| 1 | /*
|
|---|
| 2 | * Licensed to the Apache Software Foundation (ASF) under one or more
|
|---|
| 3 | * contributor license agreements. See the NOTICE file distributed with
|
|---|
| 4 | * this work for additional information regarding copyright ownership.
|
|---|
| 5 | * The ASF licenses this file to You under the Apache License, Version 2.0
|
|---|
| 6 | * (the "License"); you may not use this file except in compliance with
|
|---|
| 7 | * the License. You may obtain a copy of the License at
|
|---|
| 8 | *
|
|---|
| 9 | * http://www.apache.org/licenses/LICENSE-2.0
|
|---|
| 10 | *
|
|---|
| 11 | * Unless required by applicable law or agreed to in writing, software
|
|---|
| 12 | * distributed under the License is distributed on an "AS IS" BASIS,
|
|---|
| 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|---|
| 14 | * See the License for the specific language governing permissions and
|
|---|
| 15 | * limitations under the License.
|
|---|
| 16 | */
|
|---|
| 17 | package org.openstreetmap.josm.data.validation.routines;
|
|---|
| 18 |
|
|---|
| 19 | import java.util.Arrays;
|
|---|
| 20 |
|
|---|
| 21 | /**
|
|---|
| 22 | * <p><b>Domain name</b> validation routines.</p>
|
|---|
| 23 | *
|
|---|
| 24 | * <p>
|
|---|
| 25 | * This validator provides methods for validating Internet domain names
|
|---|
| 26 | * and top-level domains.
|
|---|
| 27 | * </p>
|
|---|
| 28 | *
|
|---|
| 29 | * <p>Domain names are evaluated according
|
|---|
| 30 | * to the standards <a href="http://www.ietf.org/rfc/rfc1034.txt">RFC1034</a>,
|
|---|
| 31 | * section 3, and <a href="http://www.ietf.org/rfc/rfc1123.txt">RFC1123</a>,
|
|---|
| 32 | * section 2.1. No accomodation is provided for the specialized needs of
|
|---|
| 33 | * other applications; if the domain name has been URL-encoded, for example,
|
|---|
| 34 | * validation will fail even though the equivalent plaintext version of the
|
|---|
| 35 | * same name would have passed.
|
|---|
| 36 | * </p>
|
|---|
| 37 | *
|
|---|
| 38 | * <p>
|
|---|
| 39 | * Validation is also provided for top-level domains (TLDs) as defined and
|
|---|
| 40 | * maintained by the Internet Assigned Numbers Authority (IANA):
|
|---|
| 41 | * </p>
|
|---|
| 42 | *
|
|---|
| 43 | * <ul>
|
|---|
| 44 | * <li>{@link #isValidInfrastructureTld} - validates infrastructure TLDs
|
|---|
| 45 | * (<code>.arpa</code>, etc.)</li>
|
|---|
| 46 | * <li>{@link #isValidGenericTld} - validates generic TLDs
|
|---|
| 47 | * (<code>.com, .org</code>, etc.)</li>
|
|---|
| 48 | * <li>{@link #isValidCountryCodeTld} - validates country code TLDs
|
|---|
| 49 | * (<code>.us, .uk, .cn</code>, etc.)</li>
|
|---|
| 50 | * </ul>
|
|---|
| 51 | *
|
|---|
| 52 | * <p>
|
|---|
| 53 | * (<b>NOTE</b>: This class does not provide IP address lookup for domain names or
|
|---|
| 54 | * methods to ensure that a given domain name matches a specific IP; see
|
|---|
| 55 | * {@link java.net.InetAddress} for that functionality.)
|
|---|
| 56 | * </p>
|
|---|
| 57 | *
|
|---|
| 58 | * @version $Revision: 1640271 $ $Date: 2014-11-18 02:32:15 2014 UTC (Tue, 18 Nov 2014) $
|
|---|
| 59 | * @since Validator 1.4
|
|---|
| 60 | */
|
|---|
| 61 | public class DomainValidator extends AbstractValidator {
|
|---|
| 62 |
|
|---|
| 63 | // Regular expression strings for hostnames (derived from RFC2396 and RFC 1123)
|
|---|
| 64 | private static final String DOMAIN_LABEL_REGEX = "\\p{Alnum}(?>[\\p{Alnum}-]*\\p{Alnum})*";
|
|---|
| 65 | private static final String TOP_LABEL_REGEX = "\\p{Alpha}{2,}";
|
|---|
| 66 | private static final String DOMAIN_NAME_REGEX =
|
|---|
| 67 | "^(?:" + DOMAIN_LABEL_REGEX + "\\.)+" + "(" + TOP_LABEL_REGEX + ")$";
|
|---|
| 68 |
|
|---|
| 69 | private final boolean allowLocal;
|
|---|
| 70 |
|
|---|
| 71 | /**
|
|---|
| 72 | * Singleton instance of this validator, which
|
|---|
| 73 | * doesn't consider local addresses as valid.
|
|---|
| 74 | */
|
|---|
| 75 | private static final DomainValidator DOMAIN_VALIDATOR = new DomainValidator(false);
|
|---|
| 76 |
|
|---|
| 77 | /**
|
|---|
| 78 | * Singleton instance of this validator, which does
|
|---|
| 79 | * consider local addresses valid.
|
|---|
| 80 | */
|
|---|
| 81 | private static final DomainValidator DOMAIN_VALIDATOR_WITH_LOCAL = new DomainValidator(true);
|
|---|
| 82 |
|
|---|
| 83 | /**
|
|---|
| 84 | * RegexValidator for matching domains.
