Home | History | Annotate | Download | only in phonenumbers
      1 /*
      2  * Copyright (C) 2009 The Libphonenumber Authors
      3  *
      4  * Licensed under the Apache License, Version 2.0 (the "License");
      5  * you may not use this file except in compliance with the License.
      6  * You may obtain a copy of the License at
      7  *
      8  * http://www.apache.org/licenses/LICENSE-2.0
      9  *
     10  * Unless required by applicable law or agreed to in writing, software
     11  * distributed under the License is distributed on an "AS IS" BASIS,
     12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
     13  * See the License for the specific language governing permissions and
     14  * limitations under the License.
     15  */
     16 
     17 package com.google.i18n.phonenumbers;
     18 
     19 import com.google.i18n.phonenumbers.Phonemetadata.NumberFormat;
     20 import com.google.i18n.phonenumbers.Phonemetadata.PhoneMetadata;
     21 import com.google.i18n.phonenumbers.Phonemetadata.PhoneNumberDesc;
     22 import com.google.i18n.phonenumbers.Phonenumber.PhoneNumber;
     23 import com.google.i18n.phonenumbers.Phonenumber.PhoneNumber.CountryCodeSource;
     24 
     25 import java.io.InputStream;
     26 import java.util.ArrayList;
     27 import java.util.Arrays;
     28 import java.util.Collections;
     29 import java.util.HashMap;
     30 import java.util.HashSet;
     31 import java.util.Iterator;
     32 import java.util.List;
     33 import java.util.Map;
     34 import java.util.Set;
     35 import java.util.logging.Level;
     36 import java.util.logging.Logger;
     37 import java.util.regex.Matcher;
     38 import java.util.regex.Pattern;
     39 
     40 /**
     41  * Utility for international phone numbers. Functionality includes formatting, parsing and
     42  * validation.
     43  *
     44  * <p>If you use this library, and want to be notified about important changes, please sign up to
     45  * our <a href="http://groups.google.com/group/libphonenumber-discuss/about">mailing list</a>.
     46  *
     47  * NOTE: A lot of methods in this class require Region Code strings. These must be provided using
     48  * ISO 3166-1 two-letter country-code format. These should be in upper-case. The list of the codes
     49  * can be found here:
     50  * http://www.iso.org/iso/country_codes/iso_3166_code_lists/country_names_and_code_elements.htm
     51  *
     52  * @author Shaopeng Jia
     53  */
     54 public class PhoneNumberUtil {
     55   // @VisibleForTesting
     56   static final MetadataLoader DEFAULT_METADATA_LOADER = new MetadataLoader() {
     57     @Override
     58     public InputStream loadMetadata(String metadataFileName) {
     59       return PhoneNumberUtil.class.getResourceAsStream(metadataFileName);
     60     }
     61   };
     62 
     63   private static final Logger logger = Logger.getLogger(PhoneNumberUtil.class.getName());
     64 
     65   /** Flags to use when compiling regular expressions for phone numbers. */
     66   static final int REGEX_FLAGS = Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE;
     67   // The minimum and maximum length of the national significant number.
     68   private static final int MIN_LENGTH_FOR_NSN = 2;
     69   // The ITU says the maximum length should be 15, but we have found longer numbers in Germany.
     70   static final int MAX_LENGTH_FOR_NSN = 17;
     71   // The maximum length of the country calling code.
     72   static final int MAX_LENGTH_COUNTRY_CODE = 3;
     73   // We don't allow input strings for parsing to be longer than 250 chars. This prevents malicious
     74   // input from overflowing the regular-expression engine.
     75   private static final int MAX_INPUT_STRING_LENGTH = 250;
     76 
     77   // Region-code for the unknown region.
     78   private static final String UNKNOWN_REGION = "ZZ";
     79 
     80   private static final int NANPA_COUNTRY_CODE = 1;
     81 
     82   // The prefix that needs to be inserted in front of a Colombian landline number when dialed from
     83   // a mobile phone in Colombia.
     84   private static final String COLOMBIA_MOBILE_TO_FIXED_LINE_PREFIX = "3";
     85 
     86   // Map of country calling codes that use a mobile token before the area code. One example of when
     87   // this is relevant is when determining the length of the national destination code, which should
     88   // be the length of the area code plus the length of the mobile token.
     89   private static final Map<Integer, String> MOBILE_TOKEN_MAPPINGS;
     90 
     91   // The PLUS_SIGN signifies the international prefix.
     92   static final char PLUS_SIGN = '+';
     93 
     94   private static final char STAR_SIGN = '*';
     95 
     96   private static final String RFC3966_EXTN_PREFIX = ";ext=";
     97   private static final String RFC3966_PREFIX = "tel:";
     98   private static final String RFC3966_PHONE_CONTEXT = ";phone-context=";
     99   private static final String RFC3966_ISDN_SUBADDRESS = ";isub=";
    100 
    101   // A map that contains characters that are essential when dialling. That means any of the
    102   // characters in this map must not be removed from a number when dialling, otherwise the call
    103   // will not reach the intended destination.
    104   private static final Map<Character, Character> DIALLABLE_CHAR_MAPPINGS;
    105 
    106   // Only upper-case variants of alpha characters are stored.
    107   private static final Map<Character, Character> ALPHA_MAPPINGS;
    108 
    109   // For performance reasons, amalgamate both into one map.
    110   private static final Map<Character, Character> ALPHA_PHONE_MAPPINGS;
    111 
    112   // Separate map of all symbols that we wish to retain when formatting alpha numbers. This
    113   // includes digits, ASCII letters and number grouping symbols such as "-" and " ".
    114   private static final Map<Character, Character> ALL_PLUS_NUMBER_GROUPING_SYMBOLS;
    115 
    116   static {
    117     HashMap<Integer, String> mobileTokenMap = new HashMap<Integer, String>();
    118     mobileTokenMap.put(52, "1");
    119     mobileTokenMap.put(54, "9");
    120     MOBILE_TOKEN_MAPPINGS = Collections.unmodifiableMap(mobileTokenMap);
    121 
    122     // Simple ASCII digits map used to populate ALPHA_PHONE_MAPPINGS and
    123     // ALL_PLUS_NUMBER_GROUPING_SYMBOLS.
    124     HashMap<Character, Character> asciiDigitMappings = new HashMap<Character, Character>();
    125     asciiDigitMappings.put('0', '0');
    126     asciiDigitMappings.put('1', '1');
    127     asciiDigitMappings.put('2', '2');
    128     asciiDigitMappings.put('3', '3');
    129     asciiDigitMappings.put('4', '4');
    130     asciiDigitMappings.put('5', '5');
    131     asciiDigitMappings.put('6', '6');
    132     asciiDigitMappings.put('7', '7');
    133     asciiDigitMappings.put('8', '8');
    134     asciiDigitMappings.put('9', '9');
    135 
    136     HashMap<Character, Character> alphaMap = new HashMap<Character, Character>(40);
    137     alphaMap.put('A', '2');
    138     alphaMap.put('B', '2');
    139     alphaMap.put('C', '2');
    140     alphaMap.put('D', '3');
    141     alphaMap.put('E', '3');
    142     alphaMap.put('F', '3');
    143     alphaMap.put('G', '4');
    144     alphaMap.put('H', '4');
    145     alphaMap.put('I', '4');
    146     alphaMap.put('J', '5');
    147     alphaMap.put('K', '5');
    148     alphaMap.put('L', '5');
    149     alphaMap.put('M', '6');
    150     alphaMap.put('N', '6');
    151     alphaMap.put('O', '6');
    152     alphaMap.put('P', '7');
    153     alphaMap.put('Q', '7');
    154     alphaMap.put('R', '7');
    155     alphaMap.put('S', '7');
    156     alphaMap.put('T', '8');
    157     alphaMap.put('U', '8');
    158     alphaMap.put('V', '8');
    159     alphaMap.put('W', '9');
    160     alphaMap.put('X', '9');
    161     alphaMap.put('Y', '9');
    162     alphaMap.put('Z', '9');
    163     ALPHA_MAPPINGS = Collections.unmodifiableMap(alphaMap);
    164 
    165     HashMap<Character, Character> combinedMap = new HashMap<Character, Character>(100);
    166     combinedMap.putAll(ALPHA_MAPPINGS);
    167     combinedMap.putAll(asciiDigitMappings);
    168     ALPHA_PHONE_MAPPINGS = Collections.unmodifiableMap(combinedMap);
    169 
    170     HashMap<Character, Character> diallableCharMap = new HashMap<Character, Character>();
    171     diallableCharMap.putAll(asciiDigitMappings);
    172     diallableCharMap.put(PLUS_SIGN, PLUS_SIGN);
    173     diallableCharMap.put('*', '*');
    174     DIALLABLE_CHAR_MAPPINGS = Collections.unmodifiableMap(diallableCharMap);
    175 
    176     HashMap<Character, Character> allPlusNumberGroupings = new HashMap<Character, Character>();
    177     // Put (lower letter -> upper letter) and (upper letter -> upper letter) mappings.
    178     for (char c : ALPHA_MAPPINGS.keySet()) {
    179       allPlusNumberGroupings.put(Character.toLowerCase(c), c);
    180       allPlusNumberGroupings.put(c, c);
    181     }
    182     allPlusNumberGroupings.putAll(asciiDigitMappings);
    183     // Put grouping symbols.
    184     allPlusNumberGroupings.put('-', '-');
    185     allPlusNumberGroupings.put('\uFF0D', '-');
    186     allPlusNumberGroupings.put('\u2010', '-');
    187     allPlusNumberGroupings.put('\u2011', '-');
    188     allPlusNumberGroupings.put('\u2012', '-');
    189     allPlusNumberGroupings.put('\u2013', '-');
    190     allPlusNumberGroupings.put('\u2014', '-');
    191     allPlusNumberGroupings.put('\u2015', '-');
    192     allPlusNumberGroupings.put('\u2212', '-');
    193     allPlusNumberGroupings.put('/', '/');
    194     allPlusNumberGroupings.put('\uFF0F', '/');
    195     allPlusNumberGroupings.put(' ', ' ');
    196     allPlusNumberGroupings.put('\u3000', ' ');
    197     allPlusNumberGroupings.put('\u2060', ' ');
    198     allPlusNumberGroupings.put('.', '.');
    199     allPlusNumberGroupings.put('\uFF0E', '.');
    200     ALL_PLUS_NUMBER_GROUPING_SYMBOLS = Collections.unmodifiableMap(allPlusNumberGroupings);
    201   }
    202 
    203   // Pattern that makes it easy to distinguish whether a region has a unique international dialing
    204   // prefix or not. If a region has a unique international prefix (e.g. 011 in USA), it will be
    205   // represented as a string that contains a sequence of ASCII digits. If there are multiple
    206   // available international prefixes in a region, they will be represented as a regex string that
    207   // always contains character(s) other than ASCII digits.
    208   // Note this regex also includes tilde, which signals waiting for the tone.
    209   private static final Pattern UNIQUE_INTERNATIONAL_PREFIX =
    210       Pattern.compile("[\\d]+(?:[~\u2053\u223C\uFF5E][\\d]+)?");
    211 
    212   // Regular expression of acceptable punctuation found in phone numbers. This excludes punctuation
    213   // found as a leading character only.
    214   // This consists of dash characters, white space characters, full stops, slashes,
    215   // square brackets, parentheses and tildes. It also includes the letter 'x' as that is found as a
    216   // placeholder for carrier information in some phone numbers. Full-width variants are also
    217   // present.
    218   static final String VALID_PUNCTUATION = "-x\u2010-\u2015\u2212\u30FC\uFF0D-\uFF0F " +
    219       "\u00A0\u00AD\u200B\u2060\u3000()\uFF08\uFF09\uFF3B\uFF3D.\\[\\]/~\u2053\u223C\uFF5E";
    220 
    221   private static final String DIGITS = "\\p{Nd}";
    222   // We accept alpha characters in phone numbers, ASCII only, upper and lower case.
    223   private static final String VALID_ALPHA =
    224       Arrays.toString(ALPHA_MAPPINGS.keySet().toArray()).replaceAll("[, \\[\\]]", "") +
    225       Arrays.toString(ALPHA_MAPPINGS.keySet().toArray()).toLowerCase().replaceAll("[, \\[\\]]", "");
    226   static final String PLUS_CHARS = "+\uFF0B";
    227   static final Pattern PLUS_CHARS_PATTERN = Pattern.compile("[" + PLUS_CHARS + "]+");
    228   private static final Pattern SEPARATOR_PATTERN = Pattern.compile("[" + VALID_PUNCTUATION + "]+");
    229   private static final Pattern CAPTURING_DIGIT_PATTERN = Pattern.compile("(" + DIGITS + ")");
    230 
    231   // Regular expression of acceptable characters that may start a phone number for the purposes of
    232   // parsing. This allows us to strip away meaningless prefixes to phone numbers that may be
    233   // mistakenly given to us. This consists of digits, the plus symbol and arabic-indic digits. This
    234   // does not contain alpha characters, although they may be used later in the number. It also does
    235   // not include other punctuation, as this will be stripped later during parsing and is of no
    236   // information value when parsing a number.
    237   private static final String VALID_START_CHAR = "[" + PLUS_CHARS + DIGITS + "]";
    238   private static final Pattern VALID_START_CHAR_PATTERN = Pattern.compile(VALID_START_CHAR);
    239 
    240   // Regular expression of characters typically used to start a second phone number for the purposes
    241   // of parsing. This allows us to strip off parts of the number that are actually the start of
    242   // another number, such as for: (530) 583-6985 x302/x2303 -> the second extension here makes this
    243   // actually two phone numbers, (530) 583-6985 x302 and (530) 583-6985 x2303. We remove the second
    244   // extension so that the first number is parsed correctly.
    245   private static final String SECOND_NUMBER_START = "[\\\\/] *x";
    246   static final Pattern SECOND_NUMBER_START_PATTERN = Pattern.compile(SECOND_NUMBER_START);
    247 
    248   // Regular expression of trailing characters that we want to remove. We remove all characters that
    249   // are not alpha or numerical characters. The hash character is retained here, as it may signify
    250   // the previous block was an extension.
    251   private static final String UNWANTED_END_CHARS = "[[\\P{N}&&\\P{L}]&&[^#]]+$";
    252   static final Pattern UNWANTED_END_CHAR_PATTERN = Pattern.compile(UNWANTED_END_CHARS);
    253 
    254   // We use this pattern to check if the phone number has at least three letters in it - if so, then
    255   // we treat it as a number where some phone-number digits are represented by letters.
    256   private static final Pattern VALID_ALPHA_PHONE_PATTERN = Pattern.compile("(?:.*?[A-Za-z]){3}.*");
    257 
    258   // Regular expression of viable phone numbers. This is location independent. Checks we have at
    259   // least three leading digits, and only valid punctuation, alpha characters and
    260   // digits in the phone number. Does not include extension data.
    261   // The symbol 'x' is allowed here as valid punctuation since it is often used as a placeholder for
    262   // carrier codes, for example in Brazilian phone numbers. We also allow multiple "+" characters at
    263   // the start.
    264   // Corresponds to the following:
    265   // [digits]{minLengthNsn}|
    266   // plus_sign*(([punctuation]|[star])*[digits]){3,}([punctuation]|[star]|[digits]|[alpha])*
    267   //
    268   // The first reg-ex is to allow short numbers (two digits long) to be parsed if they are entered
    269   // as "15" etc, but only if there is no punctuation in them. The second expression restricts the
    270   // number of digits to three or more, but then allows them to be in international form, and to
    271   // have alpha-characters and punctuation.
    272   //
    273   // Note VALID_PUNCTUATION starts with a -, so must be the first in the range.
    274   private static final String VALID_PHONE_NUMBER =
    275       DIGITS + "{" + MIN_LENGTH_FOR_NSN + "}" + "|" +
    276       "[" + PLUS_CHARS + "]*+(?:[" + VALID_PUNCTUATION + STAR_SIGN + "]*" + DIGITS + "){3,}[" +
    277       VALID_PUNCTUATION + STAR_SIGN + VALID_ALPHA + DIGITS + "]*";
    278 
    279   // Default extension prefix to use when formatting. This will be put in front of any extension
    280   // component of the number, after the main national number is formatted. For example, if you wish
    281   // the default extension formatting to be " extn: 3456", then you should specify " extn: " here
    282   // as the default extension prefix. This can be overridden by region-specific preferences.
    283   private static final String DEFAULT_EXTN_PREFIX = " ext. ";
    284 
    285   // Pattern to capture digits used in an extension. Places a maximum length of "7" for an
    286   // extension.
    287   private static final String CAPTURING_EXTN_DIGITS = "(" + DIGITS + "{1,7})";
    288   // Regexp of all possible ways to write extensions, for use when parsing. This will be run as a
    289   // case-insensitive regexp match. Wide character versions are also provided after each ASCII
    290   // version.
    291   private static final String EXTN_PATTERNS_FOR_PARSING;
    292   static final String EXTN_PATTERNS_FOR_MATCHING;
    293   static {
    294     // One-character symbols that can be used to indicate an extension.
    295     String singleExtnSymbolsForMatching = "x\uFF58#\uFF03~\uFF5E";
    296     // For parsing, we are slightly more lenient in our interpretation than for matching. Here we
    297     // allow a "comma" as a possible extension indicator. When matching, this is hardly ever used to
    298     // indicate this.
    299     String singleExtnSymbolsForParsing = "," + singleExtnSymbolsForMatching;
    300 
    301     EXTN_PATTERNS_FOR_PARSING = createExtnPattern(singleExtnSymbolsForParsing);
    302     EXTN_PATTERNS_FOR_MATCHING = createExtnPattern(singleExtnSymbolsForMatching);
    303   }
    304 
    305   /**
    306    * Helper initialiser method to create the regular-expression pattern to match extensions,
    307    * allowing the one-char extension symbols provided by {@code singleExtnSymbols}.
    308    */
    309   private static String createExtnPattern(String singleExtnSymbols) {
    310     // There are three regular expressions here. The first covers RFC 3966 format, where the
    311     // extension is added using ";ext=". The second more generic one starts with optional white
    312     // space and ends with an optional full stop (.), followed by zero or more spaces/tabs and then
    313     // the numbers themselves. The other one covers the special case of American numbers where the
    314     // extension is written with a hash at the end, such as "- 503#".
    315     // Note that the only capturing groups should be around the digits that you want to capture as
    316     // part of the extension, or else parsing will fail!
    317     // Canonical-equivalence doesn't seem to be an option with Android java, so we allow two options
    318     // for representing the accented o - the character itself, and one in the unicode decomposed
    319     // form with the combining acute accent.
    320     return (RFC3966_EXTN_PREFIX + CAPTURING_EXTN_DIGITS + "|" + "[ \u00A0\\t,]*" +
    321             "(?:e?xt(?:ensi(?:o\u0301?|\u00F3))?n?|\uFF45?\uFF58\uFF54\uFF4E?|" +
    322             "[" + singleExtnSymbols + "]|int|anexo|\uFF49\uFF4E\uFF54)" +
    323             "[:\\.\uFF0E]?[ \u00A0\\t,-]*" + CAPTURING_EXTN_DIGITS + "#?|" +
    324             "[- ]+(" + DIGITS + "{1,5})#");
    325   }
    326 
    327   // Regexp of all known extension prefixes used by different regions followed by 1 or more valid
    328   // digits, for use when parsing.
    329   private static final Pattern EXTN_PATTERN =
    330       Pattern.compile("(?:" + EXTN_PATTERNS_FOR_PARSING + ")$", REGEX_FLAGS);
    331 
    332   // We append optionally the extension pattern to the end here, as a valid phone number may
    333   // have an extension prefix appended, followed by 1 or more digits.
    334   private static final Pattern VALID_PHONE_NUMBER_PATTERN =
    335       Pattern.compile(VALID_PHONE_NUMBER + "(?:" + EXTN_PATTERNS_FOR_PARSING + ")?", REGEX_FLAGS);
    336 
    337   static final Pattern NON_DIGITS_PATTERN = Pattern.compile("(\\D+)");
    338 
    339   // The FIRST_GROUP_PATTERN was originally set to $1 but there are some countries for which the
    340   // first group is not used in the national pattern (e.g. Argentina) so the $1 group does not match
    341   // correctly.  Therefore, we use \d, so that the first group actually used in the pattern will be
    342   // matched.
