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
17package com.android.i18n.phonenumbers;
18
19import com.android.i18n.phonenumbers.Phonemetadata.NumberFormat;
20import com.android.i18n.phonenumbers.Phonemetadata.PhoneMetadata;
21import com.android.i18n.phonenumbers.Phonemetadata.PhoneMetadataCollection;
22import com.android.i18n.phonenumbers.Phonemetadata.PhoneNumberDesc;
23import com.android.i18n.phonenumbers.Phonenumber.PhoneNumber;
24import com.android.i18n.phonenumbers.Phonenumber.PhoneNumber.CountryCodeSource;
25
26import java.io.IOException;
27import java.io.InputStream;
28import java.io.ObjectInput;
29import java.io.ObjectInputStream;
30import java.util.ArrayList;
31import java.util.Arrays;
32import java.util.Collections;
33import java.util.HashMap;
34import java.util.HashSet;
35import java.util.Iterator;
36import java.util.List;
37import java.util.Map;
38import java.util.Set;
39import java.util.logging.Level;
40import java.util.logging.Logger;
41import java.util.regex.Matcher;
42import java.util.regex.Pattern;
43
44/**
45 * Utility for international phone numbers. Functionality includes formatting, parsing and
46 * validation.
47 *
48 * <p>If you use this library, and want to be notified about important changes, please sign up to
49 * our <a href="http://groups.google.com/group/libphonenumber-discuss/about">mailing list</a>.
50 *
51 * NOTE: A lot of methods in this class require Region Code strings. These must be provided using
52 * ISO 3166-1 two-letter country-code format. These should be in upper-case. The list of the codes
53 * can be found here:
54 * http://www.iso.org/iso/country_codes/iso_3166_code_lists/country_names_and_code_elements.htm
55 *
56 * @author Shaopeng Jia
57 */
58public class PhoneNumberUtil {
59  // @VisibleForTesting
60  static final MetadataLoader DEFAULT_METADATA_LOADER = new MetadataLoader() {
61    public InputStream loadMetadata(String metadataFileName) {
62      return PhoneNumberUtil.class.getResourceAsStream(metadataFileName);
63    }
64  };
65
66  private static final Logger logger = Logger.getLogger(PhoneNumberUtil.class.getName());
67
68  /** Flags to use when compiling regular expressions for phone numbers. */
69  static final int REGEX_FLAGS = Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE;
70  // The minimum and maximum length of the national significant number.
71  private static final int MIN_LENGTH_FOR_NSN = 2;
72  // The ITU says the maximum length should be 15, but we have found longer numbers in Germany.
73  static final int MAX_LENGTH_FOR_NSN = 17;
74  // The maximum length of the country calling code.
75  static final int MAX_LENGTH_COUNTRY_CODE = 3;
76  // We don't allow input strings for parsing to be longer than 250 chars. This prevents malicious
77  // input from overflowing the regular-expression engine.
78  private static final int MAX_INPUT_STRING_LENGTH = 250;
79
80  private static final String META_DATA_FILE_PREFIX =
81      "/com/android/i18n/phonenumbers/data/PhoneNumberMetadataProto";
82
83  // Region-code for the unknown region.
84  private static final String UNKNOWN_REGION = "ZZ";
85
86  private static final int NANPA_COUNTRY_CODE = 1;
87
88  // The prefix that needs to be inserted in front of a Colombian landline number when dialed from
89  // a mobile phone in Colombia.
90  private static final String COLOMBIA_MOBILE_TO_FIXED_LINE_PREFIX = "3";
91
92  // Map of country calling codes that use a mobile token before the area code. One example of when
93  // this is relevant is when determining the length of the national destination code, which should
94  // be the length of the area code plus the length of the mobile token.
95  private static final Map<Integer, String> MOBILE_TOKEN_MAPPINGS;
96
97  // The PLUS_SIGN signifies the international prefix.
98  static final char PLUS_SIGN = '+';
99
100  private static final char STAR_SIGN = '*';
101
102  private static final String RFC3966_EXTN_PREFIX = ";ext=";
103  private static final String RFC3966_PREFIX = "tel:";
104  private static final String RFC3966_PHONE_CONTEXT = ";phone-context=";
105  private static final String RFC3966_ISDN_SUBADDRESS = ";isub=";
106
107  // A map that contains characters that are essential when dialling. That means any of the
108  // characters in this map must not be removed from a number when dialling, otherwise the call
109  // will not reach the intended destination.
110  private static final Map<Character, Character> DIALLABLE_CHAR_MAPPINGS;
111
112  // Only upper-case variants of alpha characters are stored.
113  private static final Map<Character, Character> ALPHA_MAPPINGS;
114
115  // For performance reasons, amalgamate both into one map.
116  private static final Map<Character, Character> ALPHA_PHONE_MAPPINGS;
117
118  // Separate map of all symbols that we wish to retain when formatting alpha numbers. This
119  // includes digits, ASCII letters and number grouping symbols such as "-" and " ".
120  private static final Map<Character, Character> ALL_PLUS_NUMBER_GROUPING_SYMBOLS;
121
122  static {
123    HashMap<Integer, String> mobileTokenMap = new HashMap<Integer, String>();
124    mobileTokenMap.put(52, "1");
125    mobileTokenMap.put(54, "9");
126    MOBILE_TOKEN_MAPPINGS = Collections.unmodifiableMap(mobileTokenMap);
127
128    // Simple ASCII digits map used to populate ALPHA_PHONE_MAPPINGS and
129    // ALL_PLUS_NUMBER_GROUPING_SYMBOLS.
130    HashMap<Character, Character> asciiDigitMappings = new HashMap<Character, Character>();
131    asciiDigitMappings.put('0', '0');
132    asciiDigitMappings.put('1', '1');
133    asciiDigitMappings.put('2', '2');
134    asciiDigitMappings.put('3', '3');
135    asciiDigitMappings.put('4', '4');
136    asciiDigitMappings.put('5', '5');
137    asciiDigitMappings.put('6', '6');
138    asciiDigitMappings.put('7', '7');
139    asciiDigitMappings.put('8', '8');
140    asciiDigitMappings.put('9', '9');
141
142    HashMap<Character, Character> alphaMap = new HashMap<Character, Character>(40);
143    alphaMap.put('A', '2');
144    alphaMap.put('B', '2');
145    alphaMap.put('C', '2');
146    alphaMap.put('D', '3');
147    alphaMap.put('E', '3');
148    alphaMap.put('F', '3');
149    alphaMap.put('G', '4');
150    alphaMap.put('H', '4');
151    alphaMap.put('I', '4');
152    alphaMap.put('J', '5');
153    alphaMap.put('K', '5');
154    alphaMap.put('L', '5');
155    alphaMap.put('M', '6');
156    alphaMap.put('N', '6');
157    alphaMap.put('O', '6');
158    alphaMap.put('P', '7');
159    alphaMap.put('Q', '7');
160    alphaMap.put('R', '7');
161    alphaMap.put('S', '7');
162    alphaMap.put('T', '8');
163    alphaMap.put('U', '8');
164    alphaMap.put('V', '8');
165    alphaMap.put('W', '9');
166    alphaMap.put('X', '9');
167    alphaMap.put('Y', '9');
168    alphaMap.put('Z', '9');
169    ALPHA_MAPPINGS = Collections.unmodifiableMap(alphaMap);
170
171    HashMap<Character, Character> combinedMap = new HashMap<Character, Character>(100);
172    combinedMap.putAll(ALPHA_MAPPINGS);
173    combinedMap.putAll(asciiDigitMappings);
174    ALPHA_PHONE_MAPPINGS = Collections.unmodifiableMap(combinedMap);
175
176    HashMap<Character, Character> diallableCharMap = new HashMap<Character, Character>();
177    diallableCharMap.putAll(asciiDigitMappings);
178    diallableCharMap.put(PLUS_SIGN, PLUS_SIGN);
179    diallableCharMap.put('*', '*');
180    DIALLABLE_CHAR_MAPPINGS = Collections.unmodifiableMap(diallableCharMap);
181
182    HashMap<Character, Character> allPlusNumberGroupings = new HashMap<Character, Character>();
183    // Put (lower letter -> upper letter) and (upper letter -> upper letter) mappings.
184    for (char c : ALPHA_MAPPINGS.keySet()) {
185      allPlusNumberGroupings.put(Character.toLowerCase(c), c);
186      allPlusNumberGroupings.put(c, c);
187    }
188    allPlusNumberGroupings.putAll(asciiDigitMappings);
189    // Put grouping symbols.
190    allPlusNumberGroupings.put('-', '-');
191    allPlusNumberGroupings.put('\uFF0D', '-');
192    allPlusNumberGroupings.put('\u2010', '-');
193    allPlusNumberGroupings.put('\u2011', '-');
194    allPlusNumberGroupings.put('\u2012', '-');
195    allPlusNumberGroupings.put('\u2013', '-');
196    allPlusNumberGroupings.put('\u2014', '-');
197    allPlusNumberGroupings.put('\u2015', '-');
198    allPlusNumberGroupings.put('\u2212', '-');
199    allPlusNumberGroupings.put('/', '/');
200    allPlusNumberGroupings.put('\uFF0F', '/');
201    allPlusNumberGroupings.put(' ', ' ');
202    allPlusNumberGroupings.put('\u3000', ' ');
203    allPlusNumberGroupings.put('\u2060', ' ');
204    allPlusNumberGroupings.put('.', '.');
205    allPlusNumberGroupings.put('\uFF0E', '.');
206    ALL_PLUS_NUMBER_GROUPING_SYMBOLS = Collections.unmodifiableMap(allPlusNumberGroupings);
207  }
208
209  // Pattern that makes it easy to distinguish whether a region has a unique international dialing
210  // prefix or not. If a region has a unique international prefix (e.g. 011 in USA), it will be
211  // represented as a string that contains a sequence of ASCII digits. If there are multiple
212  // available international prefixes in a region, they will be represented as a regex string that
213  // always contains character(s) other than ASCII digits.
214  // Note this regex also includes tilde, which signals waiting for the tone.
215  private static final Pattern UNIQUE_INTERNATIONAL_PREFIX =
216      Pattern.compile("[\\d]+(?:[~\u2053\u223C\uFF5E][\\d]+)?");
217
218  // Regular expression of acceptable punctuation found in phone numbers. This excludes punctuation
219  // found as a leading character only.
220  // This consists of dash characters, white space characters, full stops, slashes,
221  // square brackets, parentheses and tildes. It also includes the letter 'x' as that is found as a
222  // placeholder for carrier information in some phone numbers. Full-width variants are also
223  // present.
224  static final String VALID_PUNCTUATION = "-x\u2010-\u2015\u2212\u30FC\uFF0D-\uFF0F " +
225      "\u00A0\u00AD\u200B\u2060\u3000()\uFF08\uFF09\uFF3B\uFF3D.\\[\\]/~\u2053\u223C\uFF5E";
226
227  private static final String DIGITS = "\\p{Nd}";
228  // We accept alpha characters in phone numbers, ASCII only, upper and lower case.
229  private static final String VALID_ALPHA =
230      Arrays.toString(ALPHA_MAPPINGS.keySet().toArray()).replaceAll("[, \\[\\]]", "") +
231      Arrays.toString(ALPHA_MAPPINGS.keySet().toArray()).toLowerCase().replaceAll("[, \\[\\]]", "");
232  static final String PLUS_CHARS = "+\uFF0B";
233  static final Pattern PLUS_CHARS_PATTERN = Pattern.compile("[" + PLUS_CHARS + "]+");
234  private static final Pattern SEPARATOR_PATTERN = Pattern.compile("[" + VALID_PUNCTUATION + "]+");
235  private static final Pattern CAPTURING_DIGIT_PATTERN = Pattern.compile("(" + DIGITS + ")");
236
237  // Regular expression of acceptable characters that may start a phone number for the purposes of
238  // parsing. This allows us to strip away meaningless prefixes to phone numbers that may be
239  // mistakenly given to us. This consists of digits, the plus symbol and arabic-indic digits. This
240  // does not contain alpha characters, although they may be used later in the number. It also does
241  // not include other punctuation, as this will be stripped later during parsing and is of no
242  // information value when parsing a number.