|
|---|
| 85 | */
|
|---|
| 86 | private final RegexValidator domainRegex =
|
|---|
| 87 | new RegexValidator(DOMAIN_NAME_REGEX);
|
|---|
| 88 | /**
|
|---|
| 89 | * RegexValidator for matching the a local hostname
|
|---|
| 90 | */
|
|---|
| 91 | private final RegexValidator hostnameRegex =
|
|---|
| 92 | new RegexValidator(DOMAIN_LABEL_REGEX);
|
|---|
| 93 |
|
|---|
| 94 | /**
|
|---|
| 95 | * Returns the singleton instance of this validator. It
|
|---|
| 96 | * will not consider local addresses as valid.
|
|---|
| 97 | * @return the singleton instance of this validator
|
|---|
| 98 | */
|
|---|
| 99 | public static DomainValidator getInstance() {
|
|---|
| 100 | return DOMAIN_VALIDATOR;
|
|---|
| 101 | }
|
|---|
| 102 |
|
|---|
| 103 | /**
|
|---|
| 104 | * Returns the singleton instance of this validator,
|
|---|
| 105 | * with local validation as required.
|
|---|
| 106 | * @param allowLocal Should local addresses be considered valid?
|
|---|
| 107 | * @return the singleton instance of this validator
|
|---|
| 108 | */
|
|---|
| 109 | public static DomainValidator getInstance(boolean allowLocal) {
|
|---|
| 110 | if(allowLocal) {
|
|---|
| 111 | return DOMAIN_VALIDATOR_WITH_LOCAL;
|
|---|
| 112 | }
|
|---|
| 113 | return DOMAIN_VALIDATOR;
|
|---|
| 114 | }
|
|---|
| 115 |
|
|---|
| 116 | /** Private constructor. */
|
|---|
| 117 | private DomainValidator(boolean allowLocal) {
|
|---|
| 118 | this.allowLocal = allowLocal;
|
|---|
| 119 | }
|
|---|
| 120 |
|
|---|
| 121 | /**
|
|---|
| 122 | * Returns true if the specified <code>String</code> parses
|
|---|
| 123 | * as a valid domain name with a recognized top-level domain.
|
|---|
| 124 | * The parsing is case-sensitive.
|
|---|
| 125 | * @param domain the parameter to check for domain name syntax
|
|---|
| 126 | * @return true if the parameter is a valid domain name
|
|---|
| 127 | */
|
|---|
| 128 | @Override
|
|---|
| 129 | public boolean isValid(String domain) {
|
|---|
| 130 | String[] groups = domainRegex.match(domain);
|
|---|
| 131 | if (groups != null && groups.length > 0) {
|
|---|
| 132 | return isValidTld(groups[0]);
|
|---|
| 133 | } else if(allowLocal) {
|
|---|
| 134 | if (hostnameRegex.isValid(domain)) {
|
|---|
| 135 | return true;
|
|---|
| 136 | }
|
|---|
| 137 | }
|
|---|
| 138 | return false;
|
|---|
| 139 | }
|
|---|
| 140 |
|
|---|
| 141 | /**
|
|---|
| 142 | * Returns true if the specified <code>String</code> matches any
|
|---|
| 143 | * IANA-defined top-level domain. Leading dots are ignored if present.
|
|---|
| 144 | * The search is case-sensitive.
|
|---|
| 145 | * @param tld the parameter to check for TLD status
|
|---|
| 146 | * @return true if the parameter is a TLD
|
|---|
| 147 | */
|
|---|
| 148 | public boolean isValidTld(String tld) {
|
|---|
| 149 | if(allowLocal && isValidLocalTld(tld)) {
|
|---|
| 150 | return true;
|
|---|
| 151 | }
|
|---|
| 152 | return isValidInfrastructureTld(tld)
|
|---|
| 153 | || isValidGenericTld(tld)
|
|---|
| 154 | || isValidCountryCodeTld(tld);
|
|---|
| 155 | }
|
|---|
| 156 |
|
|---|
| 157 | /**
|
|---|
| 158 | * Returns true if the specified <code>String</code> matches any
|
|---|
| 159 | * IANA-defined infrastructure top-level domain. Leading dots are
|
|---|
| 160 | * ignored if present. The search is case-sensitive.
|
|---|
| 161 | * @param iTld the parameter to check for infrastructure TLD status
|
|---|
| 162 | * @return true if the parameter is an infrastructure TLD
|
|---|
| 163 | */
|
|---|
| 164 | public boolean isValidInfrastructureTld(String iTld) {
|
|---|
| 165 | return Arrays.binarySearch(INFRASTRUCTURE_TLDS, (chompLeadingDot(iTld.toLowerCase()))) >= 0;
|
|---|
| 166 | }
|
|---|
| 167 |
|
|---|
| 168 | /**
|
|---|
| 169 | * Returns true if the specified <code>String</code> matches any
|
|---|
| 170 | * IANA-defined generic top-level domain. Leading dots are ignored
|
|---|
| 171 | * if present. The search is case-sensitive.
|
|---|
| 172 | * @param gTld the parameter to check for generic TLD status
|
|---|
| 173 | * @return true if the parameter is a generic TLD
|
|---|
| 174 | */
|
|---|
| 175 | public boolean isValidGenericTld(String gTld) {
|
|---|
| 176 | return Arrays.binarySearch(GENERIC_TLDS, chompLeadingDot(gTld.toLowerCase())) >= 0;
|
|---|
| 177 | }
|
|---|
| 178 |
|
|---|
| 179 | /**
|
|---|
| 180 | * Returns true if the specified <code>String</code> matches any
|
|---|
| 181 | * IANA-defined country code top-level domain. Leading dots are
|
|---|
| 182 | * ignored if present. The search is case-sensitive.