    343   private static final Pattern FIRST_GROUP_PATTERN = Pattern.compile("(\\$\\d)");
    344   private static final Pattern NP_PATTERN = Pattern.compile("\\$NP");
    345   private static final Pattern FG_PATTERN = Pattern.compile("\\$FG");
    346   private static final Pattern CC_PATTERN = Pattern.compile("\\$CC");
    347 
    348   // A pattern that is used to determine if the national prefix formatting rule has the first group
    349   // only, i.e., does not start with the national prefix. Note that the pattern explicitly allows
    350   // for unbalanced parentheses.
    351   private static final Pattern FIRST_GROUP_ONLY_PREFIX_PATTERN = Pattern.compile("\\(?\\$1\\)?");
    352 
    353   private static PhoneNumberUtil instance = null;
    354 
    355   public static final String REGION_CODE_FOR_NON_GEO_ENTITY = "001";
    356 
    357   /**
    358    * INTERNATIONAL and NATIONAL formats are consistent with the definition in ITU-T Recommendation
    359    * E123. For example, the number of the Google Switzerland office will be written as
    360    * "+41 44 668 1800" in INTERNATIONAL format, and as "044 668 1800" in NATIONAL format.
    361    * E164 format is as per INTERNATIONAL format but with no formatting applied, e.g.
    362    * "+41446681800". RFC3966 is as per INTERNATIONAL format, but with all spaces and other
    363    * separating symbols replaced with a hyphen, and with any phone number extension appended with
    364    * ";ext=". It also will have a prefix of "tel:" added, e.g. "tel:+41-44-668-1800".
    365    *
    366    * Note: If you are considering storing the number in a neutral format, you are highly advised to
    367    * use the PhoneNumber class.
    368    */
    369   public enum PhoneNumberFormat {
    370     E164,
    371     INTERNATIONAL,
    372     NATIONAL,
    373     RFC3966
    374   }
    375 
    376   /**
    377    * Type of phone numbers.
    378    */
    379   public enum PhoneNumberType {
    380     FIXED_LINE,
    381     MOBILE,
    382     // In some regions (e.g. the USA), it is impossible to distinguish between fixed-line and
    383     // mobile numbers by looking at the phone number itself.
    384     FIXED_LINE_OR_MOBILE,
    385     // Freephone lines
    386     TOLL_FREE,
    387     PREMIUM_RATE,
    388     // The cost of this call is shared between the caller and the recipient, and is hence typically
    389     // less than PREMIUM_RATE calls. See // http://en.wikipedia.org/wiki/Shared_Cost_Service for
    390     // more information.
    391     SHARED_COST,
    392     // Voice over IP numbers. This includes TSoIP (Telephony Service over IP).
    393     VOIP,
    394     // A personal number is associated with a particular person, and may be routed to either a
    395     // MOBILE or FIXED_LINE number. Some more information can be found here:
    396     // http://en.wikipedia.org/wiki/Personal_Numbers
    397     PERSONAL_NUMBER,
    398     PAGER,
    399     // Used for "Universal Access Numbers" or "Company Numbers". They may be further routed to
    400     // specific offices, but allow one number to be used for a company.
    401     UAN,
    402     // Used for "Voice Mail Access Numbers".
    403     VOICEMAIL,
    404     // A phone number is of type UNKNOWN when it does not fit any of the known patterns for a
    405     // specific region.
    406     UNKNOWN
    407   }
    408 
    409   /**
    410    * Types of phone number matches. See detailed description beside the isNumberMatch() method.
    411    */
    412   public enum MatchType {
    413     NOT_A_NUMBER,
    414     NO_MATCH,
    415     SHORT_NSN_MATCH,
    416     NSN_MATCH,
    417     EXACT_MATCH,
    418   }
    419 
    420   /**
    421    * Possible outcomes when testing if a PhoneNumber is possible.
    422    */
    423   public enum ValidationResult {
    424     IS_POSSIBLE,
    425     INVALID_COUNTRY_CODE,
    426     TOO_SHORT,
    427     TOO_LONG,
    428   }
    429 
    430   /**
    431    * Leniency when {@linkplain PhoneNumberUtil#findNumbers finding} potential phone numbers in text
    432    * segments. The levels here are ordered in increasing strictness.
    433    */
    434   public enum Leniency {
    435     /**
    436      * Phone numbers accepted are {@linkplain PhoneNumberUtil#isPossibleNumber(PhoneNumber)
    437      * possible}, but not necessarily {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid}.
    438      */
    439     POSSIBLE {
    440       @Override
    441       boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
    442         return util.isPossibleNumber(number);
    443       }
    444     },
    445     /**
    446      * Phone numbers accepted are {@linkplain PhoneNumberUtil#isPossibleNumber(PhoneNumber)
    447      * possible} and {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid}. Numbers written
    448      * in national format must have their national-prefix present if it is usually written for a
    449      * number of this type.
    450      */
    451     VALID {
    452       @Override
    453       boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
    454         if (!util.isValidNumber(number) ||
    455             !PhoneNumberMatcher.containsOnlyValidXChars(number, candidate, util)) {
    456           return false;
    457         }
    458         return PhoneNumberMatcher.isNationalPrefixPresentIfRequired(number, util);
    459       }
    460     },
    461     /**
    462      * Phone numbers accepted are {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid} and
    463      * are grouped in a possible way for this locale. For example, a US number written as
    464      * "65 02 53 00 00" and "650253 0000" are not accepted at this leniency level, whereas
    465      * "650 253 0000", "650 2530000" or "6502530000" are.
    466      * Numbers with more than one '/' symbol in the national significant number are also dropped at
    467      * this level.
    468      * <p>
    469      * Warning: This level might result in lower coverage especially for regions outside of country
    470      * code "+1". If you are not sure about which level to use, email the discussion group
    471      * libphonenumber-discuss (at) googlegroups.com.
    472      */
    473     STRICT_GROUPING {
    474       @Override
    475       boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
    476         if (!util.isValidNumber(number) ||
    477             !PhoneNumberMatcher.containsOnlyValidXChars(number, candidate, util) ||
    478             PhoneNumberMatcher.containsMoreThanOneSlashInNationalNumber(number, candidate) ||
    479             !PhoneNumberMatcher.isNationalPrefixPresentIfRequired(number, util)) {
    480           return false;
    481         }
    482         return PhoneNumberMatcher.checkNumberGroupingIsValid(
    483             number, candidate, util, new PhoneNumberMatcher.NumberGroupingChecker() {
    484               @Override
    485               public boolean checkGroups(PhoneNumberUtil util, PhoneNumber number,
    486                                          StringBuilder normalizedCandidate,
    487                                          String[] expectedNumberGroups) {
    488                 return PhoneNumberMatcher.allNumberGroupsRemainGrouped(
    489                     util, number, normalizedCandidate, expectedNumberGroups);
    490               }
    491             });
    492       }
    493     },
    494     /**
    495      * Phone numbers accepted are {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid} and
    496      * are grouped in the same way that we would have formatted it, or as a single block. For
    497      * example, a US number written as "650 2530000" is not accepted at this leniency level, whereas
    498      * "650 253 0000" or "6502530000" are.
    499      * Numbers with more than one '/' symbol are also dropped at this level.
    500      * <p>
    501      * Warning: This level might result in lower coverage especially for regions outside of country
    502      * code "+1". If you are not sure about which level to use, email the discussion group
    503      * libphonenumber-discuss (at) googlegroups.com.
    504      */
    505     EXACT_GROUPING {
    506       @Override
    507       boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
    508         if (!util.isValidNumber(number) ||
    509             !PhoneNumberMatcher.containsOnlyValidXChars(number, candidate, util) ||
    510             PhoneNumberMatcher.containsMoreThanOneSlashInNationalNumber(number, candidate) ||
    511             !PhoneNumberMatcher.isNationalPrefixPresentIfRequired(number, util)) {
    512           return false;
    513         }
    514         return PhoneNumberMatcher.checkNumberGroupingIsValid(
    515             number, candidate, util, new PhoneNumberMatcher.NumberGroupingChecker() {
    516               @Override
    517               public boolean checkGroups(PhoneNumberUtil util, PhoneNumber number,
    518                                          StringBuilder normalizedCandidate,
    519                                          String[] expectedNumberGroups) {
    520                 return PhoneNumberMatcher.allNumberGroupsAreExactlyPresent(
    521                     util, number, normalizedCandidate, expectedNumberGroups);
    522               }
    523             });
    524       }
    525     };
    526 
    527     /** Returns true if {@code number} is a verified number according to this leniency. */
    528     abstract boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util);
    529   }
    530 
    531   // A source of metadata for different regions.
    532   private final MetadataSource metadataSource;
    533 
    534   // A mapping from a country calling code to the region codes which denote the region represented
    535   // by that country calling code. In the case of multiple regions sharing a calling code, such as
    536   // the NANPA regions, the one indicated with "isMainCountryForCode" in the metadata should be
    537   // first.
    538   private final Map<Integer, List<String>> countryCallingCodeToRegionCodeMap;
    539 
    540   // The set of regions that share country calling code 1.
    541   // There are roughly 26 regions.
    542   // We set the initial capacity of the HashSet to 35 to offer a load factor of roughly 0.75.
    543   private final Set<String> nanpaRegions = new HashSet<String>(35);
    544 
    545   // A cache for frequently used region-specific regular expressions.
    546   // The initial capacity is set to 100 as this seems to be an optimal value for Android, based on
    547   // performance measurements.
    548   private final RegexCache regexCache = new RegexCache(100);
    549 
    550   // The set of regions the library supports.
    551   // There are roughly 240 of them and we set the initial capacity of the HashSet to 320 to offer a
    552   // load factor of roughly 0.75.
    553   private final Set<String> supportedRegions = new HashSet<String>(320);
    554 
    555   // The set of county calling codes that map to the non-geo entity region ("001"). This set
    556   // currently contains < 12 elements so the default capacity of 16 (load factor=0.75) is fine.
    557   private final Set<Integer> countryCodesForNonGeographicalRegion = new HashSet<Integer>();
    558 
    559   /**
    560    * This class implements a singleton, the constructor is only visible to facilitate testing.
    561    */
    562   // @VisibleForTesting
    563   PhoneNumberUtil(MetadataSource metadataSource,
    564       Map<Integer, List<String>> countryCallingCodeToRegionCodeMap) {
    565     this.metadataSource = metadataSource;
    566     this.countryCallingCodeToRegionCodeMap = countryCallingCodeToRegionCodeMap;
    567     for (Map.Entry<Integer, List<String>> entry : countryCallingCodeToRegionCodeMap.entrySet()) {
    568       List<String> regionCodes = entry.getValue();
    569       // We can assume that if the county calling code maps to the non-geo entity region code then
    570       // that's the only region code it maps to.
    571       if (regionCodes.size() == 1 && REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCodes.get(0))) {
    572         // This is the subset of all country codes that map to the non-geo entity region code.
    573         countryCodesForNonGeographicalRegion.add(entry.getKey());
    574       } else {
    575         // The supported regions set does not include the "001" non-geo entity region code.
    576         supportedRegions.addAll(regionCodes);
    577       }
    578     }
    579     // If the non-geo entity still got added to the set of supported regions it must be because
    580     // there are entries that list the non-geo entity alongside normal regions (which is wrong).
    581     // If we discover this, remove the non-geo entity from the set of supported regions and log.
    582     if (supportedRegions.remove(REGION_CODE_FOR_NON_GEO_ENTITY)) {
    583       logger.log(Level.WARNING, "invalid metadata " +
    584           "(country calling code was mapped to the non-geo entity as well as specific region(s))");
    585     }
    586     nanpaRegions.addAll(countryCallingCodeToRegionCodeMap.get(NANPA_COUNTRY_CODE));
    587   }
    588 
    589   /**
    590    * Attempts to extract a possible number from the string passed in. This currently strips all
    591    * leading characters that cannot be used to start a phone number. Characters that can be used to
    592    * start a phone number are defined in the VALID_START_CHAR_PATTERN. If none of these characters
    593    * are found in the number passed in, an empty string is returned. This function also attempts to
    594    * strip off any alternative extensions or endings if two or more are present, such as in the case
    595    * of: (530) 583-6985 x302/x2303. The second extension here makes this actually two phone numbers,
    596    * (530) 583-6985 x302 and (530) 583-6985 x2303. We remove the second extension so that the first
    597    * number is parsed correctly.
    598    *
    599    * @param number  the string that might contain a phone number
    600    * @return        the number, stripped of any non-phone-number prefix (such as "Tel:") or an empty
    601    *                string if no character used to start phone numbers (such as + or any digit) is
    602    *                found in the number
    603    */
    604   static String extractPossibleNumber(String number) {
    605     Matcher m = VALID_START_CHAR_PATTERN.matcher(number);
    606     if (m.find()) {
    607       number = number.substring(m.start());
    608       // Remove trailing non-alpha non-numerical characters.
    609       Matcher trailingCharsMatcher = UNWANTED_END_CHAR_PATTERN.matcher(number);
    610       if (trailingCharsMatcher.find()) {
    611         number = number.substring(0, trailingCharsMatcher.start());
    612         logger.log(Level.FINER, "Stripped trailing characters: " + number);
    613       }
    614       // Check for extra numbers at the end.
    615       Matcher secondNumber = SECOND_NUMBER_START_PATTERN.matcher(number);
    616       if (secondNumber.find()) {
    617         number = number.substring(0, secondNumber.start());
    618       }
    619       return number;
    620     } else {
    621       return "";
    622     }
    623   }
    624 
    625   /**
    626    * Checks to see if the string of characters could possibly be a phone number at all. At the
    627    * moment, checks to see that the string begins with at least 2 digits, ignoring any punctuation
    628    * commonly found in phone numbers.
    629    * This method does not require the number to be normalized in advance - but does assume that
    630    * leading non-number symbols have been removed, such as by the method extractPossibleNumber.
    631    *
    632    * @param number  string to be checked for viability as a phone number
    633    * @return        true if the number could be a phone number of some sort, otherwise false
    634    */
    635   // @VisibleForTesting
    636   static boolean isViablePhoneNumber(String number) {
    637     if (number.length() < MIN_LENGTH_FOR_NSN) {
    638       return false;
    639     }
    640     Matcher m = VALID_PHONE_NUMBER_PATTERN.matcher(number);
    641     return m.matches();
    642   }
    643 
    644   /**
    645    * Normalizes a string of characters representing a phone number. This performs the following
    646    * conversions:
    647    *   Punctuation is stripped.
    648    *   For ALPHA/VANITY numbers:
    649    *   Letters are converted to their numeric representation on a telephone keypad. The keypad
    650    *       used here is the one defined in ITU Recommendation E.161. This is only done if there are
    651    *       3 or more letters in the number, to lessen the risk that such letters are typos.
    652    *   For other numbers:
    653    *   Wide-ascii digits are converted to normal ASCII (European) digits.
    654    *   Arabic-Indic numerals are converted to European numerals.
    655    *   Spurious alpha characters are stripped.
    656    *
    657    * @param number  a string of characters representing a phone number
    658    * @return        the normalized string version of the phone number
    659    */
    660   static String normalize(String number) {
    661     Matcher m = VALID_ALPHA_PHONE_PATTERN.matcher(number);
    662     if (m.matches()) {
    663       return normalizeHelper(number, ALPHA_PHONE_MAPPINGS, true);
    664     } else {
    665       return normalizeDigitsOnly(number);
    666     }
    667   }
    668 
    669   /**
    670    * Normalizes a string of characters representing a phone number. This is a wrapper for
    671    * normalize(String number) but does in-place normalization of the StringBuilder provided.
    672    *
    673    * @param number  a StringBuilder of characters representing a phone number that will be
    674    *     normalized in place
    675    */
    676   static void normalize(StringBuilder number) {
    677     String normalizedNumber = normalize(number.toString());
    678     number.replace(0, number.length(), normalizedNumber);
    679   }
    680 
    681   /**
    682    * Normalizes a string of characters representing a phone number. This converts wide-ascii and
    683    * arabic-indic numerals to European numerals, and strips punctuation and alpha characters.
    684    *
    685    * @param number  a string of characters representing a phone number
    686    * @return        the normalized string version of the phone number
    687    */
    688   public static String normalizeDigitsOnly(String number) {
    689     return normalizeDigits(number, false /* strip non-digits */).toString();
    690   }
    691 
    692   static StringBuilder normalizeDigits(String number, boolean keepNonDigits) {
    693     StringBuilder normalizedDigits = new StringBuilder(number.length());
    694     for (char c : number.toCharArray()) {
    695       int digit = Character.digit(c, 10);
    696       if (digit != -1) {
    697         normalizedDigits.append(digit);
    698       } else if (keepNonDigits) {
    699         normalizedDigits.append(c);
    700       }
    701     }
    702     return normalizedDigits;
    703   }
    704 
    705   /**
    706    * Normalizes a string of characters representing a phone number. This strips all characters which
    707    * are not diallable on a mobile phone keypad (including all non-ASCII digits).
    708    *
    709    * @param number  a string of characters representing a phone number
    710    * @return        the normalized string version of the phone number
    711    */
    712   static String normalizeDiallableCharsOnly(String number) {
    713     return normalizeHelper(number, DIALLABLE_CHAR_MAPPINGS, true /* remove non matches */);
    714   }
    715 
    716   /**
    717    * Converts all alpha characters in a number to their respective digits on a keypad, but retains
    718    * existing formatting.
    719    */
    720   public static String convertAlphaCharactersInNumber(String number) {
    721     return normalizeHelper(number, ALPHA_PHONE_MAPPINGS, false);
    722   }
    723 
    724   /**
    725    * Gets the length of the geographical area code from the
    726    * PhoneNumber object passed in, so that clients could use it
    727    * to split a national significant number into geographical area code and subscriber number. It
    728    * works in such a way that the resultant subscriber number should be diallable, at least on some
    729    * devices. An example of how this could be used:
    730    *
    731    * <pre>{@code
    732    * PhoneNumberUtil phoneUtil = PhoneNumberUtil.getInstance();
    733    * PhoneNumber number = phoneUtil.parse("16502530000", "US");
    734    * String nationalSignificantNumber = phoneUtil.getNationalSignificantNumber(number);
    735    * String areaCode;
    736    * String subscriberNumber;
    737    *
    738    * int areaCodeLength = phoneUtil.getLengthOfGeographicalAreaCode(number);
    739    * if (areaCodeLength > 0) {
    740    *   areaCode = nationalSignificantNumber.substring(0, areaCodeLength);
    741    *   subscriberNumber = nationalSignificantNumber.substring(areaCodeLength);
    742    * } else {
    743    *   areaCode = "";
    744    *   subscriberNumber = nationalSignificantNumber;
    745    * }
    746    * }</pre>
    747    *
    748    * N.B.: area code is a very ambiguous concept, so the I18N team generally recommends against
    749    * using it for most purposes, but recommends using the more general {@code national_number}
    750    * instead. Read the following carefully before deciding to use this method:
    751    * <ul>
    752    *  <li> geographical area codes change over time, and this method honors those changes;
    753    *    therefore, it doesn't guarantee the stability of the result it produces.
    754    *  <li> subscriber numbers may not be diallable from all devices (notably mobile devices, which
    755    *    typically requires the full national_number to be dialled in most regions).
    756    *  <li> most non-geographical numbers have no area codes, including numbers from non-geographical
    757    *    entities
    758    *  <li> some geographical numbers have no area codes.
    759    * </ul>
    760    * @param number  the PhoneNumber object for which clients
    761    *     want to know the length of the area code.
    762    * @return  the length of area code of the PhoneNumber object
    763    *     passed in.
    764    */
    765   public int getLengthOfGeographicalAreaCode(PhoneNumber number) {
    766     PhoneMetadata metadata = getMetadataForRegion(getRegionCodeForNumber(number));
    767     if (metadata == null) {
    768       return 0;
    769     }
    770     // If a country doesn't use a national prefix, and this number doesn't have an Italian leading
    771     // zero, we assume it is a closed dialling plan with no area codes.
