>[\\p{Alnum}-]{0,61}\\p{Alnum})?";
     78
     79    // RFC2396 hostname = *( domainlabel "." ) toplabel [ "." ]
     80    // Note that the regex currently requires both a domain label and a top level label, whereas
     81    // the RFC does not. This is because the regex is used to detect if a TLD is present.
     82    // If the match fails, input is checked against DOMAIN_LABEL_REGEX (hostnameRegex)
     83    // RFC1123 sec 2.1 allows hostnames to start with a digit
    7284    private static final String DOMAIN_NAME_REGEX =
    73             "^(?:" + DOMAIN_LABEL_REGEX + "\\.)+" + "(" + TOP_LABEL_REGEX + "|" + TOP_LABEL_IDN_REGEX + ")$";
    74     // JOSM PATCH END
     85            "^(?:" + DOMAIN_LABEL_REGEX + "\\.)+" + "(" + TOP_LABEL_REGEX + ")\\.?$";
    7586
    7687    private final boolean allowLocal;
     
    94105            new RegexValidator(DOMAIN_NAME_REGEX);
    95106    /**
    96      * RegexValidator for matching the a local hostname
    97      */
     107     * RegexValidator for matching a local hostname
     108     */
     109    // RFC1123 sec 2.1 allows hostnames to start with a digit
    98110    private final RegexValidator hostnameRegex =
    99111            new RegexValidator(DOMAIN_LABEL_REGEX);
     
    104116     * @return the singleton instance of this validator
    105117     */
    106     public static DomainValidator getInstance() {
     118    public static synchronized DomainValidator getInstance() {
     119        inUse = true;
    107120        return DOMAIN_VALIDATOR;
    108121    }
     
    114127     * @return the singleton instance of this validator
    115128     */
    116     public static DomainValidator getInstance(boolean allowLocal) {
    117        if (allowLocal) {
    118           return DOMAIN_VALIDATOR_WITH_LOCAL;
    119        }
    120        return DOMAIN_VALIDATOR;
     129    public static synchronized DomainValidator getInstance(boolean allowLocal) {
     130        inUse = true;
     131        if (allowLocal) {
     132            return DOMAIN_VALIDATOR_WITH_LOCAL;
     133        }
     134        return DOMAIN_VALIDATOR;
    121135    }
    122136
     
    126140     */
    127141    private DomainValidator(boolean allowLocal) {
    128        this.allowLocal = allowLocal;
     142        this.allowLocal = allowLocal;
    129143    }
    130144
     
    132146     * Returns true if the specified <code>String</code> parses
    133147     * as a valid domain name with a recognized top-level domain.
    134      * The parsing is case-sensitive.
     148     * The parsing is case-insensitive.
    135149     * @param domain the parameter to check for domain name syntax
    136150     * @return true if the parameter is a valid domain name
     
    138152    @Override
    139153    public boolean isValid(String domain) {
     154        if (domain == null) {
     155            return false;
     156        }
     157        domain = unicodeToASCII(domain);
     158        // hosts must be equally reachable via punycode and Unicode
     159        // Unicode is never shorter than punycode, so check punycode
     160        // if domain did not convert, then it will be caught by ASCII
     161        // checks in the regexes below
     162        if (domain.length() > MAX_DOMAIN_LENGTH) {
     163            return false;
     164        }
    140165        String[] groups = domainRegex.match(domain);
    141166        if (groups != null && groups.length > 0) {
    142167            return isValidTld(groups[0]);
    143         } else if (allowLocal) {
    144             if (hostnameRegex.isValid(domain)) {
    145                return true;
    146             }
    147168        }
    148         return false;
     169        return allowLocal && hostnameRegex.isValid(domain);
     170    }
     171
     172    // package protected for unit test access
     173    // must agree with isValid() above
     174    boolean isValidDomainSyntax(String domain) {
     175        if (domain == null) {
     176            return false;
     177        }
     178        domain = unicodeToASCII(domain);
     179        // hosts must be equally reachable via punycode and Unicode
     180        // Unicode is never shorter than punycode, so check punycode
     181        // if domain did not convert, then it will be caught by ASCII
     182        // checks in the regexes below
     183        if (domain.length() > MAX_DOMAIN_LENGTH) {
     184            return false;
     185        }
     186        String[] groups = domainRegex.match(domain);
     187        return (groups != null && groups.length > 0)
     188                || hostnameRegex.isValid(domain);
    149189    }
    150190
     
    152192     * Returns true if the specified <code>String</code> matches any
    153193     * IANA-defined top-level domain. Leading dots are ignored if present.
    154      * The search is case-sensitive.
    155      * @param tld the parameter to check for TLD status
     194     * The search is case-insensitive.
     195     * @param tld the parameter to check for TLD status, not null
    156196     * @return true if the parameter is a TLD
    157197     */
    158198    public boolean isValidTld(String tld) {
     199        tld = unicodeToASCII(tld);
    159200        if (allowLocal && isValidLocalTld(tld)) {
    160            return true;
     201            return true;
    161202        }
    162203        return isValidInfrastructureTld(tld)
    163204                || isValidGenericTld(tld)
    164                 || isValidIdnTld(tld)
    165205                || isValidCountryCodeTld(tld);
    166206    }
     
    169209     * Returns true if the specified <code>String</code> matches any
    170210     * IANA-defined infrastructure top-level domain. Leading dots are
    171      * ignored if present. The search is case-sensitive.
    172      * @param iTld the parameter to check for infrastructure TLD status
     211     * ignored if present. The search is case-insensitive.
     212     * @param iTld the parameter to check for infrastructure TLD status, not null
    173213     * @return true if the parameter is an infrastructure TLD
    174214     */
    175215    public boolean isValidInfrastructureTld(String iTld) {
    176         return Arrays.binarySearch(INFRASTRUCTURE_TLDS, chompLeadingDot(iTld.toLowerCase(Locale.ENGLISH))) >= 0;
     216        final String key = chompLeadingDot(unicodeToASCII(iTld).toLowerCase(Locale.ENGLISH));
     217        return arrayContains(INFRASTRUCTURE_TLDS, key);
    177218    }
    178219
     
    180221     * Returns true if the specified <code>String</code> matches any
    181222     * IANA-defined generic top-level domain. Leading dots are ignored
    182      * if present. The search is case-sensitive.
    183      * @param gTld the parameter to check for generic TLD status
     223     * if present. The search is case-insensitive.
     224     * @param gTld the parameter to check for generic TLD status, not null
    184225     * @return true if the parameter is a generic TLD
    185226     */
    186227    public boolean isValidGenericTld(String gTld) {
    187         return Arrays.binarySearch(GENERIC_TLDS, chompLeadingDot(gTld.toLowerCase(Locale.ENGLISH))) >= 0;
    188     }
    189 
    190     /**
    191      * Returns true if the specified <code>String</code> matches any
    192      * IANA-defined IDN top-level domain. Leading dots are ignored
    193      * if present. The search is case-sensitive.
    194      * @param iTld the parameter to check for IDN TLD status
    195      * @return true if the parameter is an IDN TLD
    196      */
    197     public boolean isValidIdnTld(String iTld) {
    198         return Arrays.binarySearch(IDN_TLDS, chompLeadingDot(iTld.toUpperCase(Locale.ENGLISH))) >= 0;
     228        final String key = chompLeadingDot(unicodeToASCII(gTld).toLowerCase(Locale.ENGLISH));
     229        return (arrayContains(GENERIC_TLDS, key) || arrayContains(genericTLDsPlus, key))
     230                && !arrayContains(genericTLDsMinus, key);
    199231    }
    200232
     
    202234     * Returns true if the specified <code>String</code> matches any
    203235     * IANA-defined country code top-level domain. Leading dots are
    204      * ignored if present. The search is case-sensitive.
    205      * @param ccTld the parameter to check for country code TLD status
     236     * ignored if present. The search is case-insensitive.
     237     * @param ccTld the parameter to check for country code TLD status, not null
    206238     * @return true if the parameter is a country code TLD
    207239     */
    208240    public boolean isValidCountryCodeTld(String ccTld) {
    209         return Arrays.binarySearch(COUNTRY_CODE_TLDS, chompLeadingDot(ccTld.toLowerCase(Locale.ENGLISH))) >= 0;
     241        final String key = chompLeadingDot(unicodeToASCII(ccTld).toLowerCase(Locale.ENGLISH));
     242        return (arrayContains(COUNTRY_CODE_TLDS, key) || arrayContains(countryCodeTLDsPlus, key))
     243                && !arrayContains(countryCodeTLDsMinus, key);
    210244    }
    211245
     
    213247     * Returns true if the specified <code>String</code> matches any
    214248     * widely used "local" domains (localhost or localdomain). Leading dots are
    215      *  ignored if present. The search is case-sensitive.
    216      * @param iTld the parameter to check for local TLD status
     249     * ignored if present. The search is case-insensitive.
     250     * @param lTld the parameter to check for local TLD status, not null
    217251     * @return true if the parameter is an local TLD
    218252     */
    219     public boolean isValidLocalTld(String iTld) {
    220         return Arrays.binarySearch(LOCAL_TLDS, chompLeadingDot(iTld.toLowerCase(Locale.ENGLISH))) >= 0;
     253    public boolean isValidLocalTld(String lTld) {
     254        final String key = chompLeadingDot(unicodeToASCII(lTld).toLowerCase(Locale.ENGLISH));
     255        return arrayContains(LOCAL_TLDS, key);
    221256    }
    222257
     
    224259        if (str.startsWith(".")) {
    225260            return str.substring(1);
    226         } else {
    227             return str;
    228261        }
     262        return str;
    229263    }
    230264
     