|
|---|
| 183 | * @param ccTld the parameter to check for country code TLD status
|
|---|
| 184 | * @return true if the parameter is a country code TLD
|
|---|
| 185 | */
|
|---|
| 186 | public boolean isValidCountryCodeTld(String ccTld) {
|
|---|
| 187 | return Arrays.binarySearch(COUNTRY_CODE_TLDS, chompLeadingDot(ccTld.toLowerCase())) >= 0;
|
|---|
| 188 | }
|
|---|
| 189 |
|
|---|
| 190 | /**
|
|---|
| 191 | * Returns true if the specified <code>String</code> matches any
|
|---|
| 192 | * widely used "local" domains (localhost or localdomain). Leading dots are
|
|---|
| 193 | * ignored if present. The search is case-sensitive.
|
|---|
| 194 | * @param iTld the parameter to check for local TLD status
|
|---|
| 195 | * @return true if the parameter is an local TLD
|
|---|
| 196 | */
|
|---|
| 197 | public boolean isValidLocalTld(String iTld) {
|
|---|
| 198 | return Arrays.binarySearch(LOCAL_TLDS, chompLeadingDot(iTld.toLowerCase())) >= 0;
|
|---|
| 199 | }
|
|---|
| 200 |
|
|---|
| 201 | private String chompLeadingDot(String str) {
|
|---|
| 202 | if (str.startsWith(".")) {
|
|---|
| 203 | return str.substring(1);
|
|---|
| 204 | } else {
|
|---|
| 205 | return str;
|
|---|
| 206 | }
|
|---|
| 207 | }
|
|---|
| 208 |
|
|---|
| 209 | // ---------------------------------------------
|
|---|
| 210 | // ----- TLDs defined by IANA
|
|---|
| 211 | // ----- Authoritative and comprehensive list at:
|
|---|
| 212 | // ----- http://data.iana.org/TLD/tlds-alpha-by-domain.txt
|
|---|
| 213 |
|
|---|
| 214 | private static final String[] INFRASTRUCTURE_TLDS = new String[] {
|
|---|
| 215 | "arpa", // internet infrastructure
|
|---|
| 216 | "root" // diagnostic marker for non-truncated root zone
|
|---|
| 217 | };
|
|---|
| 218 |
|
|---|
| 219 | private static final String[] GENERIC_TLDS = new String[] {
|
|---|
| 220 | "abogado",
|
|---|
| 221 | "academy",
|
|---|
| 222 | "accountants",
|
|---|
| 223 | "active",
|
|---|
| 224 | "actor",
|
|---|
| 225 | "aero",
|
|---|
| 226 | "agency",
|
|---|
| 227 | "airforce",
|
|---|
| 228 | "allfinanz",
|
|---|
| 229 | "alsace",
|
|---|
| 230 | "archi",
|
|---|
| 231 | "army",
|
|---|
| 232 | "arpa",
|
|---|
| 233 | "asia",
|
|---|
| 234 | "associates",
|
|---|
| 235 | "attorney",
|
|---|
| 236 | "auction",
|
|---|
| 237 | "audio",
|
|---|
| 238 | "autos",
|
|---|
| 239 | "axa",
|
|---|
| 240 | "band",
|
|---|
| 241 | "bar",
|
|---|
| 242 | "bargains",
|
|---|
| 243 | "bayern",
|
|---|
| 244 | "beer",
|
|---|
| 245 | "berlin",
|
|---|
| 246 | "best",
|
|---|
| 247 | "bid",
|
|---|
| 248 | "bike",
|
|---|
| 249 | "bio",
|
|---|
| 250 | "biz",
|
|---|
| 251 | "black",
|
|---|
| 252 | "blackfriday",
|
|---|
| 253 | "blue",
|
|---|
| 254 | "bmw",
|
|---|
| 255 | "bnpparibas",
|
|---|
| 256 | "boo",
|
|---|
| 257 | "boutique",
|
|---|
| 258 | "brussels",
|
|---|
| 259 | "budapest",
|
|---|
| 260 | "build",
|
|---|
| 261 | "builders",
|
|---|
| 262 | "business",
|
|---|
| 263 | "buzz",
|
|---|
| 264 | "bzh",
|
|---|
| 265 | "cab",
|
|---|
| 266 | "cal",
|
|---|
| 267 | "camera",
|
|---|
| 268 | "camp",
|
|---|
| 269 | "cancerresearch",
|
|---|
| 270 | "capetown",
|
|---|
| 271 | "capital",
|
|---|
| 272 | "caravan",
|
|---|
| 273 | "cards",
|
|---|
| 274 | "care",
|
|---|
| 275 | "career",
|
|---|
| 276 | "careers",
|
|---|
| 277 | "casa",
|
|---|
| 278 | "cash",
|
|---|
| 279 | "cat",
|
|---|
| 280 | "catering",
|
|---|
| 281 | "center",
|
|---|
| 282 | "ceo",
|
|---|
| 283 | "cern",
|
|---|
| 284 | "channel",
|
|---|
| 285 | "cheap",
|
|---|
| 286 | "christmas",
|
|---|
| 287 | "chrome",
|
|---|
| 288 | "church",
|
|---|
| 289 | "citic",
|
|---|
| 290 | "city",
|
|---|
| 291 | "claims",
|
|---|
| 292 | "cleaning",
|
|---|
| 293 | "click",
|
|---|
| 294 | "clinic",
|
|---|