    772     if (!metadata.hasNationalPrefix() && !number.isItalianLeadingZero()) {
    773       return 0;
    774     }
    775 
    776     if (!isNumberGeographical(number)) {
    777       return 0;
    778     }
    779 
    780     return getLengthOfNationalDestinationCode(number);
    781   }
    782 
    783   /**
    784    * Gets the length of the national destination code (NDC) from the
    785    * PhoneNumber object passed in, so that clients could use it
    786    * to split a national significant number into NDC and subscriber number. The NDC of a phone
    787    * number is normally the first group of digit(s) right after the country calling code when the
    788    * number is formatted in the international format, if there is a subscriber number part that
    789    * follows. An example of how this could be used:
    790    *
    791    * <pre>{@code
    792    * PhoneNumberUtil phoneUtil = PhoneNumberUtil.getInstance();
    793    * PhoneNumber number = phoneUtil.parse("18002530000", "US");
    794    * String nationalSignificantNumber = phoneUtil.getNationalSignificantNumber(number);
    795    * String nationalDestinationCode;
    796    * String subscriberNumber;
    797    *
    798    * int nationalDestinationCodeLength = phoneUtil.getLengthOfNationalDestinationCode(number);
    799    * if (nationalDestinationCodeLength > 0) {
    800    *   nationalDestinationCode = nationalSignificantNumber.substring(0,
    801    *       nationalDestinationCodeLength);
    802    *   subscriberNumber = nationalSignificantNumber.substring(nationalDestinationCodeLength);
    803    * } else {
    804    *   nationalDestinationCode = "";
    805    *   subscriberNumber = nationalSignificantNumber;
    806    * }
    807    * }</pre>
    808    *
    809    * Refer to the unittests to see the difference between this function and
    810    * {@link #getLengthOfGeographicalAreaCode}.
    811    *
    812    * @param number  the PhoneNumber object for which clients
    813    *     want to know the length of the NDC.
    814    * @return  the length of NDC of the PhoneNumber object
    815    *     passed in.
    816    */
    817   public int getLengthOfNationalDestinationCode(PhoneNumber number) {
    818     PhoneNumber copiedProto;
    819     if (number.hasExtension()) {
    820       // We don't want to alter the proto given to us, but we don't want to include the extension
    821       // when we format it, so we copy it and clear the extension here.
    822       copiedProto = new PhoneNumber();
    823       copiedProto.mergeFrom(number);
    824       copiedProto.clearExtension();
    825     } else {
    826       copiedProto = number;
    827     }
    828 
    829     String nationalSignificantNumber = format(copiedProto,
    830                                               PhoneNumberUtil.PhoneNumberFormat.INTERNATIONAL);
    831     String[] numberGroups = NON_DIGITS_PATTERN.split(nationalSignificantNumber);
    832     // The pattern will start with "+COUNTRY_CODE " so the first group will always be the empty
    833     // string (before the + symbol) and the second group will be the country calling code. The third
    834     // group will be area code if it is not the last group.
    835     if (numberGroups.length <= 3) {
    836       return 0;
    837     }
    838 
    839     if (getNumberType(number) == PhoneNumberType.MOBILE) {
    840       // For example Argentinian mobile numbers, when formatted in the international format, are in
    841       // the form of +54 9 NDC XXXX.... As a result, we take the length of the third group (NDC) and
    842       // add the length of the second group (which is the mobile token), which also forms part of
    843       // the national significant number. This assumes that the mobile token is always formatted
    844       // separately from the rest of the phone number.
    845       String mobileToken = getCountryMobileToken(number.getCountryCode());
    846       if (!mobileToken.equals("")) {
    847         return numberGroups[2].length() + numberGroups[3].length();
    848       }
    849     }
    850     return numberGroups[2].length();
    851   }
    852 
    853   /**
    854    * Returns the mobile token for the provided country calling code if it has one, otherwise
    855    * returns an empty string. A mobile token is a number inserted before the area code when dialing
    856    * a mobile number from that country from abroad.
    857    *
    858    * @param countryCallingCode  the country calling code for which we want the mobile token
    859    * @return  the mobile token, as a string, for the given country calling code
    860    */
    861   public static String getCountryMobileToken(int countryCallingCode) {
    862     if (MOBILE_TOKEN_MAPPINGS.containsKey(countryCallingCode)) {
    863       return MOBILE_TOKEN_MAPPINGS.get(countryCallingCode);
    864     }
    865     return "";
    866   }
    867 
    868   /**
    869    * Normalizes a string of characters representing a phone number by replacing all characters found
    870    * in the accompanying map with the values therein, and stripping all other characters if
    871    * removeNonMatches is true.
    872    *
    873    * @param number                     a string of characters representing a phone number
    874    * @param normalizationReplacements  a mapping of characters to what they should be replaced by in
    875    *                                   the normalized version of the phone number
    876    * @param removeNonMatches           indicates whether characters that are not able to be replaced
    877    *                                   should be stripped from the number. If this is false, they
    878    *                                   will be left unchanged in the number.
    879    * @return  the normalized string version of the phone number
    880    */
    881   private static String normalizeHelper(String number,
    882                                         Map<Character, Character> normalizationReplacements,
    883                                         boolean removeNonMatches) {
    884     StringBuilder normalizedNumber = new StringBuilder(number.length());
    885     for (int i = 0; i < number.length(); i++) {
    886       char character = number.charAt(i);
    887       Character newDigit = normalizationReplacements.get(Character.toUpperCase(character));
    888       if (newDigit != null) {
    889         normalizedNumber.append(newDigit);
    890       } else if (!removeNonMatches) {
    891         normalizedNumber.append(character);
    892       }
    893       // If neither of the above are true, we remove this character.
    894     }
    895     return normalizedNumber.toString();
    896   }
    897 
    898   /**
    899    * Sets or resets the PhoneNumberUtil singleton instance. If set to null, the next call to
    900    * {@code getInstance()} will load (and return) the default instance.
    901    */
    902   // @VisibleForTesting
    903   static synchronized void setInstance(PhoneNumberUtil util) {
    904     instance = util;
    905   }
    906 
    907   /**
    908    * Convenience method to get a list of what regions the library has metadata for.
    909    */
    910   public Set<String> getSupportedRegions() {
    911     return Collections.unmodifiableSet(supportedRegions);
    912   }
    913 
    914   /**
    915    * Convenience method to get a list of what global network calling codes the library has metadata
    916    * for.
    917    */
    918   public Set<Integer> getSupportedGlobalNetworkCallingCodes() {
    919     return Collections.unmodifiableSet(countryCodesForNonGeographicalRegion);
    920   }
    921 
    922   /**
    923    * Gets a {@link PhoneNumberUtil} instance to carry out international phone number formatting,
    924    * parsing, or validation. The instance is loaded with phone number metadata for a number of most
    925    * commonly used regions.
    926    *
    927    * <p>The {@link PhoneNumberUtil} is implemented as a singleton. Therefore, calling getInstance
    928    * multiple times will only result in one instance being created.
    929    *
    930    * @return a PhoneNumberUtil instance
    931    */
    932   public static synchronized PhoneNumberUtil getInstance() {
    933     if (instance == null) {
    934       setInstance(createInstance(DEFAULT_METADATA_LOADER));
    935     }
    936     return instance;
    937   }
    938 
    939   /**
    940    * Create a new {@link PhoneNumberUtil} instance to carry out international phone number
    941    * formatting, parsing, or validation. The instance is loaded with all metadata by
    942    * using the metadataSource specified.
    943    *
    944    * This method should only be used in the rare case in which you want to manage your own
    945    * metadata loading. Calling this method multiple times is very expensive, as each time
    946    * a new instance is created from scratch. When in doubt, use {@link #getInstance}.
    947    *
    948    * @param metadataSource Customized metadata source. This should not be null.
    949    * @return a PhoneNumberUtil instance
    950    */
    951   public static PhoneNumberUtil createInstance(MetadataSource metadataSource) {
    952     if (metadataSource == null) {
    953       throw new IllegalArgumentException("metadataSource could not be null.");
    954     }
    955     return new PhoneNumberUtil(metadataSource,
    956         CountryCodeToRegionCodeMap.getCountryCodeToRegionCodeMap());
    957   }
    958 
    959   /**
    960    * Create a new {@link PhoneNumberUtil} instance to carry out international phone number
    961    * formatting, parsing, or validation. The instance is loaded with all metadata by
    962    * using the metadataLoader specified.
    963    *
    964    * This method should only be used in the rare case in which you want to manage your own
    965    * metadata loading. Calling this method multiple times is very expensive, as each time
    966    * a new instance is created from scratch. When in doubt, use {@link #getInstance}.
    967    *
    968    * @param metadataLoader Customized metadata loader. This should not be null.
    969    * @return a PhoneNumberUtil instance
    970    */
    971   public static PhoneNumberUtil createInstance(MetadataLoader metadataLoader) {
    972     if (metadataLoader == null) {
    973       throw new IllegalArgumentException("metadataLoader could not be null.");
    974     }
    975     return createInstance(new MultiFileMetadataSourceImpl(metadataLoader));
    976   }
    977 
    978   /**
    979    * Helper function to check if the national prefix formatting rule has the first group only, i.e.,
    980    * does not start with the national prefix.
    981    */
    982   static boolean formattingRuleHasFirstGroupOnly(String nationalPrefixFormattingRule) {
    983     return nationalPrefixFormattingRule.length() == 0 ||
    984         FIRST_GROUP_ONLY_PREFIX_PATTERN.matcher(nationalPrefixFormattingRule).matches();
    985   }
    986 
    987   /**
    988    * Tests whether a phone number has a geographical association. It checks if the number is
    989    * associated to a certain region in the country where it belongs to. Note that this doesn't
    990    * verify if the number is actually in use.
    991    *
    992    * A similar method is implemented as PhoneNumberOfflineGeocoder.canBeGeocoded, which performs a
    993    * looser check, since it only prevents cases where prefixes overlap for geocodable and
    994    * non-geocodable numbers. Also, if new phone number types were added, we should check if this
    995    * other method should be updated too.
    996    */
    997   boolean isNumberGeographical(PhoneNumber phoneNumber) {
    998     PhoneNumberType numberType = getNumberType(phoneNumber);
    999     // TODO: Include mobile phone numbers from countries like Indonesia, which has some
   1000     // mobile numbers that are geographical.
   1001     return numberType == PhoneNumberType.FIXED_LINE ||
   1002         numberType == PhoneNumberType.FIXED_LINE_OR_MOBILE;
   1003   }
   1004 
   1005   /**
   1006    * Helper function to check region code is not unknown or null.
   1007    */
   1008   private boolean isValidRegionCode(String regionCode) {
   1009     return regionCode != null && supportedRegions.contains(regionCode);
   1010   }
   1011 
   1012   /**
   1013    * Helper function to check the country calling code is valid.
   1014    */
   1015   private boolean hasValidCountryCallingCode(int countryCallingCode) {
   1016     return countryCallingCodeToRegionCodeMap.containsKey(countryCallingCode);
   1017   }
   1018 
   1019   /**
   1020    * Formats a phone number in the specified format using default rules. Note that this does not
   1021    * promise to produce a phone number that the user can dial from where they are - although we do
   1022    * format in either 'national' or 'international' format depending on what the client asks for, we
   1023    * do not currently support a more abbreviated format, such as for users in the same "area" who
   1024    * could potentially dial the number without area code. Note that if the phone number has a
   1025    * country calling code of 0 or an otherwise invalid country calling code, we cannot work out
   1026    * which formatting rules to apply so we return the national significant number with no formatting
   1027    * applied.
   1028    *
   1029    * @param number         the phone number to be formatted
   1030    * @param numberFormat   the format the phone number should be formatted into
   1031    * @return  the formatted phone number
   1032    */
   1033   public String format(PhoneNumber number, PhoneNumberFormat numberFormat) {
   1034     if (number.getNationalNumber() == 0 && number.hasRawInput()) {
   1035       // Unparseable numbers that kept their raw input just use that.
   1036       // This is the only case where a number can be formatted as E164 without a
   1037       // leading '+' symbol (but the original number wasn't parseable anyway).
   1038       // TODO: Consider removing the 'if' above so that unparseable
   1039       // strings without raw input format to the empty string instead of "+00"
   1040       String rawInput = number.getRawInput();
   1041       if (rawInput.length() > 0) {
   1042         return rawInput;
   1043       }
   1044     }
   1045     StringBuilder formattedNumber = new StringBuilder(20);
   1046     format(number, numberFormat, formattedNumber);
   1047     return formattedNumber.toString();
   1048   }
   1049 
   1050   /**
   1051    * Same as {@link #format(PhoneNumber, PhoneNumberFormat)}, but accepts a mutable StringBuilder as
   1052    * a parameter to decrease object creation when invoked many times.
   1053    */
   1054   public void format(PhoneNumber number, PhoneNumberFormat numberFormat,
   1055                      StringBuilder formattedNumber) {
   1056     // Clear the StringBuilder first.
   1057     formattedNumber.setLength(0);
   1058     int countryCallingCode = number.getCountryCode();
   1059     String nationalSignificantNumber = getNationalSignificantNumber(number);
   1060 
   1061     if (numberFormat == PhoneNumberFormat.E164) {
   1062       // Early exit for E164 case (even if the country calling code is invalid) since no formatting
   1063       // of the national number needs to be applied. Extensions are not formatted.
   1064       formattedNumber.append(nationalSignificantNumber);
   1065       prefixNumberWithCountryCallingCode(countryCallingCode, PhoneNumberFormat.E164,
   1066                                          formattedNumber);
   1067       return;
   1068     }
   1069     if (!hasValidCountryCallingCode(countryCallingCode)) {
   1070       formattedNumber.append(nationalSignificantNumber);
   1071       return;
   1072     }
   1073     // Note getRegionCodeForCountryCode() is used because formatting information for regions which
   1074     // share a country calling code is contained by only one region for performance reasons. For
   1075     // example, for NANPA regions it will be contained in the metadata for US.
   1076     String regionCode = getRegionCodeForCountryCode(countryCallingCode);
   1077     // Metadata cannot be null because the country calling code is valid (which means that the
   1078     // region code cannot be ZZ and must be one of our supported region codes).
   1079     PhoneMetadata metadata =
   1080         getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
   1081     formattedNumber.append(formatNsn(nationalSignificantNumber, metadata, numberFormat));
   1082     maybeAppendFormattedExtension(number, metadata, numberFormat, formattedNumber);
   1083     prefixNumberWithCountryCallingCode(countryCallingCode, numberFormat, formattedNumber);
   1084   }
   1085 
   1086   /**
   1087    * Formats a phone number in the specified format using client-defined formatting rules. Note that
   1088    * if the phone number has a country calling code of zero or an otherwise invalid country calling
   1089    * code, we cannot work out things like whether there should be a national prefix applied, or how
   1090    * to format extensions, so we return the national significant number with no formatting applied.
   1091    *
   1092    * @param number                        the phone number to be formatted
   1093    * @param numberFormat                  the format the phone number should be formatted into
   1094    * @param userDefinedFormats            formatting rules specified by clients
   1095    * @return  the formatted phone number
   1096    */
   1097   public String formatByPattern(PhoneNumber number,
   1098                                 PhoneNumberFormat numberFormat,
   1099                                 List<NumberFormat> userDefinedFormats) {
   1100     int countryCallingCode = number.getCountryCode();
   1101     String nationalSignificantNumber = getNationalSignificantNumber(number);
   1102     if (!hasValidCountryCallingCode(countryCallingCode)) {
   1103       return nationalSignificantNumber;
   1104     }
   1105     // Note getRegionCodeForCountryCode() is used because formatting information for regions which
   1106     // share a country calling code is contained by only one region for performance reasons. For
   1107     // example, for NANPA regions it will be contained in the metadata for US.
   1108     String regionCode = getRegionCodeForCountryCode(countryCallingCode);
   1109     // Metadata cannot be null because the country calling code is valid
   1110     PhoneMetadata metadata =
   1111         getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
   1112 
   1113     StringBuilder formattedNumber = new StringBuilder(20);
   1114 
   1115     NumberFormat formattingPattern =
   1116         chooseFormattingPatternForNumber(userDefinedFormats, nationalSignificantNumber);
   1117     if (formattingPattern == null) {
   1118       // If no pattern above is matched, we format the number as a whole.
   1119       formattedNumber.append(nationalSignificantNumber);
   1120     } else {
   1121       NumberFormat numFormatCopy = new NumberFormat();
   1122       // Before we do a replacement of the national prefix pattern $NP with the national prefix, we
   1123       // need to copy the rule so that subsequent replacements for different numbers have the
   1124       // appropriate national prefix.
   1125       numFormatCopy.mergeFrom(formattingPattern);
   1126       String nationalPrefixFormattingRule = formattingPattern.getNationalPrefixFormattingRule();
   1127       if (nationalPrefixFormattingRule.length() > 0) {
   1128         String nationalPrefix = metadata.getNationalPrefix();
   1129         if (nationalPrefix.length() > 0) {
   1130           // Replace $NP with national prefix and $FG with the first group ($1).
   1131           nationalPrefixFormattingRule =
   1132               NP_PATTERN.matcher(nationalPrefixFormattingRule).replaceFirst(nationalPrefix);
   1133           nationalPrefixFormattingRule =
   1134               FG_PATTERN.matcher(nationalPrefixFormattingRule).replaceFirst("\\$1");
   1135           numFormatCopy.setNationalPrefixFormattingRule(nationalPrefixFormattingRule);
   1136         } else {
   1137           // We don't want to have a rule for how to format the national prefix if there isn't one.
   1138           numFormatCopy.clearNationalPrefixFormattingRule();
   1139         }
   1140       }
   1141       formattedNumber.append(
   1142           formatNsnUsingPattern(nationalSignificantNumber, numFormatCopy, numberFormat));
   1143     }
   1144     maybeAppendFormattedExtension(number, metadata, numberFormat, formattedNumber);
   1145     prefixNumberWithCountryCallingCode(countryCallingCode, numberFormat, formattedNumber);
   1146     return formattedNumber.toString();
   1147   }
   1148 
   1149   /**
   1150    * Formats a phone number in national format for dialing using the carrier as specified in the
   1151    * {@code carrierCode}. The {@code carrierCode} will always be used regardless of whether the
   1152    * phone number already has a preferred domestic carrier code stored. If {@code carrierCode}
   1153    * contains an empty string, returns the number in national format without any carrier code.
   1154    *
   1155    * @param number  the phone number to be formatted
   1156    * @param carrierCode  the carrier selection code to be used
   1157    * @return  the formatted phone number in national format for dialing using the carrier as
   1158    *          specified in the {@code carrierCode}
   1159    */
   1160   public String formatNationalNumberWithCarrierCode(PhoneNumber number, String carrierCode) {
   1161     int countryCallingCode = number.getCountryCode();
   1162     String nationalSignificantNumber = getNationalSignificantNumber(number);
   1163     if (!hasValidCountryCallingCode(countryCallingCode)) {
   1164       return nationalSignificantNumber;
   1165     }
   1166 
   1167     // Note getRegionCodeForCountryCode() is used because formatting information for regions which
   1168     // share a country calling code is contained by only one region for performance reasons. For
   1169     // example, for NANPA regions it will be contained in the metadata for US.
   1170     String regionCode = getRegionCodeForCountryCode(countryCallingCode);
   1171     // Metadata cannot be null because the country calling code is valid.
   1172     PhoneMetadata metadata = getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
   1173 
   1174     StringBuilder formattedNumber = new StringBuilder(20);
   1175     formattedNumber.append(formatNsn(nationalSignificantNumber, metadata,
   1176                                      PhoneNumberFormat.NATIONAL, carrierCode));
   1177     maybeAppendFormattedExtension(number, metadata, PhoneNumberFormat.NATIONAL, formattedNumber);
   1178     prefixNumberWithCountryCallingCode(countryCallingCode, PhoneNumberFormat.NATIONAL,
   1179                                        formattedNumber);
   1180     return formattedNumber.toString();
   1181   }
   1182 
   1183   private PhoneMetadata getMetadataForRegionOrCallingCode(
   1184       int countryCallingCode, String regionCode) {
   1185     return REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCode)
   1186         ? getMetadataForNonGeographicalRegion(countryCallingCode)
   1187         : getMetadataForRegion(regionCode);
   1188   }
   1189 
   1190   /**
   1191    * Formats a phone number in national format for dialing using the carrier as specified in the
   1192    * preferredDomesticCarrierCode field of the PhoneNumber object passed in. If that is missing,
   1193    * use the {@code fallbackCarrierCode} passed in instead. If there is no
   1194    * {@code preferredDomesticCarrierCode}, and the {@code fallbackCarrierCode} contains an empty
   1195    * string, return the number in national format without any carrier code.
   1196    *
   1197    * <p>Use {@link #formatNationalNumberWithCarrierCode} instead if the carrier code passed in
   1198    * should take precedence over the number's {@code preferredDomesticCarrierCode} when formatting.