243  private static final String VALID_START_CHAR = "[" + PLUS_CHARS + DIGITS + "]";
244  private static final Pattern VALID_START_CHAR_PATTERN = Pattern.compile(VALID_START_CHAR);
245
246  // Regular expression of characters typically used to start a second phone number for the purposes
247  // of parsing. This allows us to strip off parts of the number that are actually the start of
248  // another number, such as for: (530) 583-6985 x302/x2303 -> the second extension here makes this
249  // actually two phone numbers, (530) 583-6985 x302 and (530) 583-6985 x2303. We remove the second
250  // extension so that the first number is parsed correctly.
251  private static final String SECOND_NUMBER_START = "[\\\\/] *x";
252  static final Pattern SECOND_NUMBER_START_PATTERN = Pattern.compile(SECOND_NUMBER_START);
253
254  // Regular expression of trailing characters that we want to remove. We remove all characters that
255  // are not alpha or numerical characters. The hash character is retained here, as it may signify
256  // the previous block was an extension.
257  private static final String UNWANTED_END_CHARS = "[[\\P{N}&&\\P{L}]&&[^#]]+$";
258  static final Pattern UNWANTED_END_CHAR_PATTERN = Pattern.compile(UNWANTED_END_CHARS);
259
260  // We use this pattern to check if the phone number has at least three letters in it - if so, then
261  // we treat it as a number where some phone-number digits are represented by letters.
262  private static final Pattern VALID_ALPHA_PHONE_PATTERN = Pattern.compile("(?:.*?[A-Za-z]){3}.*");
263
264  // Regular expression of viable phone numbers. This is location independent. Checks we have at
265  // least three leading digits, and only valid punctuation, alpha characters and
266  // digits in the phone number. Does not include extension data.
267  // The symbol 'x' is allowed here as valid punctuation since it is often used as a placeholder for
268  // carrier codes, for example in Brazilian phone numbers. We also allow multiple "+" characters at
269  // the start.
270  // Corresponds to the following:
271  // [digits]{minLengthNsn}|
272  // plus_sign*(([punctuation]|[star])*[digits]){3,}([punctuation]|[star]|[digits]|[alpha])*
273  //
274  // The first reg-ex is to allow short numbers (two digits long) to be parsed if they are entered
275  // as "15" etc, but only if there is no punctuation in them. The second expression restricts the
276  // number of digits to three or more, but then allows them to be in international form, and to
277  // have alpha-characters and punctuation.
278  //
279  // Note VALID_PUNCTUATION starts with a -, so must be the first in the range.
280  private static final String VALID_PHONE_NUMBER =
281      DIGITS + "{" + MIN_LENGTH_FOR_NSN + "}" + "|" +
282      "[" + PLUS_CHARS + "]*+(?:[" + VALID_PUNCTUATION + STAR_SIGN + "]*" + DIGITS + "){3,}[" +
283      VALID_PUNCTUATION + STAR_SIGN + VALID_ALPHA + DIGITS + "]*";
284
285  // Default extension prefix to use when formatting. This will be put in front of any extension
286  // component of the number, after the main national number is formatted. For example, if you wish
287  // the default extension formatting to be " extn: 3456", then you should specify " extn: " here
288  // as the default extension prefix. This can be overridden by region-specific preferences.
289  private static final String DEFAULT_EXTN_PREFIX = " ext. ";
290
291  // Pattern to capture digits used in an extension. Places a maximum length of "7" for an
292  // extension.
293  private static final String CAPTURING_EXTN_DIGITS = "(" + DIGITS + "{1,7})";
294  // Regexp of all possible ways to write extensions, for use when parsing. This will be run as a
295  // case-insensitive regexp match. Wide character versions are also provided after each ASCII
296  // version.
297  private static final String EXTN_PATTERNS_FOR_PARSING;
298  static final String EXTN_PATTERNS_FOR_MATCHING;
299  static {
300    // One-character symbols that can be used to indicate an extension.
301    String singleExtnSymbolsForMatching = "x\uFF58#\uFF03~\uFF5E";
302    // For parsing, we are slightly more lenient in our interpretation than for matching. Here we
303    // allow a "comma" as a possible extension indicator. When matching, this is hardly ever used to
304    // indicate this.
305    String singleExtnSymbolsForParsing = "," + singleExtnSymbolsForMatching;
306
307    EXTN_PATTERNS_FOR_PARSING = createExtnPattern(singleExtnSymbolsForParsing);
308    EXTN_PATTERNS_FOR_MATCHING = createExtnPattern(singleExtnSymbolsForMatching);
309  }
310
311  /**
312   * Helper initialiser method to create the regular-expression pattern to match extensions,
313   * allowing the one-char extension symbols provided by {@code singleExtnSymbols}.
314   */
315  private static String createExtnPattern(String singleExtnSymbols) {
316    // There are three regular expressions here. The first covers RFC 3966 format, where the
317    // extension is added using ";ext=". The second more generic one starts with optional white
318    // space and ends with an optional full stop (.), followed by zero or more spaces/tabs and then
319    // the numbers themselves. The other one covers the special case of American numbers where the
320    // extension is written with a hash at the end, such as "- 503#".
321    // Note that the only capturing groups should be around the digits that you want to capture as
322    // part of the extension, or else parsing will fail!
323    // Canonical-equivalence doesn't seem to be an option with Android java, so we allow two options
324    // for representing the accented o - the character itself, and one in the unicode decomposed
325    // form with the combining acute accent.
326    return (RFC3966_EXTN_PREFIX + CAPTURING_EXTN_DIGITS + "|" + "[ \u00A0\\t,]*" +
327            "(?:e?xt(?:ensi(?:o\u0301?|\u00F3))?n?|\uFF45?\uFF58\uFF54\uFF4E?|" +
328            "[" + singleExtnSymbols + "]|int|anexo|\uFF49\uFF4E\uFF54)" +
329            "[:\\.\uFF0E]?[ \u00A0\\t,-]*" + CAPTURING_EXTN_DIGITS + "#?|" +
330            "[- ]+(" + DIGITS + "{1,5})#");
331  }
332
333  // Regexp of all known extension prefixes used by different regions followed by 1 or more valid
334  // digits, for use when parsing.
335  private static final Pattern EXTN_PATTERN =
336      Pattern.compile("(?:" + EXTN_PATTERNS_FOR_PARSING + ")$", REGEX_FLAGS);
337
338  // We append optionally the extension pattern to the end here, as a valid phone number may
339  // have an extension prefix appended, followed by 1 or more digits.
340  private static final Pattern VALID_PHONE_NUMBER_PATTERN =
341      Pattern.compile(VALID_PHONE_NUMBER + "(?:" + EXTN_PATTERNS_FOR_PARSING + ")?", REGEX_FLAGS);
342
343  static final Pattern NON_DIGITS_PATTERN = Pattern.compile("(\\D+)");
344
345  // The FIRST_GROUP_PATTERN was originally set to $1 but there are some countries for which the
346  // first group is not used in the national pattern (e.g. Argentina) so the $1 group does not match
347  // correctly.  Therefore, we use \d, so that the first group actually used in the pattern will be
348  // matched.
349  private static final Pattern FIRST_GROUP_PATTERN = Pattern.compile("(\\$\\d)");
350  private static final Pattern NP_PATTERN = Pattern.compile("\\$NP");
351  private static final Pattern FG_PATTERN = Pattern.compile("\\$FG");
352  private static final Pattern CC_PATTERN = Pattern.compile("\\$CC");
353
354  // A pattern that is used to determine if the national prefix formatting rule has the first group
355  // only, i.e., does not start with the national prefix. Note that the pattern explicitly allows
356  // for unbalanced parentheses.
357  private static final Pattern FIRST_GROUP_ONLY_PREFIX_PATTERN = Pattern.compile("\\(?\\$1\\)?");
358
359  private static PhoneNumberUtil instance = null;
360
361  public static final String REGION_CODE_FOR_NON_GEO_ENTITY = "001";
362
363  /**
364   * INTERNATIONAL and NATIONAL formats are consistent with the definition in ITU-T Recommendation
365   * E123. For example, the number of the Google Switzerland office will be written as
366   * "+41 44 668 1800" in INTERNATIONAL format, and as "044 668 1800" in NATIONAL format.
367   * E164 format is as per INTERNATIONAL format but with no formatting applied, e.g.
368   * "+41446681800". RFC3966 is as per INTERNATIONAL format, but with all spaces and other
369   * separating symbols replaced with a hyphen, and with any phone number extension appended with
370   * ";ext=". It also will have a prefix of "tel:" added, e.g. "tel:+41-44-668-1800".
371   *
372   * Note: If you are considering storing the number in a neutral format, you are highly advised to
373   * use the PhoneNumber class.
374   */
375  public enum PhoneNumberFormat {
376    E164,
377    INTERNATIONAL,
378    NATIONAL,
379    RFC3966
380  }
381
382  /**
383   * Type of phone numbers.
384   */
385  public enum PhoneNumberType {
386    FIXED_LINE,
387    MOBILE,
388    // In some regions (e.g. the USA), it is impossible to distinguish between fixed-line and
389    // mobile numbers by looking at the phone number itself.
390    FIXED_LINE_OR_MOBILE,
391    // Freephone lines
392    TOLL_FREE,
393    PREMIUM_RATE,
394    // The cost of this call is shared between the caller and the recipient, and is hence typically
395    // less than PREMIUM_RATE calls. See // http://en.wikipedia.org/wiki/Shared_Cost_Service for
396    // more information.
397    SHARED_COST,
398    // Voice over IP numbers. This includes TSoIP (Telephony Service over IP).
399    VOIP,
400    // A personal number is associated with a particular person, and may be routed to either a
401    // MOBILE or FIXED_LINE number. Some more information can be found here:
402    // http://en.wikipedia.org/wiki/Personal_Numbers
403    PERSONAL_NUMBER,
404    PAGER,
405    // Used for "Universal Access Numbers" or "Company Numbers". They may be further routed to
406    // specific offices, but allow one number to be used for a company.
407    UAN,
408    // Used for "Voice Mail Access Numbers".
409    VOICEMAIL,
410    // A phone number is of type UNKNOWN when it does not fit any of the known patterns for a
411    // specific region.
412    UNKNOWN
413  }
414
415  /**
416   * Types of phone number matches. See detailed description beside the isNumberMatch() method.
417   */
418  public enum MatchType {
419    NOT_A_NUMBER,
420    NO_MATCH,
421    SHORT_NSN_MATCH,
422    NSN_MATCH,
423    EXACT_MATCH,
424  }
425
426  /**
427   * Possible outcomes when testing if a PhoneNumber is possible.
428   */
429  public enum ValidationResult {
430    IS_POSSIBLE,
431    INVALID_COUNTRY_CODE,
432    TOO_SHORT,
433    TOO_LONG,
434  }
435
436  /**
437   * Leniency when {@linkplain PhoneNumberUtil#findNumbers finding} potential phone numbers in text
438   * segments. The levels here are ordered in increasing strictness.
439   */
440  public enum Leniency {
441    /**
442     * Phone numbers accepted are {@linkplain PhoneNumberUtil#isPossibleNumber(PhoneNumber)
443     * possible}, but not necessarily {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid}.
444     */
445    POSSIBLE {
446      @Override
447      boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
448        return util.isPossibleNumber(number);
449      }
450    },
451    /**
452     * Phone numbers accepted are {@linkplain PhoneNumberUtil#isPossibleNumber(PhoneNumber)
453     * possible} and {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid}. Numbers written
454     * in national format must have their national-prefix present if it is usually written for a
455     * number of this type.
456     */
457    VALID {
458      @Override
459      boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
460        if (!util.isValidNumber(number) ||
461            !PhoneNumberMatcher.containsOnlyValidXChars(number, candidate, util)) {
462          return false;
463        }
464        return PhoneNumberMatcher.isNationalPrefixPresentIfRequired(number, util);
465      }
466    },
467    /**
468     * Phone numbers accepted are {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid} and
469     * are grouped in a possible way for this locale. For example, a US number written as
470     * "65 02 53 00 00" and "650253 0000" are not accepted at this leniency level, whereas
471     * "650 253 0000", "650 2530000" or "6502530000" are.
472     * Numbers with more than one '/' symbol in the national significant number are also dropped at
473     * this level.
474     * <p>
475     * Warning: This level might result in lower coverage especially for regions outside of country
476     * code "+1". If you are not sure about which level to use, email the discussion group
477     * libphonenumber-discuss@googlegroups.com.