| 295 | "clothing",
|
|---|
| 296 | "club",
|
|---|
| 297 | "codes",
|
|---|
| 298 | "coffee",
|
|---|
| 299 | "college",
|
|---|
| 300 | "cologne",
|
|---|
| 301 | "com",
|
|---|
| 302 | "community",
|
|---|
| 303 | "company",
|
|---|
| 304 | "computer",
|
|---|
| 305 | "condos",
|
|---|
| 306 | "construction",
|
|---|
| 307 | "consulting",
|
|---|
| 308 | "contractors",
|
|---|
| 309 | "cooking",
|
|---|
| 310 | "cool",
|
|---|
| 311 | "coop",
|
|---|
| 312 | "country",
|
|---|
| 313 | "credit",
|
|---|
| 314 | "creditcard",
|
|---|
| 315 | "crs",
|
|---|
| 316 | "cruises",
|
|---|
| 317 | "cuisinella",
|
|---|
| 318 | "cymru",
|
|---|
| 319 | "dad",
|
|---|
| 320 | "dance",
|
|---|
| 321 | "dating",
|
|---|
| 322 | "day",
|
|---|
| 323 | "deals",
|
|---|
| 324 | "degree",
|
|---|
| 325 | "democrat",
|
|---|
| 326 | "dental",
|
|---|
| 327 | "dentist",
|
|---|
| 328 | "desi",
|
|---|
| 329 | "diamonds",
|
|---|
| 330 | "diet",
|
|---|
| 331 | "digital",
|
|---|
| 332 | "direct",
|
|---|
| 333 | "directory",
|
|---|
| 334 | "discount",
|
|---|
| 335 | "dnp",
|
|---|
| 336 | "domains",
|
|---|
| 337 | "durban",
|
|---|
| 338 | "dvag",
|
|---|
| 339 | "eat",
|
|---|
| 340 | "edu",
|
|---|
| 341 | "education",
|
|---|
| 342 | "email",
|
|---|
| 343 | "engineer",
|
|---|
| 344 | "engineering",
|
|---|
| 345 | "enterprises",
|
|---|
| 346 | "equipment",
|
|---|
| 347 | "esq",
|
|---|
| 348 | "estate",
|
|---|
| 349 | "eus",
|
|---|
| 350 | "events",
|
|---|
| 351 | "exchange",
|
|---|
| 352 | "expert",
|
|---|
| 353 | "exposed",
|
|---|
| 354 | "fail",
|
|---|
| 355 | "farm",
|
|---|
| 356 | "feedback",
|
|---|
| 357 | "finance",
|
|---|
| 358 | "financial",
|
|---|
| 359 | "fish",
|
|---|
| 360 | "fishing",
|
|---|
| 361 | "fitness",
|
|---|
| 362 | "flights",
|
|---|
| 363 | "florist",
|
|---|
| 364 | "flsmidth",
|
|---|
| 365 | "fly",
|
|---|
| 366 | "foo",
|
|---|
| 367 | "forsale",
|
|---|
| 368 | "foundation",
|
|---|
| 369 | "frl",
|
|---|
| 370 | "frogans",
|
|---|
| 371 | "fund",
|
|---|
| 372 | "furniture",
|
|---|
| 373 | "futbol",
|
|---|
| 374 | "gal",
|
|---|
| 375 | "gallery",
|
|---|
| 376 | "gbiz",
|
|---|
| 377 | "gent",
|
|---|
| 378 | "gift",
|
|---|
| 379 | "gifts",
|
|---|
| 380 | "gives",
|
|---|
| 381 | "glass",
|
|---|
| 382 | "gle",
|
|---|
| 383 | "global",
|
|---|
| 384 | "globo",
|
|---|
| 385 | "gmail",
|
|---|
| 386 | "gmo",
|
|---|
| 387 | "gmx",
|
|---|
| 388 | "google",
|
|---|
| 389 | "gop",
|
|---|
| 390 | "gov",
|
|---|
| 391 | "graphics",
|
|---|
| 392 | "gratis",
|
|---|
| 393 | "green",
|
|---|
| 394 | "gripe",
|
|---|
| 395 | "guide",
|
|---|
| 396 | "guitars",
|
|---|
| 397 | "guru",
|
|---|
| 398 | "hamburg",
|
|---|
| 399 | "haus",
|
|---|
| 400 | "healthcare",
|
|---|
| 401 | "help",
|
|---|
| 402 | "here",
|
|---|
| 403 | "hiphop",
|
|---|
| 404 | "hiv",
|
|---|
| 405 | "holdings",
|
|---|
| 406 | "holiday",
|
|---|
| 407 | "homes",
|
|---|
| 408 | "horse",
|
|---|
| 409 | "host",
|
|---|
| 410 | "hosting",
|
|---|
| 411 | "house",
|
|---|
| 412 | "how",
|
|---|
| 413 | "ibm",
|
|---|
| 414 | "immo",
|
|---|
| 415 | "immobilien",
|
|---|
| 416 | "industries",
|
|---|
| 417 | "info",
|
|---|
| 418 | "ing",
|
|---|
| 419 | "ink",
|
|---|
| 420 | "institute",
|
|---|
| 421 | "insure",
|
|---|
| 422 | "int",
|
|---|
| 423 | "international",
|
|---|
| 424 | "investments",
|
|---|
| 425 | "jetzt",
|
|---|
| 426 | "jobs",
|
|---|
| 427 | "joburg",
|
|---|
| 428 | "juegos",
|
|---|
| 429 | "kaufen",
|
|---|
| 430 | "kim",
|
|---|
| 431 | "kitchen",
|
|---|
| 432 | "kiwi",
|
|---|
| 433 | "koeln",
|
|---|
| 434 | "krd",
|
|---|
| 435 | "kred",
|
|---|
| 436 | "lacaixa",
|
|---|
| 437 | "land",
|
|---|
| 438 | "lawyer",
|
|---|
| 439 | "lease",
|
|---|
| 440 | "lgbt",