   1199    *
   1200    * @param number  the phone number to be formatted
   1201    * @param fallbackCarrierCode  the carrier selection code to be used, if none is found in the
   1202    *     phone number itself
   1203    * @return  the formatted phone number in national format for dialing using the number's
   1204    *     {@code preferredDomesticCarrierCode}, or the {@code fallbackCarrierCode} passed in if
   1205    *     none is found
   1206    */
   1207   public String formatNationalNumberWithPreferredCarrierCode(PhoneNumber number,
   1208                                                              String fallbackCarrierCode) {
   1209     return formatNationalNumberWithCarrierCode(number, number.hasPreferredDomesticCarrierCode()
   1210                                                        ? number.getPreferredDomesticCarrierCode()
   1211                                                        : fallbackCarrierCode);
   1212   }
   1213 
   1214   /**
   1215    * Returns a number formatted in such a way that it can be dialed from a mobile phone in a
   1216    * specific region. If the number cannot be reached from the region (e.g. some countries block
   1217    * toll-free numbers from being called outside of the country), the method returns an empty
   1218    * string.
   1219    *
   1220    * @param number  the phone number to be formatted
   1221    * @param regionCallingFrom  the region where the call is being placed
   1222    * @param withFormatting  whether the number should be returned with formatting symbols, such as
   1223    *     spaces and dashes.
   1224    * @return  the formatted phone number
   1225    */
   1226   public String formatNumberForMobileDialing(PhoneNumber number, String regionCallingFrom,
   1227                                              boolean withFormatting) {
   1228     int countryCallingCode = number.getCountryCode();
   1229     if (!hasValidCountryCallingCode(countryCallingCode)) {
   1230       return number.hasRawInput() ? number.getRawInput() : "";
   1231     }
   1232 
   1233     String formattedNumber = "";
   1234     // Clear the extension, as that part cannot normally be dialed together with the main number.
   1235     PhoneNumber numberNoExt = new PhoneNumber().mergeFrom(number).clearExtension();
   1236     String regionCode = getRegionCodeForCountryCode(countryCallingCode);
   1237     PhoneNumberType numberType = getNumberType(numberNoExt);
   1238     boolean isValidNumber = (numberType != PhoneNumberType.UNKNOWN);
   1239     if (regionCallingFrom.equals(regionCode)) {
   1240       boolean isFixedLineOrMobile =
   1241           (numberType == PhoneNumberType.FIXED_LINE) || (numberType == PhoneNumberType.MOBILE) ||
   1242           (numberType == PhoneNumberType.FIXED_LINE_OR_MOBILE);
   1243       // Carrier codes may be needed in some countries. We handle this here.
   1244       if (regionCode.equals("CO") && numberType == PhoneNumberType.FIXED_LINE) {
   1245         formattedNumber =
   1246             formatNationalNumberWithCarrierCode(numberNoExt, COLOMBIA_MOBILE_TO_FIXED_LINE_PREFIX);
   1247       } else if (regionCode.equals("BR") && isFixedLineOrMobile) {
   1248         formattedNumber = numberNoExt.hasPreferredDomesticCarrierCode()
   1249             ? formattedNumber = formatNationalNumberWithPreferredCarrierCode(numberNoExt, "")
   1250             // Brazilian fixed line and mobile numbers need to be dialed with a carrier code when
   1251             // called within Brazil. Without that, most of the carriers won't connect the call.
   1252             // Because of that, we return an empty string here.
   1253             : "";
   1254       } else if (isValidNumber && regionCode.equals("HU")) {
   1255         // The national format for HU numbers doesn't contain the national prefix, because that is
   1256         // how numbers are normally written down. However, the national prefix is obligatory when
   1257         // dialing from a mobile phone, except for short numbers. As a result, we add it back here
   1258         // if it is a valid regular length phone number.
   1259         formattedNumber =
   1260             getNddPrefixForRegion(regionCode, true /* strip non-digits */) +
   1261             " " + format(numberNoExt, PhoneNumberFormat.NATIONAL);
   1262       } else if (countryCallingCode == NANPA_COUNTRY_CODE) {
   1263         // For NANPA countries, we output international format for numbers that can be dialed
   1264         // internationally, since that always works, except for numbers which might potentially be
   1265         // short numbers, which are always dialled in national format.
   1266         PhoneMetadata regionMetadata = getMetadataForRegion(regionCallingFrom);
   1267         if (canBeInternationallyDialled(numberNoExt) &&
   1268             !isShorterThanPossibleNormalNumber(regionMetadata,
   1269                 getNationalSignificantNumber(numberNoExt))) {
   1270           formattedNumber = format(numberNoExt, PhoneNumberFormat.INTERNATIONAL);
   1271         } else {
   1272           formattedNumber = format(numberNoExt, PhoneNumberFormat.NATIONAL);
   1273         }
   1274       } else {
   1275         // For non-geographical countries, and Mexican and Chilean fixed line and mobile numbers, we
   1276         // output international format for numbers that can be dialed internationally as that always
   1277         // works.
   1278         if ((regionCode.equals(REGION_CODE_FOR_NON_GEO_ENTITY) ||
   1279             // MX fixed line and mobile numbers should always be formatted in international format,
   1280             // even when dialed within MX. For national format to work, a carrier code needs to be
   1281             // used, and the correct carrier code depends on if the caller and callee are from the
   1282             // same local area. It is trickier to get that to work correctly than using
   1283             // international format, which is tested to work fine on all carriers.
   1284             // CL fixed line numbers need the national prefix when dialing in the national format,
   1285             // but don't have it when used for display. The reverse is true for mobile numbers.
   1286             // As a result, we output them in the international format to make it work.
   1287             ((regionCode.equals("MX") || regionCode.equals("CL")) &&
   1288              isFixedLineOrMobile)) &&
   1289             canBeInternationallyDialled(numberNoExt)) {
   1290           formattedNumber = format(numberNoExt, PhoneNumberFormat.INTERNATIONAL);
   1291         } else {
   1292           formattedNumber = format(numberNoExt, PhoneNumberFormat.NATIONAL);
   1293         }
   1294       }
   1295     } else if (isValidNumber && canBeInternationallyDialled(numberNoExt)) {
   1296       // We assume that short numbers are not diallable from outside their region, so if a number
   1297       // is not a valid regular length phone number, we treat it as if it cannot be internationally
   1298       // dialled.
   1299       return withFormatting ? format(numberNoExt, PhoneNumberFormat.INTERNATIONAL)
   1300                             : format(numberNoExt, PhoneNumberFormat.E164);
   1301     }
   1302     return withFormatting ? formattedNumber
   1303                           : normalizeDiallableCharsOnly(formattedNumber);
   1304   }
   1305 
   1306   /**
   1307    * Formats a phone number for out-of-country dialing purposes. If no regionCallingFrom is
   1308    * supplied, we format the number in its INTERNATIONAL format. If the country calling code is the
   1309    * same as that of the region where the number is from, then NATIONAL formatting will be applied.
   1310    *
   1311    * <p>If the number itself has a country calling code of zero or an otherwise invalid country
   1312    * calling code, then we return the number with no formatting applied.
   1313    *
   1314    * <p>Note this function takes care of the case for calling inside of NANPA and between Russia and
   1315    * Kazakhstan (who share the same country calling code). In those cases, no international prefix
   1316    * is used. For regions which have multiple international prefixes, the number in its
   1317    * INTERNATIONAL format will be returned instead.
   1318    *
   1319    * @param number               the phone number to be formatted
   1320    * @param regionCallingFrom    the region where the call is being placed
   1321    * @return  the formatted phone number
   1322    */
   1323   public String formatOutOfCountryCallingNumber(PhoneNumber number,
   1324                                                 String regionCallingFrom) {
   1325     if (!isValidRegionCode(regionCallingFrom)) {
   1326       logger.log(Level.WARNING,
   1327                  "Trying to format number from invalid region "
   1328                  + regionCallingFrom
   1329                  + ". International formatting applied.");
   1330       return format(number, PhoneNumberFormat.INTERNATIONAL);
   1331     }
   1332     int countryCallingCode = number.getCountryCode();
   1333     String nationalSignificantNumber = getNationalSignificantNumber(number);
   1334     if (!hasValidCountryCallingCode(countryCallingCode)) {
   1335       return nationalSignificantNumber;
   1336     }
   1337     if (countryCallingCode == NANPA_COUNTRY_CODE) {
   1338       if (isNANPACountry(regionCallingFrom)) {
   1339         // For NANPA regions, return the national format for these regions but prefix it with the
   1340         // country calling code.
   1341         return countryCallingCode + " " + format(number, PhoneNumberFormat.NATIONAL);
   1342       }
   1343     } else if (countryCallingCode == getCountryCodeForValidRegion(regionCallingFrom)) {
   1344       // If regions share a country calling code, the country calling code need not be dialled.
   1345       // This also applies when dialling within a region, so this if clause covers both these cases.
   1346       // Technically this is the case for dialling from La Reunion to other overseas departments of
   1347       // France (French Guiana, Martinique, Guadeloupe), but not vice versa - so we don't cover this
   1348       // edge case for now and for those cases return the version including country calling code.
   1349       // Details here: http://www.petitfute.com/voyage/225-info-pratiques-reunion
   1350       return format(number, PhoneNumberFormat.NATIONAL);
   1351     }
   1352     // Metadata cannot be null because we checked 'isValidRegionCode()' above.
   1353     PhoneMetadata metadataForRegionCallingFrom = getMetadataForRegion(regionCallingFrom);
   1354     String internationalPrefix = metadataForRegionCallingFrom.getInternationalPrefix();
   1355 
   1356     // For regions that have multiple international prefixes, the international format of the
   1357     // number is returned, unless there is a preferred international prefix.
   1358     String internationalPrefixForFormatting = "";
   1359     if (UNIQUE_INTERNATIONAL_PREFIX.matcher(internationalPrefix).matches()) {
   1360       internationalPrefixForFormatting = internationalPrefix;
   1361     } else if (metadataForRegionCallingFrom.hasPreferredInternationalPrefix()) {
   1362       internationalPrefixForFormatting =
   1363           metadataForRegionCallingFrom.getPreferredInternationalPrefix();
   1364     }
   1365 
   1366     String regionCode = getRegionCodeForCountryCode(countryCallingCode);
   1367     // Metadata cannot be null because the country calling code is valid.
   1368     PhoneMetadata metadataForRegion =
   1369         getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
   1370     String formattedNationalNumber =
   1371         formatNsn(nationalSignificantNumber, metadataForRegion, PhoneNumberFormat.INTERNATIONAL);
   1372     StringBuilder formattedNumber = new StringBuilder(formattedNationalNumber);
   1373     maybeAppendFormattedExtension(number, metadataForRegion, PhoneNumberFormat.INTERNATIONAL,
   1374                                   formattedNumber);
   1375     if (internationalPrefixForFormatting.length() > 0) {
   1376       formattedNumber.insert(0, " ").insert(0, countryCallingCode).insert(0, " ")
   1377           .insert(0, internationalPrefixForFormatting);
   1378     } else {
   1379       prefixNumberWithCountryCallingCode(countryCallingCode,
   1380                                          PhoneNumberFormat.INTERNATIONAL,
   1381                                          formattedNumber);
   1382     }
   1383     return formattedNumber.toString();
   1384   }
   1385 
   1386   /**
   1387    * Formats a phone number using the original phone number format that the number is parsed from.
   1388    * The original format is embedded in the country_code_source field of the PhoneNumber object
   1389    * passed in. If such information is missing, the number will be formatted into the NATIONAL
   1390    * format by default. When the number contains a leading zero and this is unexpected for this
   1391    * country, or we don't have a formatting pattern for the number, the method returns the raw input
   1392    * when it is available.
   1393    *
   1394    * Note this method guarantees no digit will be inserted, removed or modified as a result of
   1395    * formatting.
   1396    *
   1397    * @param number  the phone number that needs to be formatted in its original number format
   1398    * @param regionCallingFrom  the region whose IDD needs to be prefixed if the original number
   1399    *     has one
   1400    * @return  the formatted phone number in its original number format
   1401    */
   1402   public String formatInOriginalFormat(PhoneNumber number, String regionCallingFrom) {
   1403     if (number.hasRawInput() &&
   1404         (hasUnexpectedItalianLeadingZero(number) || !hasFormattingPatternForNumber(number))) {
   1405       // We check if we have the formatting pattern because without that, we might format the number
   1406       // as a group without national prefix.
   1407       return number.getRawInput();
   1408     }
   1409     if (!number.hasCountryCodeSource()) {
   1410       return format(number, PhoneNumberFormat.NATIONAL);
   1411     }
   1412     String formattedNumber;
   1413     switch (number.getCountryCodeSource()) {
   1414       case FROM_NUMBER_WITH_PLUS_SIGN:
   1415         formattedNumber = format(number, PhoneNumberFormat.INTERNATIONAL);
   1416         break;
   1417       case FROM_NUMBER_WITH_IDD:
   1418         formattedNumber = formatOutOfCountryCallingNumber(number, regionCallingFrom);
   1419         break;
   1420       case FROM_NUMBER_WITHOUT_PLUS_SIGN:
   1421         formattedNumber = format(number, PhoneNumberFormat.INTERNATIONAL).substring(1);
   1422         break;
   1423       case FROM_DEFAULT_COUNTRY:
   1424         // Fall-through to default case.
   1425       default:
   1426         String regionCode = getRegionCodeForCountryCode(number.getCountryCode());
   1427         // We strip non-digits from the NDD here, and from the raw input later, so that we can
   1428         // compare them easily.
   1429         String nationalPrefix = getNddPrefixForRegion(regionCode, true /* strip non-digits */);
   1430         String nationalFormat = format(number, PhoneNumberFormat.NATIONAL);
   1431         if (nationalPrefix == null || nationalPrefix.length() == 0) {
   1432           // If the region doesn't have a national prefix at all, we can safely return the national
   1433           // format without worrying about a national prefix being added.
   1434           formattedNumber = nationalFormat;
   1435           break;
   1436         }
   1437         // Otherwise, we check if the original number was entered with a national prefix.
   1438         if (rawInputContainsNationalPrefix(
   1439             number.getRawInput(), nationalPrefix, regionCode)) {
   1440           // If so, we can safely return the national format.
   1441           formattedNumber = nationalFormat;
   1442           break;
   1443         }
   1444         // Metadata cannot be null here because getNddPrefixForRegion() (above) returns null if
   1445         // there is no metadata for the region.
   1446         PhoneMetadata metadata = getMetadataForRegion(regionCode);
   1447         String nationalNumber = getNationalSignificantNumber(number);
   1448         NumberFormat formatRule =
   1449             chooseFormattingPatternForNumber(metadata.numberFormats(), nationalNumber);
   1450         // The format rule could still be null here if the national number was 0 and there was no
   1451         // raw input (this should not be possible for numbers generated by the phonenumber library
   1452         // as they would also not have a country calling code and we would have exited earlier).
   1453         if (formatRule == null) {
   1454           formattedNumber = nationalFormat;
   1455           break;
   1456         }
   1457         // When the format we apply to this number doesn't contain national prefix, we can just
   1458         // return the national format.
   1459         // TODO: Refactor the code below with the code in
   1460         // isNationalPrefixPresentIfRequired.
   1461         String candidateNationalPrefixRule = formatRule.getNationalPrefixFormattingRule();
   1462         // We assume that the first-group symbol will never be _before_ the national prefix.
   1463         int indexOfFirstGroup = candidateNationalPrefixRule.indexOf("$1");
   1464         if (indexOfFirstGroup <= 0) {
   1465           formattedNumber = nationalFormat;
   1466           break;
   1467         }
   1468         candidateNationalPrefixRule =
   1469             candidateNationalPrefixRule.substring(0, indexOfFirstGroup);
   1470         candidateNationalPrefixRule = normalizeDigitsOnly(candidateNationalPrefixRule);
   1471         if (candidateNationalPrefixRule.length() == 0) {
   1472           // National prefix not used when formatting this number.
   1473           formattedNumber = nationalFormat;
   1474           break;
   1475         }
   1476         // Otherwise, we need to remove the national prefix from our output.
   1477         NumberFormat numFormatCopy = new NumberFormat();
   1478         numFormatCopy.mergeFrom(formatRule);
   1479         numFormatCopy.clearNationalPrefixFormattingRule();
   1480         List<NumberFormat> numberFormats = new ArrayList<NumberFormat>(1);
   1481         numberFormats.add(numFormatCopy);
   1482         formattedNumber = formatByPattern(number, PhoneNumberFormat.NATIONAL, numberFormats);
   1483         break;
   1484     }
   1485     String rawInput = number.getRawInput();
   1486     // If no digit is inserted/removed/modified as a result of our formatting, we return the
   1487     // formatted phone number; otherwise we return the raw input the user entered.
   1488     if (formattedNumber != null && rawInput.length() > 0) {
   1489       String normalizedFormattedNumber = normalizeDiallableCharsOnly(formattedNumber);
   1490       String normalizedRawInput = normalizeDiallableCharsOnly(rawInput);
   1491       if (!normalizedFormattedNumber.equals(normalizedRawInput)) {
   1492         formattedNumber = rawInput;
   1493       }
   1494     }
   1495     return formattedNumber;
   1496   }
   1497 
   1498   // Check if rawInput, which is assumed to be in the national format, has a national prefix. The
   1499   // national prefix is assumed to be in digits-only form.
   1500   private boolean rawInputContainsNationalPrefix(String rawInput, String nationalPrefix,
   1501       String regionCode) {
   1502     String normalizedNationalNumber = normalizeDigitsOnly(rawInput);
   1503     if (normalizedNationalNumber.startsWith(nationalPrefix)) {
   1504       try {
   1505         // Some Japanese numbers (e.g. 00777123) might be mistaken to contain the national prefix
   1506         // when written without it (e.g. 0777123) if we just do prefix matching. To tackle that, we
   1507         // check the validity of the number if the assumed national prefix is removed (777123 won't
   1508         // be valid in Japan).
   1509         return isValidNumber(
   1510             parse(normalizedNationalNumber.substring(nationalPrefix.length()), regionCode));
   1511       } catch (NumberParseException e) {
   1512         return false;
   1513       }
   1514     }
   1515     return false;
   1516   }
   1517 
   1518   /**
   1519    * Returns true if a number is from a region whose national significant number couldn't contain a
   1520    * leading zero, but has the italian_leading_zero field set to true.
   1521    */
   1522   private boolean hasUnexpectedItalianLeadingZero(PhoneNumber number) {
   1523     return number.isItalianLeadingZero() && !isLeadingZeroPossible(number.getCountryCode());
   1524   }
   1525 
   1526   private boolean hasFormattingPatternForNumber(PhoneNumber number) {
   1527     int countryCallingCode = number.getCountryCode();
   1528     String phoneNumberRegion = getRegionCodeForCountryCode(countryCallingCode);
   1529     PhoneMetadata metadata =
   1530         getMetadataForRegionOrCallingCode(countryCallingCode, phoneNumberRegion);
   1531     if (metadata == null) {
   1532       return false;
   1533     }
   1534     String nationalNumber = getNationalSignificantNumber(number);
   1535     NumberFormat formatRule =
   1536         chooseFormattingPatternForNumber(metadata.numberFormats(), nationalNumber);
   1537     return formatRule != null;
   1538   }
   1539 
   1540   /**
   1541    * Formats a phone number for out-of-country dialing purposes.
   1542    *
   1543    * Note that in this version, if the number was entered originally using alpha characters and
   1544    * this version of the number is stored in raw_input, this representation of the number will be
   1545    * used rather than the digit representation. Grouping information, as specified by characters
   1546    * such as "-" and " ", will be retained.
   1547    *
   1548    * <p><b>Caveats:</b></p>
   1549    * <ul>
   1550    *  <li> This will not produce good results if the country calling code is both present in the raw
   1551    *       input _and_ is the start of the national number. This is not a problem in the regions
   1552    *       which typically use alpha numbers.
   1553    *  <li> This will also not produce good results if the raw input has any grouping information
   1554    *       within the first three digits of the national number, and if the function needs to strip
   1555    *       preceding digits/words in the raw input before these digits. Normally people group the
   1556    *       first three digits together so this is not a huge problem - and will be fixed if it
   1557    *       proves to be so.
   1558    * </ul>
   1559    *
   1560    * @param number  the phone number that needs to be formatted
   1561    * @param regionCallingFrom  the region where the call is being placed
   1562    * @return  the formatted phone number
   1563    */
   1564   public String formatOutOfCountryKeepingAlphaChars(PhoneNumber number,
   1565                                                     String regionCallingFrom) {
   1566     String rawInput = number.getRawInput();
   1567     // If there is no raw input, then we can't keep alpha characters because there aren't any.
   1568     // In this case, we return formatOutOfCountryCallingNumber.
   1569     if (rawInput.length() == 0) {
   1570       return formatOutOfCountryCallingNumber(number, regionCallingFrom);
   1571     }
   1572     int countryCode = number.getCountryCode();
   1573     if (!hasValidCountryCallingCode(countryCode)) {
   1574       return rawInput;
   1575     }
   1576     // Strip any prefix such as country calling code, IDD, that was present. We do this by comparing
   1577     // the number in raw_input with the parsed number.