    234268    // ----- http://data.iana.org/TLD/tlds-alpha-by-domain.txt
    235269
     270    // Note that the above list is in UPPER case.
     271    // The code currently converts strings to lower case (as per the tables below)
     272
     273    // IANA also provide an HTML list at http://www.iana.org/domains/root/db
     274    // Note that this contains several country code entries which are NOT in
     275    // the text file. These all have the "Not assigned" in the "Sponsoring Organisation" column
     276    // For example (as of 2015-01-02):
     277    // .bl  country-code    Not assigned
     278    // .um  country-code    Not assigned
     279
     280    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
    236281    private static final String[] INFRASTRUCTURE_TLDS = new String[] {
    237282        "arpa",               // internet infrastructure
    238         "root"                // diagnostic marker for non-truncated root zone
    239283    };
    240284
     285    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
    241286    private static final String[] GENERIC_TLDS = new String[] {
    242         "abogado",
    243         "academy",
    244         "accountants",
    245         "active",
    246         "actor",
    247         "aero",
    248         "agency",
    249         "airforce",
    250         "allfinanz",
    251         "alsace",
    252         "archi",
    253         "army",
    254         "arpa",
    255         "asia",
    256         "associates",
    257         "attorney",
    258         "auction",
    259         "audio",
    260         "autos",
    261         "axa",
    262         "band",
    263         "bar",
    264         "bargains",
    265         "bayern",
    266         "beer",
    267         "berlin",
    268         "best",
    269         "bid",
    270         "bike",
    271         "bio",
    272         "biz",
    273         "black",
    274         "blackfriday",
    275         "blue",
    276         "bmw",
    277         "bnpparibas",
    278         "boo",
    279         "boutique",
    280         "brussels",
    281         "budapest",
    282         "build",
    283         "builders",
    284         "business",
    285         "buzz",
    286         "bzh",
    287         "cab",
    288         "cal",
    289         "camera",
    290         "camp",
    291         "cancerresearch",
    292         "capetown",
    293         "capital",
    294         "caravan",
    295         "cards",
    296         "care",
    297         "career",
    298         "careers",
    299         "casa",
    300         "cash",
    301         "cat",
    302         "catering",
    303         "center",
    304         "ceo",
    305         "cern",
    306         "channel",
    307         "cheap",
    308         "christmas",
    309         "chrome",
    310         "church",
    311         "citic",
    312         "city",
    313         "claims",
    314         "cleaning",
    315         "click",
    316         "clinic",
    317         "clothing",
    318         "club",
    319         "codes",
    320         "coffee",
    321         "college",
    322         "cologne",
    323         "com",
    324         "community",
    325         "company",
    326         "computer",
    327         "condos",
    328         "construction",
    329         "consulting",
    330         "contractors",
    331         "cooking",
    332         "cool",
    333         "coop",
    334         "country",
    335         "credit",
    336         "creditcard",
    337         "crs",
    338         "cruises",
    339         "cuisinella",
    340         "cymru",
    341         "dad",
    342         "dance",
    343         "dating",
    344         "day",
    345         "deals",
    346         "degree",
    347         "democrat",
    348         "dental",
    349         "dentist",
    350         "desi",
    351         "diamonds",
    352         "diet",
    353         "digital",
    354         "direct",
    355         "directory",
    356         "discount",
    357         "dnp",
    358         "domains",
    359         "durban",
    360         "dvag",
    361         "eat",
    362         "edu",
    363         "education",
    364         "email",
    365         "engineer",
    366         "engineering",
    367         "enterprises",
    368         "equipment",
    369         "esq",
    370         "estate",
    371         "eus",
    372         "events",
    373         "exchange",
    374         "expert",
    375         "exposed",
    376         "fail",
    377         "farm",
    378         "feedback",
    379         "finance",
    380         "financial",
    381         "fish",
    382         "fishing",
    383         "fitness",
    384         "flights",
    385         "florist",
    386         "flsmidth",
    387         "fly",
    388         "foo",
    389         "forsale",
    390         "foundation",
    391         "frl",
    392         "frogans",
    393         "fund",
    394         "furniture",
    395         "futbol",
    396         "gal",
    397         "gallery",
    398         "gbiz",
    399         "gent",
    400         "gift",
    401         "gifts",
    402         "gives",
    403         "glass",
    404         "gle",
    405         "global",
    406         "globo",
    407         "gmail",
    408         "gmo",
    409         "gmx",
    410         "google",
    411         "gop",
    412         "gov",
    413         "graphics",
    414         "gratis",
    415         "green",
    416         "gripe",
    417         "guide",
    418         "guitars",
    419         "guru",
    420         "hamburg",
    421         "haus",
    422         "healthcare",
    423         "help",
    424         "here",
    425         "hiphop",
    426         "hiv",
    427         "holdings",
    428         "holiday",
    429         "homes",
    430         "horse",
    431         "host",
    432         "hosting",
    433         "house",
    434         "how",
    435         "ibm",
    436         "immo",
    437         "immobilien",
    438         "industries",
    439         "info",
    440         "ing",
    441         "ink",
    442         "institute",
    443         "insure",
    444         "int",
    445         "international",
    446         "investments",
    447         "jetzt",
    448         "jobs",
    449         "joburg",
    450         "juegos",
    451         "kaufen",
    452         "kim",
    453         "kitchen",
    454         "kiwi",
    455         "koeln",
    456         "krd",
    457         "kred",
    458         "lacaixa",
    459         "land",
    460         "lawyer",
    461         "lease",
    462         "lgbt",
    463         "life",
    464         "lighting",
    465         "limited",
    466         "limo",
    467         "link",
    468         "loans",
    469         "london",
    470         "lotto",
    471         "ltda",
    472         "luxe",
    473         "luxury",
    474         "maison",
    475         "management",
    476         "mango",
    477         "market",
    478         "marketing",
    479         "media",
    480         "meet",
    481         "melbourne",
    482         "meme",
    483         "menu",
    484         "miami",
    485         "mil",
    486         "mini",
    487         "mobi",
    488         "moda",
    489         "moe",
    490         "monash",
    491         "mortgage",
    492         "moscow",
    493         "motorcycles",
    494         "mov",
    495         "museum",
    496         "nagoya",
    497         "name",
    498         "navy",
    499         "net",
    500         "network",
    501         "neustar",
    502         "new",
    503         "nexus",
    504         "ngo",
    505         "nhk",
    506         "ninja",
    507         "nra",
    508         "nrw",
    509         "nyc",
    510         "okinawa",
    511         "ong",
    512         "onl",
    513         "ooo",
    514         "org",
    515         "organic",
    516         "otsuka",
    517         "ovh",
    518         "paris",
    519         "partners",
    520         "parts",
    521         "pharmacy",
    522         "photo",
    523         "photography",
    524         "photos",
    525         "physio",
    526         "pics",
    527         "pictures",
    528         "pink",
    529         "pizza",
    530         "place",
    531         "plumbing",
    532         "pohl",
    533         "poker",
    534         "post",
    535         "praxi",
    536         "press",
    537         "pro",
    538         "prod",
    539         "productions",
    540         "prof",
    541         "properties",
    542         "property",
    543         "pub",
    544         "qpon",
    545         "quebec",
    546         "realtor",
    547         "recipes",
    548         "red",
    549         "rehab",
    550         "reise",
    551         "reisen",
    552         "ren",
    553         "rentals",
    554         "repair",
    555         "report",
    556         "republican",
    557         "rest",
    558         "restaurant",
    559         "reviews",
    560         "rich",
    561         "rio",
    562         "rip",
    563         "rocks",
    564         "rodeo",
    565         "rsvp",
    566         "ruhr",
    567         "ryukyu",
    568         "saarland",
    569         "sarl",
    570         "sca",
    571         "scb",
    572         "schmidt",
    573         "schule",
    574         "scot",
    575         "services",
    576         "sexy",
    577         "shiksha",
    578         "shoes",
    579         "singles",
    580         "social",
    581         "software",
    582         "sohu",
    583         "solar",
    584         "solutions",
    585         "soy",
    586         "space",
    587         "spiegel",
    588         "supplies",
    589         "supply",
    590         "support",
    591         "surf",
    592         "surgery",
    593         "suzuki",
    594         "systems",
    595         "tatar",
    596         "tattoo",
    597         "tax",
    598         "technology",
    599         "tel",
    600         "tienda",
    601         "tips",
    602         "tirol",
    603         "today",
    604         "tokyo",
    605         "tools",
    606         "top",
    607         "town",
    608         "toys",
    609         "trade",
    610         "training",
    611         "travel",
    612         "tui",
    613         "university",
    614         "uno",
    615         "uol",
    616         "vacations",
    617         "vegas",
    618         "ventures",
    619         "versicherung",
    620         "vet",
    621         "viajes",
    622         "villas",
    623         "vision",
    624         "vlaanderen",
    625         "vodka",
    626         "vote",
    627         "voting",
    628         "voto",
    629         "voyage",
    630         "wales",
    631         "wang",
    632         "watch",
    633         "webcam",
    634         "website",
    635         "wed",
    636         "wedding",
    637         "whoswho",
    638         "wien",
    639         "wiki",
    640         "williamhill",
    641         "wme",
    642         "work",
    643         "works",
    644         "world",
    645         "wtc",
    646         "wtf",
    647         "xxx",
    648         "xyz",
    649         "yachts",
    650         "yandex",
    651         "yoga",
    652         "yokohama",
    653         "youtube",
    654         "zip",
    655         "zone",
     287         // Taken from Version 2016011900, Last Updated Tue Jan 19 07:07:02 2016 UTC
     288        "aaa", // aaa American Automobile Association, Inc.
     289        "aarp", // aarp AARP
     290        "abb", // abb ABB Ltd
     291        "abbott", // abbott Abbott Laboratories, Inc.
     292        "abogado", // abogado Top Level Domain Holdings Limited
     293        "academy", // academy Half Oaks, LLC
     294        "accenture", // accenture Accenture plc
     295        "accountant", // accountant dot Accountant Limited
     296        "accountants", // accountants Knob Town, LLC
     297        "aco", // aco ACO Severin Ahlmann GmbH &amp; Co. KG
     298        "active", // active The Active Network, Inc
     299        "actor", // actor United TLD Holdco Ltd.
     300        "adac", // adac Allgemeiner Deutscher Automobil-Club e.V. (ADAC)
     301        "ads", // ads Charleston Road Registry Inc.
     302        "adult", // adult ICM Registry AD LLC
     303        "aeg", // aeg Aktiebolaget Electrolux
     304        "aero", // aero Societe Internationale de Telecommunications Aeronautique (SITA INC USA)
     305        "afl", // afl Australian Football League
     306        "agency", // agency Steel Falls, LLC
     307        "aig", // aig American International Group, Inc.
     308        "airforce", // airforce United TLD Holdco Ltd.
     309        "airtel", // airtel Bharti Airtel Limited
     310        "alibaba", // alibaba Alibaba Group Holding Limited
     311        "alipay", // alipay Alibaba Group Holding Limited
     312        "allfinanz", // allfinanz Allfinanz Deutsche Vermögensberatung Aktiengesellschaft
     313        "alsace", // alsace REGION D ALSACE
     314        "amica", // amica Amica Mutual Insurance Company
     315        "amsterdam", // amsterdam Gemeente Amsterdam
     316        "analytics", // analytics Campus IP LLC
     317        "android", // android Charleston Road Registry Inc.
     318        "apartments", // apartments June Maple, LLC
     319        "app", // app Charleston Road Registry Inc.
     320        "apple", // apple Apple Inc.
     321        "aquarelle", // aquarelle Aquarelle.com
     322        "aramco", // aramco Aramco Services Company
     323        "archi", // archi STARTING DOT LIMITED
     324        "army", // army United TLD Holdco Ltd.
     325        "arte", // arte Association Relative à la Télévision Européenne G.E.I.E.
     326        "asia", // asia DotAsia Organisation Ltd.
     327        "associates", // associates Baxter Hill, LLC
     328        "attorney", // attorney United TLD Holdco, Ltd
     329        "auction", // auction United TLD HoldCo, Ltd.
     330        "audi", // audi AUDI Aktiengesellschaft
     331        "audio", // audio Uniregistry, Corp.
     332        "author", // author Amazon Registry Services, Inc.
     333        "auto", // auto Uniregistry, Corp.
     334        "autos", // autos DERAutos, LLC
     335        "axa", // axa AXA SA
     336        "azure", // azure Microsoft Corporation
     337        "baidu", // baidu Baidu, Inc.
     338        "band", // band United TLD Holdco, Ltd
     339        "bank", // bank fTLD Registry Services, LLC
     340        "bar", // bar Punto 2012 Sociedad Anonima Promotora de Inversion de Capital Variable
     341        "barcelona", // barcelona Municipi de Barcelona
     342        "barclaycard", // barclaycard Barclays Bank PLC
     343        "barclays", // barclays Barclays Bank PLC
     344        "bargains", // bargains Half Hallow, LLC
     345        "bauhaus", // bauhaus Werkhaus GmbH
     346        "bayern", // bayern Bayern Connect GmbH
     347        "bbc", // bbc British Broadcasting Corporation
     348        "bbva", // bbva BANCO BILBAO VIZCAYA ARGENTARIA, S.A.
     349        "bcn", // bcn Municipi de Barcelona
     350        "beats", // beats Beats Electronics, LLC
     351        "beer", // beer Top Level Domain Holdings Limited
     352        "bentley", // bentley Bentley Motors Limited
     353        "berlin", // berlin dotBERLIN GmbH &amp; Co. KG
     354        "best", // best BestTLD Pty Ltd
     355        "bet", // bet Afilias plc
     356        "bharti", // bharti Bharti Enterprises (Holding) Private Limited
     357        "bible", // bible American Bible Society
     358        "bid", // bid dot Bid Limited
     359        "bike", // bike Grand Hollow, LLC
     360        "bing", // bing Microsoft Corporation
     361        "bingo", // bingo Sand Cedar, LLC
     362        "bio", // bio STARTING DOT LIMITED
     363        "biz", // biz Neustar, Inc.
     364        "black", // black Afilias Limited
     365        "blackfriday", // blackfriday Uniregistry, Corp.
     366        "bloomberg", // bloomberg Bloomberg IP Holdings LLC
     367        "blue", // blue Afilias Limited
     368        "bms", // bms Bristol-Myers Squibb Company
     369        "bmw", // bmw Bayerische Motoren Werke Aktiengesellschaft
     370        "bnl", // bnl Banca Nazionale del Lavoro
     371        "bnpparibas", // bnpparibas BNP Paribas
     372        "boats", // boats DERBoats, LLC
     373        "boehringer", // boehringer Boehringer Ingelheim International GmbH
     374        "bom", // bom Núcleo de Informação e Coordenação do Ponto BR - NIC.br
     375        "bond", // bond Bond University Limited
     376        "boo", // boo Charleston Road Registry Inc.
     377        "book", // book Amazon Registry Services, Inc.
     378        "boots", // boots THE BOOTS COMPANY PLC
     379        "bosch", // bosch Robert Bosch GMBH
     380        "bostik", // bostik Bostik SA
     381        "bot", // bot Amazon Registry Services, Inc.
     382        "boutique", // boutique Over Galley, LLC
     383        "bradesco", // bradesco Banco Bradesco S.A.
     384        "bridgestone", // bridgestone Bridgestone Corporation
     385        "broadway", // broadway Celebrate Broadway, Inc.
     386        "broker", // broker DOTBROKER REGISTRY LTD
     387        "brother", // brother Brother Industries, Ltd.
     388        "brussels", // brussels DNS.be vzw