|
|---|
| 441 | "life",
|
|---|
| 442 | "lighting",
|
|---|
| 443 | "limited",
|
|---|
| 444 | "limo",
|
|---|
| 445 | "link",
|
|---|
| 446 | "loans",
|
|---|
| 447 | "london",
|
|---|
| 448 | "lotto",
|
|---|
| 449 | "ltda",
|
|---|
| 450 | "luxe",
|
|---|
| 451 | "luxury",
|
|---|
| 452 | "maison",
|
|---|
| 453 | "management",
|
|---|
| 454 | "mango",
|
|---|
| 455 | "market",
|
|---|
| 456 | "marketing",
|
|---|
| 457 | "media",
|
|---|
| 458 | "meet",
|
|---|
| 459 | "melbourne",
|
|---|
| 460 | "meme",
|
|---|
| 461 | "menu",
|
|---|
| 462 | "miami",
|
|---|
| 463 | "mil",
|
|---|
| 464 | "mini",
|
|---|
| 465 | "mobi",
|
|---|
| 466 | "moda",
|
|---|
| 467 | "moe",
|
|---|
| 468 | "monash",
|
|---|
| 469 | "mortgage",
|
|---|
| 470 | "moscow",
|
|---|
| 471 | "motorcycles",
|
|---|
| 472 | "mov",
|
|---|
| 473 | "museum",
|
|---|
| 474 | "nagoya",
|
|---|
| 475 | "name",
|
|---|
| 476 | "navy",
|
|---|
| 477 | "net",
|
|---|
| 478 | "network",
|
|---|
| 479 | "neustar",
|
|---|
| 480 | "new",
|
|---|
| 481 | "nexus",
|
|---|
| 482 | "ngo",
|
|---|
| 483 | "nhk",
|
|---|
| 484 | "ninja",
|
|---|
| 485 | "nra",
|
|---|
| 486 | "nrw",
|
|---|
| 487 | "nyc",
|
|---|
| 488 | "okinawa",
|
|---|
| 489 | "ong",
|
|---|
| 490 | "onl",
|
|---|
| 491 | "ooo",
|
|---|
| 492 | "org",
|
|---|
| 493 | "organic",
|
|---|
| 494 | "otsuka",
|
|---|
| 495 | "ovh",
|
|---|
| 496 | "paris",
|
|---|
| 497 | "partners",
|
|---|
| 498 | "parts",
|
|---|
| 499 | "pharmacy",
|
|---|
| 500 | "photo",
|
|---|
| 501 | "photography",
|
|---|
| 502 | "photos",
|
|---|
| 503 | "physio",
|
|---|
| 504 | "pics",
|
|---|
| 505 | "pictures",
|
|---|
| 506 | "pink",
|
|---|
| 507 | "pizza",
|
|---|
| 508 | "place",
|
|---|
| 509 | "plumbing",
|
|---|
| 510 | "pohl",
|
|---|
| 511 | "poker",
|
|---|
| 512 | "post",
|
|---|
| 513 | "praxi",
|
|---|
| 514 | "press",
|
|---|
| 515 | "pro",
|
|---|
| 516 | "prod",
|
|---|
| 517 | "productions",
|
|---|
| 518 | "prof",
|
|---|
| 519 | "properties",
|
|---|
| 520 | "property",
|
|---|
| 521 | "pub",
|
|---|
| 522 | "qpon",
|
|---|
| 523 | "quebec",
|
|---|
| 524 | "realtor",
|
|---|
| 525 | "recipes",
|
|---|
| 526 | "red",
|
|---|
| 527 | "rehab",
|
|---|
| 528 | "reise",
|
|---|
| 529 | "reisen",
|
|---|
| 530 | "ren",
|
|---|
| 531 | "rentals",
|
|---|
| 532 | "repair",
|
|---|
| 533 | "report",
|
|---|
| 534 | "republican",
|
|---|
| 535 | "rest",
|
|---|
| 536 | "restaurant",
|
|---|
| 537 | "reviews",
|
|---|
| 538 | "rich",
|
|---|
| 539 | "rio",
|
|---|
| 540 | "rip",
|
|---|
| 541 | "rocks",
|
|---|
| 542 | "rodeo",
|
|---|
| 543 | "rsvp",
|
|---|
| 544 | "ruhr",
|
|---|
| 545 | "ryukyu",
|
|---|
| 546 | "saarland",
|
|---|
| 547 | "sarl",
|
|---|
| 548 | "sca",
|
|---|
| 549 | "scb",
|
|---|
| 550 | "schmidt",
|
|---|
| 551 | "schule",
|
|---|
| 552 | "scot",
|
|---|
| 553 | "services",
|
|---|
| 554 | "sexy",
|
|---|
| 555 | "shiksha",
|
|---|
| 556 | "shoes",
|
|---|
| 557 | "singles",
|
|---|
| 558 | "social",
|
|---|
| 559 | "software",
|
|---|
| 560 | "sohu",
|
|---|
| 561 | "solar",
|
|---|
| 562 | "solutions",
|
|---|
| 563 | "soy",
|
|---|
| 564 | "space",
|
|---|
| 565 | "spiegel",
|
|---|
| 566 | "supplies",
|
|---|
| 567 | "supply",
|
|---|
| 568 | "support",
|
|---|
| 569 | "surf",
|
|---|
| 570 | "surgery",
|
|---|
| 571 | "suzuki",
|
|---|
| 572 | "systems",
|
|---|
| 573 | "tatar",
|
|---|
| 574 | "tattoo",
|
|---|
| 575 | "tax",
|
|---|
| 576 | "technology",
|
|---|
| 577 | "tel",
|
|---|
| 578 | "tienda",
|
|---|
| 579 | "tips",
|
|---|
| 580 | "tirol",
|
|---|
| 581 | "today",
|
|---|
| 582 | "tokyo",
|
|---|
| 583 | "tools",
|
|---|
| 584 | "top",
|
|---|
| 585 | "town",
|
|---|
| 586 | "toys",
|
|---|
| 587 | "trade",
|
|---|
| 588 | "training",