478     */
479    STRICT_GROUPING {
480      @Override
481      boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
482        if (!util.isValidNumber(number) ||
483            !PhoneNumberMatcher.containsOnlyValidXChars(number, candidate, util) ||
484            PhoneNumberMatcher.containsMoreThanOneSlashInNationalNumber(number, candidate) ||
485            !PhoneNumberMatcher.isNationalPrefixPresentIfRequired(number, util)) {
486          return false;
487        }
488        return PhoneNumberMatcher.checkNumberGroupingIsValid(
489            number, candidate, util, new PhoneNumberMatcher.NumberGroupingChecker() {
490              public boolean checkGroups(PhoneNumberUtil util, PhoneNumber number,
491                                         StringBuilder normalizedCandidate,
492                                         String[] expectedNumberGroups) {
493                return PhoneNumberMatcher.allNumberGroupsRemainGrouped(
494                    util, number, normalizedCandidate, expectedNumberGroups);
495              }
496            });
497      }
498    },
499    /**
500     * Phone numbers accepted are {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid} and
501     * are grouped in the same way that we would have formatted it, or as a single block. For
502     * example, a US number written as "650 2530000" is not accepted at this leniency level, whereas
503     * "650 253 0000" or "6502530000" are.
504     * Numbers with more than one '/' symbol are also dropped at this level.
505     * <p>
506     * Warning: This level might result in lower coverage especially for regions outside of country
507     * code "+1". If you are not sure about which level to use, email the discussion group
508     * libphonenumber-discuss@googlegroups.com.
509     */
510    EXACT_GROUPING {
511      @Override
512      boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util) {
513        if (!util.isValidNumber(number) ||
514            !PhoneNumberMatcher.containsOnlyValidXChars(number, candidate, util) ||
515            PhoneNumberMatcher.containsMoreThanOneSlashInNationalNumber(number, candidate) ||
516            !PhoneNumberMatcher.isNationalPrefixPresentIfRequired(number, util)) {
517          return false;
518        }
519        return PhoneNumberMatcher.checkNumberGroupingIsValid(
520            number, candidate, util, new PhoneNumberMatcher.NumberGroupingChecker() {
521              public boolean checkGroups(PhoneNumberUtil util, PhoneNumber number,
522                                         StringBuilder normalizedCandidate,
523                                         String[] expectedNumberGroups) {
524                return PhoneNumberMatcher.allNumberGroupsAreExactlyPresent(
525                    util, number, normalizedCandidate, expectedNumberGroups);
526              }
527            });
528      }
529    };
530
531    /** Returns true if {@code number} is a verified number according to this leniency. */
532    abstract boolean verify(PhoneNumber number, String candidate, PhoneNumberUtil util);
533  }
534
535  // A mapping from a country calling code to the region codes which denote the region represented
536  // by that country calling code. In the case of multiple regions sharing a calling code, such as
537  // the NANPA regions, the one indicated with "isMainCountryForCode" in the metadata should be
538  // first.
539  private final Map<Integer, List<String>> countryCallingCodeToRegionCodeMap;
540
541  // The set of regions that share country calling code 1.
542  // There are roughly 26 regions.
543  // We set the initial capacity of the HashSet to 35 to offer a load factor of roughly 0.75.
544  private final Set<String> nanpaRegions = new HashSet<String>(35);
545
546  // A mapping from a region code to the PhoneMetadata for that region.
547  // Note: Synchronization, though only needed for the Android version of the library, is used in
548  // all versions for consistency.
549  private final Map<String, PhoneMetadata> regionToMetadataMap =
550      Collections.synchronizedMap(new HashMap<String, PhoneMetadata>());
551
552  // A mapping from a country calling code for a non-geographical entity to the PhoneMetadata for
553  // that country calling code. Examples of the country calling codes include 800 (International
554  // Toll Free Service) and 808 (International Shared Cost Service).
555  // Note: Synchronization, though only needed for the Android version of the library, is used in
556  // all versions for consistency.
557  private final Map<Integer, PhoneMetadata> countryCodeToNonGeographicalMetadataMap =
558      Collections.synchronizedMap(new HashMap<Integer, PhoneMetadata>());
559
560  // A cache for frequently used region-specific regular expressions.
561  // The initial capacity is set to 100 as this seems to be an optimal value for Android, based on
562  // performance measurements.
563  private final RegexCache regexCache = new RegexCache(100);
564
565  // The set of regions the library supports.
566  // There are roughly 240 of them and we set the initial capacity of the HashSet to 320 to offer a
567  // load factor of roughly 0.75.
568  private final Set<String> supportedRegions = new HashSet<String>(320);
569
570  // The set of county calling codes that map to the non-geo entity region ("001"). This set
571  // currently contains < 12 elements so the default capacity of 16 (load factor=0.75) is fine.
572  private final Set<Integer> countryCodesForNonGeographicalRegion = new HashSet<Integer>();
573
574  // The prefix of the metadata files from which region data is loaded.
575  private final String currentFilePrefix;
576  // The metadata loader used to inject alternative metadata sources.
577  private final MetadataLoader metadataLoader;
578
579  /**
580   * This class implements a singleton, the constructor is only visible to facilitate testing.
581   */
582  // @VisibleForTesting
583  PhoneNumberUtil(String filePrefix, MetadataLoader metadataLoader,
584      Map<Integer, List<String>> countryCallingCodeToRegionCodeMap) {
585    this.currentFilePrefix = filePrefix;
586    this.metadataLoader = metadataLoader;
587    this.countryCallingCodeToRegionCodeMap = countryCallingCodeToRegionCodeMap;
588    for (Map.Entry<Integer, List<String>> entry : countryCallingCodeToRegionCodeMap.entrySet()) {
589      List<String> regionCodes = entry.getValue();
590      // We can assume that if the county calling code maps to the non-geo entity region code then
591      // that's the only region code it maps to.
592      if (regionCodes.size() == 1 && REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCodes.get(0))) {
593        // This is the subset of all country codes that map to the non-geo entity region code.
594        countryCodesForNonGeographicalRegion.add(entry.getKey());
595      } else {
596        // The supported regions set does not include the "001" non-geo entity region code.
597        supportedRegions.addAll(regionCodes);
598      }
599    }
600    // If the non-geo entity still got added to the set of supported regions it must be because
601    // there are entries that list the non-geo entity alongside normal regions (which is wrong).
602    // If we discover this, remove the non-geo entity from the set of supported regions and log.
603    if (supportedRegions.remove(REGION_CODE_FOR_NON_GEO_ENTITY)) {
604      logger.log(Level.WARNING, "invalid metadata " +
605          "(country calling code was mapped to the non-geo entity as well as specific region(s))");
606    }
607    nanpaRegions.addAll(countryCallingCodeToRegionCodeMap.get(NANPA_COUNTRY_CODE));
608  }
609
610  // @VisibleForTesting
611  void loadMetadataFromFile(String filePrefix, String regionCode, int countryCallingCode,
612      MetadataLoader metadataLoader) {
613    boolean isNonGeoRegion = REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCode);
614    String fileName = filePrefix + "_" +
615        (isNonGeoRegion ? String.valueOf(countryCallingCode) : regionCode);
616    InputStream source = metadataLoader.loadMetadata(fileName);
617    if (source == null) {
618      logger.log(Level.SEVERE, "missing metadata: " + fileName);
619      throw new IllegalStateException("missing metadata: " + fileName);
620    }
621    ObjectInputStream in = null;
622    try {
623      in = new ObjectInputStream(source);
624      PhoneMetadataCollection metadataCollection = loadMetadataAndCloseInput(in);
625      List<PhoneMetadata> metadataList = metadataCollection.getMetadataList();
626      if (metadataList.isEmpty()) {
627        logger.log(Level.SEVERE, "empty metadata: " + fileName);
628        throw new IllegalStateException("empty metadata: " + fileName);
629      }
630      if (metadataList.size() > 1) {
631        logger.log(Level.WARNING, "invalid metadata (too many entries): " + fileName);
632      }
633      PhoneMetadata metadata = metadataList.get(0);
634      if (isNonGeoRegion) {
635        countryCodeToNonGeographicalMetadataMap.put(countryCallingCode, metadata);
636      } else {
637        regionToMetadataMap.put(regionCode, metadata);
638      }
639    } catch (IOException e) {
640      logger.log(Level.SEVERE, "cannot load/parse metadata: " + fileName, e);
641      throw new RuntimeException("cannot load/parse metadata: " + fileName, e);
642    }
643  }
644
645  /**
646   * Loads the metadata protocol buffer from the given stream and closes the stream afterwards. Any
647   * exceptions that occur while reading the stream are propagated (though exceptions that occur
648   * when the stream is closed will be ignored).
649   *
650   * @param source  the non-null stream from which metadata is to be read.
651   * @return        the loaded metadata protocol buffer.
652   */
653  private static PhoneMetadataCollection loadMetadataAndCloseInput(ObjectInputStream source) {
654    PhoneMetadataCollection metadataCollection = new PhoneMetadataCollection();
655    try {
656      metadataCollection.readExternal(source);
657    } catch (IOException e) {
658      logger.log(Level.WARNING, "error reading input (ignored)", e);
659    } finally {
660      try {
661        source.close();
662      } catch (IOException e) {
663        logger.log(Level.WARNING, "error closing input stream (ignored)", e);
664      } finally {
665        return metadataCollection;
666      }
667    }
668  }
669
670  /**
671   * Attempts to extract a possible number from the string passed in. This currently strips all
672   * leading characters that cannot be used to start a phone number. Characters that can be used to
673   * start a phone number are defined in the VALID_START_CHAR_PATTERN. If none of these characters
674   * are found in the number passed in, an empty string is returned. This function also attempts to
675   * strip off any alternative extensions or endings if two or more are present, such as in the case
676   * of: (530) 583-6985 x302/x2303. The second extension here makes this actually two phone numbers,
677   * (530) 583-6985 x302 and (530) 583-6985 x2303. We remove the second extension so that the first
678   * number is parsed correctly.
679   *
680   * @param number  the string that might contain a phone number
681   * @return        the number, stripped of any non-phone-number prefix (such as "Tel:") or an empty
682   *                string if no character used to start phone numbers (such as + or any digit) is
683   *                found in the number
684   */
685  static String extractPossibleNumber(String number) {
686    Matcher m = VALID_START_CHAR_PATTERN.matcher(number);
687    if (m.find()) {
688      number = number.substring(m.start());
689      // Remove trailing non-alpha non-numerical characters.
690      Matcher trailingCharsMatcher = UNWANTED_END_CHAR_PATTERN.matcher(number);
691      if (trailingCharsMatcher.find()) {
692        number = number.substring(0, trailingCharsMatcher.start());
693        logger.log(Level.FINER, "Stripped trailing characters: " + number);
694      }
695      // Check for extra numbers at the end.
696      Matcher secondNumber = SECOND_NUMBER_START_PATTERN.matcher(number);
697      if (secondNumber.find()) {
698        number = number.substring(0, secondNumber.start());
699      }
700      return number;
701    } else {
702      return "";
703    }
704  }
705
706  /**
707   * Checks to see if the string of characters could possibly be a phone number at all. At the
708   * moment, checks to see that the string begins with at least 2 digits, ignoring any punctuation
709   * commonly found in phone numbers.
710   * This method does not require the number to be normalized in advance - but does assume that
711   * leading non-number symbols have been removed, such as by the method extractPossibleNumber.
712   *
713   * @param number  string to be checked for viability as a phone number
714   * @return        true if the number could be a phone number of some sort, otherwise false
715   */
716  // @VisibleForTesting
717  static boolean isViablePhoneNumber(String number) {
718    if (number.length() < MIN_LENGTH_FOR_NSN) {
719      return false;
720    }
721    Matcher m = VALID_PHONE_NUMBER_PATTERN.matcher(number);
722    return m.matches();
723  }
724
725  /**
726   * Normalizes a string of characters representing a phone number. This performs the following
727   * conversions:
728   *   Punctuation is stripped.
729   *   For ALPHA/VANITY numbers:
730   *   Letters are converted to their numeric representation on a telephone keypad. The keypad
731   *       used here is the one defined in ITU Recommendation E.161. This is only done if there are
732   *       3 or more letters in the number, to lessen the risk that such letters are typos.
733   *   For other numbers:
734   *   Wide-ascii digits are converted to normal ASCII (European) digits.
735   *   Arabic-Indic numerals are converted to European numerals.
736   *   Spurious alpha characters are stripped.