     389        "budapest", // budapest Top Level Domain Holdings Limited
     390        "bugatti", // bugatti Bugatti International SA
     391        "build", // build Plan Bee LLC
     392        "builders", // builders Atomic Madison, LLC
     393        "business", // business Spring Cross, LLC
     394        "buy", // buy Amazon Registry Services, INC
     395        "buzz", // buzz DOTSTRATEGY CO.
     396        "bzh", // bzh Association www.bzh
     397        "cab", // cab Half Sunset, LLC
     398        "cafe", // cafe Pioneer Canyon, LLC
     399        "cal", // cal Charleston Road Registry Inc.
     400        "call", // call Amazon Registry Services, Inc.
     401        "camera", // camera Atomic Maple, LLC
     402        "camp", // camp Delta Dynamite, LLC
     403        "cancerresearch", // cancerresearch Australian Cancer Research Foundation
     404        "canon", // canon Canon Inc.
     405        "capetown", // capetown ZA Central Registry NPC trading as ZA Central Registry
     406        "capital", // capital Delta Mill, LLC
     407        "car", // car Cars Registry Limited
     408        "caravan", // caravan Caravan International, Inc.
     409        "cards", // cards Foggy Hollow, LLC
     410        "care", // care Goose Cross, LLC
     411        "career", // career dotCareer LLC
     412        "careers", // careers Wild Corner, LLC
     413        "cars", // cars Uniregistry, Corp.
     414        "cartier", // cartier Richemont DNS Inc.
     415        "casa", // casa Top Level Domain Holdings Limited
     416        "cash", // cash Delta Lake, LLC
     417        "casino", // casino Binky Sky, LLC
     418        "cat", // cat Fundacio puntCAT
     419        "catering", // catering New Falls. LLC
     420        "cba", // cba COMMONWEALTH BANK OF AUSTRALIA
     421        "cbn", // cbn The Christian Broadcasting Network, Inc.
     422        "ceb", // ceb The Corporate Executive Board Company
     423        "center", // center Tin Mill, LLC
     424        "ceo", // ceo CEOTLD Pty Ltd
     425        "cern", // cern European Organization for Nuclear Research (&quot;CERN&quot;)
     426        "cfa", // cfa CFA Institute
     427        "cfd", // cfd DOTCFD REGISTRY LTD
     428        "chanel", // chanel Chanel International B.V.
     429        "channel", // channel Charleston Road Registry Inc.
     430        "chat", // chat Sand Fields, LLC
     431        "cheap", // cheap Sand Cover, LLC
     432        "chloe", // chloe Richemont DNS Inc.
     433        "christmas", // christmas Uniregistry, Corp.
     434        "chrome", // chrome Charleston Road Registry Inc.
     435        "church", // church Holly Fileds, LLC
     436        "cipriani", // cipriani Hotel Cipriani Srl
     437        "circle", // circle Amazon Registry Services, Inc.
     438        "cisco", // cisco Cisco Technology, Inc.
     439        "citic", // citic CITIC Group Corporation
     440        "city", // city Snow Sky, LLC
     441        "cityeats", // cityeats Lifestyle Domain Holdings, Inc.
     442        "claims", // claims Black Corner, LLC
     443        "cleaning", // cleaning Fox Shadow, LLC
     444        "click", // click Uniregistry, Corp.
     445        "clinic", // clinic Goose Park, LLC
     446        "clinique", // clinique The Estée Lauder Companies Inc.
     447        "clothing", // clothing Steel Lake, LLC
     448        "cloud", // cloud ARUBA S.p.A.
     449        "club", // club .CLUB DOMAINS, LLC
     450        "clubmed", // clubmed Club Méditerranée S.A.
     451        "coach", // coach Koko Island, LLC
     452        "codes", // codes Puff Willow, LLC
     453        "coffee", // coffee Trixy Cover, LLC
     454        "college", // college XYZ.COM LLC
     455        "cologne", // cologne NetCologne Gesellschaft für Telekommunikation mbH
     456        "com", // com VeriSign Global Registry Services
     457        "commbank", // commbank COMMONWEALTH BANK OF AUSTRALIA
     458        "community", // community Fox Orchard, LLC
     459        "company", // company Silver Avenue, LLC
     460        "compare", // compare iSelect Ltd
     461        "computer", // computer Pine Mill, LLC
     462        "comsec", // comsec VeriSign, Inc.
     463        "condos", // condos Pine House, LLC
     464        "construction", // construction Fox Dynamite, LLC
     465        "consulting", // consulting United TLD Holdco, LTD.
     466        "contact", // contact Top Level Spectrum, Inc.
     467        "contractors", // contractors Magic Woods, LLC
     468        "cooking", // cooking Top Level Domain Holdings Limited
     469        "cool", // cool Koko Lake, LLC
     470        "coop", // coop DotCooperation LLC
     471        "corsica", // corsica Collectivité Territoriale de Corse
     472        "country", // country Top Level Domain Holdings Limited
     473        "coupon", // coupon Amazon Registry Services, Inc.
     474        "coupons", // coupons Black Island, LLC
     475        "courses", // courses OPEN UNIVERSITIES AUSTRALIA PTY LTD
     476        "credit", // credit Snow Shadow, LLC
     477        "creditcard", // creditcard Binky Frostbite, LLC
     478        "creditunion", // creditunion CUNA Performance Resources, LLC
     479        "cricket", // cricket dot Cricket Limited
     480        "crown", // crown Crown Equipment Corporation
     481        "crs", // crs Federated Co-operatives Limited
     482        "cruises", // cruises Spring Way, LLC
     483        "csc", // csc Alliance-One Services, Inc.
     484        "cuisinella", // cuisinella SALM S.A.S.
     485        "cymru", // cymru Nominet UK
     486        "cyou", // cyou Beijing Gamease Age Digital Technology Co., Ltd.
     487        "dabur", // dabur Dabur India Limited
     488        "dad", // dad Charleston Road Registry Inc.
     489        "dance", // dance United TLD Holdco Ltd.
     490        "date", // date dot Date Limited
     491        "dating", // dating Pine Fest, LLC
     492        "datsun", // datsun NISSAN MOTOR CO., LTD.
     493        "day", // day Charleston Road Registry Inc.
     494        "dclk", // dclk Charleston Road Registry Inc.
     495        "dealer", // dealer Dealer Dot Com, Inc.
     496        "deals", // deals Sand Sunset, LLC
     497        "degree", // degree United TLD Holdco, Ltd
     498        "delivery", // delivery Steel Station, LLC
     499        "dell", // dell Dell Inc.
     500        "deloitte", // deloitte Deloitte Touche Tohmatsu
     501        "delta", // delta Delta Air Lines, Inc.
     502        "democrat", // democrat United TLD Holdco Ltd.
     503        "dental", // dental Tin Birch, LLC
     504        "dentist", // dentist United TLD Holdco, Ltd
     505        "desi", // desi Desi Networks LLC
     506        "design", // design Top Level Design, LLC
     507        "dev", // dev Charleston Road Registry Inc.
     508        "diamonds", // diamonds John Edge, LLC
     509        "diet", // diet Uniregistry, Corp.
     510        "digital", // digital Dash Park, LLC
     511        "direct", // direct Half Trail, LLC
     512        "directory", // directory Extra Madison, LLC
     513        "discount", // discount Holly Hill, LLC
     514        "dnp", // dnp Dai Nippon Printing Co., Ltd.
     515        "docs", // docs Charleston Road Registry Inc.
     516        "dog", // dog Koko Mill, LLC
     517        "doha", // doha Communications Regulatory Authority (CRA)
     518        "domains", // domains Sugar Cross, LLC
     519        "doosan", // doosan Doosan Corporation
     520        "download", // download dot Support Limited
     521        "drive", // drive Charleston Road Registry Inc.
     522        "dubai", // dubai Dubai Smart Government Department
     523        "durban", // durban ZA Central Registry NPC trading as ZA Central Registry
     524        "dvag", // dvag Deutsche Vermögensberatung Aktiengesellschaft DVAG
     525        "earth", // earth Interlink Co., Ltd.
     526        "eat", // eat Charleston Road Registry Inc.
     527        "edeka", // edeka EDEKA Verband kaufmännischer Genossenschaften e.V.
     528        "edu", // edu EDUCAUSE
     529        "education", // education Brice Way, LLC
     530        "email", // email Spring Madison, LLC
     531        "emerck", // emerck Merck KGaA
     532        "energy", // energy Binky Birch, LLC
     533        "engineer", // engineer United TLD Holdco Ltd.
     534        "engineering", // engineering Romeo Canyon
     535        "enterprises", // enterprises Snow Oaks, LLC
     536        "epson", // epson Seiko Epson Corporation
     537        "equipment", // equipment Corn Station, LLC
     538        "erni", // erni ERNI Group Holding AG
     539        "esq", // esq Charleston Road Registry Inc.
     540        "estate", // estate Trixy Park, LLC
     541        "eurovision", // eurovision European Broadcasting Union (EBU)
     542        "eus", // eus Puntueus Fundazioa
     543        "events", // events Pioneer Maple, LLC
     544        "everbank", // everbank EverBank
     545        "exchange", // exchange Spring Falls, LLC
     546        "expert", // expert Magic Pass, LLC
     547        "exposed", // exposed Victor Beach, LLC
     548        "express", // express Sea Sunset, LLC
     549        "fage", // fage Fage International S.A.
     550        "fail", // fail Atomic Pipe, LLC
     551        "fairwinds", // fairwinds FairWinds Partners, LLC
     552        "faith", // faith dot Faith Limited
     553        "family", // family United TLD Holdco Ltd.
     554        "fan", // fan Asiamix Digital Ltd
     555        "fans", // fans Asiamix Digital Limited
     556        "farm", // farm Just Maple, LLC
     557        "fashion", // fashion Top Level Domain Holdings Limited
     558        "fast", // fast Amazon Registry Services, Inc.
     559        "feedback", // feedback Top Level Spectrum, Inc.
     560        "ferrero", // ferrero Ferrero Trading Lux S.A.
     561        "film", // film Motion Picture Domain Registry Pty Ltd
     562        "final", // final Núcleo de Informação e Coordenação do Ponto BR - NIC.br
     563        "finance", // finance Cotton Cypress, LLC
     564        "financial", // financial Just Cover, LLC
     565        "firestone", // firestone Bridgestone Corporation
     566        "firmdale", // firmdale Firmdale Holdings Limited
     567        "fish", // fish Fox Woods, LLC
     568        "fishing", // fishing Top Level Domain Holdings Limited
     569        "fit", // fit Minds + Machines Group Limited
     570        "fitness", // fitness Brice Orchard, LLC
     571        "flickr", // flickr Yahoo! Domain Services Inc.
     572        "flights", // flights Fox Station, LLC
     573        "florist", // florist Half Cypress, LLC
     574        "flowers", // flowers Uniregistry, Corp.
     575        "flsmidth", // flsmidth FLSmidth A/S
     576        "fly", // fly Charleston Road Registry Inc.
     577        "foo", // foo Charleston Road Registry Inc.
     578        "football", // football Foggy Farms, LLC
     579        "ford", // ford Ford Motor Company
     580        "forex", // forex DOTFOREX REGISTRY LTD
     581        "forsale", // forsale United TLD Holdco, LLC
     582        "forum", // forum Fegistry, LLC
     583        "foundation", // foundation John Dale, LLC
     584        "fox", // fox FOX Registry, LLC
     585        "fresenius", // fresenius Fresenius Immobilien-Verwaltungs-GmbH
     586        "frl", // frl FRLregistry B.V.
     587        "frogans", // frogans OP3FT
     588        "frontier", // frontier Frontier Communications Corporation
     589        "fund", // fund John Castle, LLC
     590        "furniture", // furniture Lone Fields, LLC
     591        "futbol", // futbol United TLD Holdco, Ltd.
     592        "fyi", // fyi Silver Tigers, LLC
     593        "gal", // gal Asociación puntoGAL
     594        "gallery", // gallery Sugar House, LLC
     595        "gallup", // gallup Gallup, Inc.
     596        "game", // game Uniregistry, Corp.
     597        "garden", // garden Top Level Domain Holdings Limited
     598        "gbiz", // gbiz Charleston Road Registry Inc.
     599        "gdn", // gdn Joint Stock Company "Navigation-information systems"
     600        "gea", // gea GEA Group Aktiengesellschaft
     601        "gent", // gent COMBELL GROUP NV/SA
     602        "genting", // genting Resorts World Inc. Pte. Ltd.
     603        "ggee", // ggee GMO Internet, Inc.
     604        "gift", // gift Uniregistry, Corp.
     605        "gifts", // gifts Goose Sky, LLC
     606        "gives", // gives United TLD Holdco Ltd.
     607        "giving", // giving Giving Limited
     608        "glass", // glass Black Cover, LLC
     609        "gle", // gle Charleston Road Registry Inc.
     610        "global", // global Dot Global Domain Registry Limited
     611        "globo", // globo Globo Comunicação e Participações S.A
     612        "gmail", // gmail Charleston Road Registry Inc.
     613        "gmo", // gmo GMO Internet, Inc.
     614        "gmx", // gmx 1&amp;1 Mail &amp; Media GmbH
     615        "gold", // gold June Edge, LLC
     616        "goldpoint", // goldpoint YODOBASHI CAMERA CO.,LTD.
     617        "golf", // golf Lone Falls, LLC
     618        "goo", // goo NTT Resonant Inc.
     619        "goog", // goog Charleston Road Registry Inc.
     620        "google", // google Charleston Road Registry Inc.
     621        "gop", // gop Republican State Leadership Committee, Inc.
     622        "got", // got Amazon Registry Services, Inc.
     623        "gov", // gov General Services Administration Attn: QTDC, 2E08 (.gov Domain Registration)
     624        "grainger", // grainger Grainger Registry Services, LLC
     625        "graphics", // graphics Over Madison, LLC
     626        "gratis", // gratis Pioneer Tigers, LLC
     627        "green", // green Afilias Limited
     628        "gripe", // gripe Corn Sunset, LLC
     629        "group", // group Romeo Town, LLC
     630        "gucci", // gucci Guccio Gucci S.p.a.
     631        "guge", // guge Charleston Road Registry Inc.
     632        "guide", // guide Snow Moon, LLC
     633        "guitars", // guitars Uniregistry, Corp.
     634        "guru", // guru Pioneer Cypress, LLC
     635        "hamburg", // hamburg Hamburg Top-Level-Domain GmbH
     636        "hangout", // hangout Charleston Road Registry Inc.
     637        "haus", // haus United TLD Holdco, LTD.
     638        "hdfcbank", // hdfcbank HDFC Bank Limited
     639        "health", // health DotHealth, LLC
     640        "healthcare", // healthcare Silver Glen, LLC
     641        "help", // help Uniregistry, Corp.
     642        "helsinki", // helsinki City of Helsinki
     643        "here", // here Charleston Road Registry Inc.
     644        "hermes", // hermes Hermes International
     645        "hiphop", // hiphop Uniregistry, Corp.
     646        "hitachi", // hitachi Hitachi, Ltd.
     647        "hiv", // hiv dotHIV gemeinnuetziger e.V.
     648        "hockey", // hockey Half Willow, LLC
     649        "holdings", // holdings John Madison, LLC
     650        "holiday", // holiday Goose Woods, LLC
     651        "homedepot", // homedepot Homer TLC, Inc.
     652        "homes", // homes DERHomes, LLC
     653        "honda", // honda Honda Motor Co., Ltd.
     654        "horse", // horse Top Level Domain Holdings Limited
     655        "host", // host DotHost Inc.
     656        "hosting", // hosting Uniregistry, Corp.
     657        "hoteles", // hoteles Travel Reservations SRL
     658        "hotmail", // hotmail Microsoft Corporation
     659        "house", // house Sugar Park, LLC
     660        "how", // how Charleston Road Registry Inc.
     661        "hsbc", // hsbc HSBC Holdings PLC
     662        "hyundai", // hyundai Hyundai Motor Company
     663        "ibm", // ibm International Business Machines Corporation
     664        "icbc", // icbc Industrial and Commercial Bank of China Limited
     665        "ice", // ice IntercontinentalExchange, Inc.
     666        "icu", // icu One.com A/S
     667        "ifm", // ifm ifm electronic gmbh
     668        "iinet", // iinet Connect West Pty. Ltd.
     669        "immo", // immo Auburn Bloom, LLC
     670        "immobilien", // immobilien United TLD Holdco Ltd.
     671        "industries", // industries Outer House, LLC
     672        "infiniti", // infiniti NISSAN MOTOR CO., LTD.
     673        "info", // info Afilias Limited
     674        "ing", // ing Charleston Road Registry Inc.
     675        "ink", // ink Top Level Design, LLC
     676        "institute", // institute Outer Maple, LLC
     677        "insurance", // insurance fTLD Registry Services LLC
     678        "insure", // insure Pioneer Willow, LLC
     679        "int", // int Internet Assigned Numbers Authority
     680        "international", // international Wild Way, LLC
     681        "investments", // investments Holly Glen, LLC
     682        "ipiranga", // ipiranga Ipiranga Produtos de Petroleo S.A.
     683        "irish", // irish Dot-Irish LLC
     684        "iselect", // iselect iSelect Ltd
     685        "ist", // ist Istanbul Metropolitan Municipality
     686        "istanbul", // istanbul Istanbul Metropolitan Municipality / Medya A.S.
     687        "itau", // itau Itau Unibanco Holding S.A.
     688        "iwc", // iwc Richemont DNS Inc.
     689        "jaguar", // jaguar Jaguar Land Rover Ltd
     690        "java", // java Oracle Corporation
     691        "jcb", // jcb JCB Co., Ltd.
     692        "jetzt", // jetzt New TLD Company AB
     693        "jewelry", // jewelry Wild Bloom, LLC