|
|---|
| 589 | "travel",
|
|---|
| 590 | "tui",
|
|---|
| 591 | "university",
|
|---|
| 592 | "uno",
|
|---|
| 593 | "uol",
|
|---|
| 594 | "vacations",
|
|---|
| 595 | "vegas",
|
|---|
| 596 | "ventures",
|
|---|
| 597 | "versicherung",
|
|---|
| 598 | "vet",
|
|---|
| 599 | "viajes",
|
|---|
| 600 | "villas",
|
|---|
| 601 | "vision",
|
|---|
| 602 | "vlaanderen",
|
|---|
| 603 | "vodka",
|
|---|
| 604 | "vote",
|
|---|
| 605 | "voting",
|
|---|
| 606 | "voto",
|
|---|
| 607 | "voyage",
|
|---|
| 608 | "wales",
|
|---|
| 609 | "wang",
|
|---|
| 610 | "watch",
|
|---|
| 611 | "webcam",
|
|---|
| 612 | "website",
|
|---|
| 613 | "wed",
|
|---|
| 614 | "wedding",
|
|---|
| 615 | "whoswho",
|
|---|
| 616 | "wien",
|
|---|
| 617 | "wiki",
|
|---|
| 618 | "williamhill",
|
|---|
| 619 | "wme",
|
|---|
| 620 | "work",
|
|---|
| 621 | "works",
|
|---|
| 622 | "world",
|
|---|
| 623 | "wtc",
|
|---|
| 624 | "wtf",
|
|---|
| 625 | "xxx",
|
|---|
| 626 | "xyz",
|
|---|
| 627 | "yachts",
|
|---|
| 628 | "yandex",
|
|---|
| 629 | "yoga",
|
|---|
| 630 | "yokohama",
|
|---|
| 631 | "youtube",
|
|---|
| 632 | "zip",
|
|---|
| 633 | "zone",
|
|---|
| 634 | };
|
|---|
| 635 |
|
|---|
| 636 | private static final String[] COUNTRY_CODE_TLDS = new String[] {
|
|---|
| 637 | "ac", // Ascension Island
|
|---|
| 638 | "ad", // Andorra
|
|---|
| 639 | "ae", // United Arab Emirates
|
|---|
| 640 | "af", // Afghanistan
|
|---|
| 641 | "ag", // Antigua and Barbuda
|
|---|
| 642 | "ai", // Anguilla
|
|---|
| 643 | "al", // Albania
|
|---|
| 644 | "am", // Armenia
|
|---|
| 645 | "an", // Netherlands Antilles
|
|---|
| 646 | "ao", // Angola
|
|---|
| 647 | "aq", // Antarctica
|
|---|
| 648 | "ar", // Argentina
|
|---|
| 649 | "as", // American Samoa
|
|---|
| 650 | "at", // Austria
|
|---|
| 651 | "au", // Australia (includes Ashmore and Cartier Islands and Coral Sea Islands)
|
|---|
| 652 | "aw", // Aruba
|
|---|
| 653 | "ax", // Åland
|
|---|
| 654 | "az", // Azerbaijan
|
|---|
| 655 | "ba", // Bosnia and Herzegovina
|
|---|
| 656 | "bb", // Barbados
|
|---|
| 657 | "bd", // Bangladesh
|
|---|
| 658 | "be", // Belgium
|
|---|
| 659 | "bf", // Burkina Faso
|
|---|
| 660 | "bg", // Bulgaria
|
|---|
| 661 | "bh", // Bahrain
|
|---|
| 662 | "bi", // Burundi
|
|---|
| 663 | "bj", // Benin
|
|---|
| 664 | "bm", // Bermuda
|
|---|
| 665 | "bn", // Brunei Darussalam
|
|---|
| 666 | "bo", // Bolivia
|
|---|
| 667 | "br", // Brazil
|
|---|
| 668 | "bs", // Bahamas
|
|---|
| 669 | "bt", // Bhutan
|
|---|
| 670 | "bv", // Bouvet Island
|
|---|
| 671 | "bw", // Botswana
|
|---|
| 672 | "by", // Belarus
|
|---|
| 673 | "bz", // Belize
|
|---|
| 674 | "ca", // Canada
|
|---|
| 675 | "cc", // Cocos (Keeling) Islands
|
|---|
| 676 | "cd", // Democratic Republic of the Congo (formerly Zaire)
|
|---|
| 677 | "cf", // Central African Republic
|
|---|
| 678 | "cg", // Republic of the Congo
|
|---|
| 679 | "ch", // Switzerland
|
|---|
| 680 | "ci", // Côte d'Ivoire
|
|---|
| 681 | "ck", // Cook Islands
|
|---|
| 682 | "cl", // Chile
|
|---|
| 683 | "cm", // Cameroon
|
|---|
| 684 | "cn", // China, mainland
|
|---|
| 685 | "co", // Colombia
|
|---|
| 686 | "cr", // Costa Rica
|
|---|
| 687 | "cu", // Cuba
|
|---|
| 688 | "cv", // Cape Verde
|
|---|
| 689 | "cw", // Curaçao
|
|---|
| 690 | "cx", // Christmas Island
|
|---|
| 691 | "cy", // Cyprus
|
|---|
| 692 | "cz", // Czech Republic
|
|---|
| 693 | "de", // Germany
|
|---|
| 694 | "dj", // Djibouti
|
|---|
| 695 | "dk", // Denmark
|
|---|
| 696 | "dm", // Dominica
|
|---|
| 697 | "do", // Dominican Republic
|
|---|
| 698 | "dz", // Algeria
|
|---|
| 699 | "ec", // Ecuador
|
|---|
| 700 | "ee", // Estonia
|
|---|
| 701 | "eg", // Egypt
|
|---|
| 702 | "er", // Eritrea
|
|---|
| 703 | "es", // Spain