737   *
738   * @param number  a string of characters representing a phone number
739   * @return        the normalized string version of the phone number
740   */
741  static String normalize(String number) {
742    Matcher m = VALID_ALPHA_PHONE_PATTERN.matcher(number);
743    if (m.matches()) {
744      return normalizeHelper(number, ALPHA_PHONE_MAPPINGS, true);
745    } else {
746      return normalizeDigitsOnly(number);
747    }
748  }
749
750  /**
751   * Normalizes a string of characters representing a phone number. This is a wrapper for
752   * normalize(String number) but does in-place normalization of the StringBuilder provided.
753   *
754   * @param number  a StringBuilder of characters representing a phone number that will be
755   *     normalized in place
756   */
757  static void normalize(StringBuilder number) {
758    String normalizedNumber = normalize(number.toString());
759    number.replace(0, number.length(), normalizedNumber);
760  }
761
762  /**
763   * Normalizes a string of characters representing a phone number. This converts wide-ascii and
764   * arabic-indic numerals to European numerals, and strips punctuation and alpha characters.
765   *
766   * @param number  a string of characters representing a phone number
767   * @return        the normalized string version of the phone number
768   */
769  public static String normalizeDigitsOnly(String number) {
770    return normalizeDigits(number, false /* strip non-digits */).toString();
771  }
772
773  static StringBuilder normalizeDigits(String number, boolean keepNonDigits) {
774    StringBuilder normalizedDigits = new StringBuilder(number.length());
775    for (char c : number.toCharArray()) {
776      int digit = Character.digit(c, 10);
777      if (digit != -1) {
778        normalizedDigits.append(digit);
779      } else if (keepNonDigits) {
780        normalizedDigits.append(c);
781      }
782    }
783    return normalizedDigits;
784  }
785
786  /**
787   * Normalizes a string of characters representing a phone number. This strips all characters which
788   * are not diallable on a mobile phone keypad (including all non-ASCII digits).
789   *
790   * @param number  a string of characters representing a phone number
791   * @return        the normalized string version of the phone number
792   */
793  static String normalizeDiallableCharsOnly(String number) {
794    return normalizeHelper(number, DIALLABLE_CHAR_MAPPINGS, true /* remove non matches */);
795  }
796
797  /**
798   * Converts all alpha characters in a number to their respective digits on a keypad, but retains
799   * existing formatting.
800   */
801  public static String convertAlphaCharactersInNumber(String number) {
802    return normalizeHelper(number, ALPHA_PHONE_MAPPINGS, false);
803  }
804
805  /**
806   * Gets the length of the geographical area code from the
807   * PhoneNumber object passed in, so that clients could use it
808   * to split a national significant number into geographical area code and subscriber number. It
809   * works in such a way that the resultant subscriber number should be diallable, at least on some
810   * devices. An example of how this could be used:
811   *
812   * <pre>
813   * PhoneNumberUtil phoneUtil = PhoneNumberUtil.getInstance();
814   * PhoneNumber number = phoneUtil.parse("16502530000", "US");
815   * String nationalSignificantNumber = phoneUtil.getNationalSignificantNumber(number);
816   * String areaCode;
817   * String subscriberNumber;
818   *
819   * int areaCodeLength = phoneUtil.getLengthOfGeographicalAreaCode(number);
820   * if (areaCodeLength > 0) {
821   *   areaCode = nationalSignificantNumber.substring(0, areaCodeLength);
822   *   subscriberNumber = nationalSignificantNumber.substring(areaCodeLength);
823   * } else {
824   *   areaCode = "";
825   *   subscriberNumber = nationalSignificantNumber;
826   * }
827   * </pre>
828   *
829   * N.B.: area code is a very ambiguous concept, so the I18N team generally recommends against
830   * using it for most purposes, but recommends using the more general {@code national_number}
831   * instead. Read the following carefully before deciding to use this method:
832   * <ul>
833   *  <li> geographical area codes change over time, and this method honors those changes;
834   *    therefore, it doesn't guarantee the stability of the result it produces.
835   *  <li> subscriber numbers may not be diallable from all devices (notably mobile devices, which
836   *    typically requires the full national_number to be dialled in most regions).
837   *  <li> most non-geographical numbers have no area codes, including numbers from non-geographical
838   *    entities
839   *  <li> some geographical numbers have no area codes.
840   * </ul>
841   * @param number  the PhoneNumber object for which clients
842   *     want to know the length of the area code.
843   * @return  the length of area code of the PhoneNumber object
844   *     passed in.
845   */
846  public int getLengthOfGeographicalAreaCode(PhoneNumber number) {
847    PhoneMetadata metadata = getMetadataForRegion(getRegionCodeForNumber(number));
848    if (metadata == null) {
849      return 0;
850    }
851    // If a country doesn't use a national prefix, and this number doesn't have an Italian leading
852    // zero, we assume it is a closed dialling plan with no area codes.
853    if (!metadata.hasNationalPrefix() && !number.isItalianLeadingZero()) {
854      return 0;
855    }
856
857    if (!isNumberGeographical(number)) {
858      return 0;
859    }
860
861    return getLengthOfNationalDestinationCode(number);
862  }
863
864  /**
865   * Gets the length of the national destination code (NDC) from the
866   * PhoneNumber object passed in, so that clients could use it
867   * to split a national significant number into NDC and subscriber number. The NDC of a phone
868   * number is normally the first group of digit(s) right after the country calling code when the
869   * number is formatted in the international format, if there is a subscriber number part that
870   * follows. An example of how this could be used:
871   *
872   * <pre>
873   * PhoneNumberUtil phoneUtil = PhoneNumberUtil.getInstance();
874   * PhoneNumber number = phoneUtil.parse("18002530000", "US");
875   * String nationalSignificantNumber = phoneUtil.getNationalSignificantNumber(number);
876   * String nationalDestinationCode;
877   * String subscriberNumber;
878   *
879   * int nationalDestinationCodeLength = phoneUtil.getLengthOfNationalDestinationCode(number);
880   * if (nationalDestinationCodeLength > 0) {
881   *   nationalDestinationCode = nationalSignificantNumber.substring(0,
882   *       nationalDestinationCodeLength);
883   *   subscriberNumber = nationalSignificantNumber.substring(nationalDestinationCodeLength);
884   * } else {
885   *   nationalDestinationCode = "";
886   *   subscriberNumber = nationalSignificantNumber;
887   * }
888   * </pre>
889   *
890   * Refer to the unittests to see the difference between this function and
891   * {@link #getLengthOfGeographicalAreaCode}.
892   *
893   * @param number  the PhoneNumber object for which clients
894   *     want to know the length of the NDC.
895   * @return  the length of NDC of the PhoneNumber object
896   *     passed in.
897   */
898  public int getLengthOfNationalDestinationCode(PhoneNumber number) {
899    PhoneNumber copiedProto;
900    if (number.hasExtension()) {
901      // We don't want to alter the proto given to us, but we don't want to include the extension
902      // when we format it, so we copy it and clear the extension here.
903      copiedProto = new PhoneNumber();
904      copiedProto.mergeFrom(number);
905      copiedProto.clearExtension();
906    } else {
907      copiedProto = number;
908    }
909
910    String nationalSignificantNumber = format(copiedProto,
911                                              PhoneNumberUtil.PhoneNumberFormat.INTERNATIONAL);
912    String[] numberGroups = NON_DIGITS_PATTERN.split(nationalSignificantNumber);
913    // The pattern will start with "+COUNTRY_CODE " so the first group will always be the empty
914    // string (before the + symbol) and the second group will be the country calling code. The third
915    // group will be area code if it is not the last group.
916    if (numberGroups.length <= 3) {
917      return 0;
918    }
919
920    if (getNumberType(number) == PhoneNumberType.MOBILE) {
921      // For example Argentinian mobile numbers, when formatted in the international format, are in
922      // the form of +54 9 NDC XXXX.... As a result, we take the length of the third group (NDC) and
923      // add the length of the second group (which is the mobile token), which also forms part of
924      // the national significant number. This assumes that the mobile token is always formatted
925      // separately from the rest of the phone number.
926      String mobileToken = getCountryMobileToken(number.getCountryCode());
927      if (!mobileToken.equals("")) {
928        return numberGroups[2].length() + numberGroups[3].length();
929      }
930    }
931    return numberGroups[2].length();
932  }
933
934  /**
935   * Returns the mobile token for the provided country calling code if it has one, otherwise
936   * returns an empty string. A mobile token is a number inserted before the area code when dialing
937   * a mobile number from that country from abroad.
938   *
939   * @param countryCallingCode  the country calling code for which we want the mobile token
940   * @return  the mobile token, as a string, for the given country calling code
941   */
942  public static String getCountryMobileToken(int countryCallingCode) {
943    if (MOBILE_TOKEN_MAPPINGS.containsKey(countryCallingCode)) {
944      return MOBILE_TOKEN_MAPPINGS.get(countryCallingCode);
945    }
946    return "";
947  }
948
949  /**
950   * Normalizes a string of characters representing a phone number by replacing all characters found
951   * in the accompanying map with the values therein, and stripping all other characters if
952   * removeNonMatches is true.
953   *
954   * @param number                     a string of characters representing a phone number
955   * @param normalizationReplacements  a mapping of characters to what they should be replaced by in
956   *                                   the normalized version of the phone number
957   * @param removeNonMatches           indicates whether characters that are not able to be replaced
958   *                                   should be stripped from the number. If this is false, they
959   *                                   will be left unchanged in the number.
960   * @return  the normalized string version of the phone number
961   */
962  private static String normalizeHelper(String number,
963                                        Map<Character, Character> normalizationReplacements,
964                                        boolean removeNonMatches) {
965    StringBuilder normalizedNumber = new StringBuilder(number.length());
966    for (int i = 0; i < number.length(); i++) {
967      char character = number.charAt(i);
968      Character newDigit = normalizationReplacements.get(Character.toUpperCase(character));
969      if (newDigit != null) {
970        normalizedNumber.append(newDigit);
971      } else if (!removeNonMatches) {
972        normalizedNumber.append(character);
973      }
974      // If neither of the above are true, we remove this character.
975    }
976    return normalizedNumber.toString();
977  }
978
979  /**
980   * Sets or resets the PhoneNumberUtil singleton instance. If set to null, the next call to
981   * {@code getInstance()} will load (and return) the default instance.
982   */
983  // @VisibleForTesting
984  static synchronized void setInstance(PhoneNumberUtil util) {
985    instance = util;
986  }
987
988  /**
989   * Convenience method to get a list of what regions the library has metadata for.
990   */
991  public Set<String> getSupportedRegions() {
992    return Collections.unmodifiableSet(supportedRegions);
993  }
994
995  /**
996   * Convenience method to get a list of what global network calling codes the library has metadata
997   * for.
998   */
999  public Set<Integer> getSupportedGlobalNetworkCallingCodes() {
1000    return Collections.unmodifiableSet(countryCodesForNonGeographicalRegion);
1001  }
1002
1003  /**
1004   * Gets a {@link PhoneNumberUtil} instance to carry out international phone number formatting,
1005   * parsing, or validation. The instance is loaded with phone number metadata for a number of most
1006   * commonly used regions.
1007   *
1008   * <p>The {@link PhoneNumberUtil} is implemented as a singleton. Therefore, calling getInstance
1009   * multiple times will only result in one instance being created.
1010   *
1011   * @return a PhoneNumberUtil instance
1012   */
1013  public static synchronized PhoneNumberUtil getInstance() {
1014    if (instance == null) {
1015      setInstance(createInstance(DEFAULT_METADATA_LOADER));
1016    }
1017    return instance;
1018  }
1019
1020  /**
1021   * Create a new {@link PhoneNumberUtil} instance to carry out international phone number
1022   * formatting, parsing, or validation. The instance is loaded with all metadata by
1023   * using the metadataLoader specified.
1024   *
1025   * This method should only be used in the rare case in which you want to manage your own
1026   * metadata loading. Calling this method multiple times is very expensive, as each time
1027   * a new instance is created from scratch. When in doubt, use {@link #getInstance}.
1028   *
1029   * @param metadataLoader Customized metadata loader. If null, default metadata loader will
1030   *     be used. This should not be null.