     694        "jlc", // jlc Richemont DNS Inc.
     695        "jll", // jll Jones Lang LaSalle Incorporated
     696        "jmp", // jmp Matrix IP LLC
     697        "jobs", // jobs Employ Media LLC
     698        "joburg", // joburg ZA Central Registry NPC trading as ZA Central Registry
     699        "jot", // jot Amazon Registry Services, Inc.
     700        "joy", // joy Amazon Registry Services, Inc.
     701        "jprs", // jprs Japan Registry Services Co., Ltd.
     702        "juegos", // juegos Uniregistry, Corp.
     703        "kaufen", // kaufen United TLD Holdco Ltd.
     704        "kddi", // kddi KDDI CORPORATION
     705        "kfh", // kfh Kuwait Finance House
     706        "kia", // kia KIA MOTORS CORPORATION
     707        "kim", // kim Afilias Limited
     708        "kinder", // kinder Ferrero Trading Lux S.A.
     709        "kitchen", // kitchen Just Goodbye, LLC
     710        "kiwi", // kiwi DOT KIWI LIMITED
     711        "koeln", // koeln NetCologne Gesellschaft für Telekommunikation mbH
     712        "komatsu", // komatsu Komatsu Ltd.
     713        "kpn", // kpn Koninklijke KPN N.V.
     714        "krd", // krd KRG Department of Information Technology
     715        "kred", // kred KredTLD Pty Ltd
     716        "kyoto", // kyoto Academic Institution: Kyoto Jyoho Gakuen
     717        "lacaixa", // lacaixa CAIXA D&#39;ESTALVIS I PENSIONS DE BARCELONA
     718        "lamborghini", // lamborghini Automobili Lamborghini S.p.A.
     719        "lamer", // lamer The Estée Lauder Companies Inc.
     720        "lancaster", // lancaster LANCASTER
     721        "land", // land Pine Moon, LLC
     722        "landrover", // landrover Jaguar Land Rover Ltd
     723        "lanxess", // lanxess LANXESS Corporation
     724        "lasalle", // lasalle Jones Lang LaSalle Incorporated
     725        "lat", // lat ECOM-LAC Federación de Latinoamérica y el Caribe para Internet y el Comercio Electrónico
     726        "latrobe", // latrobe La Trobe University
     727        "law", // law Minds + Machines Group Limited
     728        "lawyer", // lawyer United TLD Holdco, Ltd
     729        "lds", // lds IRI Domain Management, LLC
     730        "lease", // lease Victor Trail, LLC
     731        "leclerc", // leclerc A.C.D. LEC Association des Centres Distributeurs Edouard Leclerc
     732        "legal", // legal Blue Falls, LLC
     733        "lexus", // lexus TOYOTA MOTOR CORPORATION
     734        "lgbt", // lgbt Afilias Limited
     735        "liaison", // liaison Liaison Technologies, Incorporated
     736        "lidl", // lidl Schwarz Domains und Services GmbH &amp; Co. KG
     737        "life", // life Trixy Oaks, LLC
     738        "lifeinsurance", // lifeinsurance American Council of Life Insurers
     739        "lifestyle", // lifestyle Lifestyle Domain Holdings, Inc.
     740        "lighting", // lighting John McCook, LLC
     741        "like", // like Amazon Registry Services, Inc.
     742        "limited", // limited Big Fest, LLC
     743        "limo", // limo Hidden Frostbite, LLC
     744        "lincoln", // lincoln Ford Motor Company
     745        "linde", // linde Linde Aktiengesellschaft
     746        "link", // link Uniregistry, Corp.
     747        "live", // live United TLD Holdco Ltd.
     748        "living", // living Lifestyle Domain Holdings, Inc.
     749        "lixil", // lixil LIXIL Group Corporation
     750        "loan", // loan dot Loan Limited
     751        "loans", // loans June Woods, LLC
     752        "lol", // lol Uniregistry, Corp.
     753        "london", // london Dot London Domains Limited
     754        "lotte", // lotte Lotte Holdings Co., Ltd.
     755        "lotto", // lotto Afilias Limited
     756        "love", // love Merchant Law Group LLP
     757        "ltd", // ltd Over Corner, LLC
     758        "ltda", // ltda InterNetX Corp.
     759        "lupin", // lupin LUPIN LIMITED
     760        "luxe", // luxe Top Level Domain Holdings Limited
     761        "luxury", // luxury Luxury Partners LLC
     762        "madrid", // madrid Comunidad de Madrid
     763        "maif", // maif Mutuelle Assurance Instituteur France (MAIF)
     764        "maison", // maison Victor Frostbite, LLC
     765        "makeup", // makeup L&#39;Oréal
     766        "man", // man MAN SE
     767        "management", // management John Goodbye, LLC
     768        "mango", // mango PUNTO FA S.L.
     769        "market", // market Unitied TLD Holdco, Ltd
     770        "marketing", // marketing Fern Pass, LLC
     771        "markets", // markets DOTMARKETS REGISTRY LTD
     772        "marriott", // marriott Marriott Worldwide Corporation
     773        "mba", // mba Lone Hollow, LLC
     774        "med", // med Medistry LLC
     775        "media", // media Grand Glen, LLC
     776        "meet", // meet Afilias Limited
     777        "melbourne", // melbourne The Crown in right of the State of Victoria
     778        "meme", // meme Charleston Road Registry Inc.
     779        "memorial", // memorial Dog Beach, LLC
     780        "men", // men Exclusive Registry Limited
     781        "menu", // menu Wedding TLD2, LLC
     782        "meo", // meo PT Comunicacoes S.A.
     783        "miami", // miami Top Level Domain Holdings Limited
     784        "microsoft", // microsoft Microsoft Corporation
     785        "mil", // mil DoD Network Information Center
     786        "mini", // mini Bayerische Motoren Werke Aktiengesellschaft
     787        "mma", // mma MMA IARD
     788        "mobi", // mobi Afilias Technologies Limited dba dotMobi
     789        "mobily", // mobily GreenTech Consultancy Company W.L.L.
     790        "moda", // moda United TLD Holdco Ltd.
     791        "moe", // moe Interlink Co., Ltd.
     792        "moi", // moi Amazon Registry Services, Inc.
     793        "mom", // mom Uniregistry, Corp.
     794        "monash", // monash Monash University
     795        "money", // money Outer McCook, LLC
     796        "montblanc", // montblanc Richemont DNS Inc.
     797        "mormon", // mormon IRI Domain Management, LLC (&quot;Applicant&quot;)
     798        "mortgage", // mortgage United TLD Holdco, Ltd
     799        "moscow", // moscow Foundation for Assistance for Internet Technologies and Infrastructure Development (FAITID)
     800        "motorcycles", // motorcycles DERMotorcycles, LLC
     801        "mov", // mov Charleston Road Registry Inc.
     802        "movie", // movie New Frostbite, LLC
     803        "movistar", // movistar Telefónica S.A.
     804        "mtn", // mtn MTN Dubai Limited
     805        "mtpc", // mtpc Mitsubishi Tanabe Pharma Corporation
     806        "mtr", // mtr MTR Corporation Limited
     807        "museum", // museum Museum Domain Management Association
     808        "mutuelle", // mutuelle Fédération Nationale de la Mutualité Française
     809        "nadex", // nadex Nadex Domains, Inc
     810        "nagoya", // nagoya GMO Registry, Inc.
     811        "name", // name VeriSign Information Services, Inc.
     812        "natura", // natura NATURA COSMÉTICOS S.A.
     813        "navy", // navy United TLD Holdco Ltd.
     814        "nec", // nec NEC Corporation
     815        "net", // net VeriSign Global Registry Services
     816        "netbank", // netbank COMMONWEALTH BANK OF AUSTRALIA
     817        "network", // network Trixy Manor, LLC
     818        "neustar", // neustar NeuStar, Inc.
     819        "new", // new Charleston Road Registry Inc.
     820        "news", // news United TLD Holdco Ltd.
     821        "nexus", // nexus Charleston Road Registry Inc.
     822        "ngo", // ngo Public Interest Registry
     823        "nhk", // nhk Japan Broadcasting Corporation (NHK)
     824        "nico", // nico DWANGO Co., Ltd.
     825        "nikon", // nikon NIKON CORPORATION
     826        "ninja", // ninja United TLD Holdco Ltd.
     827        "nissan", // nissan NISSAN MOTOR CO., LTD.
     828        "nokia", // nokia Nokia Corporation
     829        "norton", // norton Symantec Corporation
     830        "nowruz", // nowruz Asia Green IT System Bilgisayar San. ve Tic. Ltd. Sti.
     831        "nra", // nra NRA Holdings Company, INC.
     832        "nrw", // nrw Minds + Machines GmbH
     833        "ntt", // ntt NIPPON TELEGRAPH AND TELEPHONE CORPORATION
     834        "nyc", // nyc The City of New York by and through the New York City Department of Information Technology &amp; Telecommunications
     835        "obi", // obi OBI Group Holding SE &amp; Co. KGaA
     836        "office", // office Microsoft Corporation
     837        "okinawa", // okinawa BusinessRalliart inc.
     838        "omega", // omega The Swatch Group Ltd
     839        "one", // one One.com A/S
     840        "ong", // ong Public Interest Registry
     841        "onl", // onl I-REGISTRY Ltd., Niederlassung Deutschland
     842        "online", // online DotOnline Inc.
     843        "ooo", // ooo INFIBEAM INCORPORATION LIMITED
     844        "oracle", // oracle Oracle Corporation
     845        "orange", // orange Orange Brand Services Limited
     846        "org", // org Public Interest Registry (PIR)
     847        "organic", // organic Afilias Limited
     848        "origins", // origins The Estée Lauder Companies Inc.
     849        "osaka", // osaka Interlink Co., Ltd.
     850        "otsuka", // otsuka Otsuka Holdings Co., Ltd.
     851        "ovh", // ovh OVH SAS
     852        "page", // page Charleston Road Registry Inc.
     853        "pamperedchef", // pamperedchef The Pampered Chef, Ltd.
     854        "panerai", // panerai Richemont DNS Inc.
     855        "paris", // paris City of Paris
     856        "pars", // pars Asia Green IT System Bilgisayar San. ve Tic. Ltd. Sti.
     857        "partners", // partners Magic Glen, LLC
     858        "parts", // parts Sea Goodbye, LLC
     859        "party", // party Blue Sky Registry Limited
     860        "pet", // pet Afilias plc
     861        "pharmacy", // pharmacy National Association of Boards of Pharmacy
     862        "philips", // philips Koninklijke Philips N.V.
     863        "photo", // photo Uniregistry, Corp.
     864        "photography", // photography Sugar Glen, LLC
     865        "photos", // photos Sea Corner, LLC
     866        "physio", // physio PhysBiz Pty Ltd
     867        "piaget", // piaget Richemont DNS Inc.
     868        "pics", // pics Uniregistry, Corp.
     869        "pictet", // pictet Pictet Europe S.A.
     870        "pictures", // pictures Foggy Sky, LLC
     871        "pid", // pid Top Level Spectrum, Inc.
     872        "pin", // pin Amazon Registry Services, Inc.
     873        "ping", // ping Ping Registry Provider, Inc.
     874        "pink", // pink Afilias Limited
     875        "pizza", // pizza Foggy Moon, LLC
     876        "place", // place Snow Galley, LLC
     877        "play", // play Charleston Road Registry Inc.
     878        "playstation", // playstation Sony Computer Entertainment Inc.
     879        "plumbing", // plumbing Spring Tigers, LLC
     880        "plus", // plus Sugar Mill, LLC
     881        "pohl", // pohl Deutsche Vermögensberatung Aktiengesellschaft DVAG
     882        "poker", // poker Afilias Domains No. 5 Limited
     883        "porn", // porn ICM Registry PN LLC
     884        "post", // post Universal Postal Union
     885        "praxi", // praxi Praxi S.p.A.
     886        "press", // press DotPress Inc.
     887        "pro", // pro Registry Services Corporation dba RegistryPro
     888        "prod", // prod Charleston Road Registry Inc.
     889        "productions", // productions Magic Birch, LLC
     890        "prof", // prof Charleston Road Registry Inc.
     891        "promo", // promo Afilias plc
     892        "properties", // properties Big Pass, LLC
     893        "property", // property Uniregistry, Corp.
     894        "protection", // protection XYZ.COM LLC
     895        "pub", // pub United TLD Holdco Ltd.
     896        "pwc", // pwc PricewaterhouseCoopers LLP
     897        "qpon", // qpon dotCOOL, Inc.
     898        "quebec", // quebec PointQuébec Inc
     899        "quest", // quest Quest ION Limited
     900        "racing", // racing Premier Registry Limited
     901        "read", // read Amazon Registry Services, Inc.
     902        "realtor", // realtor Real Estate Domains LLC
     903        "realty", // realty Fegistry, LLC
     904        "recipes", // recipes Grand Island, LLC
     905        "red", // red Afilias Limited
     906        "redstone", // redstone Redstone Haute Couture Co., Ltd.
     907        "redumbrella", // redumbrella Travelers TLD, LLC
     908        "rehab", // rehab United TLD Holdco Ltd.
     909        "reise", // reise Foggy Way, LLC
     910        "reisen", // reisen New Cypress, LLC
     911        "reit", // reit National Association of Real Estate Investment Trusts, Inc.
     912        "ren", // ren Beijing Qianxiang Wangjing Technology Development Co., Ltd.
     913        "rent", // rent XYZ.COM LLC
     914        "rentals", // rentals Big Hollow,LLC
     915        "repair", // repair Lone Sunset, LLC
     916        "report", // report Binky Glen, LLC
     917        "republican", // republican United TLD Holdco Ltd.
     918        "rest", // rest Punto 2012 Sociedad Anonima Promotora de Inversion de Capital Variable
     919        "restaurant", // restaurant Snow Avenue, LLC
     920        "review", // review dot Review Limited
     921        "reviews", // reviews United TLD Holdco, Ltd.
     922        "rexroth", // rexroth Robert Bosch GMBH
     923        "rich", // rich I-REGISTRY Ltd., Niederlassung Deutschland
     924        "ricoh", // ricoh Ricoh Company, Ltd.
     925        "rio", // rio Empresa Municipal de Informática SA - IPLANRIO
     926        "rip", // rip United TLD Holdco Ltd.
     927        "rocher", // rocher Ferrero Trading Lux S.A.
     928        "rocks", // rocks United TLD Holdco, LTD.
     929        "rodeo", // rodeo Top Level Domain Holdings Limited
     930        "room", // room Amazon Registry Services, Inc.
     931        "rsvp", // rsvp Charleston Road Registry Inc.
     932        "ruhr", // ruhr regiodot GmbH &amp; Co. KG
     933        "run", // run Snow Park, LLC
     934        "rwe", // rwe RWE AG
     935        "ryukyu", // ryukyu BusinessRalliart inc.
     936        "saarland", // saarland dotSaarland GmbH
     937        "safe", // safe Amazon Registry Services, Inc.
     938        "safety", // safety Safety Registry Services, LLC.
     939        "sakura", // sakura SAKURA Internet Inc.
     940        "sale", // sale United TLD Holdco, Ltd
     941        "salon", // salon Outer Orchard, LLC
     942        "samsung", // samsung SAMSUNG SDS CO., LTD
     943        "sandvik", // sandvik Sandvik AB
     944        "sandvikcoromant", // sandvikcoromant Sandvik AB
     945        "sanofi", // sanofi Sanofi
     946        "sap", // sap SAP AG
     947        "sapo", // sapo PT Comunicacoes S.A.
     948        "sarl", // sarl Delta Orchard, LLC
     949        "sas", // sas Research IP LLC
     950        "saxo", // saxo Saxo Bank A/S
     951        "sbs", // sbs SPECIAL BROADCASTING SERVICE CORPORATION
     952        "sca", // sca SVENSKA CELLULOSA AKTIEBOLAGET SCA (publ)
     953        "scb", // scb The Siam Commercial Bank Public Company Limited (&quot;SCB&quot;)
     954        "schaeffler", // schaeffler Schaeffler Technologies AG &amp; Co. KG
     955        "schmidt", // schmidt SALM S.A.S.
     956        "scholarships", // scholarships Scholarships.com, LLC
     957        "school", // school Little Galley, LLC
     958        "schule", // schule Outer Moon, LLC
     959        "schwarz", // schwarz Schwarz Domains und Services GmbH &amp; Co. KG
     960        "science", // science dot Science Limited
     961        "scor", // scor SCOR SE
     962        "scot", // scot Dot Scot Registry Limited
     963        "seat", // seat SEAT, S.A. (Sociedad Unipersonal)
     964        "security", // security XYZ.COM LLC
     965        "seek", // seek Seek Limited
     966        "select", // select iSelect Ltd
     967        "sener", // sener Sener Ingeniería y Sistemas, S.A.
     968        "services", // services Fox Castle, LLC
     969        "seven", // seven Seven West Media Ltd
     970        "sew", // sew SEW-EURODRIVE GmbH &amp; Co KG
     971        "sex", // sex ICM Registry SX LLC
     972        "sexy", // sexy Uniregistry, Corp.
     973        "sfr", // sfr Societe Francaise du Radiotelephone - SFR
     974        "sharp", // sharp Sharp Corporation
     975        "shell", // shell Shell Information Technology International Inc
     976        "shia", // shia Asia Green IT System Bilgisayar San. ve Tic. Ltd. Sti.
     977        "shiksha", // shiksha Afilias Limited
     978        "shoes", // shoes Binky Galley, LLC
     979        "show", // show Snow Beach, LLC
     980        "shriram", // shriram Shriram Capital Ltd.
     981        "singles", // singles Fern Madison, LLC
     982        "site", // site DotSite Inc.
     983        "ski", // ski STARTING DOT LIMITED
     984        "skin", // skin L&#39;Oréal
     985        "sky", // sky Sky International AG
     986        "skype", // skype Microsoft Corporation
     987        "smile", // smile Amazon Registry Services, Inc.
     988        "sncf", // sncf SNCF (Société Nationale des Chemins de fer Francais)
     989        "soccer", // soccer Foggy Shadow, LLC
     990        "social", // social United TLD Holdco Ltd.
     991        "softbank", // softbank SoftBank Group Corp.
     992        "software", // software United TLD Holdco, Ltd