|
|---|
| 704 | "et", // Ethiopia
|
|---|
| 705 | "eu", // European Union
|
|---|
| 706 | "fi", // Finland
|
|---|
| 707 | "fj", // Fiji
|
|---|
| 708 | "fk", // Falkland Islands
|
|---|
| 709 | "fm", // Federated States of Micronesia
|
|---|
| 710 | "fo", // Faroe Islands
|
|---|
| 711 | "fr", // France
|
|---|
| 712 | "ga", // Gabon
|
|---|
| 713 | "gb", // Great Britain (United Kingdom)
|
|---|
| 714 | "gd", // Grenada
|
|---|
| 715 | "ge", // Georgia
|
|---|
| 716 | "gf", // French Guiana
|
|---|
| 717 | "gg", // Guernsey
|
|---|
| 718 | "gh", // Ghana
|
|---|
| 719 | "gi", // Gibraltar
|
|---|
| 720 | "gl", // Greenland
|
|---|
| 721 | "gm", // The Gambia
|
|---|
| 722 | "gn", // Guinea
|
|---|
| 723 | "gp", // Guadeloupe
|
|---|
| 724 | "gq", // Equatorial Guinea
|
|---|
| 725 | "gr", // Greece
|
|---|
| 726 | "gs", // South Georgia and the South Sandwich Islands
|
|---|
| 727 | "gt", // Guatemala
|
|---|
| 728 | "gu", // Guam
|
|---|
| 729 | "gw", // Guinea-Bissau
|
|---|
| 730 | "gy", // Guyana
|
|---|
| 731 | "hk", // Hong Kong
|
|---|
| 732 | "hm", // Heard Island and McDonald Islands
|
|---|
| 733 | "hn", // Honduras
|
|---|
| 734 | "hr", // Croatia (Hrvatska)
|
|---|
| 735 | "ht", // Haiti
|
|---|
| 736 | "hu", // Hungary
|
|---|
| 737 | "id", // Indonesia
|
|---|
| 738 | "ie", // Ireland (Éire)
|
|---|
| 739 | "il", // Israel
|
|---|
| 740 | "im", // Isle of Man
|
|---|
| 741 | "in", // India
|
|---|
| 742 | "io", // British Indian Ocean Territory
|
|---|
| 743 | "iq", // Iraq
|
|---|
| 744 | "ir", // Iran
|
|---|
| 745 | "is", // Iceland
|
|---|
| 746 | "it", // Italy
|
|---|
| 747 | "je", // Jersey
|
|---|
| 748 | "jm", // Jamaica
|
|---|
| 749 | "jo", // Jordan
|
|---|
| 750 | "jp", // Japan
|
|---|
| 751 | "ke", // Kenya
|
|---|
| 752 | "kg", // Kyrgyzstan
|
|---|
| 753 | "kh", // Cambodia (Khmer)
|
|---|
| 754 | "ki", // Kiribati
|
|---|
| 755 | "km", // Comoros
|
|---|
| 756 | "kn", // Saint Kitts and Nevis
|
|---|
| 757 | "kp", // North Korea
|
|---|
| 758 | "kr", // South Korea
|
|---|
| 759 | "kw", // Kuwait
|
|---|
| 760 | "ky", // Cayman Islands
|
|---|
| 761 | "kz", // Kazakhstan
|
|---|
| 762 | "la", // Laos (currently being marketed as the official domain for Los Angeles)
|
|---|
| 763 | "lb", // Lebanon
|
|---|
| 764 | "lc", // Saint Lucia
|
|---|
| 765 | "li", // Liechtenstein
|
|---|
| 766 | "lk", // Sri Lanka
|
|---|
| 767 | "lr", // Liberia
|
|---|
| 768 | "ls", // Lesotho
|
|---|
| 769 | "lt", // Lithuania
|
|---|
| 770 | "lu", // Luxembourg
|
|---|
| 771 | "lv", // Latvia
|
|---|
| 772 | "ly", // Libya
|
|---|
| 773 | "ma", // Morocco
|
|---|
| 774 | "mc", // Monaco
|
|---|
| 775 | "md", // Moldova
|
|---|
| 776 | "me", // Montenegro
|
|---|
| 777 | "mg", // Madagascar
|
|---|
| 778 | "mh", // Marshall Islands
|
|---|
| 779 | "mk", // Republic of Macedonia
|
|---|
| 780 | "ml", // Mali
|
|---|
| 781 | "mm", // Myanmar
|
|---|
| 782 | "mn", // Mongolia
|
|---|
| 783 | "mo", // Macau
|
|---|
| 784 | "mp", // Northern Mariana Islands
|
|---|
| 785 | "mq", // Martinique
|
|---|
| 786 | "mr", // Mauritania
|
|---|
| 787 | "ms", // Montserrat
|
|---|
| 788 | "mt", // Malta
|
|---|
| 789 | "mu", // Mauritius
|
|---|
| 790 | "mv", // Maldives
|
|---|
| 791 | "mw", // Malawi
|
|---|
| 792 | "mx", // Mexico
|
|---|
| 793 | "my", // Malaysia
|
|---|
| 794 | "mz", // Mozambique
|
|---|
| 795 | "na", // Namibia
|
|---|
| 796 | "nc", // New Caledonia
|
|---|
| 797 | "ne", // Niger
|
|---|
| 798 | "nf", // Norfolk Island
|
|---|
| 799 | "ng", // Nigeria
|
|---|
| 800 | "ni", // Nicaragua
|
|---|
| 801 | "nl", // Netherlands
|
|---|
| 802 | "no", // Norway
|
|---|
| 803 | "np", // Nepal
|
|---|
| 804 | "nr", // Nauru
|
|---|
| 805 | "nu", // Niue
|
|---|