1031   * @return a PhoneNumberUtil instance
1032   */
1033  public static PhoneNumberUtil createInstance(MetadataLoader metadataLoader) {
1034    if (metadataLoader == null) {
1035      throw new IllegalArgumentException("metadataLoader could not be null.");
1036    }
1037    return new PhoneNumberUtil(META_DATA_FILE_PREFIX, metadataLoader,
1038        CountryCodeToRegionCodeMap.getCountryCodeToRegionCodeMap());
1039  }
1040
1041  /**
1042   * Helper function to check if the national prefix formatting rule has the first group only, i.e.,
1043   * does not start with the national prefix.
1044   */
1045  static boolean formattingRuleHasFirstGroupOnly(String nationalPrefixFormattingRule) {
1046    return nationalPrefixFormattingRule.length() == 0 ||
1047        FIRST_GROUP_ONLY_PREFIX_PATTERN.matcher(nationalPrefixFormattingRule).matches();
1048  }
1049
1050  /**
1051   * Tests whether a phone number has a geographical association. It checks if the number is
1052   * associated to a certain region in the country where it belongs to. Note that this doesn't
1053   * verify if the number is actually in use.
1054   *
1055   * A similar method is implemented as PhoneNumberOfflineGeocoder.canBeGeocoded, which performs a
1056   * looser check, since it only prevents cases where prefixes overlap for geocodable and
1057   * non-geocodable numbers. Also, if new phone number types were added, we should check if this
1058   * other method should be updated too.
1059   */
1060  boolean isNumberGeographical(PhoneNumber phoneNumber) {
1061    PhoneNumberType numberType = getNumberType(phoneNumber);
1062    // TODO: Include mobile phone numbers from countries like Indonesia, which has some
1063    // mobile numbers that are geographical.
1064    return numberType == PhoneNumberType.FIXED_LINE ||
1065        numberType == PhoneNumberType.FIXED_LINE_OR_MOBILE;
1066  }
1067
1068  /**
1069   * Helper function to check region code is not unknown or null.
1070   */
1071  private boolean isValidRegionCode(String regionCode) {
1072    return regionCode != null && supportedRegions.contains(regionCode);
1073  }
1074
1075  /**
1076   * Helper function to check the country calling code is valid.
1077   */
1078  private boolean hasValidCountryCallingCode(int countryCallingCode) {
1079    return countryCallingCodeToRegionCodeMap.containsKey(countryCallingCode);
1080  }
1081
1082  /**
1083   * Formats a phone number in the specified format using default rules. Note that this does not
1084   * promise to produce a phone number that the user can dial from where they are - although we do
1085   * format in either 'national' or 'international' format depending on what the client asks for, we
1086   * do not currently support a more abbreviated format, such as for users in the same "area" who
1087   * could potentially dial the number without area code. Note that if the phone number has a
1088   * country calling code of 0 or an otherwise invalid country calling code, we cannot work out
1089   * which formatting rules to apply so we return the national significant number with no formatting
1090   * applied.
1091   *
1092   * @param number         the phone number to be formatted
1093   * @param numberFormat   the format the phone number should be formatted into
1094   * @return  the formatted phone number
1095   */
1096  public String format(PhoneNumber number, PhoneNumberFormat numberFormat) {
1097    if (number.getNationalNumber() == 0 && number.hasRawInput()) {
1098      // Unparseable numbers that kept their raw input just use that.
1099      // This is the only case where a number can be formatted as E164 without a
1100      // leading '+' symbol (but the original number wasn't parseable anyway).
1101      // TODO: Consider removing the 'if' above so that unparseable
1102      // strings without raw input format to the empty string instead of "+00"
1103      String rawInput = number.getRawInput();
1104      if (rawInput.length() > 0) {
1105        return rawInput;
1106      }
1107    }
1108    StringBuilder formattedNumber = new StringBuilder(20);
1109    format(number, numberFormat, formattedNumber);
1110    return formattedNumber.toString();
1111  }
1112
1113  /**
1114   * Same as {@link #format(PhoneNumber, PhoneNumberFormat)}, but accepts a mutable StringBuilder as
1115   * a parameter to decrease object creation when invoked many times.
1116   */
1117  public void format(PhoneNumber number, PhoneNumberFormat numberFormat,
1118                     StringBuilder formattedNumber) {
1119    // Clear the StringBuilder first.
1120    formattedNumber.setLength(0);
1121    int countryCallingCode = number.getCountryCode();
1122    String nationalSignificantNumber = getNationalSignificantNumber(number);
1123
1124    if (numberFormat == PhoneNumberFormat.E164) {
1125      // Early exit for E164 case (even if the country calling code is invalid) since no formatting
1126      // of the national number needs to be applied. Extensions are not formatted.
1127      formattedNumber.append(nationalSignificantNumber);
1128      prefixNumberWithCountryCallingCode(countryCallingCode, PhoneNumberFormat.E164,
1129                                         formattedNumber);
1130      return;
1131    }
1132    if (!hasValidCountryCallingCode(countryCallingCode)) {
1133      formattedNumber.append(nationalSignificantNumber);
1134      return;
1135    }
1136    // Note getRegionCodeForCountryCode() is used because formatting information for regions which
1137    // share a country calling code is contained by only one region for performance reasons. For
1138    // example, for NANPA regions it will be contained in the metadata for US.
1139    String regionCode = getRegionCodeForCountryCode(countryCallingCode);
1140    // Metadata cannot be null because the country calling code is valid (which means that the
1141    // region code cannot be ZZ and must be one of our supported region codes).
1142    PhoneMetadata metadata =
1143        getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
1144    formattedNumber.append(formatNsn(nationalSignificantNumber, metadata, numberFormat));
1145    maybeAppendFormattedExtension(number, metadata, numberFormat, formattedNumber);
1146    prefixNumberWithCountryCallingCode(countryCallingCode, numberFormat, formattedNumber);
1147  }
1148
1149  /**
1150   * Formats a phone number in the specified format using client-defined formatting rules. Note that
1151   * if the phone number has a country calling code of zero or an otherwise invalid country calling
1152   * code, we cannot work out things like whether there should be a national prefix applied, or how
1153   * to format extensions, so we return the national significant number with no formatting applied.
1154   *
1155   * @param number                        the phone number to be formatted
1156   * @param numberFormat                  the format the phone number should be formatted into
1157   * @param userDefinedFormats            formatting rules specified by clients
1158   * @return  the formatted phone number
1159   */
1160  public String formatByPattern(PhoneNumber number,
1161                                PhoneNumberFormat numberFormat,
1162                                List<NumberFormat> userDefinedFormats) {
1163    int countryCallingCode = number.getCountryCode();
1164    String nationalSignificantNumber = getNationalSignificantNumber(number);
1165    if (!hasValidCountryCallingCode(countryCallingCode)) {
1166      return nationalSignificantNumber;
1167    }
1168    // Note getRegionCodeForCountryCode() is used because formatting information for regions which
1169    // share a country calling code is contained by only one region for performance reasons. For
1170    // example, for NANPA regions it will be contained in the metadata for US.
1171    String regionCode = getRegionCodeForCountryCode(countryCallingCode);
1172    // Metadata cannot be null because the country calling code is valid
1173    PhoneMetadata metadata =
1174        getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
1175
1176    StringBuilder formattedNumber = new StringBuilder(20);
1177
1178    NumberFormat formattingPattern =
1179        chooseFormattingPatternForNumber(userDefinedFormats, nationalSignificantNumber);
1180    if (formattingPattern == null) {
1181      // If no pattern above is matched, we format the number as a whole.
1182      formattedNumber.append(nationalSignificantNumber);
1183    } else {
1184      NumberFormat numFormatCopy = new NumberFormat();
1185      // Before we do a replacement of the national prefix pattern $NP with the national prefix, we
1186      // need to copy the rule so that subsequent replacements for different numbers have the
1187      // appropriate national prefix.
1188      numFormatCopy.mergeFrom(formattingPattern);
1189      String nationalPrefixFormattingRule = formattingPattern.getNationalPrefixFormattingRule();
1190      if (nationalPrefixFormattingRule.length() > 0) {
1191        String nationalPrefix = metadata.getNationalPrefix();
1192        if (nationalPrefix.length() > 0) {
1193          // Replace $NP with national prefix and $FG with the first group ($1).
1194          nationalPrefixFormattingRule =
1195              NP_PATTERN.matcher(nationalPrefixFormattingRule).replaceFirst(nationalPrefix);
1196          nationalPrefixFormattingRule =
1197              FG_PATTERN.matcher(nationalPrefixFormattingRule).replaceFirst("\\$1");
1198          numFormatCopy.setNationalPrefixFormattingRule(nationalPrefixFormattingRule);
1199        } else {
1200          // We don't want to have a rule for how to format the national prefix if there isn't one.
1201          numFormatCopy.clearNationalPrefixFormattingRule();
1202        }
1203      }
1204      formattedNumber.append(
1205          formatNsnUsingPattern(nationalSignificantNumber, numFormatCopy, numberFormat));
1206    }
1207    maybeAppendFormattedExtension(number, metadata, numberFormat, formattedNumber);
1208    prefixNumberWithCountryCallingCode(countryCallingCode, numberFormat, formattedNumber);
1209    return formattedNumber.toString();
1210  }
1211
1212  /**
1213   * Formats a phone number in national format for dialing using the carrier as specified in the
1214   * {@code carrierCode}. The {@code carrierCode} will always be used regardless of whether the
1215   * phone number already has a preferred domestic carrier code stored. If {@code carrierCode}
1216   * contains an empty string, returns the number in national format without any carrier code.
1217   *
1218   * @param number  the phone number to be formatted
1219   * @param carrierCode  the carrier selection code to be used
1220   * @return  the formatted phone number in national format for dialing using the carrier as
1221   *          specified in the {@code carrierCode}
1222   */
1223  public String formatNationalNumberWithCarrierCode(PhoneNumber number, String carrierCode) {
1224    int countryCallingCode = number.getCountryCode();
1225    String nationalSignificantNumber = getNationalSignificantNumber(number);
1226    if (!hasValidCountryCallingCode(countryCallingCode)) {
1227      return nationalSignificantNumber;
1228    }
1229
1230    // Note getRegionCodeForCountryCode() is used because formatting information for regions which
1231    // share a country calling code is contained by only one region for performance reasons. For
1232    // example, for NANPA regions it will be contained in the metadata for US.
1233    String regionCode = getRegionCodeForCountryCode(countryCallingCode);
1234    // Metadata cannot be null because the country calling code is valid.
1235    PhoneMetadata metadata = getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
1236
1237    StringBuilder formattedNumber = new StringBuilder(20);
1238    formattedNumber.append(formatNsn(nationalSignificantNumber, metadata,
1239                                     PhoneNumberFormat.NATIONAL, carrierCode));
1240    maybeAppendFormattedExtension(number, metadata, PhoneNumberFormat.NATIONAL, formattedNumber);
1241    prefixNumberWithCountryCallingCode(countryCallingCode, PhoneNumberFormat.NATIONAL,
1242                                       formattedNumber);
1243    return formattedNumber.toString();
1244  }
1245
1246  private PhoneMetadata getMetadataForRegionOrCallingCode(
1247      int countryCallingCode, String regionCode) {
1248    return REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCode)
1249        ? getMetadataForNonGeographicalRegion(countryCallingCode)
1250        : getMetadataForRegion(regionCode);
1251  }
1252
1253  /**
1254   * Formats a phone number in national format for dialing using the carrier as specified in the
1255   * preferredDomesticCarrierCode field of the PhoneNumber object passed in. If that is missing,
1256   * use the {@code fallbackCarrierCode} passed in instead. If there is no
1257   * {@code preferredDomesticCarrierCode}, and the {@code fallbackCarrierCode} contains an empty
1258   * string, return the number in national format without any carrier code.
1259   *
1260   * <p>Use {@link #formatNationalNumberWithCarrierCode} instead if the carrier code passed in
1261   * should take precedence over the number's {@code preferredDomesticCarrierCode} when formatting.
1262   *
1263   * @param number  the phone number to be formatted
1264   * @param fallbackCarrierCode  the carrier selection code to be used, if none is found in the
1265   *     phone number itself
1266   * @return  the formatted phone number in national format for dialing using the number's
1267   *     {@code preferredDomesticCarrierCode}, or the {@code fallbackCarrierCode} passed in if
1268   *     none is found
1269   */
1270  public String formatNationalNumberWithPreferredCarrierCode(PhoneNumber number,
1271                                                             String fallbackCarrierCode) {
1272    return formatNationalNumberWithCarrierCode(number, number.hasPreferredDomesticCarrierCode()
1273                                                       ? number.getPreferredDomesticCarrierCode()
1274                                                       : fallbackCarrierCode);
1275  }
1276
1277  /**
1278   * Returns a number formatted in such a way that it can be dialed from a mobile phone in a
1279   * specific region. If the number cannot be reached from the region (e.g. some countries block
1280   * toll-free numbers from being called outside of the country), the method returns an empty
1281   * string.