   1578     // To do this, first we normalize punctuation. We retain number grouping symbols such as " "
   1579     // only.
   1580     rawInput = normalizeHelper(rawInput, ALL_PLUS_NUMBER_GROUPING_SYMBOLS, true);
   1581     // Now we trim everything before the first three digits in the parsed number. We choose three
   1582     // because all valid alpha numbers have 3 digits at the start - if it does not, then we don't
   1583     // trim anything at all. Similarly, if the national number was less than three digits, we don't
   1584     // trim anything at all.
   1585     String nationalNumber = getNationalSignificantNumber(number);
   1586     if (nationalNumber.length() > 3) {
   1587       int firstNationalNumberDigit = rawInput.indexOf(nationalNumber.substring(0, 3));
   1588       if (firstNationalNumberDigit != -1) {
   1589         rawInput = rawInput.substring(firstNationalNumberDigit);
   1590       }
   1591     }
   1592     PhoneMetadata metadataForRegionCallingFrom = getMetadataForRegion(regionCallingFrom);
   1593     if (countryCode == NANPA_COUNTRY_CODE) {
   1594       if (isNANPACountry(regionCallingFrom)) {
   1595         return countryCode + " " + rawInput;
   1596       }
   1597     } else if (metadataForRegionCallingFrom != null &&
   1598                countryCode == getCountryCodeForValidRegion(regionCallingFrom)) {
   1599       NumberFormat formattingPattern =
   1600           chooseFormattingPatternForNumber(metadataForRegionCallingFrom.numberFormats(),
   1601                                            nationalNumber);
   1602       if (formattingPattern == null) {
   1603         // If no pattern above is matched, we format the original input.
   1604         return rawInput;
   1605       }
   1606       NumberFormat newFormat = new NumberFormat();
   1607       newFormat.mergeFrom(formattingPattern);
   1608       // The first group is the first group of digits that the user wrote together.
   1609       newFormat.setPattern("(\\d+)(.*)");
   1610       // Here we just concatenate them back together after the national prefix has been fixed.
   1611       newFormat.setFormat("$1$2");
   1612       // Now we format using this pattern instead of the default pattern, but with the national
   1613       // prefix prefixed if necessary.
   1614       // This will not work in the cases where the pattern (and not the leading digits) decide
   1615       // whether a national prefix needs to be used, since we have overridden the pattern to match
   1616       // anything, but that is not the case in the metadata to date.
   1617       return formatNsnUsingPattern(rawInput, newFormat, PhoneNumberFormat.NATIONAL);
   1618     }
   1619     String internationalPrefixForFormatting = "";
   1620     // If an unsupported region-calling-from is entered, or a country with multiple international
   1621     // prefixes, the international format of the number is returned, unless there is a preferred
   1622     // international prefix.
   1623     if (metadataForRegionCallingFrom != null) {
   1624       String internationalPrefix = metadataForRegionCallingFrom.getInternationalPrefix();
   1625       internationalPrefixForFormatting =
   1626           UNIQUE_INTERNATIONAL_PREFIX.matcher(internationalPrefix).matches()
   1627           ? internationalPrefix
   1628           : metadataForRegionCallingFrom.getPreferredInternationalPrefix();
   1629     }
   1630     StringBuilder formattedNumber = new StringBuilder(rawInput);
   1631     String regionCode = getRegionCodeForCountryCode(countryCode);
   1632     // Metadata cannot be null because the country calling code is valid.
   1633     PhoneMetadata metadataForRegion = getMetadataForRegionOrCallingCode(countryCode, regionCode);
   1634     maybeAppendFormattedExtension(number, metadataForRegion,
   1635                                   PhoneNumberFormat.INTERNATIONAL, formattedNumber);
   1636     if (internationalPrefixForFormatting.length() > 0) {
   1637       formattedNumber.insert(0, " ").insert(0, countryCode).insert(0, " ")
   1638           .insert(0, internationalPrefixForFormatting);
   1639     } else {
   1640       // Invalid region entered as country-calling-from (so no metadata was found for it) or the
   1641       // region chosen has multiple international dialling prefixes.
   1642       logger.log(Level.WARNING,
   1643                  "Trying to format number from invalid region "
   1644                  + regionCallingFrom
   1645                  + ". International formatting applied.");
   1646       prefixNumberWithCountryCallingCode(countryCode,
   1647                                          PhoneNumberFormat.INTERNATIONAL,
   1648                                          formattedNumber);
   1649     }
   1650     return formattedNumber.toString();
   1651   }
   1652 
   1653   /**
   1654    * Gets the national significant number of the a phone number. Note a national significant number
   1655    * doesn't contain a national prefix or any formatting.
   1656    *
   1657    * @param number  the phone number for which the national significant number is needed
   1658    * @return  the national significant number of the PhoneNumber object passed in
   1659    */
   1660   public String getNationalSignificantNumber(PhoneNumber number) {
   1661     // If leading zero(s) have been set, we prefix this now. Note this is not a national prefix.
   1662     StringBuilder nationalNumber = new StringBuilder();
   1663     if (number.isItalianLeadingZero()) {
   1664       char[] zeros = new char[number.getNumberOfLeadingZeros()];
   1665       Arrays.fill(zeros, '0');
   1666       nationalNumber.append(new String(zeros));
   1667     }
   1668     nationalNumber.append(number.getNationalNumber());
   1669     return nationalNumber.toString();
   1670   }
   1671 
   1672   /**
   1673    * A helper function that is used by format and formatByPattern.
   1674    */
   1675   private void prefixNumberWithCountryCallingCode(int countryCallingCode,
   1676                                                   PhoneNumberFormat numberFormat,
   1677                                                   StringBuilder formattedNumber) {
   1678     switch (numberFormat) {
   1679       case E164:
   1680         formattedNumber.insert(0, countryCallingCode).insert(0, PLUS_SIGN);
   1681         return;
   1682       case INTERNATIONAL:
   1683         formattedNumber.insert(0, " ").insert(0, countryCallingCode).insert(0, PLUS_SIGN);
   1684         return;
   1685       case RFC3966:
   1686         formattedNumber.insert(0, "-").insert(0, countryCallingCode).insert(0, PLUS_SIGN)
   1687             .insert(0, RFC3966_PREFIX);
   1688         return;
   1689       case NATIONAL:
   1690       default:
   1691         return;
   1692     }
   1693   }
   1694 
   1695   // Simple wrapper of formatNsn for the common case of no carrier code.
   1696   private String formatNsn(String number, PhoneMetadata metadata, PhoneNumberFormat numberFormat) {
   1697     return formatNsn(number, metadata, numberFormat, null);
   1698   }
   1699 
   1700   // Note in some regions, the national number can be written in two completely different ways
   1701   // depending on whether it forms part of the NATIONAL format or INTERNATIONAL format. The
   1702   // numberFormat parameter here is used to specify which format to use for those cases. If a
   1703   // carrierCode is specified, this will be inserted into the formatted string to replace $CC.
   1704   private String formatNsn(String number,
   1705                            PhoneMetadata metadata,
   1706                            PhoneNumberFormat numberFormat,
   1707                            String carrierCode) {
   1708     List<NumberFormat> intlNumberFormats = metadata.intlNumberFormats();
   1709     // When the intlNumberFormats exists, we use that to format national number for the
   1710     // INTERNATIONAL format instead of using the numberDesc.numberFormats.
   1711     List<NumberFormat> availableFormats =
   1712         (intlNumberFormats.size() == 0 || numberFormat == PhoneNumberFormat.NATIONAL)
   1713         ? metadata.numberFormats()
   1714         : metadata.intlNumberFormats();
   1715     NumberFormat formattingPattern = chooseFormattingPatternForNumber(availableFormats, number);
   1716     return (formattingPattern == null)
   1717         ? number
   1718         : formatNsnUsingPattern(number, formattingPattern, numberFormat, carrierCode);
   1719   }
   1720 
   1721   NumberFormat chooseFormattingPatternForNumber(List<NumberFormat> availableFormats,
   1722                                                 String nationalNumber) {
   1723     for (NumberFormat numFormat : availableFormats) {
   1724       int size = numFormat.leadingDigitsPatternSize();
   1725       if (size == 0 || regexCache.getPatternForRegex(
   1726               // We always use the last leading_digits_pattern, as it is the most detailed.
   1727               numFormat.getLeadingDigitsPattern(size - 1)).matcher(nationalNumber).lookingAt()) {
   1728         Matcher m = regexCache.getPatternForRegex(numFormat.getPattern()).matcher(nationalNumber);
   1729         if (m.matches()) {
   1730           return numFormat;
   1731         }
   1732       }
   1733     }
   1734     return null;
   1735   }
   1736 
   1737   // Simple wrapper of formatNsnUsingPattern for the common case of no carrier code.
   1738   String formatNsnUsingPattern(String nationalNumber,
   1739                                NumberFormat formattingPattern,
   1740                                PhoneNumberFormat numberFormat) {
   1741     return formatNsnUsingPattern(nationalNumber, formattingPattern, numberFormat, null);
   1742   }
   1743 
   1744   // Note that carrierCode is optional - if null or an empty string, no carrier code replacement
   1745   // will take place.
   1746   private String formatNsnUsingPattern(String nationalNumber,
   1747                                        NumberFormat formattingPattern,
   1748                                        PhoneNumberFormat numberFormat,
   1749                                        String carrierCode) {
   1750     String numberFormatRule = formattingPattern.getFormat();
   1751     Matcher m =
   1752         regexCache.getPatternForRegex(formattingPattern.getPattern()).matcher(nationalNumber);
   1753     String formattedNationalNumber = "";
   1754     if (numberFormat == PhoneNumberFormat.NATIONAL &&
   1755         carrierCode != null && carrierCode.length() > 0 &&
   1756         formattingPattern.getDomesticCarrierCodeFormattingRule().length() > 0) {
   1757       // Replace the $CC in the formatting rule with the desired carrier code.
   1758       String carrierCodeFormattingRule = formattingPattern.getDomesticCarrierCodeFormattingRule();
   1759       carrierCodeFormattingRule =
   1760           CC_PATTERN.matcher(carrierCodeFormattingRule).replaceFirst(carrierCode);
   1761       // Now replace the $FG in the formatting rule with the first group and the carrier code
   1762       // combined in the appropriate way.
   1763       numberFormatRule = FIRST_GROUP_PATTERN.matcher(numberFormatRule)
   1764           .replaceFirst(carrierCodeFormattingRule);
   1765       formattedNationalNumber = m.replaceAll(numberFormatRule);
   1766     } else {
   1767       // Use the national prefix formatting rule instead.
   1768       String nationalPrefixFormattingRule = formattingPattern.getNationalPrefixFormattingRule();
   1769       if (numberFormat == PhoneNumberFormat.NATIONAL &&
   1770           nationalPrefixFormattingRule != null &&
   1771           nationalPrefixFormattingRule.length() > 0) {
   1772         Matcher firstGroupMatcher = FIRST_GROUP_PATTERN.matcher(numberFormatRule);
   1773         formattedNationalNumber =
   1774             m.replaceAll(firstGroupMatcher.replaceFirst(nationalPrefixFormattingRule));
   1775       } else {
   1776         formattedNationalNumber = m.replaceAll(numberFormatRule);
   1777       }
   1778     }
   1779     if (numberFormat == PhoneNumberFormat.RFC3966) {
   1780       // Strip any leading punctuation.
   1781       Matcher matcher = SEPARATOR_PATTERN.matcher(formattedNationalNumber);
   1782       if (matcher.lookingAt()) {
   1783         formattedNationalNumber = matcher.replaceFirst("");
   1784       }
   1785       // Replace the rest with a dash between each number group.
   1786       formattedNationalNumber = matcher.reset(formattedNationalNumber).replaceAll("-");
   1787     }
   1788     return formattedNationalNumber;
   1789   }
   1790 
   1791   /**
   1792    * Gets a valid number for the specified region.
   1793    *
   1794    * @param regionCode  the region for which an example number is needed
   1795    * @return  a valid fixed-line number for the specified region. Returns null when the metadata
   1796    *    does not contain such information, or the region 001 is passed in. For 001 (representing
   1797    *    non-geographical numbers), call {@link #getExampleNumberForNonGeoEntity} instead.
   1798    */
   1799   public PhoneNumber getExampleNumber(String regionCode) {
   1800     return getExampleNumberForType(regionCode, PhoneNumberType.FIXED_LINE);
   1801   }
   1802 
   1803   /**
   1804    * Gets a valid number for the specified region and number type.
   1805    *
   1806    * @param regionCode  the region for which an example number is needed
   1807    * @param type  the type of number that is needed
   1808    * @return  a valid number for the specified region and type. Returns null when the metadata
   1809    *     does not contain such information or if an invalid region or region 001 was entered.
   1810    *     For 001 (representing non-geographical numbers), call
   1811    *     {@link #getExampleNumberForNonGeoEntity} instead.
   1812    */
   1813   public PhoneNumber getExampleNumberForType(String regionCode, PhoneNumberType type) {
   1814     // Check the region code is valid.
   1815     if (!isValidRegionCode(regionCode)) {
   1816       logger.log(Level.WARNING, "Invalid or unknown region code provided: " + regionCode);
   1817       return null;
   1818     }
   1819     PhoneNumberDesc desc = getNumberDescByType(getMetadataForRegion(regionCode), type);
   1820     try {
   1821       if (desc.hasExampleNumber()) {
   1822         return parse(desc.getExampleNumber(), regionCode);
   1823       }
   1824     } catch (NumberParseException e) {
   1825       logger.log(Level.SEVERE, e.toString());
   1826     }
   1827     return null;
   1828   }
   1829 
   1830   /**
   1831    * Gets a valid number for the specified country calling code for a non-geographical entity.
   1832    *
   1833    * @param countryCallingCode  the country calling code for a non-geographical entity
   1834    * @return  a valid number for the non-geographical entity. Returns null when the metadata
   1835    *    does not contain such information, or the country calling code passed in does not belong
   1836    *    to a non-geographical entity.
   1837    */
   1838   public PhoneNumber getExampleNumberForNonGeoEntity(int countryCallingCode) {
   1839     PhoneMetadata metadata = getMetadataForNonGeographicalRegion(countryCallingCode);
   1840     if (metadata != null) {
   1841       PhoneNumberDesc desc = metadata.getGeneralDesc();
   1842       try {
   1843         if (desc.hasExampleNumber()) {
   1844           return parse("+" + countryCallingCode + desc.getExampleNumber(), "ZZ");
   1845         }
   1846       } catch (NumberParseException e) {
   1847         logger.log(Level.SEVERE, e.toString());
   1848       }
   1849     } else {
   1850       logger.log(Level.WARNING,
   1851                  "Invalid or unknown country calling code provided: " + countryCallingCode);
   1852     }
   1853     return null;
   1854   }
   1855 
   1856   /**
   1857    * Appends the formatted extension of a phone number to formattedNumber, if the phone number had
   1858    * an extension specified.
   1859    */
   1860   private void maybeAppendFormattedExtension(PhoneNumber number, PhoneMetadata metadata,
   1861                                              PhoneNumberFormat numberFormat,
   1862                                              StringBuilder formattedNumber) {
   1863     if (number.hasExtension() && number.getExtension().length() > 0) {
   1864       if (numberFormat == PhoneNumberFormat.RFC3966) {
   1865         formattedNumber.append(RFC3966_EXTN_PREFIX).append(number.getExtension());
   1866       } else {
   1867         if (metadata.hasPreferredExtnPrefix()) {
   1868           formattedNumber.append(metadata.getPreferredExtnPrefix()).append(number.getExtension());
   1869         } else {
   1870           formattedNumber.append(DEFAULT_EXTN_PREFIX).append(number.getExtension());
   1871         }
   1872       }
   1873     }
   1874   }
   1875 
   1876   PhoneNumberDesc getNumberDescByType(PhoneMetadata metadata, PhoneNumberType type) {
   1877     switch (type) {
   1878       case PREMIUM_RATE:
   1879         return metadata.getPremiumRate();
   1880       case TOLL_FREE:
   1881         return metadata.getTollFree();
   1882       case MOBILE:
   1883         return metadata.getMobile();
   1884       case FIXED_LINE:
   1885       case FIXED_LINE_OR_MOBILE:
   1886         return metadata.getFixedLine();
   1887       case SHARED_COST:
   1888         return metadata.getSharedCost();
   1889       case VOIP:
   1890         return metadata.getVoip();
   1891       case PERSONAL_NUMBER:
   1892         return metadata.getPersonalNumber();
   1893       case PAGER:
   1894         return metadata.getPager();
   1895       case UAN:
   1896         return metadata.getUan();
   1897       case VOICEMAIL:
   1898         return metadata.getVoicemail();
   1899       default:
   1900         return metadata.getGeneralDesc();
   1901     }
   1902   }
   1903 
   1904   /**
   1905    * Gets the type of a phone number.
   1906    *
   1907    * @param number  the phone number that we want to know the type
   1908    * @return  the type of the phone number
   1909    */
   1910   public PhoneNumberType getNumberType(PhoneNumber number) {
   1911     String regionCode = getRegionCodeForNumber(number);
   1912     PhoneMetadata metadata = getMetadataForRegionOrCallingCode(number.getCountryCode(), regionCode);
   1913     if (metadata == null) {
   1914       return PhoneNumberType.UNKNOWN;
   1915     }
   1916     String nationalSignificantNumber = getNationalSignificantNumber(number);
   1917     return getNumberTypeHelper(nationalSignificantNumber, metadata);
   1918   }
   1919 
   1920   private PhoneNumberType getNumberTypeHelper(String nationalNumber, PhoneMetadata metadata) {
   1921     if (!isNumberMatchingDesc(nationalNumber, metadata.getGeneralDesc())) {
   1922       return PhoneNumberType.UNKNOWN;
   1923     }
   1924 
   1925     if (isNumberMatchingDesc(nationalNumber, metadata.getPremiumRate())) {
   1926       return PhoneNumberType.PREMIUM_RATE;
   1927     }
   1928     if (isNumberMatchingDesc(nationalNumber, metadata.getTollFree())) {
   1929       return PhoneNumberType.TOLL_FREE;
   1930     }
   1931     if (isNumberMatchingDesc(nationalNumber, metadata.getSharedCost())) {
   1932       return PhoneNumberType.SHARED_COST;
   1933     }
   1934     if (isNumberMatchingDesc(nationalNumber, metadata.getVoip())) {
   1935       return PhoneNumberType.VOIP;
   1936     }
   1937     if (isNumberMatchingDesc(nationalNumber, metadata.getPersonalNumber())) {
   1938       return PhoneNumberType.PERSONAL_NUMBER;
   1939     }
   1940     if (isNumberMatchingDesc(nationalNumber, metadata.getPager())) {
   1941       return PhoneNumberType.PAGER;
   1942     }
   1943     if (isNumberMatchingDesc(nationalNumber, metadata.getUan())) {
   1944       return PhoneNumberType.UAN;
   1945     }
   1946     if (isNumberMatchingDesc(nationalNumber, metadata.getVoicemail())) {
   1947       return PhoneNumberType.VOICEMAIL;
   1948     }
   1949 
   1950     boolean isFixedLine = isNumberMatchingDesc(nationalNumber, metadata.getFixedLine());
   1951     if (isFixedLine) {
   1952       if (metadata.isSameMobileAndFixedLinePattern()) {
   1953         return PhoneNumberType.FIXED_LINE_OR_MOBILE;
   1954       } else if (isNumberMatchingDesc(nationalNumber, metadata.getMobile())) {
   1955         return PhoneNumberType.FIXED_LINE_OR_MOBILE;
   1956       }
   1957       return PhoneNumberType.FIXED_LINE;
   1958     }
   1959     // Otherwise, test to see if the number is mobile. Only do this if certain that the patterns for
   1960     // mobile and fixed line aren't the same.
   1961     if (!metadata.isSameMobileAndFixedLinePattern() &&
   1962         isNumberMatchingDesc(nationalNumber, metadata.getMobile())) {
   1963       return PhoneNumberType.MOBILE;
   1964     }
   1965     return PhoneNumberType.UNKNOWN;
   1966   }
   1967 
   1968   /**
   1969    * Returns the metadata for the given region code or {@code null} if the region code is invalid
   1970    * or unknown.