     993        "sohu", // sohu Sohu.com Limited
     994        "solar", // solar Ruby Town, LLC
     995        "solutions", // solutions Silver Cover, LLC
     996        "sony", // sony Sony Corporation
     997        "soy", // soy Charleston Road Registry Inc.
     998        "space", // space DotSpace Inc.
     999        "spiegel", // spiegel SPIEGEL-Verlag Rudolf Augstein GmbH &amp; Co. KG
     1000        "spot", // spot Amazon Registry Services, Inc.
     1001        "spreadbetting", // spreadbetting DOTSPREADBETTING REGISTRY LTD
     1002        "srl", // srl InterNetX Corp.
     1003        "stada", // stada STADA Arzneimittel AG
     1004        "star", // star Star India Private Limited
     1005        "starhub", // starhub StarHub Limited
     1006        "statefarm", // statefarm State Farm Mutual Automobile Insurance Company
     1007        "statoil", // statoil Statoil ASA
     1008        "stc", // stc Saudi Telecom Company
     1009        "stcgroup", // stcgroup Saudi Telecom Company
     1010        "stockholm", // stockholm Stockholms kommun
     1011        "storage", // storage Self Storage Company LLC
     1012        "studio", // studio United TLD Holdco Ltd.
     1013        "study", // study OPEN UNIVERSITIES AUSTRALIA PTY LTD
     1014        "style", // style Binky Moon, LLC
     1015        "sucks", // sucks Vox Populi Registry Ltd.
     1016        "supplies", // supplies Atomic Fields, LLC
     1017        "supply", // supply Half Falls, LLC
     1018        "support", // support Grand Orchard, LLC
     1019        "surf", // surf Top Level Domain Holdings Limited
     1020        "surgery", // surgery Tin Avenue, LLC
     1021        "suzuki", // suzuki SUZUKI MOTOR CORPORATION
     1022        "swatch", // swatch The Swatch Group Ltd
     1023        "swiss", // swiss Swiss Confederation
     1024        "sydney", // sydney State of New South Wales, Department of Premier and Cabinet
     1025        "symantec", // symantec Symantec Corporation
     1026        "systems", // systems Dash Cypress, LLC
     1027        "tab", // tab Tabcorp Holdings Limited
     1028        "taipei", // taipei Taipei City Government
     1029        "taobao", // taobao Alibaba Group Holding Limited
     1030        "tatamotors", // tatamotors Tata Motors Ltd
     1031        "tatar", // tatar Limited Liability Company "Coordination Center of Regional Domain of Tatarstan Republic"
     1032        "tattoo", // tattoo Uniregistry, Corp.
     1033        "tax", // tax Storm Orchard, LLC
     1034        "taxi", // taxi Pine Falls, LLC
     1035        "tci", // tci Asia Green IT System Bilgisayar San. ve Tic. Ltd. Sti.
     1036        "team", // team Atomic Lake, LLC
     1037        "tech", // tech Dot Tech LLC
     1038        "technology", // technology Auburn Falls, LLC
     1039        "tel", // tel Telnic Ltd.
     1040        "telefonica", // telefonica Telefónica S.A.
     1041        "temasek", // temasek Temasek Holdings (Private) Limited
     1042        "tennis", // tennis Cotton Bloom, LLC
     1043        "thd", // thd Homer TLC, Inc.
     1044        "theater", // theater Blue Tigers, LLC
     1045        "theatre", // theatre XYZ.COM LLC
     1046        "tickets", // tickets Accent Media Limited
     1047        "tienda", // tienda Victor Manor, LLC
     1048        "tiffany", // tiffany Tiffany and Company
     1049        "tips", // tips Corn Willow, LLC
     1050        "tires", // tires Dog Edge, LLC
     1051        "tirol", // tirol punkt Tirol GmbH
     1052        "tmall", // tmall Alibaba Group Holding Limited
     1053        "today", // today Pearl Woods, LLC
     1054        "tokyo", // tokyo GMO Registry, Inc.
     1055        "tools", // tools Pioneer North, LLC
     1056        "top", // top Jiangsu Bangning Science &amp; Technology Co.,Ltd.
     1057        "toray", // toray Toray Industries, Inc.
     1058        "toshiba", // toshiba TOSHIBA Corporation
     1059        "tours", // tours Sugar Station, LLC
     1060        "town", // town Koko Moon, LLC
     1061        "toyota", // toyota TOYOTA MOTOR CORPORATION
     1062        "toys", // toys Pioneer Orchard, LLC
     1063        "trade", // trade Elite Registry Limited
     1064        "trading", // trading DOTTRADING REGISTRY LTD
     1065        "training", // training Wild Willow, LLC
     1066        "travel", // travel Tralliance Registry Management Company, LLC.
     1067        "travelers", // travelers Travelers TLD, LLC
     1068        "travelersinsurance", // travelersinsurance Travelers TLD, LLC
     1069        "trust", // trust Artemis Internet Inc
     1070        "trv", // trv Travelers TLD, LLC
     1071        "tube", // tube Latin American Telecom LLC
     1072        "tui", // tui TUI AG
     1073        "tushu", // tushu Amazon Registry Services, Inc.
     1074        "tvs", // tvs T V SUNDRAM IYENGAR  &amp; SONS PRIVATE LIMITED
     1075        "ubs", // ubs UBS AG
     1076        "unicom", // unicom China United Network Communications Corporation Limited
     1077        "university", // university Little Station, LLC
     1078        "uno", // uno Dot Latin LLC
     1079        "uol", // uol UBN INTERNET LTDA.
     1080        "vacations", // vacations Atomic Tigers, LLC
     1081        "vana", // vana Lifestyle Domain Holdings, Inc.
     1082        "vegas", // vegas Dot Vegas, Inc.
     1083        "ventures", // ventures Binky Lake, LLC
     1084        "verisign", // verisign VeriSign, Inc.
     1085        "versicherung", // versicherung dotversicherung-registry GmbH
     1086        "vet", // vet United TLD Holdco, Ltd
     1087        "viajes", // viajes Black Madison, LLC
     1088        "video", // video United TLD Holdco, Ltd
     1089        "villas", // villas New Sky, LLC
     1090        "vin", // vin Holly Shadow, LLC
     1091        "vip", // vip Minds + Machines Group Limited
     1092        "virgin", // virgin Virgin Enterprises Limited
     1093        "vision", // vision Koko Station, LLC
     1094        "vista", // vista Vistaprint Limited
     1095        "vistaprint", // vistaprint Vistaprint Limited
     1096        "viva", // viva Saudi Telecom Company
     1097        "vlaanderen", // vlaanderen DNS.be vzw
     1098        "vodka", // vodka Top Level Domain Holdings Limited
     1099        "volkswagen", // volkswagen Volkswagen Group of America Inc.
     1100        "vote", // vote Monolith Registry LLC
     1101        "voting", // voting Valuetainment Corp.
     1102        "voto", // voto Monolith Registry LLC
     1103        "voyage", // voyage Ruby House, LLC
     1104        "wales", // wales Nominet UK
     1105        "walter", // walter Sandvik AB
     1106        "wang", // wang Zodiac Registry Limited
     1107        "wanggou", // wanggou Amazon Registry Services, Inc.
     1108        "watch", // watch Sand Shadow, LLC
     1109        "watches", // watches Richemont DNS Inc.
     1110        "weather", // weather The Weather Channel, LLC
     1111        "weatherchannel", // weatherchannel The Weather Channel, LLC
     1112        "webcam", // webcam dot Webcam Limited
     1113        "weber", // weber Saint-Gobain Weber SA
     1114        "website", // website DotWebsite Inc.
     1115        "wed", // wed Atgron, Inc.
     1116        "wedding", // wedding Top Level Domain Holdings Limited
     1117        "weir", // weir Weir Group IP Limited
     1118        "whoswho", // whoswho Who&#39;s Who Registry
     1119        "wien", // wien punkt.wien GmbH
     1120        "wiki", // wiki Top Level Design, LLC
     1121        "williamhill", // williamhill William Hill Organization Limited
     1122        "win", // win First Registry Limited
     1123        "windows", // windows Microsoft Corporation
     1124        "wine", // wine June Station, LLC
     1125        "wme", // wme William Morris Endeavor Entertainment, LLC
     1126        "wolterskluwer", // wolterskluwer Wolters Kluwer N.V.
     1127        "work", // work Top Level Domain Holdings Limited
     1128        "works", // works Little Dynamite, LLC
     1129        "world", // world Bitter Fields, LLC
     1130        "wtc", // wtc World Trade Centers Association, Inc.
     1131        "wtf", // wtf Hidden Way, LLC
     1132        "xbox", // xbox Microsoft Corporation
     1133        "xerox", // xerox Xerox DNHC LLC
     1134        "xin", // xin Elegant Leader Limited
     1135        "xn--11b4c3d", // कॉम VeriSign Sarl
     1136        "xn--1ck2e1b", // セール Amazon Registry Services, Inc.
     1137        "xn--1qqw23a", // 佛山 Guangzhou YU Wei Information Technology Co., Ltd.
     1138        "xn--30rr7y", // 慈善 Excellent First Limited
     1139        "xn--3bst00m", // 集团 Eagle Horizon Limited
     1140        "xn--3ds443g", // 在线 TLD REGISTRY LIMITED
     1141        "xn--3pxu8k", // 点看 VeriSign Sarl
     1142        "xn--42c2d9a", // คอม VeriSign Sarl
     1143        "xn--45q11c", // 八卦 Zodiac Scorpio Limited
     1144        "xn--4gbrim", // موقع Suhub Electronic Establishment
     1145        "xn--55qw42g", // 公益 China Organizational Name Administration Center
     1146        "xn--55qx5d", // 公司 Computer Network Information Center of Chinese Academy of Sciences (China Internet Network Information Center)
     1147        "xn--6frz82g", // 移动 Afilias Limited
     1148        "xn--6qq986b3xl", // 我爱你 Tycoon Treasure Limited
     1149        "xn--80adxhks", // москва Foundation for Assistance for Internet Technologies and Infrastructure Development (FAITID)
     1150        "xn--80asehdb", // онлайн CORE Association
     1151        "xn--80aswg", // сайт CORE Association
     1152        "xn--8y0a063a", // 联通 China United Network Communications Corporation Limited
     1153        "xn--9dbq2a", // קום VeriSign Sarl
     1154        "xn--9et52u", // 时尚 RISE VICTORY LIMITED
     1155        "xn--b4w605ferd", // 淡马锡 Temasek Holdings (Private) Limited
     1156        "xn--bck1b9a5dre4c", // ファッション Amazon Registry Services, Inc.
     1157        "xn--c1avg", // орг Public Interest Registry
     1158        "xn--c2br7g", // नेट VeriSign Sarl
     1159        "xn--cck2b3b", // ストア Amazon Registry Services, Inc.
     1160        "xn--cg4bki", // 삼성 SAMSUNG SDS CO., LTD
     1161        "xn--czr694b", // 商标 HU YI GLOBAL INFORMATION RESOURCES(HOLDING) COMPANY.HONGKONG LIMITED
     1162        "xn--czrs0t", // 商店 Wild Island, LLC
     1163        "xn--czru2d", // 商城 Zodiac Aquarius Limited
     1164        "xn--d1acj3b", // дети The Foundation for Network Initiatives “The Smart Internet”
     1165        "xn--eckvdtc9d", // ポイント Amazon Registry Services, Inc.
     1166        "xn--efvy88h", // 新闻 Xinhua News Agency Guangdong Branch 新华通讯社广东分社
     1167        "xn--estv75g", // 工行 Industrial and Commercial Bank of China Limited
     1168        "xn--fhbei", // كوم VeriSign Sarl
     1169        "xn--fiq228c5hs", // 中文网 TLD REGISTRY LIMITED
     1170        "xn--fiq64b", // 中信 CITIC Group Corporation
     1171        "xn--fjq720a", // 娱乐 Will Bloom, LLC
     1172        "xn--flw351e", // 谷歌 Charleston Road Registry Inc.
     1173        "xn--g2xx48c", // 购物 Minds + Machines Group Limited
     1174        "xn--gckr3f0f", // クラウド Amazon Registry Services, Inc.
     1175        "xn--hxt814e", // 网店 Zodiac Libra Limited
     1176        "xn--i1b6b1a6a2e", // संगठन Public Interest Registry
     1177        "xn--imr513n", // 餐厅 HU YI GLOBAL INFORMATION RESOURCES (HOLDING) COMPANY. HONGKONG LIMITED
     1178        "xn--io0a7i", // 网络 Computer Network Information Center of Chinese Academy of Sciences (China Internet Network Information Center)
     1179        "xn--j1aef", // ком VeriSign Sarl
     1180        "xn--jlq61u9w7b", // 诺基亚 Nokia Corporation
     1181        "xn--kcrx77d1x4a", // 飞利浦 Koninklijke Philips N.V.
     1182        "xn--kpu716f", // 手表 Richemont DNS Inc.
     1183        "xn--kput3i", // 手机 Beijing RITT-Net Technology Development Co., Ltd
     1184        "xn--mgba3a3ejt", // ارامكو Aramco Services Company
     1185        "xn--mgbab2bd", // بازار CORE Association
     1186        "xn--mgbb9fbpob", // موبايلي GreenTech Consultancy Company W.L.L.
     1187        "xn--mgbt3dhd", // همراه Asia Green IT System Bilgisayar San. ve Tic. Ltd. Sti.
     1188        "xn--mk1bu44c", // 닷컴 VeriSign Sarl
     1189        "xn--mxtq1m", // 政府 Net-Chinese Co., Ltd.
     1190        "xn--ngbc5azd", // شبكة International Domain Registry Pty. Ltd.
     1191        "xn--ngbe9e0a", // بيتك Kuwait Finance House
     1192        "xn--nqv7f", // 机构 Public Interest Registry
     1193        "xn--nqv7fs00ema", // 组织机构 Public Interest Registry
     1194        "xn--nyqy26a", // 健康 Stable Tone Limited
     1195        "xn--p1acf", // рус Rusnames Limited
     1196        "xn--pbt977c", // 珠宝 Richemont DNS Inc.
     1197        "xn--pssy2u", // 大拿 VeriSign Sarl
     1198        "xn--q9jyb4c", // みんな Charleston Road Registry Inc.
     1199        "xn--qcka1pmc", // グーグル Charleston Road Registry Inc.
     1200        "xn--rhqv96g", // 世界 Stable Tone Limited
     1201        "xn--rovu88b", // 書籍 Amazon EU S.à r.l.
     1202        "xn--ses554g", // 网址 KNET Co., Ltd
     1203        "xn--t60b56a", // 닷넷 VeriSign Sarl
     1204        "xn--tckwe", // コム VeriSign Sarl
     1205        "xn--unup4y", // 游戏 Spring Fields, LLC
     1206        "xn--vermgensberater-ctb", // VERMöGENSBERATER Deutsche Vermögensberatung Aktiengesellschaft DVAG
     1207        "xn--vermgensberatung-pwb", // VERMöGENSBERATUNG Deutsche Vermögensberatung Aktiengesellschaft DVAG
     1208        "xn--vhquv", // 企业 Dash McCook, LLC
     1209        "xn--vuq861b", // 信息 Beijing Tele-info Network Technology Co., Ltd.
     1210        "xn--xhq521b", // 广东 Guangzhou YU Wei Information Technology Co., Ltd.
     1211        "xn--zfr164b", // 政务 China Organizational Name Administration Center
     1212        "xperia", // xperia Sony Mobile Communications AB
     1213        "xxx", // xxx ICM Registry LLC
     1214        "xyz", // xyz XYZ.COM LLC
     1215        "yachts", // yachts DERYachts, LLC
     1216        "yahoo", // yahoo Yahoo! Domain Services Inc.
     1217        "yamaxun", // yamaxun Amazon Registry Services, Inc.
     1218        "yandex", // yandex YANDEX, LLC
     1219        "yodobashi", // yodobashi YODOBASHI CAMERA CO.,LTD.
     1220        "yoga", // yoga Top Level Domain Holdings Limited
     1221        "yokohama", // yokohama GMO Registry, Inc.
     1222        "youtube", // youtube Charleston Road Registry Inc.
     1223        "zara", // zara Industria de Diseño Textil, S.A. (INDITEX, S.A.)
     1224        "zero", // zero Amazon Registry Services, Inc.
     1225        "zip", // zip Charleston Road Registry Inc.
     1226        "zone", // zone Outer Falls, LLC
     1227        "zuerich", // zuerich Kanton Zürich (Canton of Zurich)
    6561228    };
    6571229
    658     // JOSM PATCH BEGIN
    659     // see #10862 - list of IDN TLDs taken from IANA on 2014-12-18
    660     private static final String[] IDN_TLDS = new String[] {
    661         "XN--1QQW23A",
    662         "XN--3BST00M",
    663         "XN--3DS443G",
    664         "XN--3E0B707E",
    665         "XN--45BRJ9C",
    666         "XN--45Q11C",
    667         "XN--4GBRIM",
    668         "XN--55QW42G",
    669         "XN--55QX5D",
    670         "XN--6FRZ82G",
    671         "XN--6QQ986B3XL",
    672         "XN--80ADXHKS",
    673         "XN--80AO21A",
    674         "XN--80ASEHDB",
    675         "XN--80ASWG",
    676         "XN--90A3AC",
    677         "XN--C1AVG",
    678         "XN--CG4BKI",
    679         "XN--CLCHC0EA0B2G2A9GCD",
    680         "XN--CZR694B",
    681         "XN--CZRS0T",
    682         "XN--CZRU2D",
    683         "XN--D1ACJ3B",
    684         "XN--D1ALF",
    685         "XN--FIQ228C5HS",
    686         "XN--FIQ64B",
    687         "XN--FIQS8S",
    688         "XN--FIQZ9S",
    689         "XN--FLW351E",
    690         "XN--FPCRJ9C3D",
    691         "XN--FZC2C9E2C",
    692         "XN--GECRJ9C",
    693         "XN--H2BRJ9C",
    694         "XN--HXT814E",
    695         "XN--I1B6B1A6A2E",
    696         "XN--IO0A7I",
    697         "XN--J1AMH",
    698         "XN--J6W193G",
    699         "XN--KPRW13D",
    700         "XN--KPRY57D",
    701         "XN--KPUT3I",
    702         "XN--L1ACC",
    703         "XN--LGBBAT1AD8J",
    704         "XN--MGB9AWBF",
    705         "XN--MGBA3A4F16A",
    706         "XN--MGBAAM7A8H",
    707         "XN--MGBAB2BD",
    708         "XN--MGBAYH7GPA",
    709         "XN--MGBBH1A71E",
    710         "XN--MGBC0A9AZCG",
    711         "XN--MGBERP4A5D4AR",
    712         "XN--MGBX4CD0AB",
    713         "XN--NGBC5AZD",
    714         "XN--NODE",
    715         "XN--NQV7F",
    716         "XN--NQV7FS00EMA",
    717         "XN--O3CW4H",
    718         "XN--OGBPF8FL",
    719         "XN--P1ACF",
    720         "XN--P1AI",
    721         "XN--PGBS0DH",
    722         "XN--Q9JYB4C",
    723         "XN--QCKA1PMC",
    724         "XN--RHQV96G",
    725         "XN--S9BRJ9C",
    726         "XN--SES554G",
    727         "XN--UNUP4Y",
    728         "XN--VERMGENSBERATER-CTB",
    729         "XN--VERMGENSBERATUNG-PWB",
    730         "XN--VHQUV",
    731         "XN--WGBH1C",
    732         "XN--WGBL6A",
    733         "XN--XHQ521B",
    734         "XN--XKC2AL3HYE2A",
    735         "XN--XKC2DL3A5EE0H",
    736         "XN--YFRO4I67O",
    737         "XN--YGBI2AMMX",
    738         "XN--ZFR164B",
    739     };
    740     // END JOSM PATCH
    741 
     1230    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
    7421231    private static final String[] COUNTRY_CODE_TLDS = new String[] {
    7431232        "ac",                 // Ascension Island
     