1282   *
1283   * @param number  the phone number to be formatted
1284   * @param regionCallingFrom  the region where the call is being placed
1285   * @param withFormatting  whether the number should be returned with formatting symbols, such as
1286   *     spaces and dashes.
1287   * @return  the formatted phone number
1288   */
1289  public String formatNumberForMobileDialing(PhoneNumber number, String regionCallingFrom,
1290                                             boolean withFormatting) {
1291    int countryCallingCode = number.getCountryCode();
1292    if (!hasValidCountryCallingCode(countryCallingCode)) {
1293      return number.hasRawInput() ? number.getRawInput() : "";
1294    }
1295
1296    String formattedNumber = "";
1297    // Clear the extension, as that part cannot normally be dialed together with the main number.
1298    PhoneNumber numberNoExt = new PhoneNumber().mergeFrom(number).clearExtension();
1299    String regionCode = getRegionCodeForCountryCode(countryCallingCode);
1300    PhoneNumberType numberType = getNumberType(numberNoExt);
1301    boolean isValidNumber = (numberType != PhoneNumberType.UNKNOWN);
1302    if (regionCallingFrom.equals(regionCode)) {
1303      boolean isFixedLineOrMobile =
1304          (numberType == PhoneNumberType.FIXED_LINE) || (numberType == PhoneNumberType.MOBILE) ||
1305          (numberType == PhoneNumberType.FIXED_LINE_OR_MOBILE);
1306      // Carrier codes may be needed in some countries. We handle this here.
1307      if (regionCode.equals("CO") && numberType == PhoneNumberType.FIXED_LINE) {
1308        formattedNumber =
1309            formatNationalNumberWithCarrierCode(numberNoExt, COLOMBIA_MOBILE_TO_FIXED_LINE_PREFIX);
1310      } else if (regionCode.equals("BR") && isFixedLineOrMobile) {
1311        formattedNumber = numberNoExt.hasPreferredDomesticCarrierCode()
1312            ? formattedNumber = formatNationalNumberWithPreferredCarrierCode(numberNoExt, "")
1313            // Brazilian fixed line and mobile numbers need to be dialed with a carrier code when
1314            // called within Brazil. Without that, most of the carriers won't connect the call.
1315            // Because of that, we return an empty string here.
1316            : "";
1317      } else if (isValidNumber && regionCode.equals("HU")) {
1318        // The national format for HU numbers doesn't contain the national prefix, because that is
1319        // how numbers are normally written down. However, the national prefix is obligatory when
1320        // dialing from a mobile phone, except for short numbers. As a result, we add it back here
1321        // if it is a valid regular length phone number.
1322        formattedNumber =
1323            getNddPrefixForRegion(regionCode, true /* strip non-digits */) +
1324            " " + format(numberNoExt, PhoneNumberFormat.NATIONAL);
1325      } else if (countryCallingCode == NANPA_COUNTRY_CODE) {
1326        // For NANPA countries, we output international format for numbers that can be dialed
1327        // internationally, since that always works, except for numbers which might potentially be
1328        // short numbers, which are always dialled in national format.
1329        PhoneMetadata regionMetadata = getMetadataForRegion(regionCallingFrom);
1330        if (canBeInternationallyDialled(numberNoExt) &&
1331            !isShorterThanPossibleNormalNumber(regionMetadata,
1332                getNationalSignificantNumber(numberNoExt))) {
1333          formattedNumber = format(numberNoExt, PhoneNumberFormat.INTERNATIONAL);
1334        } else {
1335          formattedNumber = format(numberNoExt, PhoneNumberFormat.NATIONAL);
1336        }
1337      } else {
1338        // For non-geographical countries, and Mexican and Chilean fixed line and mobile numbers, we
1339        // output international format for numbers that can be dialed internationally as that always
1340        // works.
1341        if ((regionCode.equals(REGION_CODE_FOR_NON_GEO_ENTITY) ||
1342            // MX fixed line and mobile numbers should always be formatted in international format,
1343            // even when dialed within MX. For national format to work, a carrier code needs to be
1344            // used, and the correct carrier code depends on if the caller and callee are from the
1345            // same local area. It is trickier to get that to work correctly than using
1346            // international format, which is tested to work fine on all carriers.
1347            // CL fixed line numbers need the national prefix when dialing in the national format,
1348            // but don't have it when used for display. The reverse is true for mobile numbers.
1349            // As a result, we output them in the international format to make it work.
1350            ((regionCode.equals("MX") || regionCode.equals("CL")) &&
1351             isFixedLineOrMobile)) &&
1352            canBeInternationallyDialled(numberNoExt)) {
1353          formattedNumber = format(numberNoExt, PhoneNumberFormat.INTERNATIONAL);
1354        } else {
1355          formattedNumber = format(numberNoExt, PhoneNumberFormat.NATIONAL);
1356        }
1357      }
1358    } else if (isValidNumber && canBeInternationallyDialled(numberNoExt)) {
1359      // We assume that short numbers are not diallable from outside their region, so if a number
1360      // is not a valid regular length phone number, we treat it as if it cannot be internationally
1361      // dialled.
1362      return withFormatting ? format(numberNoExt, PhoneNumberFormat.INTERNATIONAL)
1363                            : format(numberNoExt, PhoneNumberFormat.E164);
1364    }
1365    return withFormatting ? formattedNumber
1366                          : normalizeDiallableCharsOnly(formattedNumber);
1367  }
1368
1369  /**
1370   * Formats a phone number for out-of-country dialing purposes. If no regionCallingFrom is
1371   * supplied, we format the number in its INTERNATIONAL format. If the country calling code is the
1372   * same as that of the region where the number is from, then NATIONAL formatting will be applied.
1373   *
1374   * <p>If the number itself has a country calling code of zero or an otherwise invalid country
1375   * calling code, then we return the number with no formatting applied.
1376   *
1377   * <p>Note this function takes care of the case for calling inside of NANPA and between Russia and
1378   * Kazakhstan (who share the same country calling code). In those cases, no international prefix
1379   * is used. For regions which have multiple international prefixes, the number in its
1380   * INTERNATIONAL format will be returned instead.
1381   *
1382   * @param number               the phone number to be formatted
1383   * @param regionCallingFrom    the region where the call is being placed
1384   * @return  the formatted phone number
1385   */
1386  public String formatOutOfCountryCallingNumber(PhoneNumber number,
1387                                                String regionCallingFrom) {
1388    if (!isValidRegionCode(regionCallingFrom)) {
1389      logger.log(Level.WARNING,
1390                 "Trying to format number from invalid region "
1391                 + regionCallingFrom
1392                 + ". International formatting applied.");
1393      return format(number, PhoneNumberFormat.INTERNATIONAL);
1394    }
1395    int countryCallingCode = number.getCountryCode();
1396    String nationalSignificantNumber = getNationalSignificantNumber(number);
1397    if (!hasValidCountryCallingCode(countryCallingCode)) {
1398      return nationalSignificantNumber;
1399    }
1400    if (countryCallingCode == NANPA_COUNTRY_CODE) {
1401      if (isNANPACountry(regionCallingFrom)) {
1402        // For NANPA regions, return the national format for these regions but prefix it with the
1403        // country calling code.
1404        return countryCallingCode + " " + format(number, PhoneNumberFormat.NATIONAL);
1405      }
1406    } else if (countryCallingCode == getCountryCodeForValidRegion(regionCallingFrom)) {
1407      // If regions share a country calling code, the country calling code need not be dialled.
1408      // This also applies when dialling within a region, so this if clause covers both these cases.
1409      // Technically this is the case for dialling from La Reunion to other overseas departments of
1410      // France (French Guiana, Martinique, Guadeloupe), but not vice versa - so we don't cover this
1411      // edge case for now and for those cases return the version including country calling code.
1412      // Details here: http://www.petitfute.com/voyage/225-info-pratiques-reunion
1413      return format(number, PhoneNumberFormat.NATIONAL);
1414    }
1415    // Metadata cannot be null because we checked 'isValidRegionCode()' above.
1416    PhoneMetadata metadataForRegionCallingFrom = getMetadataForRegion(regionCallingFrom);
1417    String internationalPrefix = metadataForRegionCallingFrom.getInternationalPrefix();
1418
1419    // For regions that have multiple international prefixes, the international format of the
1420    // number is returned, unless there is a preferred international prefix.
1421    String internationalPrefixForFormatting = "";
1422    if (UNIQUE_INTERNATIONAL_PREFIX.matcher(internationalPrefix).matches()) {
1423      internationalPrefixForFormatting = internationalPrefix;
1424    } else if (metadataForRegionCallingFrom.hasPreferredInternationalPrefix()) {
1425      internationalPrefixForFormatting =
1426          metadataForRegionCallingFrom.getPreferredInternationalPrefix();
1427    }
1428
1429    String regionCode = getRegionCodeForCountryCode(countryCallingCode);
1430    // Metadata cannot be null because the country calling code is valid.
1431    PhoneMetadata metadataForRegion =
1432        getMetadataForRegionOrCallingCode(countryCallingCode, regionCode);
1433    String formattedNationalNumber =
1434        formatNsn(nationalSignificantNumber, metadataForRegion, PhoneNumberFormat.INTERNATIONAL);
1435    StringBuilder formattedNumber = new StringBuilder(formattedNationalNumber);
1436    maybeAppendFormattedExtension(number, metadataForRegion, PhoneNumberFormat.INTERNATIONAL,
1437                                  formattedNumber);
1438    if (internationalPrefixForFormatting.length() > 0) {
1439      formattedNumber.insert(0, " ").insert(0, countryCallingCode).insert(0, " ")
1440          .insert(0, internationalPrefixForFormatting);
1441    } else {
1442      prefixNumberWithCountryCallingCode(countryCallingCode,
1443                                         PhoneNumberFormat.INTERNATIONAL,
1444                                         formattedNumber);
1445    }
1446    return formattedNumber.toString();
1447  }
1448
1449  /**
1450   * Formats a phone number using the original phone number format that the number is parsed from.
1451   * The original format is embedded in the country_code_source field of the PhoneNumber object
1452   * passed in. If such information is missing, the number will be formatted into the NATIONAL
1453   * format by default. When the number contains a leading zero and this is unexpected for this
1454   * country, or we don't have a formatting pattern for the number, the method returns the raw input
1455   * when it is available.
1456   *
1457   * Note this method guarantees no digit will be inserted, removed or modified as a result of
1458   * formatting.
1459   *
1460   * @param number  the phone number that needs to be formatted in its original number format
1461   * @param regionCallingFrom  the region whose IDD needs to be prefixed if the original number
1462   *     has one
1463   * @return  the formatted phone number in its original number format
1464   */
1465  public String formatInOriginalFormat(PhoneNumber number, String regionCallingFrom) {
1466    if (number.hasRawInput() &&
1467        (hasUnexpectedItalianLeadingZero(number) || !hasFormattingPatternForNumber(number))) {
1468      // We check if we have the formatting pattern because without that, we might format the number
1469      // as a group without national prefix.
1470      return number.getRawInput();
1471    }
1472    if (!number.hasCountryCodeSource()) {
1473      return format(number, PhoneNumberFormat.NATIONAL);
1474    }
1475    String formattedNumber;
1476    switch (number.getCountryCodeSource()) {
1477      case FROM_NUMBER_WITH_PLUS_SIGN:
1478        formattedNumber = format(number, PhoneNumberFormat.INTERNATIONAL);
1479        break;
1480      case FROM_NUMBER_WITH_IDD:
1481        formattedNumber = formatOutOfCountryCallingNumber(number, regionCallingFrom);
1482        break;
1483      case FROM_NUMBER_WITHOUT_PLUS_SIGN:
1484        formattedNumber = format(number, PhoneNumberFormat.INTERNATIONAL).substring(1);
1485        break;
1486      case FROM_DEFAULT_COUNTRY:
1487        // Fall-through to default case.