   1971    */
   1972   PhoneMetadata getMetadataForRegion(String regionCode) {
   1973     if (!isValidRegionCode(regionCode)) {
   1974       return null;
   1975     }
   1976     return metadataSource.getMetadataForRegion(regionCode);
   1977   }
   1978 
   1979   PhoneMetadata getMetadataForNonGeographicalRegion(int countryCallingCode) {
   1980     if (!countryCallingCodeToRegionCodeMap.containsKey(countryCallingCode)) {
   1981       return null;
   1982     }
   1983     return metadataSource.getMetadataForNonGeographicalRegion(countryCallingCode);
   1984   }
   1985 
   1986   boolean isNumberPossibleForDesc(String nationalNumber, PhoneNumberDesc numberDesc) {
   1987     Matcher possibleNumberPatternMatcher =
   1988         regexCache.getPatternForRegex(numberDesc.getPossibleNumberPattern())
   1989             .matcher(nationalNumber);
   1990     return possibleNumberPatternMatcher.matches();
   1991   }
   1992 
   1993   boolean isNumberMatchingDesc(String nationalNumber, PhoneNumberDesc numberDesc) {
   1994     Matcher nationalNumberPatternMatcher =
   1995         regexCache.getPatternForRegex(numberDesc.getNationalNumberPattern())
   1996             .matcher(nationalNumber);
   1997     return isNumberPossibleForDesc(nationalNumber, numberDesc) &&
   1998         nationalNumberPatternMatcher.matches();
   1999   }
   2000 
   2001   /**
   2002    * Tests whether a phone number matches a valid pattern. Note this doesn't verify the number
   2003    * is actually in use, which is impossible to tell by just looking at a number itself.
   2004    *
   2005    * @param number       the phone number that we want to validate
   2006    * @return  a boolean that indicates whether the number is of a valid pattern
   2007    */
   2008   public boolean isValidNumber(PhoneNumber number) {
   2009     String regionCode = getRegionCodeForNumber(number);
   2010     return isValidNumberForRegion(number, regionCode);
   2011   }
   2012 
   2013   /**
   2014    * Tests whether a phone number is valid for a certain region. Note this doesn't verify the number
   2015    * is actually in use, which is impossible to tell by just looking at a number itself. If the
   2016    * country calling code is not the same as the country calling code for the region, this
   2017    * immediately exits with false. After this, the specific number pattern rules for the region are
   2018    * examined. This is useful for determining for example whether a particular number is valid for
   2019    * Canada, rather than just a valid NANPA number.
   2020    * Warning: In most cases, you want to use {@link #isValidNumber} instead. For example, this
   2021    * method will mark numbers from British Crown dependencies such as the Isle of Man as invalid for
   2022    * the region "GB" (United Kingdom), since it has its own region code, "IM", which may be
   2023    * undesirable.
   2024    *
   2025    * @param number       the phone number that we want to validate
   2026    * @param regionCode   the region that we want to validate the phone number for
   2027    * @return  a boolean that indicates whether the number is of a valid pattern
   2028    */
   2029   public boolean isValidNumberForRegion(PhoneNumber number, String regionCode) {
   2030     int countryCode = number.getCountryCode();
   2031     PhoneMetadata metadata = getMetadataForRegionOrCallingCode(countryCode, regionCode);
   2032     if ((metadata == null) ||
   2033         (!REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCode) &&
   2034          countryCode != getCountryCodeForValidRegion(regionCode))) {
   2035       // Either the region code was invalid, or the country calling code for this number does not
   2036       // match that of the region code.
   2037       return false;
   2038     }
   2039     String nationalSignificantNumber = getNationalSignificantNumber(number);
   2040     return getNumberTypeHelper(nationalSignificantNumber, metadata) != PhoneNumberType.UNKNOWN;
   2041   }
   2042 
   2043   /**
   2044    * Returns the region where a phone number is from. This could be used for geocoding at the region
   2045    * level.
   2046    *
   2047    * @param number  the phone number whose origin we want to know
   2048    * @return  the region where the phone number is from, or null if no region matches this calling
   2049    *     code
   2050    */
   2051   public String getRegionCodeForNumber(PhoneNumber number) {
   2052     int countryCode = number.getCountryCode();
   2053     List<String> regions = countryCallingCodeToRegionCodeMap.get(countryCode);
   2054     if (regions == null) {
   2055       String numberString = getNationalSignificantNumber(number);
   2056       logger.log(Level.INFO,
   2057                  "Missing/invalid country_code (" + countryCode + ") for number " + numberString);
   2058       return null;
   2059     }
   2060     if (regions.size() == 1) {
   2061       return regions.get(0);
   2062     } else {
   2063       return getRegionCodeForNumberFromRegionList(number, regions);
   2064     }
   2065   }
   2066 
   2067   private String getRegionCodeForNumberFromRegionList(PhoneNumber number,
   2068                                                       List<String> regionCodes) {
   2069     String nationalNumber = getNationalSignificantNumber(number);
   2070     for (String regionCode : regionCodes) {
   2071       // If leadingDigits is present, use this. Otherwise, do full validation.
   2072       // Metadata cannot be null because the region codes come from the country calling code map.
   2073       PhoneMetadata metadata = getMetadataForRegion(regionCode);
   2074       if (metadata.hasLeadingDigits()) {
   2075         if (regexCache.getPatternForRegex(metadata.getLeadingDigits())
   2076                 .matcher(nationalNumber).lookingAt()) {
   2077           return regionCode;
   2078         }
   2079       } else if (getNumberTypeHelper(nationalNumber, metadata) != PhoneNumberType.UNKNOWN) {
   2080         return regionCode;
   2081       }
   2082     }
   2083     return null;
   2084   }
   2085 
   2086   /**
   2087    * Returns the region code that matches the specific country calling code. In the case of no
   2088    * region code being found, ZZ will be returned. In the case of multiple regions, the one
   2089    * designated in the metadata as the "main" region for this calling code will be returned. If the
   2090    * countryCallingCode entered is valid but doesn't match a specific region (such as in the case of
   2091    * non-geographical calling codes like 800) the value "001" will be returned (corresponding to
   2092    * the value for World in the UN M.49 schema).
   2093    */
   2094   public String getRegionCodeForCountryCode(int countryCallingCode) {
   2095     List<String> regionCodes = countryCallingCodeToRegionCodeMap.get(countryCallingCode);
   2096     return regionCodes == null ? UNKNOWN_REGION : regionCodes.get(0);
   2097   }
   2098 
   2099   /**
   2100    * Returns a list with the region codes that match the specific country calling code. For
   2101    * non-geographical country calling codes, the region code 001 is returned. Also, in the case
   2102    * of no region code being found, an empty list is returned.
   2103    */
   2104   public List<String> getRegionCodesForCountryCode(int countryCallingCode) {
   2105     List<String> regionCodes = countryCallingCodeToRegionCodeMap.get(countryCallingCode);
   2106     return Collections.unmodifiableList(regionCodes == null ? new ArrayList<String>(0)
   2107                                                             : regionCodes);
   2108   }
   2109 
   2110   /**
   2111    * Returns the country calling code for a specific region. For example, this would be 1 for the
   2112    * United States, and 64 for New Zealand.
   2113    *
   2114    * @param regionCode  the region that we want to get the country calling code for
   2115    * @return  the country calling code for the region denoted by regionCode
   2116    */
   2117   public int getCountryCodeForRegion(String regionCode) {
   2118     if (!isValidRegionCode(regionCode)) {
   2119       logger.log(Level.WARNING,
   2120                  "Invalid or missing region code ("
   2121                   + ((regionCode == null) ? "null" : regionCode)
   2122                   + ") provided.");
   2123       return 0;
   2124     }
   2125     return getCountryCodeForValidRegion(regionCode);
   2126   }
   2127 
   2128   /**
   2129    * Returns the country calling code for a specific region. For example, this would be 1 for the
   2130    * United States, and 64 for New Zealand. Assumes the region is already valid.
   2131    *
   2132    * @param regionCode  the region that we want to get the country calling code for
   2133    * @return  the country calling code for the region denoted by regionCode
   2134    * @throws IllegalArgumentException if the region is invalid
   2135    */
   2136   private int getCountryCodeForValidRegion(String regionCode) {
   2137     PhoneMetadata metadata = getMetadataForRegion(regionCode);
   2138     if (metadata == null) {
   2139       throw new IllegalArgumentException("Invalid region code: " + regionCode);
   2140     }
   2141     return metadata.getCountryCode();
   2142   }
   2143 
   2144   /**
   2145    * Returns the national dialling prefix for a specific region. For example, this would be 1 for
   2146    * the United States, and 0 for New Zealand. Set stripNonDigits to true to strip symbols like "~"
   2147    * (which indicates a wait for a dialling tone) from the prefix returned. If no national prefix is
   2148    * present, we return null.
   2149    *
   2150    * <p>Warning: Do not use this method for do-your-own formatting - for some regions, the
   2151    * national dialling prefix is used only for certain types of numbers. Use the library's
   2152    * formatting functions to prefix the national prefix when required.
   2153    *
   2154    * @param regionCode  the region that we want to get the dialling prefix for
   2155    * @param stripNonDigits  true to strip non-digits from the national dialling prefix
   2156    * @return  the dialling prefix for the region denoted by regionCode
   2157    */
   2158   public String getNddPrefixForRegion(String regionCode, boolean stripNonDigits) {
   2159     PhoneMetadata metadata = getMetadataForRegion(regionCode);
   2160     if (metadata == null) {
   2161       logger.log(Level.WARNING,
   2162                  "Invalid or missing region code ("
   2163                   + ((regionCode == null) ? "null" : regionCode)
   2164                   + ") provided.");
   2165       return null;
   2166     }
   2167     String nationalPrefix = metadata.getNationalPrefix();
   2168     // If no national prefix was found, we return null.
   2169     if (nationalPrefix.length() == 0) {
   2170       return null;
   2171     }
   2172     if (stripNonDigits) {
   2173       // Note: if any other non-numeric symbols are ever used in national prefixes, these would have
   2174       // to be removed here as well.
   2175       nationalPrefix = nationalPrefix.replace("~", "");
   2176     }
   2177     return nationalPrefix;
   2178   }
   2179 
   2180   /**
   2181    * Checks if this is a region under the North American Numbering Plan Administration (NANPA).
   2182    *
   2183    * @return  true if regionCode is one of the regions under NANPA
   2184    */
   2185   public boolean isNANPACountry(String regionCode) {
   2186     return nanpaRegions.contains(regionCode);
   2187   }
   2188 
   2189   /**
   2190    * Checks whether the country calling code is from a region whose national significant number
   2191    * could contain a leading zero. An example of such a region is Italy. Returns false if no
   2192    * metadata for the country is found.
   2193    */
   2194   boolean isLeadingZeroPossible(int countryCallingCode) {
   2195     PhoneMetadata mainMetadataForCallingCode =
   2196         getMetadataForRegionOrCallingCode(countryCallingCode,
   2197                                           getRegionCodeForCountryCode(countryCallingCode));
   2198     if (mainMetadataForCallingCode == null) {
   2199       return false;
   2200     }
   2201     return mainMetadataForCallingCode.isLeadingZeroPossible();
   2202   }
   2203 
   2204   /**
   2205    * Checks if the number is a valid vanity (alpha) number such as 800 MICROSOFT. A valid vanity
   2206    * number will start with at least 3 digits and will have three or more alpha characters. This
   2207    * does not do region-specific checks - to work out if this number is actually valid for a region,
   2208    * it should be parsed and methods such as {@link #isPossibleNumberWithReason} and
   2209    * {@link #isValidNumber} should be used.
   2210    *
   2211    * @param number  the number that needs to be checked
   2212    * @return  true if the number is a valid vanity number
   2213    */
   2214   public boolean isAlphaNumber(String number) {
   2215     if (!isViablePhoneNumber(number)) {
   2216       // Number is too short, or doesn't match the basic phone number pattern.
   2217       return false;
   2218     }
   2219     StringBuilder strippedNumber = new StringBuilder(number);
   2220     maybeStripExtension(strippedNumber);
   2221     return VALID_ALPHA_PHONE_PATTERN.matcher(strippedNumber).matches();
   2222   }
   2223 
   2224   /**
   2225    * Convenience wrapper around {@link #isPossibleNumberWithReason}. Instead of returning the reason
   2226    * for failure, this method returns a boolean value.
   2227    * @param number  the number that needs to be checked
   2228    * @return  true if the number is possible
   2229    */
   2230   public boolean isPossibleNumber(PhoneNumber number) {
   2231     return isPossibleNumberWithReason(number) == ValidationResult.IS_POSSIBLE;
   2232   }
   2233 
   2234   /**
   2235    * Helper method to check a number against a particular pattern and determine whether it matches,
   2236    * or is too short or too long. Currently, if a number pattern suggests that numbers of length 7
   2237    * and 10 are possible, and a number in between these possible lengths is entered, such as of
   2238    * length 8, this will return TOO_LONG.
   2239    */
   2240   private ValidationResult testNumberLengthAgainstPattern(Pattern numberPattern, String number) {
   2241     Matcher numberMatcher = numberPattern.matcher(number);
   2242     if (numberMatcher.matches()) {
   2243       return ValidationResult.IS_POSSIBLE;
   2244     }
   2245     if (numberMatcher.lookingAt()) {
   2246       return ValidationResult.TOO_LONG;
   2247     } else {
   2248       return ValidationResult.TOO_SHORT;
   2249     }
   2250   }
   2251 
   2252   /**
   2253    * Helper method to check whether a number is too short to be a regular length phone number in a
   2254    * region.
   2255    */
   2256   private boolean isShorterThanPossibleNormalNumber(PhoneMetadata regionMetadata, String number) {
   2257     Pattern possibleNumberPattern = regexCache.getPatternForRegex(
   2258         regionMetadata.getGeneralDesc().getPossibleNumberPattern());
   2259     return testNumberLengthAgainstPattern(possibleNumberPattern, number) ==
   2260         ValidationResult.TOO_SHORT;
   2261   }
   2262 
   2263   /**
   2264    * Check whether a phone number is a possible number. It provides a more lenient check than
   2265    * {@link #isValidNumber} in the following sense:
   2266    *<ol>
   2267    * <li> It only checks the length of phone numbers. In particular, it doesn't check starting
   2268    *      digits of the number.
   2269    * <li> It doesn't attempt to figure out the type of the number, but uses general rules which
   2270    *      applies to all types of phone numbers in a region. Therefore, it is much faster than
   2271    *      isValidNumber.
   2272    * <li> For fixed line numbers, many regions have the concept of area code, which together with
   2273    *      subscriber number constitute the national significant number. It is sometimes okay to dial
   2274    *      the subscriber number only when dialing in the same area. This function will return
   2275    *      true if the subscriber-number-only version is passed in. On the other hand, because
   2276    *      isValidNumber validates using information on both starting digits (for fixed line
   2277    *      numbers, that would most likely be area codes) and length (obviously includes the
   2278    *      length of area codes for fixed line numbers), it will return false for the
   2279    *      subscriber-number-only version.
   2280    * </ol>
   2281    * @param number  the number that needs to be checked
   2282    * @return  a ValidationResult object which indicates whether the number is possible
   2283    */
   2284   public ValidationResult isPossibleNumberWithReason(PhoneNumber number) {
   2285     String nationalNumber = getNationalSignificantNumber(number);
   2286     int countryCode = number.getCountryCode();
   2287     // Note: For Russian Fed and NANPA numbers, we just use the rules from the default region (US or
   2288     // Russia) since the getRegionCodeForNumber will not work if the number is possible but not
   2289     // valid. This would need to be revisited if the possible number pattern ever differed between
   2290     // various regions within those plans.
   2291     if (!hasValidCountryCallingCode(countryCode)) {
   2292       return ValidationResult.INVALID_COUNTRY_CODE;
   2293     }
   2294     String regionCode = getRegionCodeForCountryCode(countryCode);
   2295     // Metadata cannot be null because the country calling code is valid.
   2296     PhoneMetadata metadata = getMetadataForRegionOrCallingCode(countryCode, regionCode);
   2297     Pattern possibleNumberPattern =
   2298         regexCache.getPatternForRegex(metadata.getGeneralDesc().getPossibleNumberPattern());
   2299     return testNumberLengthAgainstPattern(possibleNumberPattern, nationalNumber);
   2300   }
   2301 
   2302   /**
   2303    * Check whether a phone number is a possible number given a number in the form of a string, and
   2304    * the region where the number could be dialed from. It provides a more lenient check than
   2305    * {@link #isValidNumber}. See {@link #isPossibleNumber(PhoneNumber)} for details.
   2306    *
   2307    * <p>This method first parses the number, then invokes {@link #isPossibleNumber(PhoneNumber)}
   2308    * with the resultant PhoneNumber object.
   2309    *
   2310    * @param number  the number that needs to be checked, in the form of a string
   2311    * @param regionDialingFrom  the region that we are expecting the number to be dialed from.
   2312    *     Note this is different from the region where the number belongs.  For example, the number
   2313    *     +1 650 253 0000 is a number that belongs to US. When written in this form, it can be
   2314    *     dialed from any region. When it is written as 00 1 650 253 0000, it can be dialed from any
   2315    *     region which uses an international dialling prefix of 00. When it is written as
   2316    *     650 253 0000, it can only be dialed from within the US, and when written as 253 0000, it
   2317    *     can only be dialed from within a smaller area in the US (Mountain View, CA, to be more
   2318    *     specific).
   2319    * @return  true if the number is possible
   2320    */
   2321   public boolean isPossibleNumber(String number, String regionDialingFrom) {
   2322     try {
   2323       return isPossibleNumber(parse(number, regionDialingFrom));
   2324     } catch (NumberParseException e) {
   2325       return false;
   2326     }
   2327   }
   2328 
   2329   /**
   2330    * Attempts to extract a valid number from a phone number that is too long to be valid, and resets
   2331    * the PhoneNumber object passed in to that valid version. If no valid number could be extracted,
   2332    * the PhoneNumber object passed in will not be modified.
   2333    * @param number a PhoneNumber object which contains a number that is too long to be valid.
   2334    * @return  true if a valid phone number can be successfully extracted.
   2335    */
   2336   public boolean truncateTooLongNumber(PhoneNumber number) {
   2337     if (isValidNumber(number)) {
   2338       return true;
   2339     }
   2340     PhoneNumber numberCopy = new PhoneNumber();
   2341     numberCopy.mergeFrom(number);
   2342     long nationalNumber = number.getNationalNumber();
   2343     do {
   2344       nationalNumber /= 10;
   2345       numberCopy.setNationalNumber(nationalNumber);
   2346       if (isPossibleNumberWithReason(numberCopy) == ValidationResult.TOO_SHORT ||
   2347           nationalNumber == 0) {
   2348         return false;
   2349       }
   2350     } while (!isValidNumber(numberCopy));
   2351     number.setNationalNumber(nationalNumber);
   2352     return true;
   2353   }
   2354 
   2355   /**
   2356    * Gets an {@link com.google.i18n.phonenumbers.AsYouTypeFormatter} for the specific region.
   2357    *
   2358    * @param regionCode  the region where the phone number is being entered
   2359    * @return  an {@link com.google.i18n.phonenumbers.AsYouTypeFormatter} object, which can be used
   2360    *     to format phone numbers in the specific region "as you type"
   2361    */
   2362   public AsYouTypeFormatter getAsYouTypeFormatter(String regionCode) {
   2363     return new AsYouTypeFormatter(regionCode);
   2364   }
   2365 
   2366   // Extracts country calling code from fullNumber, returns it and places the remaining number in
   2367   // nationalNumber. It assumes that the leading plus sign or IDD has already been removed. Returns
   2368   // 0 if fullNumber doesn't start with a valid country calling code, and leaves nationalNumber
   2369   // unmodified.
   2370   int extractCountryCode(StringBuilder fullNumber, StringBuilder nationalNumber) {
   2371     if ((fullNumber.length() == 0) || (fullNumber.charAt(0) == '0')) {
   2372       // Country codes do not begin with a '0'.
   2373       return 0;
   2374     }
   2375     int potentialCountryCode;
   2376     int numberLength = fullNumber.length();
   2377     for (int i = 1; i <= MAX_LENGTH_COUNTRY_CODE && i <= numberLength; i++) {
   2378       potentialCountryCode = Integer.parseInt(fullNumber.substring(0, i));
   2379       if (countryCallingCodeToRegionCodeMap.containsKey(potentialCountryCode)) {
   2380         nationalNumber.append(fullNumber.substring(i));
   2381         return potentialCountryCode;
   2382       }
   2383     }
   2384     return 0;
   2385   }
   2386 
   2387   /**
   2388    * Tries to extract a country calling code from a number. This method will return zero if no
   2389    * country calling code is considered to be present. Country calling codes are extracted in the
   2390    * following ways:
   2391    * <ul>
   2392    *  <li> by stripping the international dialing prefix of the region the person is dialing from,
   2393    *       if this is present in the number, and looking at the next digits
   2394    *  <li> by stripping the '+' sign if present and then looking at the next digits
   2395    *  <li> by comparing the start of the number and the country calling code of the default region.