    7491238        "al",                 // Albania
    7501239        "am",                 // Armenia
    751         "an",                 // Netherlands Antilles
     1240//        "an",                 // Netherlands Antilles (retired)
    7521241        "ao",                 // Angola
    7531242        "aq",                 // Antarctica
     
    9641453        "tn",                 // Tunisia
    9651454        "to",                 // Tonga
    966         "tp",                 // East Timor
     1455//        "tp",                 // East Timor (Retired)
    9671456        "tr",                 // Turkey
    9681457        "tt",                 // Trinidad and Tobago
     
    9731462        "ug",                 // Uganda
    9741463        "uk",                 // United Kingdom
    975         "um",                 // United States Minor Outlying Islands
    9761464        "us",                 // United States of America
    9771465        "uy",                 // Uruguay
     
    9861474        "wf",                 // Wallis and Futuna
    9871475        "ws",                 // Samoa (formerly Western Samoa)
     1476        "xn--3e0b707e", // 한국 KISA (Korea Internet &amp; Security Agency)
     1477        "xn--45brj9c", // ভারত National Internet Exchange of India
     1478        "xn--80ao21a", // қаз Association of IT Companies of Kazakhstan
     1479        "xn--90a3ac", // срб Serbian National Internet Domain Registry (RNIDS)
     1480        "xn--90ais", // ??? Reliable Software Inc.
     1481        "xn--clchc0ea0b2g2a9gcd", // சிங்கப்பூர் Singapore Network Information Centre (SGNIC) Pte Ltd
     1482        "xn--d1alf", // мкд Macedonian Academic Research Network Skopje
     1483        "xn--e1a4c", // ею EURid vzw/asbl
     1484        "xn--fiqs8s", // 中国 China Internet Network Information Center
     1485        "xn--fiqz9s", // 中國 China Internet Network Information Center
     1486        "xn--fpcrj9c3d", // భారత్ National Internet Exchange of India
     1487        "xn--fzc2c9e2c", // ලංකා LK Domain Registry
     1488        "xn--gecrj9c", // ભારત National Internet Exchange of India
     1489        "xn--h2brj9c", // भारत National Internet Exchange of India
     1490        "xn--j1amh", // укр Ukrainian Network Information Centre (UANIC), Inc.
     1491        "xn--j6w193g", // 香港 Hong Kong Internet Registration Corporation Ltd.
     1492        "xn--kprw13d", // 台湾 Taiwan Network Information Center (TWNIC)
     1493        "xn--kpry57d", // 台灣 Taiwan Network Information Center (TWNIC)
     1494        "xn--l1acc", // мон Datacom Co.,Ltd
     1495        "xn--lgbbat1ad8j", // الجزائر CERIST
     1496        "xn--mgb9awbf", // عمان Telecommunications Regulatory Authority (TRA)
     1497        "xn--mgba3a4f16a", // ایران Institute for Research in Fundamental Sciences (IPM)
     1498        "xn--mgbaam7a8h", // امارات Telecommunications Regulatory Authority (TRA)
     1499        "xn--mgbayh7gpa", // الاردن National Information Technology Center (NITC)
     1500        "xn--mgbbh1a71e", // بھارت National Internet Exchange of India
     1501        "xn--mgbc0a9azcg", // المغرب Agence Nationale de Réglementation des Télécommunications (ANRT)
     1502        "xn--mgberp4a5d4ar", // السعودية Communications and Information Technology Commission
     1503        "xn--mgbpl2fh", // ????? Sudan Internet Society
     1504        "xn--mgbtx2b", // عراق Communications and Media Commission (CMC)
     1505        "xn--mgbx4cd0ab", // مليسيا MYNIC Berhad
     1506        "xn--mix891f", // 澳門 Bureau of Telecommunications Regulation (DSRT)
     1507        "xn--node", // გე Information Technologies Development Center (ITDC)
     1508        "xn--o3cw4h", // ไทย Thai Network Information Center Foundation
     1509        "xn--ogbpf8fl", // سورية National Agency for Network Services (NANS)
     1510        "xn--p1ai", // рф Coordination Center for TLD RU
     1511        "xn--pgbs0dh", // تونس Agence Tunisienne d&#39;Internet
     1512        "xn--qxam", // ελ ICS-FORTH GR
     1513        "xn--s9brj9c", // ਭਾਰਤ National Internet Exchange of India
     1514        "xn--wgbh1c", // مصر National Telecommunication Regulatory Authority - NTRA
     1515        "xn--wgbl6a", // قطر Communications Regulatory Authority
     1516        "xn--xkc2al3hye2a", // இலங்கை LK Domain Registry
     1517        "xn--xkc2dl3a5ee0h", // இந்தியா National Internet Exchange of India
     1518        "xn--y9a3aq", // ??? Internet Society
     1519        "xn--yfro4i67o", // 新加坡 Singapore Network Information Centre (SGNIC) Pte Ltd
     1520        "xn--ygbi2ammx", // فلسطين Ministry of Telecom &amp; Information Technology (MTIT)
    9881521        "ye",                 // Yemen
    9891522        "yt",                 // Mayotte
    990         "yu",                 // Serbia and Montenegro (originally Yugoslavia)
    9911523        "za",                 // South Africa
    9921524        "zm",                 // Zambia
     
    9941526    };
    9951527
     1528    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
    9961529    private static final String[] LOCAL_TLDS = new String[] {
     1530       "localdomain",         // Also widely used as localhost.localdomain
    9971531       "localhost",           // RFC2606 defined
    998        "localdomain"          // Also widely used as localhost.localdomain
    999    };
    1000 
    1001     static {
    1002         Arrays.sort(INFRASTRUCTURE_TLDS);
    1003         Arrays.sort(COUNTRY_CODE_TLDS);
    1004         Arrays.sort(GENERIC_TLDS);
    1005         Arrays.sort(IDN_TLDS);
    1006         Arrays.sort(LOCAL_TLDS);
     1532    };
     1533
     1534    // Additional arrays to supplement or override the built in ones.
     1535    // The PLUS arrays are valid keys, the MINUS arrays are invalid keys
     1536
     1537    /*
     1538     * This field is used to detect whether the getInstance has been called.
     1539     * After this, the method updateTLDOverride is not allowed to be called.
     1540     * This field does not need to be volatile since it is only accessed from
     1541     * synchronized methods.
     1542     */
     1543    private static boolean inUse = false;
     1544
     1545    /*
     1546     * These arrays are mutable, but they don't need to be volatile.
     1547     * They can only be updated by the updateTLDOverride method, and any readers must get an instance
     1548     * using the getInstance methods which are all (now) synchronised.
     1549     */
     1550    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
     1551    private static volatile String[] countryCodeTLDsPlus = EMPTY_STRING_ARRAY;
     1552
     1553    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
     1554    private static volatile String[] genericTLDsPlus = EMPTY_STRING_ARRAY;
     1555
     1556    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
     1557    private static volatile String[] countryCodeTLDsMinus = EMPTY_STRING_ARRAY;
     1558
     1559    // WARNING: this array MUST be sorted, otherwise it cannot be searched reliably using binary search
     1560    private static volatile String[] genericTLDsMinus = EMPTY_STRING_ARRAY;
     1561
     1562    /**
     1563     * enum used by {@link DomainValidator#updateTLDOverride(ArrayType, String[])}
     1564     * to determine which override array to update / fetch
     1565     * @since 1.5.0
     1566     * @since 1.5.1 made public and added read-only array references
     1567     */
     1568    public enum ArrayType {
     1569        /** Update (or get a copy of) the GENERIC_TLDS_PLUS table containing additonal generic TLDs */
     1570        GENERIC_PLUS,
     1571        /** Update (or get a copy of) the GENERIC_TLDS_MINUS table containing deleted generic TLDs */
     1572        GENERIC_MINUS,
     1573        /** Update (or get a copy of) the COUNTRY_CODE_TLDS_PLUS table containing additonal country code TLDs */
     1574        COUNTRY_CODE_PLUS,
     1575        /** Update (or get a copy of) the COUNTRY_CODE_TLDS_MINUS table containing deleted country code TLDs */
     1576        COUNTRY_CODE_MINUS,
     1577        /** Get a copy of the generic TLDS table */
     1578        GENERIC_RO,
     1579        /** Get a copy of the country code table */
     1580        COUNTRY_CODE_RO,
     1581        /** Get a copy of the infrastructure table */
     1582        INFRASTRUCTURE_RO,
     1583        /** Get a copy of the local table */
     1584        LOCAL_RO
     1585    }
     1586
     1587    // For use by unit test code only
     1588    static synchronized void clearTLDOverrides() {
     1589        inUse = false;
     1590        countryCodeTLDsPlus = EMPTY_STRING_ARRAY;
     1591        countryCodeTLDsMinus = EMPTY_STRING_ARRAY;
     1592        genericTLDsPlus = EMPTY_STRING_ARRAY;
     1593        genericTLDsMinus = EMPTY_STRING_ARRAY;
     1594    }
     1595
     1596    /**
     1597     * Update one of the TLD override arrays.
     1598     * This must only be done at program startup, before any instances are accessed using getInstance.
     1599     * <p>
     1600     * For example:
     1601     * <p>
     1602     * <code>DomainValidator.updateTLDOverride(ArrayType.GENERIC_PLUS, new String[]{"apache"})}</code>
     1603     * <p>
     1604     * To clear an override array, provide an empty array.
     1605     *
     1606     * @param table the table to update, see {@link DomainValidator.ArrayType}
     1607     * Must be one of the following
     1608     * <ul>
     1609     * <li>COUNTRY_CODE_MINUS</li>
     1610     * <li>COUNTRY_CODE_PLUS</li>
     1611     * <li>GENERIC_MINUS</li>
     1612     * <li>GENERIC_PLUS</li>
     1613     * </ul>
     1614     * @param tlds the array of TLDs, must not be null
     1615     * @throws IllegalStateException if the method is called after getInstance
     1616     * @throws IllegalArgumentException if one of the read-only tables is requested
     1617     * @since 1.5.0
     1618     */
     1619    public static synchronized void updateTLDOverride(ArrayType table, String[] tlds) {
     1620        if (inUse) {
     1621            throw new IllegalStateException("Can only invoke this method before calling getInstance");
     1622        }
     1623        String[] copy = new String[tlds.length];
     1624        // Comparisons are always done with lower-case entries
     1625        for (int i = 0; i < tlds.length; i++) {
     1626            copy[i] = tlds[i].toLowerCase(Locale.ENGLISH);
     1627        }
     1628        Arrays.sort(copy);
     1629        switch(table) {
     1630        case COUNTRY_CODE_MINUS:
     1631            countryCodeTLDsMinus = copy;
     1632            break;
     1633        case COUNTRY_CODE_PLUS:
     1634            countryCodeTLDsPlus = copy;
     1635            break;
     1636        case GENERIC_MINUS:
     1637            genericTLDsMinus = copy;
     1638            break;
     1639        case GENERIC_PLUS:
     1640            genericTLDsPlus = copy;
     1641            break;
     1642        case COUNTRY_CODE_RO:
     1643        case GENERIC_RO:
     1644        case INFRASTRUCTURE_RO:
     1645        case LOCAL_RO:
     1646            throw new IllegalArgumentException("Cannot update the table: " + table);
     1647        default:
     1648            throw new IllegalArgumentException("Unexpected enum value: " + table);
     1649        }
     1650    }
     1651
     1652    /**
     1653     * Get a copy of the internal array.
     1654     * @param table the array type (any of the enum values)
     1655     * @return a copy of the array
     1656     * @throws IllegalArgumentException if the table type is unexpected (should not happen)
     1657     * @since 1.5.1
     1658     */
     1659    public static String[] getTLDEntries(ArrayType table) {
     1660        final String[] array;
     1661        switch(table) {
     1662        case COUNTRY_CODE_MINUS:
     1663            array = countryCodeTLDsMinus;
     1664            break;
     1665        case COUNTRY_CODE_PLUS:
     1666            array = countryCodeTLDsPlus;
     1667            break;
     1668        case GENERIC_MINUS:
     1669            array = genericTLDsMinus;
     1670            break;
     1671        case GENERIC_PLUS:
     1672            array = genericTLDsPlus;
     1673            break;
     1674        case GENERIC_RO:
     1675            array = GENERIC_TLDS;
     1676            break;
     1677        case COUNTRY_CODE_RO:
     1678            array = COUNTRY_CODE_TLDS;
     1679            break;
     1680        case INFRASTRUCTURE_RO:
     1681            array = INFRASTRUCTURE_TLDS;
     1682            break;
     1683        case LOCAL_RO:
     1684            array = LOCAL_TLDS;
     1685            break;
     1686        default:
     1687            throw new IllegalArgumentException("Unexpected enum value: " + table);
     1688        }
     1689        return Arrays.copyOf(array, array.length); // clone the array
     1690    }
     1691
     1692    /**
     1693     * Converts potentially Unicode input to punycode.
     1694     * If conversion fails, returns the original input.
     1695     *
     1696     * @param input the string to convert, not null
     1697     * @return converted input, or original input if conversion fails
     1698     */
     1699    // Needed by UrlValidator
     1700    static String unicodeToASCII(String input) {
     1701        if (isOnlyASCII(input)) { // skip possibly expensive processing
     1702            return input;
     1703        }
     1704        try {
     1705            final String ascii = IDN.toASCII(input);
     1706            if (IDNBUGHOLDER.IDN_TOASCII_PRESERVES_TRAILING_DOTS) {
     1707                return ascii;
     1708            }
     1709            final int length = input.length();
     1710            if (length == 0) { // check there is a last character
     1711                return input;
     1712            }
     1713            // RFC3490 3.1. 1)
     1714            //            Whenever dots are used as label separators, the following
     1715            //            characters MUST be recognized as dots: U+002E (full stop), U+3002
     1716            //            (ideographic full stop), U+FF0E (fullwidth full stop), U+FF61
     1717            //            (halfwidth ideographic full stop).
     1718            char lastChar = input.charAt(length-1); // fetch original last char
     1719            switch(lastChar) {
     1720                case '\u002E': // "." full stop
     1721                case '\u3002': // ideographic full stop
     1722                case '\uFF0E': // fullwidth full stop
     1723                case '\uFF61': // halfwidth ideographic full stop
     1724                    return ascii + "."; // restore the missing stop
     1725                default:
     1726                    return ascii;
     1727            }
     1728        } catch (IllegalArgumentException e) { // input is not valid
     1729            return input;
     1730        }
     1731    }
     1732
     1733    private static class IDNBUGHOLDER {
     1734        private static boolean keepsTrailingDot() {
     1735            final String input = "a."; // must be a valid name
     1736            return input.equals(IDN.toASCII(input));
     1737        }
     1738
     1739        private static final boolean IDN_TOASCII_PRESERVES_TRAILING_DOTS = keepsTrailingDot();
     1740    }
     1741
     1742    /*
     1743     * Check if input contains only ASCII
     1744     * Treats null as all ASCII
     1745     */
     1746    private static boolean isOnlyASCII(String input) {
     1747        if (input == null) {
     1748            return true;
     1749        }
     1750        for (int i = 0; i < input.length(); i++) {
     1751            if (input.charAt(i) > 0x7F) { // CHECKSTYLE IGNORE MagicNumber
     1752                return false;
     1753            }
     1754        }
     1755        return true;
     1756    }
     1757
     1758    /**
     1759     * Check if a sorted array contains the specified key
     1760     *
     1761     * @param sortedArray the array to search
     1762     * @param key the key to find
     1763     * @return {@code true} if the array contains the key
     1764     */
     1765    private static boolean arrayContains(String[] sortedArray, String key) {
     1766        return Arrays.binarySearch(sortedArray, key) >= 0;
    10071767    }
    10081768}
  • trunk/src/org/openstreetmap/josm/data/validation/routines/EmailValidator.java

    r8510 r9853  
    2525 * <p>Perform email validations.</p>
    2626 * <p>
    27  * This class is a Singleton; you can retrieve the instance via the getInstance() method.
    28  * </p>
    29  * <p>
    3027 * Based on a script by <a href="mailto:stamhankar@hotmail.com">Sandeep V. Tamhankar</a>
    3128 * http://javascript.internet.com
     
    3330 * <p>
    3431 * This implementation is not guaranteed to catch all possible errors in an email address.
    35  * For example, an address like nobody@noplace.somedog will pass validator, even though there
    36  * is no TLD "somedog"
    3732 * </p>.
    3833 *
    39  * @version $Revision: 1608584 $ $Date: 2014-07-07 19:54:07 UTC (Mon, 07 Jul 2014) $
     34 * @version $Revision: 1723573 $
    4035 * @since Validator 1.4
    4136 */
     
    4338
    4439    private static final String SPECIAL_CHARS = "\\p{Cntrl}\\(\\)<>@,;:'\\\\\\\"\\.\\[\\]";
    45     private static final String VALID_CHARS = "[^\\s" + SPECIAL_CHARS + "]";
    46     private static final String QUOTED_USER = "(\"[^\"]*\")";
     40    private static final String VALID_CHARS = "(\\\\.)|[^\\s" + SPECIAL_CHARS + "]";
     41    private static final String QUOTED_USER = "(\"(\\\\\"|[^\"])*\")";
    4742    private static final String WORD = "((" + VALID_CHARS + "|')+|" + QUOTED_USER + ")";
    4843
    49     private static final String LEGAL_ASCII_REGEX = "^\\p{ASCII}+$";
    5044    private static final String EMAIL_REGEX = "^\\s*?(.+)@(.+?)\\s*$";
    5145    private static final String IP_DOMAIN_REGEX = "^\\[(.*)\\]$";
    5246    private static final String USER_REGEX = "^\\s*" + WORD + "(\\." + WORD + ")*$";
    5347
    54     private static final Pattern MATCH_ASCII_PATTERN = Pattern.compile(LEGAL_ASCII_REGEX);
    5548    private static final Pattern EMAIL_PATTERN = Pattern.compile(EMAIL_REGEX);
    5649    private static final Pattern IP_DOMAIN_PATTERN = Pattern.compile(IP_DOMAIN_REGEX);
    5750    private static final Pattern USER_PATTERN = Pattern.compile(USER_REGEX);
    5851
     52    private static final int MAX_USERNAME_LEN = 64;
     53
    5954    private final boolean allowLocal;
     55    private final boolean allowTld;
    6056
    6157    /**
     
    6359     *  doesn't consider local addresses as valid.
    6460     */
    65     private static final EmailValidator EMAIL_VALIDATOR = new EmailValidator(false);
     61    private static final EmailValidator EMAIL_VALIDATOR = new EmailValidator(false, false);
     62
     63    /**
     64     * Singleton instance of this class, which
     65     *  doesn't consider local addresses as valid.
     66     */
     67    private static final EmailValidator EMAIL_VALIDATOR_WITH_TLD = new EmailValidator(false, true);
    6668
    6769    /**
     
    6971     *  consider local addresses valid.
    7072     */
    71     private static final EmailValidator EMAIL_VALIDATOR_WITH_LOCAL = new EmailValidator(true);
     73    private static final EmailValidator EMAIL_VALIDATOR_WITH_LOCAL = new EmailValidator(true, false);
     74
     75
     76    /**
     77     * Singleton instance of this class, which does
     78     *  consider local addresses valid.
     79     */
     80    private static final EmailValidator EMAIL_VALIDATOR_WITH_LOCAL_WITH_TLD = new EmailValidator(true, true);
    7281
    7382    /**
     