| 806 | "nz", // New Zealand
|
|---|
| 807 | "om", // Oman
|
|---|
| 808 | "pa", // Panama
|
|---|
| 809 | "pe", // Peru
|
|---|
| 810 | "pf", // French Polynesia With Clipperton Island
|
|---|
| 811 | "pg", // Papua New Guinea
|
|---|
| 812 | "ph", // Philippines
|
|---|
| 813 | "pk", // Pakistan
|
|---|
| 814 | "pl", // Poland
|
|---|
| 815 | "pm", // Saint-Pierre and Miquelon
|
|---|
| 816 | "pn", // Pitcairn Islands
|
|---|
| 817 | "pr", // Puerto Rico
|
|---|
| 818 | "ps", // Palestinian territories (PA-controlled West Bank and Gaza Strip)
|
|---|
| 819 | "pt", // Portugal
|
|---|
| 820 | "pw", // Palau
|
|---|
| 821 | "py", // Paraguay
|
|---|
| 822 | "qa", // Qatar
|
|---|
| 823 | "re", // Réunion
|
|---|
| 824 | "ro", // Romania
|
|---|
| 825 | "rs", // Serbia
|
|---|
| 826 | "ru", // Russia
|
|---|
| 827 | "rw", // Rwanda
|
|---|
| 828 | "sa", // Saudi Arabia
|
|---|
| 829 | "sb", // Solomon Islands
|
|---|
| 830 | "sc", // Seychelles
|
|---|
| 831 | "sd", // Sudan
|
|---|
| 832 | "se", // Sweden
|
|---|
| 833 | "sg", // Singapore
|
|---|
| 834 | "sh", // Saint Helena
|
|---|
| 835 | "si", // Slovenia
|
|---|
| 836 | "sj", // Svalbard and Jan Mayen Islands Not in use (Norwegian dependencies; see .no)
|
|---|
| 837 | "sk", // Slovakia
|
|---|
| 838 | "sl", // Sierra Leone
|
|---|
| 839 | "sm", // San Marino
|
|---|
| 840 | "sn", // Senegal
|
|---|
| 841 | "so", // Somalia
|
|---|
| 842 | "sr", // Suriname
|
|---|
| 843 | "st", // São Tomé and Príncipe
|
|---|
| 844 | "su", // Soviet Union (deprecated)
|
|---|
| 845 | "sv", // El Salvador
|
|---|
| 846 | "sx", // Sint Maarten
|
|---|
| 847 | "sy", // Syria
|
|---|
| 848 | "sz", // Swaziland
|
|---|
| 849 | "tc", // Turks and Caicos Islands
|
|---|
| 850 | "td", // Chad
|
|---|
| 851 | "tf", // French Southern and Antarctic Lands
|
|---|
| 852 | "tg", // Togo
|
|---|
| 853 | "th", // Thailand
|
|---|
| 854 | "tj", // Tajikistan
|
|---|
| 855 | "tk", // Tokelau
|
|---|
| 856 | "tl", // East Timor (deprecated old code)
|
|---|
| 857 | "tm", // Turkmenistan
|
|---|
| 858 | "tn", // Tunisia
|
|---|
| 859 | "to", // Tonga
|
|---|
| 860 | "tp", // East Timor
|
|---|
| 861 | "tr", // Turkey
|
|---|
| 862 | "tt", // Trinidad and Tobago
|
|---|
| 863 | "tv", // Tuvalu
|
|---|
| 864 | "tw", // Taiwan, Republic of China
|
|---|
| 865 | "tz", // Tanzania
|
|---|
| 866 | "ua", // Ukraine
|
|---|
| 867 | "ug", // Uganda
|
|---|
| 868 | "uk", // United Kingdom
|
|---|
| 869 | "um", // United States Minor Outlying Islands
|
|---|
| 870 | "us", // United States of America
|
|---|
| 871 | "uy", // Uruguay
|
|---|
| 872 | "uz", // Uzbekistan
|
|---|
| 873 | "va", // Vatican City State
|
|---|
| 874 | "vc", // Saint Vincent and the Grenadines
|
|---|
| 875 | "ve", // Venezuela
|
|---|
| 876 | "vg", // British Virgin Islands
|
|---|
| 877 | "vi", // U.S. Virgin Islands
|
|---|
| 878 | "vn", // Vietnam
|
|---|
| 879 | "vu", // Vanuatu
|
|---|
| 880 | "wf", // Wallis and Futuna
|
|---|
| 881 | "ws", // Samoa (formerly Western Samoa)
|
|---|
| 882 | "ye", // Yemen
|
|---|
| 883 | "yt", // Mayotte
|
|---|
| 884 | "yu", // Serbia and Montenegro (originally Yugoslavia)
|
|---|
| 885 | "za", // South Africa
|
|---|
| 886 | "zm", // Zambia
|
|---|
| 887 | "zw", // Zimbabwe
|
|---|
| 888 | };
|
|---|
| 889 |
|
|---|
| 890 | private static final String[] LOCAL_TLDS = new String[] {
|
|---|
| 891 | "localhost", // RFC2606 defined
|
|---|
| 892 | "localdomain" // Also widely used as localhost.localdomain
|
|---|
| 893 | };
|
|---|
| 894 |
|
|---|
| 895 | static {
|
|---|
| 896 | Arrays.sort(INFRASTRUCTURE_TLDS);
|
|---|
| 897 | Arrays.sort(COUNTRY_CODE_TLDS);
|
|---|
| 898 | Arrays.sort(GENERIC_TLDS);
|
|---|
| 899 | Arrays.sort(LOCAL_TLDS);
|
|---|
| 900 | }
|
|---|
| 901 | }
|
|---|