   2396    *       If the number is not considered possible for the numbering plan of the default region
   2397    *       initially, but starts with the country calling code of this region, validation will be
   2398    *       reattempted after stripping this country calling code. If this number is considered a
   2399    *       possible number, then the first digits will be considered the country calling code and
   2400    *       removed as such.
   2401    * </ul>
   2402    * It will throw a NumberParseException if the number starts with a '+' but the country calling
   2403    * code supplied after this does not match that of any known region.
   2404    *
   2405    * @param number  non-normalized telephone number that we wish to extract a country calling
   2406    *     code from - may begin with '+'
   2407    * @param defaultRegionMetadata  metadata about the region this number may be from
   2408    * @param nationalNumber  a string buffer to store the national significant number in, in the case
   2409    *     that a country calling code was extracted. The number is appended to any existing contents.
   2410    *     If no country calling code was extracted, this will be left unchanged.
   2411    * @param keepRawInput  true if the country_code_source and preferred_carrier_code fields of
   2412    *     phoneNumber should be populated.
   2413    * @param phoneNumber  the PhoneNumber object where the country_code and country_code_source need
   2414    *     to be populated. Note the country_code is always populated, whereas country_code_source is
   2415    *     only populated when keepCountryCodeSource is true.
   2416    * @return  the country calling code extracted or 0 if none could be extracted
   2417    */
   2418   // @VisibleForTesting
   2419   int maybeExtractCountryCode(String number, PhoneMetadata defaultRegionMetadata,
   2420                               StringBuilder nationalNumber, boolean keepRawInput,
   2421                               PhoneNumber phoneNumber)
   2422       throws NumberParseException {
   2423     if (number.length() == 0) {
   2424       return 0;
   2425     }
   2426     StringBuilder fullNumber = new StringBuilder(number);
   2427     // Set the default prefix to be something that will never match.
   2428     String possibleCountryIddPrefix = "NonMatch";
   2429     if (defaultRegionMetadata != null) {
   2430       possibleCountryIddPrefix = defaultRegionMetadata.getInternationalPrefix();
   2431     }
   2432 
   2433     CountryCodeSource countryCodeSource =
   2434         maybeStripInternationalPrefixAndNormalize(fullNumber, possibleCountryIddPrefix);
   2435     if (keepRawInput) {
   2436       phoneNumber.setCountryCodeSource(countryCodeSource);
   2437     }
   2438     if (countryCodeSource != CountryCodeSource.FROM_DEFAULT_COUNTRY) {
   2439       if (fullNumber.length() <= MIN_LENGTH_FOR_NSN) {
   2440         throw new NumberParseException(NumberParseException.ErrorType.TOO_SHORT_AFTER_IDD,
   2441                                        "Phone number had an IDD, but after this was not "
   2442                                        + "long enough to be a viable phone number.");
   2443       }
   2444       int potentialCountryCode = extractCountryCode(fullNumber, nationalNumber);
   2445       if (potentialCountryCode != 0) {
   2446         phoneNumber.setCountryCode(potentialCountryCode);
   2447         return potentialCountryCode;
   2448       }
   2449 
   2450       // If this fails, they must be using a strange country calling code that we don't recognize,
   2451       // or that doesn't exist.
   2452       throw new NumberParseException(NumberParseException.ErrorType.INVALID_COUNTRY_CODE,
   2453                                      "Country calling code supplied was not recognised.");
   2454     } else if (defaultRegionMetadata != null) {
   2455       // Check to see if the number starts with the country calling code for the default region. If
   2456       // so, we remove the country calling code, and do some checks on the validity of the number
   2457       // before and after.
   2458       int defaultCountryCode = defaultRegionMetadata.getCountryCode();
   2459       String defaultCountryCodeString = String.valueOf(defaultCountryCode);
   2460       String normalizedNumber = fullNumber.toString();
   2461       if (normalizedNumber.startsWith(defaultCountryCodeString)) {
   2462         StringBuilder potentialNationalNumber =
   2463             new StringBuilder(normalizedNumber.substring(defaultCountryCodeString.length()));
   2464         PhoneNumberDesc generalDesc = defaultRegionMetadata.getGeneralDesc();
   2465         Pattern validNumberPattern =
   2466             regexCache.getPatternForRegex(generalDesc.getNationalNumberPattern());
   2467         maybeStripNationalPrefixAndCarrierCode(
   2468             potentialNationalNumber, defaultRegionMetadata, null /* Don't need the carrier code */);
   2469         Pattern possibleNumberPattern =
   2470             regexCache.getPatternForRegex(generalDesc.getPossibleNumberPattern());
   2471         // If the number was not valid before but is valid now, or if it was too long before, we
   2472         // consider the number with the country calling code stripped to be a better result and
   2473         // keep that instead.
   2474         if ((!validNumberPattern.matcher(fullNumber).matches() &&
   2475              validNumberPattern.matcher(potentialNationalNumber).matches()) ||
   2476              testNumberLengthAgainstPattern(possibleNumberPattern, fullNumber.toString())
   2477                   == ValidationResult.TOO_LONG) {
   2478           nationalNumber.append(potentialNationalNumber);
   2479           if (keepRawInput) {
   2480             phoneNumber.setCountryCodeSource(CountryCodeSource.FROM_NUMBER_WITHOUT_PLUS_SIGN);
   2481           }
   2482           phoneNumber.setCountryCode(defaultCountryCode);
   2483           return defaultCountryCode;
   2484         }
   2485       }
   2486     }
   2487     // No country calling code present.
   2488     phoneNumber.setCountryCode(0);
   2489     return 0;
   2490   }
   2491 
   2492   /**
   2493    * Strips the IDD from the start of the number if present. Helper function used by
   2494    * maybeStripInternationalPrefixAndNormalize.
   2495    */
   2496   private boolean parsePrefixAsIdd(Pattern iddPattern, StringBuilder number) {
   2497     Matcher m = iddPattern.matcher(number);
   2498     if (m.lookingAt()) {
   2499       int matchEnd = m.end();
   2500       // Only strip this if the first digit after the match is not a 0, since country calling codes
   2501       // cannot begin with 0.
   2502       Matcher digitMatcher = CAPTURING_DIGIT_PATTERN.matcher(number.substring(matchEnd));
   2503       if (digitMatcher.find()) {
   2504         String normalizedGroup = normalizeDigitsOnly(digitMatcher.group(1));
   2505         if (normalizedGroup.equals("0")) {
   2506           return false;
   2507         }
   2508       }
   2509       number.delete(0, matchEnd);
   2510       return true;
   2511     }
   2512     return false;
   2513   }
   2514 
   2515   /**
   2516    * Strips any international prefix (such as +, 00, 011) present in the number provided, normalizes
   2517    * the resulting number, and indicates if an international prefix was present.
   2518    *
   2519    * @param number  the non-normalized telephone number that we wish to strip any international
   2520    *     dialing prefix from.
   2521    * @param possibleIddPrefix  the international direct dialing prefix from the region we
   2522    *     think this number may be dialed in
   2523    * @return  the corresponding CountryCodeSource if an international dialing prefix could be
   2524    *     removed from the number, otherwise CountryCodeSource.FROM_DEFAULT_COUNTRY if the number did
   2525    *     not seem to be in international format.
   2526    */
   2527   // @VisibleForTesting
   2528   CountryCodeSource maybeStripInternationalPrefixAndNormalize(
   2529       StringBuilder number,
   2530       String possibleIddPrefix) {
   2531     if (number.length() == 0) {
   2532       return CountryCodeSource.FROM_DEFAULT_COUNTRY;
   2533     }
   2534     // Check to see if the number begins with one or more plus signs.
   2535     Matcher m = PLUS_CHARS_PATTERN.matcher(number);
   2536     if (m.lookingAt()) {
   2537       number.delete(0, m.end());
   2538       // Can now normalize the rest of the number since we've consumed the "+" sign at the start.
   2539       normalize(number);
   2540       return CountryCodeSource.FROM_NUMBER_WITH_PLUS_SIGN;
   2541     }
   2542     // Attempt to parse the first digits as an international prefix.
   2543     Pattern iddPattern = regexCache.getPatternForRegex(possibleIddPrefix);
   2544     normalize(number);
   2545     return parsePrefixAsIdd(iddPattern, number)
   2546            ? CountryCodeSource.FROM_NUMBER_WITH_IDD
   2547            : CountryCodeSource.FROM_DEFAULT_COUNTRY;
   2548   }
   2549 
   2550   /**
   2551    * Strips any national prefix (such as 0, 1) present in the number provided.
   2552    *
   2553    * @param number  the normalized telephone number that we wish to strip any national
   2554    *     dialing prefix from
   2555    * @param metadata  the metadata for the region that we think this number is from
   2556    * @param carrierCode  a place to insert the carrier code if one is extracted
   2557    * @return true if a national prefix or carrier code (or both) could be extracted.
   2558    */
   2559   // @VisibleForTesting
   2560   boolean maybeStripNationalPrefixAndCarrierCode(
   2561       StringBuilder number, PhoneMetadata metadata, StringBuilder carrierCode) {
   2562     int numberLength = number.length();
   2563     String possibleNationalPrefix = metadata.getNationalPrefixForParsing();
   2564     if (numberLength == 0 || possibleNationalPrefix.length() == 0) {
   2565       // Early return for numbers of zero length.
   2566       return false;
   2567     }
   2568     // Attempt to parse the first digits as a national prefix.
   2569     Matcher prefixMatcher = regexCache.getPatternForRegex(possibleNationalPrefix).matcher(number);
   2570     if (prefixMatcher.lookingAt()) {
   2571       Pattern nationalNumberRule =
   2572           regexCache.getPatternForRegex(metadata.getGeneralDesc().getNationalNumberPattern());
   2573       // Check if the original number is viable.
   2574       boolean isViableOriginalNumber = nationalNumberRule.matcher(number).matches();
   2575       // prefixMatcher.group(numOfGroups) == null implies nothing was captured by the capturing
   2576       // groups in possibleNationalPrefix; therefore, no transformation is necessary, and we just
   2577       // remove the national prefix.
   2578       int numOfGroups = prefixMatcher.groupCount();
   2579       String transformRule = metadata.getNationalPrefixTransformRule();
   2580       if (transformRule == null || transformRule.length() == 0 ||
   2581           prefixMatcher.group(numOfGroups) == null) {
   2582         // If the original number was viable, and the resultant number is not, we return.
   2583         if (isViableOriginalNumber &&
   2584             !nationalNumberRule.matcher(number.substring(prefixMatcher.end())).matches()) {
   2585           return false;
   2586         }
   2587         if (carrierCode != null && numOfGroups > 0 && prefixMatcher.group(numOfGroups) != null) {
   2588           carrierCode.append(prefixMatcher.group(1));
   2589         }
   2590         number.delete(0, prefixMatcher.end());
   2591         return true;
   2592       } else {
   2593         // Check that the resultant number is still viable. If not, return. Check this by copying
   2594         // the string buffer and making the transformation on the copy first.
   2595         StringBuilder transformedNumber = new StringBuilder(number);
   2596         transformedNumber.replace(0, numberLength, prefixMatcher.replaceFirst(transformRule));
   2597         if (isViableOriginalNumber &&
   2598             !nationalNumberRule.matcher(transformedNumber.toString()).matches()) {
   2599           return false;
   2600         }
   2601         if (carrierCode != null && numOfGroups > 1) {
   2602           carrierCode.append(prefixMatcher.group(1));
   2603         }
   2604         number.replace(0, number.length(), transformedNumber.toString());
   2605         return true;
   2606       }
   2607     }
   2608     return false;
   2609   }
   2610 
   2611   /**
   2612    * Strips any extension (as in, the part of the number dialled after the call is connected,
   2613    * usually indicated with extn, ext, x or similar) from the end of the number, and returns it.
   2614    *
   2615    * @param number  the non-normalized telephone number that we wish to strip the extension from
   2616    * @return        the phone extension
   2617    */
   2618   // @VisibleForTesting
   2619   String maybeStripExtension(StringBuilder number) {
   2620     Matcher m = EXTN_PATTERN.matcher(number);
   2621     // If we find a potential extension, and the number preceding this is a viable number, we assume
   2622     // it is an extension.
   2623     if (m.find() && isViablePhoneNumber(number.substring(0, m.start()))) {
   2624       // The numbers are captured into groups in the regular expression.
   2625       for (int i = 1, length = m.groupCount(); i <= length; i++) {
   2626         if (m.group(i) != null) {
   2627           // We go through the capturing groups until we find one that captured some digits. If none
   2628           // did, then we will return the empty string.
   2629           String extension = m.group(i);
   2630           number.delete(m.start(), number.length());
   2631           return extension;
   2632         }
   2633       }
   2634     }
   2635     return "";
   2636   }
   2637 
   2638   /**
   2639    * Checks to see that the region code used is valid, or if it is not valid, that the number to
   2640    * parse starts with a + symbol so that we can attempt to infer the region from the number.
   2641    * Returns false if it cannot use the region provided and the region cannot be inferred.
   2642    */
   2643   private boolean checkRegionForParsing(String numberToParse, String defaultRegion) {
   2644     if (!isValidRegionCode(defaultRegion)) {
   2645       // If the number is null or empty, we can't infer the region.
   2646       if ((numberToParse == null) || (numberToParse.length() == 0) ||
   2647           !PLUS_CHARS_PATTERN.matcher(numberToParse).lookingAt()) {
   2648         return false;
   2649       }
   2650     }
   2651     return true;
   2652   }
   2653 
   2654   /**
   2655    * Parses a string and returns it in proto buffer format. This method will throw a
   2656    * {@link com.google.i18n.phonenumbers.NumberParseException} if the number is not considered to be
   2657    * a possible number. Note that validation of whether the number is actually a valid number for a
   2658    * particular region is not performed. This can be done separately with {@link #isValidNumber}.
   2659    *
   2660    * @param numberToParse     number that we are attempting to parse. This can contain formatting
   2661    *                          such as +, ( and -, as well as a phone number extension. It can also
   2662    *                          be provided in RFC3966 format.
   2663    * @param defaultRegion     region that we are expecting the number to be from. This is only used
   2664    *                          if the number being parsed is not written in international format.
   2665    *                          The country_code for the number in this case would be stored as that
   2666    *                          of the default region supplied. If the number is guaranteed to
   2667    *                          start with a '+' followed by the country calling code, then
   2668    *                          "ZZ" or null can be supplied.
   2669    * @return                  a phone number proto buffer filled with the parsed number
   2670    * @throws NumberParseException  if the string is not considered to be a viable phone number or if
   2671    *                               no default region was supplied and the number is not in
   2672    *                               international format (does not start with +)
   2673    */
   2674   public PhoneNumber parse(String numberToParse, String defaultRegion)
   2675       throws NumberParseException {
   2676     PhoneNumber phoneNumber = new PhoneNumber();
   2677     parse(numberToParse, defaultRegion, phoneNumber);
   2678     return phoneNumber;
   2679   }
   2680 
   2681   /**
   2682    * Same as {@link #parse(String, String)}, but accepts mutable PhoneNumber as a parameter to
   2683    * decrease object creation when invoked many times.
   2684    */
   2685   public void parse(String numberToParse, String defaultRegion, PhoneNumber phoneNumber)
   2686       throws NumberParseException {
   2687     parseHelper(numberToParse, defaultRegion, false, true, phoneNumber);
   2688   }
   2689 
   2690   /**
   2691    * Parses a string and returns it in proto buffer format. This method differs from {@link #parse}
   2692    * in that it always populates the raw_input field of the protocol buffer with numberToParse as
   2693    * well as the country_code_source field.
   2694    *
   2695    * @param numberToParse     number that we are attempting to parse. This can contain formatting
   2696    *                          such as +, ( and -, as well as a phone number extension.
   2697    * @param defaultRegion     region that we are expecting the number to be from. This is only used
   2698    *                          if the number being parsed is not written in international format.
   2699    *                          The country calling code for the number in this case would be stored
   2700    *                          as that of the default region supplied.
   2701    * @return                  a phone number proto buffer filled with the parsed number
   2702    * @throws NumberParseException  if the string is not considered to be a viable phone number or if
   2703    *                               no default region was supplied
   2704    */
   2705   public PhoneNumber parseAndKeepRawInput(String numberToParse, String defaultRegion)
   2706       throws NumberParseException {
   2707     PhoneNumber phoneNumber = new PhoneNumber();
   2708     parseAndKeepRawInput(numberToParse, defaultRegion, phoneNumber);
   2709     return phoneNumber;
   2710   }
   2711 
   2712   /**
   2713    * Same as{@link #parseAndKeepRawInput(String, String)}, but accepts a mutable PhoneNumber as
   2714    * a parameter to decrease object creation when invoked many times.
   2715    */
   2716   public void parseAndKeepRawInput(String numberToParse, String defaultRegion,
   2717                                    PhoneNumber phoneNumber)
   2718       throws NumberParseException {
   2719     parseHelper(numberToParse, defaultRegion, true, true, phoneNumber);
   2720   }
   2721 
   2722   /**
   2723    * Returns an iterable over all {@link PhoneNumberMatch PhoneNumberMatches} in {@code text}. This
   2724    * is a shortcut for {@link #findNumbers(CharSequence, String, Leniency, long)
   2725    * getMatcher(text, defaultRegion, Leniency.VALID, Long.MAX_VALUE)}.
   2726    *
   2727    * @param text              the text to search for phone numbers, null for no text
   2728    * @param defaultRegion     region that we are expecting the number to be from. This is only used
   2729    *                          if the number being parsed is not written in international format. The
   2730    *                          country_code for the number in this case would be stored as that of
   2731    *                          the default region supplied. May be null if only international
   2732    *                          numbers are expected.
   2733    */
   2734   public Iterable<PhoneNumberMatch> findNumbers(CharSequence text, String defaultRegion) {
   2735     return findNumbers(text, defaultRegion, Leniency.VALID, Long.MAX_VALUE);
   2736   }
   2737 
   2738   /**
   2739    * Returns an iterable over all {@link PhoneNumberMatch PhoneNumberMatches} in {@code text}.
   2740    *
   2741    * @param text              the text to search for phone numbers, null for no text
   2742    * @param defaultRegion     region that we are expecting the number to be from. This is only used
   2743    *                          if the number being parsed is not written in international format. The
   2744    *                          country_code for the number in this case would be stored as that of
   2745    *                          the default region supplied. May be null if only international
   2746    *                          numbers are expected.
   2747    * @param leniency          the leniency to use when evaluating candidate phone numbers
   2748    * @param maxTries          the maximum number of invalid numbers to try before giving up on the
   2749    *                          text. This is to cover degenerate cases where the text has a lot of
   2750    *                          false positives in it. Must be {@code >= 0}.
   2751    */
   2752   public Iterable<PhoneNumberMatch> findNumbers(
   2753       final CharSequence text, final String defaultRegion, final Leniency leniency,
   2754       final long maxTries) {
   2755 
   2756     return new Iterable<PhoneNumberMatch>() {
   2757       @Override
   2758       public Iterator<PhoneNumberMatch> iterator() {
   2759         return new PhoneNumberMatcher(
   2760             PhoneNumberUtil.this, text, defaultRegion, leniency, maxTries);
   2761       }
   2762     };
   2763   }
   2764 
   2765   /**
   2766    * A helper function to set the values related to leading zeros in a PhoneNumber.
   2767    */
   2768   static void setItalianLeadingZerosForPhoneNumber(String nationalNumber, PhoneNumber phoneNumber) {
   2769     if (nationalNumber.length() > 1 && nationalNumber.charAt(0) == '0') {
   2770       phoneNumber.setItalianLeadingZero(true);
   2771       int numberOfLeadingZeros = 1;
   2772       // Note that if the national number is all "0"s, the last "0" is not counted as a leading
   2773       // zero.
   2774       while (numberOfLeadingZeros < nationalNumber.length() - 1 &&
   2775              nationalNumber.charAt(numberOfLeadingZeros) == '0') {
   2776         numberOfLeadingZeros++;
   2777       }
   2778       if (numberOfLeadingZeros != 1) {
   2779         phoneNumber.setNumberOfLeadingZeros(numberOfLeadingZeros);
   2780       }
   2781     }
   2782   }
   2783 
   2784   /**
   2785    * Parses a string and fills up the phoneNumber. This method is the same as the public
   2786    * parse() method, with the exception that it allows the default region to be null, for use by
   2787    * isNumberMatch(). checkRegion should be set to false if it is permitted for the default region
   2788    * to be null or unknown ("ZZ").