    8594     *
    8695     * @param allowLocal Should local addresses be considered valid?
     96     * @param allowTld Should TLDs be allowed?
    8797     * @return singleton instance of this validator
    8898     */
     99    public static EmailValidator getInstance(boolean allowLocal, boolean allowTld) {
     100        if (allowLocal) {
     101            if (allowTld) {
     102                return EMAIL_VALIDATOR_WITH_LOCAL_WITH_TLD;
     103            } else {
     104                return EMAIL_VALIDATOR_WITH_LOCAL;
     105            }
     106        } else {
     107            if (allowTld) {
     108                return EMAIL_VALIDATOR_WITH_TLD;
     109            } else {
     110                return EMAIL_VALIDATOR;
     111            }
     112        }
     113    }
     114
     115    /**
     116     * Returns the Singleton instance of this validator,
     117     *  with local validation as required.
     118     *
     119     * @param allowLocal Should local addresses be considered valid?
     120     * @return singleton instance of this validator
     121     */
    89122    public static EmailValidator getInstance(boolean allowLocal) {
    90         if (allowLocal) {
    91            return EMAIL_VALIDATOR_WITH_LOCAL;
    92         }
    93         return EMAIL_VALIDATOR;
     123        return getInstance(allowLocal, false);
     124    }
     125
     126    /**
     127     * Protected constructor for subclasses to use.
     128     *
     129     * @param allowLocal Should local addresses be considered valid?
     130     * @param allowTld Should TLDs be allowed?
     131     */
     132    protected EmailValidator(boolean allowLocal, boolean allowTld) {
     133        super();
     134        this.allowLocal = allowLocal;
     135        this.allowTld = allowTld;
    94136    }
    95137
     
    102144        super();
    103145        this.allowLocal = allowLocal;
     146        this.allowTld = false;
    104147    }
    105148
     
    117160        }
    118161
    119         Matcher asciiMatcher = MATCH_ASCII_PATTERN.matcher(email);
    120         if (!asciiMatcher.matches()) {
    121             setErrorMessage(tr("E-mail address contains non-ascii characters"));
    122             setFix(email.replaceAll("[^\\p{ASCII}]+", ""));
     162        if (email.endsWith(".")) { // check this first - it's cheap!
     163            setErrorMessage(tr("E-mail address is invalid"));
    123164            return false;
    124165        }
     
    131172        }
    132173
    133         if (email.endsWith(".")) {
    134             setErrorMessage(tr("E-mail address is invalid"));
    135             return false;
    136         }
    137 
    138174        String username = emailMatcher.group(1);
    139175        if (!isValidUser(username)) {
     
    154190     * Returns true if the domain component of an email address is valid.
    155191     *
    156      * @param domain being validated.
     192     * @param domain being validated, may be in IDN format
    157193     * @return true if the email address's domain is valid.
    158194     */
     
    165201                    InetAddressValidator.getInstance();
    166202            return inetAddressValidator.isValid(ipDomainMatcher.group(1));
     203        }
     204        // Domain is symbolic name
     205        DomainValidator domainValidator =
     206                DomainValidator.getInstance(allowLocal);
     207        if (allowTld) {
     208            return domainValidator.isValid(domain) || (!domain.startsWith(".") && domainValidator.isValidTld(domain));
    167209        } else {
    168             // Domain is symbolic name
    169             DomainValidator domainValidator =
    170                     DomainValidator.getInstance(allowLocal);
    171             return domainValidator.isValid(domain) ||
    172                     domainValidator.isValidTld(domain);
     210            return domainValidator.isValid(domain);
    173211        }
    174212    }
     
    181219     */
    182220    protected boolean isValidUser(String user) {
     221
     222        if (user == null || user.length() > MAX_USERNAME_LEN) {
     223            return false;
     224        }
     225
    183226        return USER_PATTERN.matcher(user).matches();
    184227    }
  • trunk/src/org/openstreetmap/josm/data/validation/routines/InetAddressValidator.java

    r8510 r9853  
    1717package org.openstreetmap.josm.data.validation.routines;
    1818
     19import java.util.ArrayList;
     20import java.util.Arrays;
     21import java.util.List;
     22
    1923/**
    2024 * <p><b>InetAddress</b> validation and conversion routines (<code>java.net.InetAddress</code>).</p>
     
    2630 * </p>
    2731 *
    28  * @version $Revision: 1227719 $
     32 * @version $Revision: 1715439 $
    2933 * @since Validator 1.4
    3034 */
    3135public class InetAddressValidator extends AbstractValidator {
    3236
     37    private static final int IPV4_MAX_OCTET_VALUE = 255;
     38
     39    private static final int MAX_UNSIGNED_SHORT = 0xffff;
     40
     41    private static final int BASE_16 = 16;
     42
    3343    private static final String IPV4_REGEX =
    3444            "^(\\d{1,3})\\.(\\d{1,3})\\.(\\d{1,3})\\.(\\d{1,3})$";
     45
     46    // Max number of hex groups (separated by :) in an IPV6 address
     47    private static final int IPV6_MAX_HEX_GROUPS = 8;
     48
     49    // Max hex digits in each IPv6 group
     50    private static final int IPV6_MAX_HEX_DIGITS_PER_GROUP = 4;
    3551
    3652    /**
     
    5773    @Override
    5874    public boolean isValid(String inetAddress) {
    59         return isValidInet4Address(inetAddress);
     75        return isValidInet4Address(inetAddress) || isValidInet6Address(inetAddress);
    6076    }
    6177
     
    6985        String[] groups = ipv4Validator.match(inet4Address);
    7086
    71         if (groups == null) return false;
     87        if (groups == null) {
     88            return false;
     89        }
    7290
    7391        // verify that address subgroups are legal
    74         for (int i = 0; i <= 3; i++) {
    75             String ipSegment = groups[i];
     92        for (String ipSegment : groups) {
    7693            if (ipSegment == null || ipSegment.isEmpty()) {
    7794                return false;
     
    86103            }
    87104
    88             if (iIpSegment > 255) {
     105            if (iIpSegment > IPV4_MAX_OCTET_VALUE) {
     106                return false;
     107            }
     108
     109            if (ipSegment.length() > 1 && ipSegment.startsWith("0")) {
    89110                return false;
    90111            }
     
    94115        return true;
    95116    }
     117
     118    /**
     119     * Validates an IPv6 address. Returns true if valid.
     120     * @param inet6Address the IPv6 address to validate
     121     * @return true if the argument contains a valid IPv6 address
     122     *
     123     * @since 1.4.1
     124     */
     125    public boolean isValidInet6Address(String inet6Address) {
     126        boolean containsCompressedZeroes = inet6Address.contains("::");
     127        if (containsCompressedZeroes && (inet6Address.indexOf("::") != inet6Address.lastIndexOf("::"))) {
     128            return false;
     129        }
     130        if ((inet6Address.startsWith(":") && !inet6Address.startsWith("::"))
     131                || (inet6Address.endsWith(":") && !inet6Address.endsWith("::"))) {
     132            return false;
     133        }
     134        String[] octets = inet6Address.split(":");
     135        if (containsCompressedZeroes) {
     136            List<String> octetList = new ArrayList<>(Arrays.asList(octets));
     137            if (inet6Address.endsWith("::")) {
     138                // String.split() drops ending empty segments
     139                octetList.add("");
     140            } else if (inet6Address.startsWith("::") && !octetList.isEmpty()) {
     141                octetList.remove(0);
     142            }
     143            octets = octetList.toArray(new String[octetList.size()]);
     144        }
     145        if (octets.length > IPV6_MAX_HEX_GROUPS) {
     146            return false;
     147        }
     148        int validOctets = 0;
     149        int emptyOctets = 0;
     150        for (int index = 0; index < octets.length; index++) {
     151            String octet = octets[index];
     152            if (octet.length() == 0) {
     153                emptyOctets++;
     154                if (emptyOctets > 1) {
     155                    return false;
     156                }
     157            } else {
     158                emptyOctets = 0;
     159                if (octet.contains(".")) { // contains is Java 1.5+
     160                    if (!inet6Address.endsWith(octet)) {
     161                        return false;
     162                    }
     163                    if (index > octets.length - 1 || index > 6) { // TODO magic number (sort of)
     164                        // IPV4 occupies last two octets
     165                        return false;
     166                    }
     167                    if (!isValidInet4Address(octet)) {
     168                        return false;
     169                    }
     170                    validOctets += 2;
     171                    continue;
     172                }
     173                if (octet.length() > IPV6_MAX_HEX_DIGITS_PER_GROUP) {
     174                    return false;
     175                }
     176                int octetInt = 0;
     177                try {
     178                    octetInt = Integer.valueOf(octet, BASE_16).intValue();
     179                } catch (NumberFormatException e) {
     180                    return false;
     181                }
     182                if (octetInt < 0 || octetInt > MAX_UNSIGNED_SHORT) {
     183                    return false;
     184                }
     185            }
     186            validOctets++;
     187        }
     188        if (validOctets < IPV6_MAX_HEX_GROUPS && !containsCompressedZeroes) {
     189            return false;
     190        }
     191        return true;
     192    }
    96193}
  • trunk/src/org/openstreetmap/josm/data/validation/routines/RegexValidator.java

    r9249 r9853  
    2828 * a validator which does <i>case in-sensitive</i> validation for a set of regular
    2929 * expressions:
     30 * </p>
    3031 * <pre>
    31  *         String[] regexs = new String[] {...};
    32  *         RegexValidator validator = new RegexValidator(regexs, false);
     32 * <code>
     33 * String[] regexs = new String[] {...};
     34 * RegexValidator validator = new RegexValidator(regexs, false);
     35 * </code>
    3336 * </pre>
     37 *
    3438 * <ul>
    35  *   <li>Validate <code>true</code> or <code>false</code>:
    36  *   <ul>
    37  *     <li><code>boolean valid = validator.isValid(value);</code></li>
    38  *   </ul></li>
    39  *   <li>Validate returning an aggregated String of the matched groups:
    40  *   <ul>
    41  *     <li><code>String result = validator.validate(value);</code></li>
    42  *   </ul></li>
    43  *   <li>Validate returning the matched groups:
    44  *   <ul>
    45  *     <li><code>String[] result = validator.match(value);</code></li>
    46  *   </ul></li>
     39 *   <li>Validate <code>true</code> or <code>false</code>:</li>
     40 *   <li>
     41 *     <ul>
     42 *       <li><code>boolean valid = validator.isValid(value);</code></li>
     43 *     </ul>
     44 *   </li>
     45 *   <li>Validate returning an aggregated String of the matched groups:</li>
     46 *   <li>
     47 *     <ul>
     48 *       <li><code>String result = validator.validate(value);</code></li>
     49 *     </ul>
     50 *   </li>
     51 *   <li>Validate returning the matched groups:</li>
     52 *   <li>
     53 *     <ul>
     54 *       <li><code>String[] result = validator.match(value);</code></li>
     55 *     </ul>
     56 *   </li>
    4757 * </ul>
     58 *
     59 * <b>Note that patterns are matched against the entire input.</b>
     60 *
    4861 * <p>
    4962 * Cached instances pre-compile and re-use {@link Pattern}(s) - which according
    5063 * to the {@link Pattern} API are safe to use in a multi-threaded environment.
    51  *
    52  * @version $Revision: 1227719 $ $Date: 2012-01-05 18:45:51 +0100 (Thu, 05 Jan 2012) $
     64 * </p>
     65 *
     66 * @version $Revision: 1713331 $
    5367 * @since Validator 1.4
    5468 */
  • trunk/src/org/openstreetmap/josm/data/validation/routines/UrlValidator.java

    r8795 r9853  
    1919import static org.openstreetmap.josm.tools.I18n.tr;
    2020
    21 import java.util.Arrays;
     21import java.net.URI;
     22import java.net.URISyntaxException;
    2223import java.util.Collections;
    2324import java.util.HashSet;
     25import java.util.Locale;
    2426import java.util.Set;
    2527import java.util.regex.Matcher;
     
    2830/**
    2931 * <p><b>URL Validation</b> routines.</p>
    30  * Behavior of validation is modified by passing in options:<ul>
     32 * Behavior of validation is modified by passing in options:
     33 * <ul>
    3134 * <li>ALLOW_2_SLASHES - [FALSE]  Allows double '/' characters in the path
    3235 * component.</li>
     
    3639 * considered valid schemes.  Enabling this option will let any scheme pass validation.</li>
    3740 * </ul>
     41 *
    3842 * <p>Originally based in on php script by Debbie Dyer, validation.php v1.2b, Date: 03/07/02,
    3943 * http://javascript.internet.com. However, this validation now bears little resemblance
     
    6468 *  </pre>
    6569 *
    66  * @version $Revision: 1640269 $ $Date: 2014-11-18 02:28:56 UTC (Tue, 18 Nov 2014) $
     70 * @version $Revision: 1715435 $
    6771 * @see
    6872 * <a href="http://www.ietf.org/rfc/rfc2396.txt">
     
    96100     *  a {@link RegexValidator} instead ({@link #UrlValidator(RegexValidator, long)})
    97101     */
    98     public static final long ALLOW_LOCAL_URLS = 1 << 3;
    99 
    100     // Drop numeric, and  "+-." for now
    101     private static final String AUTHORITY_CHARS_REGEX = "\\p{Alnum}\\-\\.";
     102    public static final long ALLOW_LOCAL_URLS = 1 << 3; // CHECKSTYLE IGNORE MagicNumber
    102103
    103104    /**
     
    106107    private static final String URL_REGEX =
    107108            "^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\\?([^#]*))?(#(.*))?";
    108 
     109    //        12            3  4          5       6   7        8 9
    109110    private static final Pattern URL_PATTERN = Pattern.compile(URL_REGEX);
    110111
     
    126127
    127128    /**
    128      * Protocol (ie. http:, ftp:,https:).
     129     * Protocol scheme (e.g. http, ftp, https).
    129130     */
    130131    private static final String SCHEME_REGEX = "^\\p{Alpha}[\\p{Alnum}\\+\\-\\.]*";
    131132    private static final Pattern SCHEME_PATTERN = Pattern.compile(SCHEME_REGEX);
    132133
     134    // Drop numeric, and  "+-." for now
     135    // TODO does not allow for optional userinfo.
     136    // Validation of character set is done by isValidAuthority
     137    private static final String AUTHORITY_CHARS_REGEX = "\\p{Alnum}\\-\\."; // allows for IPV4 but not IPV6
     138    private static final String IPV6_REGEX = "[0-9a-fA-F:]+"; // do this as separate match because : could cause ambiguity with port prefix
     139
     140    // userinfo    = *( unreserved / pct-encoded / sub-delims / ":" )
     141    // unreserved    = ALPHA / DIGIT / "-" / "." / "_" / "~"
     142    // sub-delims    = "!" / "$" / "&" / "'" / "(" / ")" / "*" / "+" / "," / ";" / "="
     143    // We assume that password has the same valid chars as user info
     144    private static final String USERINFO_CHARS_REGEX = "[a-zA-Z0-9%-._~!$&'()*+,;=]";
     145    // since neither ':' nor '@' are allowed chars, we don't need to use non-greedy matching
     146    private static final String USERINFO_FIELD_REGEX =
     147            USERINFO_CHARS_REGEX + "+:" + // At least one character for the name
     148            USERINFO_CHARS_REGEX + "*@"; // password may be absent
    133149    private static final String AUTHORITY_REGEX =
    134             "^([" + AUTHORITY_CHARS_REGEX + "]*)(:\\d*)?(.*)?";
    135 
     150            "(?:\\[("+IPV6_REGEX+")\\]|(?:(?:"+USERINFO_FIELD_REGEX+")?([" + AUTHORITY_CHARS_REGEX + "]*)))(:\\d*)?(.*)?";
     151    //             1                          e.g. user:pass@          2                                   3       4
    136152    private static final Pattern AUTHORITY_PATTERN = Pattern.compile(AUTHORITY_REGEX);
    137153
    138     private static final int PARSE_AUTHORITY_HOST_IP = 1;
    139 
    140     private static final int PARSE_AUTHORITY_PORT = 2;
    141 
    142     /**
    143      * Should always be empty.
    144      */
    145     private static final int PARSE_AUTHORITY_EXTRA = 3;
     154    private static final int PARSE_AUTHORITY_IPV6 = 1;
     155
     156    private static final int PARSE_AUTHORITY_HOST_IP = 2; // excludes userinfo, if present
     157
     158    // Not needed, because it is validated by AUTHORITY_REGEX
     159//    private static final int PARSE_AUTHORITY_PORT = 3;
     160
     161    /**
     162     * Should always be empty. The code currently allows spaces.
     163     */
     164    private static final int PARSE_AUTHORITY_EXTRA = 4;
    146165
    147166    private static final String PATH_REGEX = "^(/[-\\w:@&?=+,.!/~*'%$_;\\(\\)]*)?$";
     
    151170    private static final Pattern QUERY_PATTERN = Pattern.compile(QUERY_REGEX);
    152171
    153     private static final String LEGAL_ASCII_REGEX = "^\\p{ASCII}+$";
    154     private static final Pattern ASCII_PATTERN = Pattern.compile(LEGAL_ASCII_REGEX);
    155 
    156     private static final String PORT_REGEX = "^:(\\d{1,5})$";
    157     private static final Pattern PORT_PATTERN = Pattern.compile(PORT_REGEX);
    158 
    159172    /**
    160173     * Holds the set of current validation options.
     