1488      default:
1489        String regionCode = getRegionCodeForCountryCode(number.getCountryCode());
1490        // We strip non-digits from the NDD here, and from the raw input later, so that we can
1491        // compare them easily.
1492        String nationalPrefix = getNddPrefixForRegion(regionCode, true /* strip non-digits */);
1493        String nationalFormat = format(number, PhoneNumberFormat.NATIONAL);
1494        if (nationalPrefix == null || nationalPrefix.length() == 0) {
1495          // If the region doesn't have a national prefix at all, we can safely return the national
1496          // format without worrying about a national prefix being added.
1497          formattedNumber = nationalFormat;
1498          break;
1499        }
1500        // Otherwise, we check if the original number was entered with a national prefix.
1501        if (rawInputContainsNationalPrefix(
1502            number.getRawInput(), nationalPrefix, regionCode)) {
1503          // If so, we can safely return the national format.
1504          formattedNumber = nationalFormat;
1505          break;
1506        }
1507        // Metadata cannot be null here because getNddPrefixForRegion() (above) returns null if
1508        // there is no metadata for the region.
1509        PhoneMetadata metadata = getMetadataForRegion(regionCode);
1510        String nationalNumber = getNationalSignificantNumber(number);
1511        NumberFormat formatRule =
1512            chooseFormattingPatternForNumber(metadata.numberFormats(), nationalNumber);
1513        // The format rule could still be null here if the national number was 0 and there was no
1514        // raw input (this should not be possible for numbers generated by the phonenumber library
1515        // as they would also not have a country calling code and we would have exited earlier).
1516        if (formatRule == null) {
1517          formattedNumber = nationalFormat;
1518          break;
1519        }
1520        // When the format we apply to this number doesn't contain national prefix, we can just
1521        // return the national format.
1522        // TODO: Refactor the code below with the code in
1523        // isNationalPrefixPresentIfRequired.
1524        String candidateNationalPrefixRule = formatRule.getNationalPrefixFormattingRule();
1525        // We assume that the first-group symbol will never be _before_ the national prefix.
1526        int indexOfFirstGroup = candidateNationalPrefixRule.indexOf("$1");
1527        if (indexOfFirstGroup <= 0) {
1528          formattedNumber = nationalFormat;
1529          break;
1530        }
1531        candidateNationalPrefixRule =
1532            candidateNationalPrefixRule.substring(0, indexOfFirstGroup);
1533        candidateNationalPrefixRule = normalizeDigitsOnly(candidateNationalPrefixRule);
1534        if (candidateNationalPrefixRule.length() == 0) {
1535          // National prefix not used when formatting this number.
1536          formattedNumber = nationalFormat;
1537          break;
1538        }
1539        // Otherwise, we need to remove the national prefix from our output.
1540        NumberFormat numFormatCopy = new NumberFormat();
1541        numFormatCopy.mergeFrom(formatRule);
1542        numFormatCopy.clearNationalPrefixFormattingRule();
1543        List<NumberFormat> numberFormats = new ArrayList<NumberFormat>(1);
1544        numberFormats.add(numFormatCopy);
1545        formattedNumber = formatByPattern(number, PhoneNumberFormat.NATIONAL, numberFormats);
1546        break;
1547    }
1548    String rawInput = number.getRawInput();
1549    // If no digit is inserted/removed/modified as a result of our formatting, we return the
1550    // formatted phone number; otherwise we return the raw input the user entered.
1551    if (formattedNumber != null && rawInput.length() > 0) {
1552      String normalizedFormattedNumber = normalizeDiallableCharsOnly(formattedNumber);
1553      String normalizedRawInput = normalizeDiallableCharsOnly(rawInput);
1554      if (!normalizedFormattedNumber.equals(normalizedRawInput)) {
1555        formattedNumber = rawInput;
1556      }
1557    }
1558    return formattedNumber;
1559  }
1560
1561  // Check if rawInput, which is assumed to be in the national format, has a national prefix. The
1562  // national prefix is assumed to be in digits-only form.
1563  private boolean rawInputContainsNationalPrefix(String rawInput, String nationalPrefix,
1564      String regionCode) {
1565    String normalizedNationalNumber = normalizeDigitsOnly(rawInput);
1566    if (normalizedNationalNumber.startsWith(nationalPrefix)) {
1567      try {
1568        // Some Japanese numbers (e.g. 00777123) might be mistaken to contain the national prefix
1569        // when written without it (e.g. 0777123) if we just do prefix matching. To tackle that, we
1570        // check the validity of the number if the assumed national prefix is removed (777123 won't
1571        // be valid in Japan).
1572        return isValidNumber(
1573            parse(normalizedNationalNumber.substring(nationalPrefix.length()), regionCode));
1574      } catch (NumberParseException e) {
1575        return false;
1576      }
1577    }
1578    return false;
1579  }
1580
1581  /**
1582   * Returns true if a number is from a region whose national significant number couldn't contain a
1583   * leading zero, but has the italian_leading_zero field set to true.
1584   */
1585  private boolean hasUnexpectedItalianLeadingZero(PhoneNumber number) {
1586    return number.isItalianLeadingZero() && !isLeadingZeroPossible(number.getCountryCode());
1587  }
1588
1589  private boolean hasFormattingPatternForNumber(PhoneNumber number) {
1590    int countryCallingCode = number.getCountryCode();
1591    String phoneNumberRegion = getRegionCodeForCountryCode(countryCallingCode);
1592    PhoneMetadata metadata =
1593        getMetadataForRegionOrCallingCode(countryCallingCode, phoneNumberRegion);
1594    if (metadata == null) {
1595      return false;
1596    }
1597    String nationalNumber = getNationalSignificantNumber(number);
1598    NumberFormat formatRule =
1599        chooseFormattingPatternForNumber(metadata.numberFormats(), nationalNumber);
1600    return formatRule != null;
1601  }
1602
1603  /**
1604   * Formats a phone number for out-of-country dialing purposes.
1605   *
1606   * Note that in this version, if the number was entered originally using alpha characters and
1607   * this version of the number is stored in raw_input, this representation of the number will be
1608   * used rather than the digit representation. Grouping information, as specified by characters
1609   * such as "-" and " ", will be retained.
1610   *
1611   * <p><b>Caveats:</b></p>
1612   * <ul>
1613   *  <li> This will not produce good results if the country calling code is both present in the raw
1614   *       input _and_ is the start of the national number. This is not a problem in the regions
1615   *       which typically use alpha numbers.
1616   *  <li> This will also not produce good results if the raw input has any grouping information
1617   *       within the first three digits of the national number, and if the function needs to strip
1618   *       preceding digits/words in the raw input before these digits. Normally people group the
1619   *       first three digits together so this is not a huge problem - and will be fixed if it
1620   *       proves to be so.
1621   * </ul>
1622   *
1623   * @param number  the phone number that needs to be formatted
1624   * @param regionCallingFrom  the region where the call is being placed
1625   * @return  the formatted phone number
1626   */
1627  public String formatOutOfCountryKeepingAlphaChars(PhoneNumber number,
1628                                                    String regionCallingFrom) {
1629    String rawInput = number.getRawInput();
1630    // If there is no raw input, then we can't keep alpha characters because there aren't any.
1631    // In this case, we return formatOutOfCountryCallingNumber.
1632    if (rawInput.length() == 0) {
1633      return formatOutOfCountryCallingNumber(number, regionCallingFrom);
1634    }
1635    int countryCode = number.getCountryCode();
1636    if (!hasValidCountryCallingCode(countryCode)) {
1637      return rawInput;
1638    }
1639    // Strip any prefix such as country calling code, IDD, that was present. We do this by comparing
1640    // the number in raw_input with the parsed number.
1641    // To do this, first we normalize punctuation. We retain number grouping symbols such as " "
1642    // only.
1643    rawInput = normalizeHelper(rawInput, ALL_PLUS_NUMBER_GROUPING_SYMBOLS, true);
1644    // Now we trim everything before the first three digits in the parsed number. We choose three
1645    // because all valid alpha numbers have 3 digits at the start - if it does not, then we don't
1646    // trim anything at all. Similarly, if the national number was less than three digits, we don't
1647    // trim anything at all.
1648    String nationalNumber = getNationalSignificantNumber(number);
1649    if (nationalNumber.length() > 3) {
1650      int firstNationalNumberDigit = rawInput.indexOf(nationalNumber.substring(0, 3));
1651      if (firstNationalNumberDigit != -1) {
1652        rawInput = rawInput.substring(firstNationalNumberDigit);
1653      }
1654    }
1655    PhoneMetadata metadataForRegionCallingFrom = getMetadataForRegion(regionCallingFrom);
1656    if (countryCode == NANPA_COUNTRY_CODE) {
1657      if (isNANPACountry(regionCallingFrom)) {
1658        return countryCode + " " + rawInput;
1659      }
1660    } else if (metadataForRegionCallingFrom != null &&
1661               countryCode == getCountryCodeForValidRegion(regionCallingFrom)) {
1662      NumberFormat formattingPattern =
1663          chooseFormattingPatternForNumber(metadataForRegionCallingFrom.numberFormats(),
1664                                           nationalNumber);
1665      if (formattingPattern == null) {
1666        // If no pattern above is matched, we format the original input.
1667        return rawInput;
1668      }
1669      NumberFormat newFormat = new NumberFormat();
1670      newFormat.mergeFrom(formattingPattern);
1671      // The first group is the first group of digits that the user wrote together.
1672      newFormat.setPattern("(\\d+)(.*)");
1673      // Here we just concatenate them back together after the national prefix has been fixed.
1674      newFormat.setFormat("$1$2");
1675      // Now we format using this pattern instead of the default pattern, but with the national
1676      // prefix prefixed if necessary.
1677      // This will not work in the cases where the pattern (and not the leading digits) decide
1678      // whether a national prefix needs to be used, since we have overridden the pattern to match
1679      // anything, but that is not the case in the metadata to date.
1680      return formatNsnUsingPattern(rawInput, newFormat, PhoneNumberFormat.NATIONAL);
1681    }
1682    String internationalPrefixForFormatting = "";
1683    // If an unsupported region-calling-from is entered, or a country with multiple international
1684    // prefixes, the international format of the number is returned, unless there is a preferred
1685    // international prefix.
1686    if (metadataForRegionCallingFrom != null) {
1687      String internationalPrefix = metadataForRegionCallingFrom.getInternationalPrefix();
1688      internationalPrefixForFormatting =
1689          UNIQUE_INTERNATIONAL_PREFIX.matcher(internationalPrefix).matches()
1690          ? internationalPrefix
1691          : metadataForRegionCallingFrom.getPreferredInternationalPrefix();
1692    }
1693    StringBuilder formattedNumber = new StringBuilder(rawInput);
1694    String regionCode = getRegionCodeForCountryCode(countryCode);
1695    // Metadata cannot be null because the country calling code is valid.
1696    PhoneMetadata metadataForRegion = getMetadataForRegionOrCallingCode(countryCode, regionCode);
1697    maybeAppendFormattedExtension(number, metadataForRegion,
1698                                  PhoneNumberFormat.INTERNATIONAL, formattedNumber);
1699    if (internationalPrefixForFormatting.length() > 0) {
1700      formattedNumber.insert(0, " ").insert(0, countryCode).insert(0, " ")
1701          .insert(0, internationalPrefixForFormatting);
1702    } else {
1703      // Invalid region entered as country-calling-from (so no metadata was found for it) or the
1704      // region chosen has multiple international dialling prefixes.
1705      logger.log(Level.WARNING,
1706                 "Trying to format number from invalid region "
1707                 + regionCallingFrom
1708                 + ". International formatting applied.");
1709      prefixNumberWithCountryCallingCode(countryCode,
1710                                         PhoneNumberFormat.INTERNATIONAL,
1711                                         formattedNumber);
1712    }
1713    return formattedNumber.toString();
1714  }
1715
1716  /**
1717   * Gets the national significant number of the a phone number. Note a national significant number
1718   * doesn't contain a national prefix or any formatting.
1719   *
1720   * @param number  the phone number for which the national significant number is needed
1721   * @return  the national significant number of the PhoneNumber object passed in
1722   */
1723  public String getNationalSignificantNumber(PhoneNumber number) {
1724    // If leading zero(s) have been set, we prefix this now. Note this is not a national prefix.