   2789    */
   2790   private void parseHelper(String numberToParse, String defaultRegion, boolean keepRawInput,
   2791                            boolean checkRegion, PhoneNumber phoneNumber)
   2792       throws NumberParseException {
   2793     if (numberToParse == null) {
   2794       throw new NumberParseException(NumberParseException.ErrorType.NOT_A_NUMBER,
   2795                                      "The phone number supplied was null.");
   2796     } else if (numberToParse.length() > MAX_INPUT_STRING_LENGTH) {
   2797       throw new NumberParseException(NumberParseException.ErrorType.TOO_LONG,
   2798                                      "The string supplied was too long to parse.");
   2799     }
   2800 
   2801     StringBuilder nationalNumber = new StringBuilder();
   2802     buildNationalNumberForParsing(numberToParse, nationalNumber);
   2803 
   2804     if (!isViablePhoneNumber(nationalNumber.toString())) {
   2805       throw new NumberParseException(NumberParseException.ErrorType.NOT_A_NUMBER,
   2806                                      "The string supplied did not seem to be a phone number.");
   2807     }
   2808 
   2809     // Check the region supplied is valid, or that the extracted number starts with some sort of +
   2810     // sign so the number's region can be determined.
   2811     if (checkRegion && !checkRegionForParsing(nationalNumber.toString(), defaultRegion)) {
   2812       throw new NumberParseException(NumberParseException.ErrorType.INVALID_COUNTRY_CODE,
   2813                                      "Missing or invalid default region.");
   2814     }
   2815 
   2816     if (keepRawInput) {
   2817       phoneNumber.setRawInput(numberToParse);
   2818     }
   2819     // Attempt to parse extension first, since it doesn't require region-specific data and we want
   2820     // to have the non-normalised number here.
   2821     String extension = maybeStripExtension(nationalNumber);
   2822     if (extension.length() > 0) {
   2823       phoneNumber.setExtension(extension);
   2824     }
   2825 
   2826     PhoneMetadata regionMetadata = getMetadataForRegion(defaultRegion);
   2827     // Check to see if the number is given in international format so we know whether this number is
   2828     // from the default region or not.
   2829     StringBuilder normalizedNationalNumber = new StringBuilder();
   2830     int countryCode = 0;
   2831     try {
   2832       // TODO: This method should really just take in the string buffer that has already
   2833       // been created, and just remove the prefix, rather than taking in a string and then
   2834       // outputting a string buffer.
   2835       countryCode = maybeExtractCountryCode(nationalNumber.toString(), regionMetadata,
   2836                                             normalizedNationalNumber, keepRawInput, phoneNumber);
   2837     } catch (NumberParseException e) {
   2838       Matcher matcher = PLUS_CHARS_PATTERN.matcher(nationalNumber.toString());
   2839       if (e.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE &&
   2840           matcher.lookingAt()) {
   2841         // Strip the plus-char, and try again.
   2842         countryCode = maybeExtractCountryCode(nationalNumber.substring(matcher.end()),
   2843                                               regionMetadata, normalizedNationalNumber,
   2844                                               keepRawInput, phoneNumber);
   2845         if (countryCode == 0) {
   2846           throw new NumberParseException(NumberParseException.ErrorType.INVALID_COUNTRY_CODE,
   2847                                          "Could not interpret numbers after plus-sign.");
   2848         }
   2849       } else {
   2850         throw new NumberParseException(e.getErrorType(), e.getMessage());
   2851       }
   2852     }
   2853     if (countryCode != 0) {
   2854       String phoneNumberRegion = getRegionCodeForCountryCode(countryCode);
   2855       if (!phoneNumberRegion.equals(defaultRegion)) {
   2856         // Metadata cannot be null because the country calling code is valid.
   2857         regionMetadata = getMetadataForRegionOrCallingCode(countryCode, phoneNumberRegion);
   2858       }
   2859     } else {
   2860       // If no extracted country calling code, use the region supplied instead. The national number
   2861       // is just the normalized version of the number we were given to parse.
   2862       normalize(nationalNumber);
   2863       normalizedNationalNumber.append(nationalNumber);
   2864       if (defaultRegion != null) {
   2865         countryCode = regionMetadata.getCountryCode();
   2866         phoneNumber.setCountryCode(countryCode);
   2867       } else if (keepRawInput) {
   2868         phoneNumber.clearCountryCodeSource();
   2869       }
   2870     }
   2871     if (normalizedNationalNumber.length() < MIN_LENGTH_FOR_NSN) {
   2872       throw new NumberParseException(NumberParseException.ErrorType.TOO_SHORT_NSN,
   2873                                      "The string supplied is too short to be a phone number.");
   2874     }
   2875     if (regionMetadata != null) {
   2876       StringBuilder carrierCode = new StringBuilder();
   2877       StringBuilder potentialNationalNumber = new StringBuilder(normalizedNationalNumber);
   2878       maybeStripNationalPrefixAndCarrierCode(potentialNationalNumber, regionMetadata, carrierCode);
   2879       // We require that the NSN remaining after stripping the national prefix and carrier code be
   2880       // of a possible length for the region. Otherwise, we don't do the stripping, since the
   2881       // original number could be a valid short number.
   2882       if (!isShorterThanPossibleNormalNumber(regionMetadata, potentialNationalNumber.toString())) {
   2883         normalizedNationalNumber = potentialNationalNumber;
   2884         if (keepRawInput) {
   2885           phoneNumber.setPreferredDomesticCarrierCode(carrierCode.toString());
   2886         }
   2887       }
   2888     }
   2889     int lengthOfNationalNumber = normalizedNationalNumber.length();
   2890     if (lengthOfNationalNumber < MIN_LENGTH_FOR_NSN) {
   2891       throw new NumberParseException(NumberParseException.ErrorType.TOO_SHORT_NSN,
   2892                                      "The string supplied is too short to be a phone number.");
   2893     }
   2894     if (lengthOfNationalNumber > MAX_LENGTH_FOR_NSN) {
   2895       throw new NumberParseException(NumberParseException.ErrorType.TOO_LONG,
   2896                                      "The string supplied is too long to be a phone number.");
   2897     }
   2898     setItalianLeadingZerosForPhoneNumber(normalizedNationalNumber.toString(), phoneNumber);
   2899     phoneNumber.setNationalNumber(Long.parseLong(normalizedNationalNumber.toString()));
   2900   }
   2901 
   2902   /**
   2903    * Converts numberToParse to a form that we can parse and write it to nationalNumber if it is
   2904    * written in RFC3966; otherwise extract a possible number out of it and write to nationalNumber.
   2905    */
   2906   private void buildNationalNumberForParsing(String numberToParse, StringBuilder nationalNumber) {
   2907     int indexOfPhoneContext = numberToParse.indexOf(RFC3966_PHONE_CONTEXT);
   2908     if (indexOfPhoneContext > 0) {
   2909       int phoneContextStart = indexOfPhoneContext + RFC3966_PHONE_CONTEXT.length();
   2910       // If the phone context contains a phone number prefix, we need to capture it, whereas domains
   2911       // will be ignored.
   2912       if (numberToParse.charAt(phoneContextStart) == PLUS_SIGN) {
   2913         // Additional parameters might follow the phone context. If so, we will remove them here
   2914         // because the parameters after phone context are not important for parsing the
   2915         // phone number.
   2916         int phoneContextEnd = numberToParse.indexOf(';', phoneContextStart);
   2917         if (phoneContextEnd > 0) {
   2918           nationalNumber.append(numberToParse.substring(phoneContextStart, phoneContextEnd));
   2919         } else {
   2920           nationalNumber.append(numberToParse.substring(phoneContextStart));
   2921         }
   2922       }
   2923 
   2924       // Now append everything between the "tel:" prefix and the phone-context. This should include
   2925       // the national number, an optional extension or isdn-subaddress component. Note we also
   2926       // handle the case when "tel:" is missing, as we have seen in some of the phone number inputs.
   2927       // In that case, we append everything from the beginning.
   2928       int indexOfRfc3966Prefix = numberToParse.indexOf(RFC3966_PREFIX);
   2929       int indexOfNationalNumber = (indexOfRfc3966Prefix >= 0) ?
   2930           indexOfRfc3966Prefix + RFC3966_PREFIX.length() : 0;
   2931       nationalNumber.append(numberToParse.substring(indexOfNationalNumber, indexOfPhoneContext));
   2932     } else {
   2933       // Extract a possible number from the string passed in (this strips leading characters that
   2934       // could not be the start of a phone number.)
   2935       nationalNumber.append(extractPossibleNumber(numberToParse));
   2936     }
   2937 
   2938     // Delete the isdn-subaddress and everything after it if it is present. Note extension won't
   2939     // appear at the same time with isdn-subaddress according to paragraph 5.3 of the RFC3966 spec,
   2940     int indexOfIsdn = nationalNumber.indexOf(RFC3966_ISDN_SUBADDRESS);
   2941     if (indexOfIsdn > 0) {
   2942       nationalNumber.delete(indexOfIsdn, nationalNumber.length());
   2943     }
   2944     // If both phone context and isdn-subaddress are absent but other parameters are present, the
   2945     // parameters are left in nationalNumber. This is because we are concerned about deleting
   2946     // content from a potential number string when there is no strong evidence that the number is
   2947     // actually written in RFC3966.
   2948   }
   2949 
   2950   /**
   2951    * Takes two phone numbers and compares them for equality.
   2952    *
   2953    * <p>Returns EXACT_MATCH if the country_code, NSN, presence of a leading zero for Italian numbers
   2954    * and any extension present are the same.
   2955    * Returns NSN_MATCH if either or both has no region specified, and the NSNs and extensions are
   2956    * the same.
   2957    * Returns SHORT_NSN_MATCH if either or both has no region specified, or the region specified is
   2958    * the same, and one NSN could be a shorter version of the other number. This includes the case
   2959    * where one has an extension specified, and the other does not.
   2960    * Returns NO_MATCH otherwise.
   2961    * For example, the numbers +1 345 657 1234 and 657 1234 are a SHORT_NSN_MATCH.
   2962    * The numbers +1 345 657 1234 and 345 657 are a NO_MATCH.
   2963    *
   2964    * @param firstNumberIn  first number to compare
   2965    * @param secondNumberIn  second number to compare
   2966    *
   2967    * @return  NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH or EXACT_MATCH depending on the level of equality
   2968    *     of the two numbers, described in the method definition.
   2969    */
   2970   public MatchType isNumberMatch(PhoneNumber firstNumberIn, PhoneNumber secondNumberIn) {
   2971     // Make copies of the phone number so that the numbers passed in are not edited.
   2972     PhoneNumber firstNumber = new PhoneNumber();
   2973     firstNumber.mergeFrom(firstNumberIn);
   2974     PhoneNumber secondNumber = new PhoneNumber();
   2975     secondNumber.mergeFrom(secondNumberIn);
   2976     // First clear raw_input, country_code_source and preferred_domestic_carrier_code fields and any
   2977     // empty-string extensions so that we can use the proto-buffer equality method.
   2978     firstNumber.clearRawInput();
   2979     firstNumber.clearCountryCodeSource();
   2980     firstNumber.clearPreferredDomesticCarrierCode();
   2981     secondNumber.clearRawInput();
   2982     secondNumber.clearCountryCodeSource();
   2983     secondNumber.clearPreferredDomesticCarrierCode();
   2984     if (firstNumber.hasExtension() &&
   2985         firstNumber.getExtension().length() == 0) {
   2986         firstNumber.clearExtension();
   2987     }
   2988     if (secondNumber.hasExtension() &&
   2989         secondNumber.getExtension().length() == 0) {
   2990         secondNumber.clearExtension();
   2991     }
   2992     // Early exit if both had extensions and these are different.
   2993     if (firstNumber.hasExtension() && secondNumber.hasExtension() &&
   2994         !firstNumber.getExtension().equals(secondNumber.getExtension())) {
   2995       return MatchType.NO_MATCH;
   2996     }
   2997     int firstNumberCountryCode = firstNumber.getCountryCode();
   2998     int secondNumberCountryCode = secondNumber.getCountryCode();
   2999     // Both had country_code specified.
   3000     if (firstNumberCountryCode != 0 && secondNumberCountryCode != 0) {
   3001       if (firstNumber.exactlySameAs(secondNumber)) {
   3002         return MatchType.EXACT_MATCH;
   3003       } else if (firstNumberCountryCode == secondNumberCountryCode &&
   3004                  isNationalNumberSuffixOfTheOther(firstNumber, secondNumber)) {
   3005         // A SHORT_NSN_MATCH occurs if there is a difference because of the presence or absence of
   3006         // an 'Italian leading zero', the presence or absence of an extension, or one NSN being a
   3007         // shorter variant of the other.
   3008         return MatchType.SHORT_NSN_MATCH;
   3009       }
   3010       // This is not a match.
   3011       return MatchType.NO_MATCH;
   3012     }
   3013     // Checks cases where one or both country_code fields were not specified. To make equality
   3014     // checks easier, we first set the country_code fields to be equal.
   3015     firstNumber.setCountryCode(secondNumberCountryCode);
   3016     // If all else was the same, then this is an NSN_MATCH.
   3017     if (firstNumber.exactlySameAs(secondNumber)) {
   3018       return MatchType.NSN_MATCH;
   3019     }
   3020     if (isNationalNumberSuffixOfTheOther(firstNumber, secondNumber)) {
   3021       return MatchType.SHORT_NSN_MATCH;
   3022     }
   3023     return MatchType.NO_MATCH;
   3024   }
   3025 
   3026   // Returns true when one national number is the suffix of the other or both are the same.
   3027   private boolean isNationalNumberSuffixOfTheOther(PhoneNumber firstNumber,
   3028                                                    PhoneNumber secondNumber) {
   3029     String firstNumberNationalNumber = String.valueOf(firstNumber.getNationalNumber());
   3030     String secondNumberNationalNumber = String.valueOf(secondNumber.getNationalNumber());
   3031     // Note that endsWith returns true if the numbers are equal.
   3032     return firstNumberNationalNumber.endsWith(secondNumberNationalNumber) ||
   3033            secondNumberNationalNumber.endsWith(firstNumberNationalNumber);
   3034   }
   3035 
   3036   /**
   3037    * Takes two phone numbers as strings and compares them for equality. This is a convenience
   3038    * wrapper for {@link #isNumberMatch(PhoneNumber, PhoneNumber)}. No default region is known.
   3039    *
   3040    * @param firstNumber  first number to compare. Can contain formatting, and can have country
   3041    *     calling code specified with + at the start.
   3042    * @param secondNumber  second number to compare. Can contain formatting, and can have country
   3043    *     calling code specified with + at the start.
   3044    * @return  NOT_A_NUMBER, NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH, EXACT_MATCH. See
   3045    *     {@link #isNumberMatch(PhoneNumber, PhoneNumber)} for more details.
   3046    */
   3047   public MatchType isNumberMatch(String firstNumber, String secondNumber) {
   3048     try {
   3049       PhoneNumber firstNumberAsProto = parse(firstNumber, UNKNOWN_REGION);
   3050       return isNumberMatch(firstNumberAsProto, secondNumber);
   3051     } catch (NumberParseException e) {
   3052       if (e.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE) {
   3053         try {
   3054           PhoneNumber secondNumberAsProto = parse(secondNumber, UNKNOWN_REGION);
   3055           return isNumberMatch(secondNumberAsProto, firstNumber);
   3056         } catch (NumberParseException e2) {
   3057           if (e2.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE) {
   3058             try {
   3059               PhoneNumber firstNumberProto = new PhoneNumber();
   3060               PhoneNumber secondNumberProto = new PhoneNumber();
   3061               parseHelper(firstNumber, null, false, false, firstNumberProto);
   3062               parseHelper(secondNumber, null, false, false, secondNumberProto);
   3063               return isNumberMatch(firstNumberProto, secondNumberProto);
   3064             } catch (NumberParseException e3) {
   3065               // Fall through and return MatchType.NOT_A_NUMBER.
   3066             }
   3067           }
   3068         }
   3069       }
   3070     }
   3071     // One or more of the phone numbers we are trying to match is not a viable phone number.
   3072     return MatchType.NOT_A_NUMBER;
   3073   }
   3074 
   3075   /**
   3076    * Takes two phone numbers and compares them for equality. This is a convenience wrapper for
   3077    * {@link #isNumberMatch(PhoneNumber, PhoneNumber)}. No default region is known.
   3078    *
   3079    * @param firstNumber  first number to compare in proto buffer format.
   3080    * @param secondNumber  second number to compare. Can contain formatting, and can have country
   3081    *     calling code specified with + at the start.
   3082    * @return  NOT_A_NUMBER, NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH, EXACT_MATCH. See
   3083    *     {@link #isNumberMatch(PhoneNumber, PhoneNumber)} for more details.
   3084    */
   3085   public MatchType isNumberMatch(PhoneNumber firstNumber, String secondNumber) {
   3086     // First see if the second number has an implicit country calling code, by attempting to parse
   3087     // it.
   3088     try {
   3089       PhoneNumber secondNumberAsProto = parse(secondNumber, UNKNOWN_REGION);
   3090       return isNumberMatch(firstNumber, secondNumberAsProto);
   3091     } catch (NumberParseException e) {
   3092       if (e.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE) {
   3093         // The second number has no country calling code. EXACT_MATCH is no longer possible.
   3094         // We parse it as if the region was the same as that for the first number, and if
   3095         // EXACT_MATCH is returned, we replace this with NSN_MATCH.
   3096         String firstNumberRegion = getRegionCodeForCountryCode(firstNumber.getCountryCode());
   3097         try {
   3098           if (!firstNumberRegion.equals(UNKNOWN_REGION)) {
   3099             PhoneNumber secondNumberWithFirstNumberRegion = parse(secondNumber, firstNumberRegion);
   3100             MatchType match = isNumberMatch(firstNumber, secondNumberWithFirstNumberRegion);
   3101             if (match == MatchType.EXACT_MATCH) {
   3102               return MatchType.NSN_MATCH;
   3103             }
   3104             return match;
   3105           } else {
   3106             // If the first number didn't have a valid country calling code, then we parse the
   3107             // second number without one as well.
   3108             PhoneNumber secondNumberProto = new PhoneNumber();
   3109             parseHelper(secondNumber, null, false, false, secondNumberProto);
   3110             return isNumberMatch(firstNumber, secondNumberProto);
   3111           }
   3112         } catch (NumberParseException e2) {
   3113           // Fall-through to return NOT_A_NUMBER.
   3114         }
   3115       }
   3116     }
   3117     // One or more of the phone numbers we are trying to match is not a viable phone number.
   3118     return MatchType.NOT_A_NUMBER;
   3119   }
   3120 
   3121   /**
   3122    * Returns true if the number can be dialled from outside the region, or unknown. If the number
   3123    * can only be dialled from within the region, returns false. Does not check the number is a valid
   3124    * number. Note that, at the moment, this method does not handle short numbers.
   3125    * TODO: Make this method public when we have enough metadata to make it worthwhile.
   3126    *
   3127    * @param number  the phone-number for which we want to know whether it is diallable from
   3128    *     outside the region
   3129    */
   3130   // @VisibleForTesting
   3131   boolean canBeInternationallyDialled(PhoneNumber number) {
   3132     PhoneMetadata metadata = getMetadataForRegion(getRegionCodeForNumber(number));
   3133     if (metadata == null) {
   3134       // Note numbers belonging to non-geographical entities (e.g. +800 numbers) are always
   3135       // internationally diallable, and will be caught here.
   3136       return true;
   3137     }
   3138     String nationalSignificantNumber = getNationalSignificantNumber(number);
   3139     return !isNumberMatchingDesc(nationalSignificantNumber, metadata.getNoInternationalDialling());
   3140   }
   3141 
   3142   /**
   3143    * Returns true if the supplied region supports mobile number portability. Returns false for
   3144    * invalid, unknown or regions that don't support mobile number portability.
   3145    *
   3146    * @param regionCode  the region for which we want to know whether it supports mobile number
   3147    *                    portability or not.
   3148    */
   3149   public boolean isMobileNumberPortableRegion(String regionCode) {
   3150     PhoneMetadata metadata = getMetadataForRegion(regionCode);
   3151     if (metadata == null) {
   3152       logger.log(Level.WARNING, "Invalid or unknown region code provided: " + regionCode);
   3153       return false;
   3154     }
   3155     return metadata.isMobileNumberPortableRegion();
   3156   }
   3157 }
   3158