    165178     * The set of schemes that are allowed to be in a URL.
    166179     */
    167     private final Set<String> allowedSchemes;
     180    private final Set<String> allowedSchemes; // Must be lower-case
    168181
    169182    /**
     
    176189     * If no schemes are provided, default to this set.
    177190     */
    178     private static final String[] DEFAULT_SCHEMES = {"http", "https", "ftp"};
     191    private static final String[] DEFAULT_SCHEMES = {"http", "https", "ftp"}; // Must be lower-case
    179192
    180193    /**
     
    222235    /**
    223236     * Behavior of validation is modified by passing in options:
    224      * @param schemes The set of valid schemes.
     237     * @param schemes The set of valid schemes. Ignored if the ALLOW_ALL_SCHEMES option is set.
    225238     * @param options The options should be set using the public constants declared in
    226239     * this class.  To set multiple options you simply add them together.  For example,
     
    233246    /**
    234247     * Initialize a UrlValidator with the given validation options.
     248     * @param authorityValidator Regular expression validator used to validate the authority part
     249     * This allows the user to override the standard set of domains.
     250     * @param options Validation options. Set using the public constants of this class.
     251     * To set multiple options, simply add them together:
     252     * <p><code>ALLOW_2_SLASHES + NO_FRAGMENTS</code></p>
     253     * enables both of those options.
     254     */
     255    public UrlValidator(RegexValidator authorityValidator, long options) {
     256        this(null, authorityValidator, options);
     257    }
     258
     259    /**
     260     * Customizable constructor. Validation behavior is modifed by passing in options.
     261     * @param schemes the set of valid schemes. Ignored if the ALLOW_ALL_SCHEMES option is set.
    235262     * @param authorityValidator Regular expression validator used to validate the authority part
    236263     * @param options Validation options. Set using the public constants of this class.
     
    239266     * enables both of those options.
    240267     */
    241     public UrlValidator(RegexValidator authorityValidator, long options) {
    242         this(null, authorityValidator, options);
    243     }
    244 
    245     /**
    246      * Customizable constructor. Validation behavior is modifed by passing in options.
    247      * @param schemes the set of valid schemes
    248      * @param authorityValidator Regular expression validator used to validate the authority part
    249      * @param options Validation options. Set using the public constants of this class.
    250      * To set multiple options, simply add them together:
    251      * <p><code>ALLOW_2_SLASHES + NO_FRAGMENTS</code></p>
    252      * enables both of those options.
    253      */
    254268    public UrlValidator(String[] schemes, RegexValidator authorityValidator, long options) {
    255269        this.options = options;
    256270
    257271        if (isOn(ALLOW_ALL_SCHEMES)) {
    258             this.allowedSchemes = Collections.emptySet();
     272            allowedSchemes = Collections.emptySet();
    259273        } else {
    260274            if (schemes == null) {
    261275                schemes = DEFAULT_SCHEMES;
    262276            }
    263             this.allowedSchemes = new HashSet<>();
    264             this.allowedSchemes.addAll(Arrays.asList(schemes));
     277            allowedSchemes = new HashSet<>(schemes.length);
     278            for (int i = 0; i < schemes.length; i++) {
     279                allowedSchemes.add(schemes[i].toLowerCase(Locale.ENGLISH));
     280            }
    265281        }
    266282
    267283        this.authorityValidator = authorityValidator;
    268 
    269284    }
    270285
    271286    /**
    272287     * <p>Checks if a field has a valid url address.</p>
     288     *
     289     * Note that the method calls #isValidAuthority()
     290     * which checks that the domain is valid.
    273291     *
    274292     * @param value The value validation is being performed on.  A <code>null</code>
     
    282300        }
    283301
    284         if (!ASCII_PATTERN.matcher(value).matches()) {
    285             setErrorMessage(tr("URL contains non-ascii characters"));
    286             setFix(value.replaceAll("[^\\p{ASCII}]+", ""));
    287             return false;
    288         }
    289 
    290302        // Check the whole url address structure
    291303        Matcher urlMatcher = URL_PATTERN.matcher(value);
     
    302314
    303315        String authority = urlMatcher.group(PARSE_URL_AUTHORITY);
    304         if ("file".equals(scheme) && "".equals(authority)) {
    305            // Special case - file: allows an empty authority
    306         } else {
    307            // Validate the authority
    308            if (!isValidAuthority(authority)) {
    309                setErrorMessage(tr("URL contains an invalid authority: {0}", authority));
    310                return false;
     316        if ("file".equals(scheme)) { // Special case - file: allows an empty authority
     317            if (!"".equals(authority)) {
     318                if (authority.contains(":")) { // but cannot allow trailing :
     319                    setErrorMessage(tr("URL contains an invalid authority: {0}", authority));
     320                    return false;
     321                }
     322            }
     323            // drop through to continue validation
     324        } else { // not file:
     325            // Validate the authority
     326            if (!isValidAuthority(authority)) {
     327                setErrorMessage(tr("URL contains an invalid authority: {0}", authority));
     328                return false;
    311329            }
    312330        }
     
    335353    /**
    336354     * Validate scheme. If schemes[] was initialized to a non null,
    337      * then only those scheme's are allowed.  Note this is slightly different
    338      * than for the constructor.
     355     * then only those schemes are allowed.
     356     * Otherwise the default schemes are "http", "https", "ftp".
     357     * Matching is case-blind.
    339358     * @param scheme The scheme to validate.  A <code>null</code> value is considered
    340359     * invalid.
     
    346365        }
    347366
     367        // TODO could be removed if external schemes were checked in the ctor before being stored
    348368        if (!SCHEME_PATTERN.matcher(scheme).matches()) {
    349369            return false;
    350370        }
    351371
    352         if (isOff(ALLOW_ALL_SCHEMES)) {
    353 
    354             if (!this.allowedSchemes.contains(scheme)) {
    355                 return false;
    356             }
     372        if (isOff(ALLOW_ALL_SCHEMES) && !allowedSchemes.contains(scheme.toLowerCase(Locale.ENGLISH))) {
     373            return false;
    357374        }
    358375
     
    363380     * Returns true if the authority is properly formatted.  An authority is the combination
    364381     * of hostname and port.  A <code>null</code> authority value is considered invalid.
    365      * @param authority Authority value to validate.
     382     * Note: this implementation validates the domain unless a RegexValidator was provided.
     383     * If a RegexValidator was supplied and it matches, then the authority is regarded
     384     * as valid with no further checks, otherwise the method checks against the
     385     * AUTHORITY_PATTERN and the DomainValidator (ALLOW_LOCAL_URLS)
     386     * @param authority Authority value to validate, alllows IDN
    366387     * @return true if authority (hostname and port) is valid.
    367388     */
     
    372393
    373394        // check manual authority validation if specified
    374         if (authorityValidator != null) {
    375             if (authorityValidator.isValid(authority)) {
    376                 return true;
    377             }
    378         }
    379 
    380         Matcher authorityMatcher = AUTHORITY_PATTERN.matcher(authority);
     395        if (authorityValidator != null && authorityValidator.isValid(authority)) {
     396            return true;
     397        }
     398        // convert to ASCII if possible
     399        final String authorityASCII = DomainValidator.unicodeToASCII(authority);
     400
     401        Matcher authorityMatcher = AUTHORITY_PATTERN.matcher(authorityASCII);
    381402        if (!authorityMatcher.matches()) {
    382403            return false;
    383404        }
    384405
    385         String hostLocation = authorityMatcher.group(PARSE_AUTHORITY_HOST_IP);
    386         // check if authority is hostname or IP address:
    387         // try a hostname first since that's much more likely
    388         DomainValidator domainValidator = DomainValidator.getInstance(isOn(ALLOW_LOCAL_URLS));
    389         if (!domainValidator.isValid(hostLocation)) {
    390             // try an IP address
    391             InetAddressValidator inetAddressValidator =
    392                 InetAddressValidator.getInstance();
    393             if (!inetAddressValidator.isValid(hostLocation)) {
    394                 // isn't either one, so the URL is invalid
    395                 return false;
    396             }
    397         }
    398 
    399         String port = authorityMatcher.group(PARSE_AUTHORITY_PORT);
    400         if (port != null) {
    401             if (!PORT_PATTERN.matcher(port).matches()) {
    402                 return false;
     406        // We have to process IPV6 separately because that is parsed in a different group
     407        String ipv6 = authorityMatcher.group(PARSE_AUTHORITY_IPV6);
     408        if (ipv6 != null) {
     409            InetAddressValidator inetAddressValidator = InetAddressValidator.getInstance();
     410                if (!inetAddressValidator.isValidInet6Address(ipv6)) {
     411                    return false;
     412                }
     413        } else {
     414            String hostLocation = authorityMatcher.group(PARSE_AUTHORITY_HOST_IP);
     415            // check if authority is hostname or IP address:
     416            // try a hostname first since that's much more likely
     417            DomainValidator domainValidator = DomainValidator.getInstance(isOn(ALLOW_LOCAL_URLS));
     418            if (!domainValidator.isValid(hostLocation)) {
     419                // try an IPv4 address
     420                InetAddressValidator inetAddressValidator = InetAddressValidator.getInstance();
     421                if (!inetAddressValidator.isValidInet4Address(hostLocation)) {
     422                    // isn't IPv4, so the URL is invalid
     423                    return false;
     424                }
    403425            }
    404426        }
     
    426448        }
    427449
     450        try {
     451            URI uri = new URI(null, null, path, null);
     452            String norm = uri.normalize().getPath();
     453            if (norm.startsWith("/../") // Trying to go via the parent dir
     454             || norm.equals("/..")) {   // Trying to go to the parent dir
     455                return false;
     456            }
     457        } catch (URISyntaxException e) {
     458            return false;
     459        }
     460
    428461        int slash2Count = countToken("//", path);
    429462        if (isOff(ALLOW_2_SLASHES) && (slash2Count > 0)) {
    430463            return false;
    431         }
    432 
    433         int slashCount = countToken("/", path);
    434         int dot2Count = countToken("..", path);
    435         if (dot2Count > 0) {
    436             if ((slashCount - slash2Count - 1) <= dot2Count) {
    437                 return false;
    438             }
    439464        }
    440465
     
    496521     */
    497522    private boolean isOn(long flag) {
    498         return (this.options & flag) > 0;
     523        return (options & flag) > 0;
    499524    }
    500525
     
    508533     */
    509534    private boolean isOff(long flag) {
    510         return (this.options & flag) == 0;
     535        return (options & flag) == 0;
     536    }
     537
     538    // Unit test access to pattern matcher
     539    Matcher matchURL(String value) {
     540        return URL_PATTERN.matcher(value);
    511541    }
    512542}
  • trunk/src/org/openstreetmap/josm/data/validation/routines/package.html

    r7489 r9853  
    2020</head>
    2121<body bgcolor="white">
    22     <p>This package contains <i>independant</i> validation routines adapted from Apache Commons Validator 1.4.0.</p>
     22    <p>This package contains <i>independant</i> validation routines adapted from Apache Commons Validator 1.5.0.</p>
    2323<h1>Table of Contents</h1>
    2424
     
    3636</ul>
    3737
    38 <a id="overview"></a>
     38<a name="overview"></a>
    3939<h1>1. Overview</h1>
    4040<p>
     
    4242</p>
    4343    <ul>
    44        <li>To provide standard, independant validation routines/functions.</li>
     44       <li>To provide standard, independent validation routines/functions.</li>
    4545       <li>To provide a <i>mini</i> framework for Validation.</li>
    4646    </ul>
    4747<p>
    4848   This package has been created, since version 1.3.0, in an attempt to clearly
    49    separate these two concerns and is the location for the standard, independant
     49   separate these two concerns and is the location for the standard, independent
    5050   validation routines/functions in <em>Commons Validator</em>.
    5151</p>
     
    5656</p>
    5757
    58 <a id="other"></a>
     58<a name="other"></a>
    5959<h1>2. Validators</h1>
    6060
    61 <a id="other.overview"></a>
     61<a name="other.overview"></a>
    6262<h3>2.1 Overview</h3>
    6363<p>
     
    7777</ul>
    7878
    79 <a id="other.regex"></a>
     79<a name="other.regex"></a>
    8080<h3>2.2 Regular Expression Validation</h3>
    8181<p>
     
    157157</p>
    158158
    159 <a id="other.inet"></a>
     159<a name="other.inet"></a>
    160160<h3>2.3 IP Address Validation</h3>
    161161
     
    179179</pre>
    180180
    181 <a id="other.email"></a>
     181<a name="other.email"></a>
    182182<h3>2.4 Email Address Validation</h3>
    183183
     
    203203</pre>
    204204
    205 <a id="other.url"></a>
     205<a name="other.url"></a>
    206206<h3>2.5 URL Validation</h3>
    207207
     
    244244</pre>
    245245
    246 <a id="other.domain"></a>
     246<a name="other.domain"></a>
    247247<h3>2.6 Domain Name Validation</h3>
    248248
  • trunk/test/unit/org/openstreetmap/josm/data/validation/tests/InternetTagsTest.java

    r7937 r9853  
    3131
    3232    /**
    33      * Test of "Internet Tags" validation test.
     33     * Test of valid URLs.
    3434     */
    3535    @Test
    36     public void test() {
    37 
    38         // Valid URLs
     36    public void testValidUrls() {
    3937        testUrl("url", "www.domain.com", true);                                // No protocol
    4038        testUrl("url", "http://josm.openstreetmap.de", true);                  // Simple HTTP
     
    4745        testUrl("website", "http://золотаяцепь.рф", true);                     // see #10862: IDN URL in Unicode form
    4846        testUrl("website", "http://золотаяцепь.рф/", true);                    // see #10862: IDN URL in Unicode form + slash
     47        testUrl("website", "http://www.dasideenreich.online", true);           // see #12257: new TLD added August 19, 2015
     48    }
    4949
    50         // Invalid URLs
     50    /**
     51     * Test of invalid URLs.
     52     */
     53    @Test
     54    public void testInvalidUrls() {
    5155        testUrl("url", "something://www.domain.com", false);                   // invalid protocol
    5256        testUrl("url", "http://www.domain.invalidtld", false);                 // invalid TLD
     57    }
    5358
    54         // Valid E-mails
     59    /**
     60     * Test of valid e-mails.
     61     */
     62    @Test
     63    public void testValidEmails() {
    5564        testEmail("email", "contact@www.domain.com", true);                    // Simple email
    5665        testEmail("contact:email", "john.doe@other-domain.org", true);         // Key with : + dash in domain
     66    }
    5767
    58         // Invalid E-mails
     68    /**
     69     * Test of invalid e-mails.
     70     */
     71    @Test
     72    public void testInvalidEmails() {
    5973        testEmail("email", "contact at www.domain.com", false);                // No @
    6074        testEmail("contact:email", "john.doe@other-domain.invalidtld", false); // invalid TLD
  • trunk/tools/checkstyle/josm_filters.xml

    r9297 r9853  
    99  <suppress checks="HeaderCheck" files="NTV2SubGrid\.java" />
    1010  <suppress checks="HeaderCheck" files="NTV2Util\.java" />
    11   <suppress checks="HeaderCheck" files="DomainValidator\.java" />
    12   <suppress checks="HeaderCheck" files="EmailValidator\.java" />
    13   <suppress checks="HeaderCheck" files="InetAddressValidator\.java" />
    14   <suppress checks="HeaderCheck" files="RegexValidator\.java" />
    15   <suppress checks="HeaderCheck" files="UrlValidator\.java" />
     11  <suppress checks="HeaderCheck" files="DomainValidator(Test|IT)?\.java" />
     12  <suppress checks="HeaderCheck" files="EmailValidator(Test)?\.java" />
     13  <suppress checks="HeaderCheck" files="InetAddressValidator(Test)?\.java" />
     14  <suppress checks="HeaderCheck" files="RegexValidator(Test)?\.java" />
     15  <suppress checks="HeaderCheck" files="UrlValidator(Test)?\.java" />
     16  <suppress checks="HeaderCheck" files="ResultPair\.java" />
    1617  <suppress checks="HeaderCheck" files="Entities\.java" />
    1718  <suppress checks="HeaderCheck" files="FileDrop\.java" />
Note: See TracChangeset for help on using the changeset viewer.