1725    StringBuilder nationalNumber = new StringBuilder();
1726    if (number.isItalianLeadingZero()) {
1727      char[] zeros = new char[number.getNumberOfLeadingZeros()];
1728      Arrays.fill(zeros, '0');
1729      nationalNumber.append(new String(zeros));
1730    }
1731    nationalNumber.append(number.getNationalNumber());
1732    return nationalNumber.toString();
1733  }
1734
1735  /**
1736   * A helper function that is used by format and formatByPattern.
1737   */
1738  private void prefixNumberWithCountryCallingCode(int countryCallingCode,
1739                                                  PhoneNumberFormat numberFormat,
1740                                                  StringBuilder formattedNumber) {
1741    switch (numberFormat) {
1742      case E164:
1743        formattedNumber.insert(0, countryCallingCode).insert(0, PLUS_SIGN);
1744        return;
1745      case INTERNATIONAL:
1746        formattedNumber.insert(0, " ").insert(0, countryCallingCode).insert(0, PLUS_SIGN);
1747        return;
1748      case RFC3966:
1749        formattedNumber.insert(0, "-").insert(0, countryCallingCode).insert(0, PLUS_SIGN)
1750            .insert(0, RFC3966_PREFIX);
1751        return;
1752      case NATIONAL:
1753      default:
1754        return;
1755    }
1756  }
1757
1758  // Simple wrapper of formatNsn for the common case of no carrier code.
1759  private String formatNsn(String number, PhoneMetadata metadata, PhoneNumberFormat numberFormat) {
1760    return formatNsn(number, metadata, numberFormat, null);
1761  }
1762
1763  // Note in some regions, the national number can be written in two completely different ways
1764  // depending on whether it forms part of the NATIONAL format or INTERNATIONAL format. The
1765  // numberFormat parameter here is used to specify which format to use for those cases. If a
1766  // carrierCode is specified, this will be inserted into the formatted string to replace $CC.
1767  private String formatNsn(String number,
1768                           PhoneMetadata metadata,
1769                           PhoneNumberFormat numberFormat,
1770                           String carrierCode) {
1771    List<NumberFormat> intlNumberFormats = metadata.intlNumberFormats();
1772    // When the intlNumberFormats exists, we use that to format national number for the
1773    // INTERNATIONAL format instead of using the numberDesc.numberFormats.
1774    List<NumberFormat> availableFormats =
1775        (intlNumberFormats.size() == 0 || numberFormat == PhoneNumberFormat.NATIONAL)
1776        ? metadata.numberFormats()
1777        : metadata.intlNumberFormats();
1778    NumberFormat formattingPattern = chooseFormattingPatternForNumber(availableFormats, number);
1779    return (formattingPattern == null)
1780        ? number
1781        : formatNsnUsingPattern(number, formattingPattern, numberFormat, carrierCode);
1782  }
1783
1784  NumberFormat chooseFormattingPatternForNumber(List<NumberFormat> availableFormats,
1785                                                String nationalNumber) {
1786    for (NumberFormat numFormat : availableFormats) {
1787      int size = numFormat.leadingDigitsPatternSize();
1788      if (size == 0 || regexCache.getPatternForRegex(
1789              // We always use the last leading_digits_pattern, as it is the most detailed.
1790              numFormat.getLeadingDigitsPattern(size - 1)).matcher(nationalNumber).lookingAt()) {
1791        Matcher m = regexCache.getPatternForRegex(numFormat.getPattern()).matcher(nationalNumber);
1792        if (m.matches()) {
1793          return numFormat;
1794        }
1795      }
1796    }
1797    return null;
1798  }
1799
1800  // Simple wrapper of formatNsnUsingPattern for the common case of no carrier code.
1801  String formatNsnUsingPattern(String nationalNumber,
1802                               NumberFormat formattingPattern,
1803                               PhoneNumberFormat numberFormat) {
1804    return formatNsnUsingPattern(nationalNumber, formattingPattern, numberFormat, null);
1805  }
1806
1807  // Note that carrierCode is optional - if null or an empty string, no carrier code replacement
1808  // will take place.
1809  private String formatNsnUsingPattern(String nationalNumber,
1810                                       NumberFormat formattingPattern,
1811                                       PhoneNumberFormat numberFormat,
1812                                       String carrierCode) {
1813    String numberFormatRule = formattingPattern.getFormat();
1814    Matcher m =
1815        regexCache.getPatternForRegex(formattingPattern.getPattern()).matcher(nationalNumber);
1816    String formattedNationalNumber = "";
1817    if (numberFormat == PhoneNumberFormat.NATIONAL &&
1818        carrierCode != null && carrierCode.length() > 0 &&
1819        formattingPattern.getDomesticCarrierCodeFormattingRule().length() > 0) {
1820      // Replace the $CC in the formatting rule with the desired carrier code.
1821      String carrierCodeFormattingRule = formattingPattern.getDomesticCarrierCodeFormattingRule();
1822      carrierCodeFormattingRule =
1823          CC_PATTERN.matcher(carrierCodeFormattingRule).replaceFirst(carrierCode);
1824      // Now replace the $FG in the formatting rule with the first group and the carrier code
1825      // combined in the appropriate way.
1826      numberFormatRule = FIRST_GROUP_PATTERN.matcher(numberFormatRule)
1827          .replaceFirst(carrierCodeFormattingRule);
1828      formattedNationalNumber = m.replaceAll(numberFormatRule);
1829    } else {
1830      // Use the national prefix formatting rule instead.
1831      String nationalPrefixFormattingRule = formattingPattern.getNationalPrefixFormattingRule();
1832      if (numberFormat == PhoneNumberFormat.NATIONAL &&
1833          nationalPrefixFormattingRule != null &&
1834          nationalPrefixFormattingRule.length() > 0) {
1835        Matcher firstGroupMatcher = FIRST_GROUP_PATTERN.matcher(numberFormatRule);
1836        formattedNationalNumber =
1837            m.replaceAll(firstGroupMatcher.replaceFirst(nationalPrefixFormattingRule));
1838      } else {
1839        formattedNationalNumber = m.replaceAll(numberFormatRule);
1840      }
1841    }
1842    if (numberFormat == PhoneNumberFormat.RFC3966) {
1843      // Strip any leading punctuation.
1844      Matcher matcher = SEPARATOR_PATTERN.matcher(formattedNationalNumber);
1845      if (matcher.lookingAt()) {
1846        formattedNationalNumber = matcher.replaceFirst("");
1847      }
1848      // Replace the rest with a dash between each number group.
1849      formattedNationalNumber = matcher.reset(formattedNationalNumber).replaceAll("-");
1850    }
1851    return formattedNationalNumber;
1852  }
1853
1854  /**
1855   * Gets a valid number for the specified region.
1856   *
1857   * @param regionCode  the region for which an example number is needed
1858   * @return  a valid fixed-line number for the specified region. Returns null when the metadata
1859   *    does not contain such information, or the region 001 is passed in. For 001 (representing
1860   *    non-geographical numbers), call {@link #getExampleNumberForNonGeoEntity} instead.
1861   */
1862  public PhoneNumber getExampleNumber(String regionCode) {
1863    return getExampleNumberForType(regionCode, PhoneNumberType.FIXED_LINE);
1864  }
1865
1866  /**
1867   * Gets a valid number for the specified region and number type.
1868   *
1869   * @param regionCode  the region for which an example number is needed
1870   * @param type  the type of number that is needed
1871   * @return  a valid number for the specified region and type. Returns null when the metadata
1872   *     does not contain such information or if an invalid region or region 001 was entered.
1873   *     For 001 (representing non-geographical numbers), call
1874   *     {@link #getExampleNumberForNonGeoEntity} instead.
1875   */
1876  public PhoneNumber getExampleNumberForType(String regionCode, PhoneNumberType type) {
1877    // Check the region code is valid.
1878    if (!isValidRegionCode(regionCode)) {
1879      logger.log(Level.WARNING, "Invalid or unknown region code provided: " + regionCode);
1880      return null;
1881    }
1882    PhoneNumberDesc desc = getNumberDescByType(getMetadataForRegion(regionCode), type);
1883    try {
1884      if (desc.hasExampleNumber()) {
1885        return parse(desc.getExampleNumber(), regionCode);
1886      }
1887    } catch (NumberParseException e) {
1888      logger.log(Level.SEVERE, e.toString());
1889    }
1890    return null;
1891  }
1892
1893  /**
1894   * Gets a valid number for the specified country calling code for a non-geographical entity.
1895   *
1896   * @param countryCallingCode  the country calling code for a non-geographical entity
1897   * @return  a valid number for the non-geographical entity. Returns null when the metadata
1898   *    does not contain such information, or the country calling code passed in does not belong
1899   *    to a non-geographical entity.
1900   */
1901  public PhoneNumber getExampleNumberForNonGeoEntity(int countryCallingCode) {
1902    PhoneMetadata metadata = getMetadataForNonGeographicalRegion(countryCallingCode);
1903    if (metadata != null) {
1904      PhoneNumberDesc desc = metadata.getGeneralDesc();
1905      try {
1906        if (desc.hasExampleNumber()) {
1907          return parse("+" + countryCallingCode + desc.getExampleNumber(), "ZZ");
1908        }
1909      } catch (NumberParseException e) {
1910        logger.log(Level.SEVERE, e.toString());
1911      }
1912    } else {
1913      logger.log(Level.WARNING,
1914                 "Invalid or unknown country calling code provided: " + countryCallingCode);
1915    }
1916    return null;
1917  }
1918
1919  /**
1920   * Appends the formatted extension of a phone number to formattedNumber, if the phone number had
1921   * an extension specified.
1922   */
1923  private void maybeAppendFormattedExtension(PhoneNumber number, PhoneMetadata metadata,
1924                                             PhoneNumberFormat numberFormat,
1925                                             StringBuilder formattedNumber) {
1926    if (number.hasExtension() && number.getExtension().length() > 0) {
1927      if (numberFormat == PhoneNumberFormat.RFC3966) {
1928        formattedNumber.append(RFC3966_EXTN_PREFIX).append(number.getExtension());
1929      } else {
1930        if (metadata.hasPreferredExtnPrefix()) {
1931          formattedNumber.append(metadata.getPreferredExtnPrefix()).append(number.getExtension());
1932        } else {
1933          formattedNumber.append(DEFAULT_EXTN_PREFIX).append(number.getExtension());
1934        }
1935      }
1936    }
1937  }
1938
1939  PhoneNumberDesc getNumberDescByType(PhoneMetadata metadata, PhoneNumberType type) {
1940    switch (type) {
1941      case PREMIUM_RATE:
1942        return metadata.getPremiumRate();
1943      case TOLL_FREE:
1944        return metadata.getTollFree();
1945      case MOBILE:
1946        return metadata.getMobile();
1947      case FIXED_LINE:
1948      case FIXED_LINE_OR_MOBILE:
1949        return metadata.getFixedLine();
1950      case SHARED_COST:
1951        return metadata.getSharedCost();
1952      case VOIP:
1953        return metadata.getVoip();
1954      case PERSONAL_NUMBER:
1955        return metadata.getPersonalNumber();
1956      case PAGER:
1957        return metadata.getPager();
1958      case UAN:
1959        return metadata.getUan();
1960      case VOICEMAIL:
1961        return metadata.getVoicemail();
1962      default:
1963        return metadata.getGeneralDesc();
1964    }
1965  }
1966
1967  /**
1968   * Gets the type of a phone number.
1969   *
1970   * @param number  the phone number that we want to know the type
1971   * @return  the type of the phone number
1972   */
1973  public PhoneNumberType getNumberType(PhoneNumber number) {
1974    String regionCode = getRegionCodeForNumber(number);
1975    PhoneMetadata metadata = getMetadataForRegionOrCallingCode(number.getCountryCode(), regionCode);
1976    if (metadata == null) {
1977      return PhoneNumberType.UNKNOWN;
1978    }
1979    String nationalSignificantNumber = getNationalSignificantNumber(number);
1980    return getNumberTypeHelper(nationalSignificantNumber, metadata);
1981  }
1982
1983  private PhoneNumberType getNumberTypeHelper(String nationalNumber, PhoneMetadata metadata) {
1984    PhoneNumberDesc generalNumberDesc = metadata.getGeneralDesc();
1985    if (!generalNumberDesc.hasNationalNumberPattern() ||
1986        !isNumberMatchingDesc(nationalNumber, generalNumberDesc)) {
1987      return PhoneNumberType.UNKNOWN;
1988    }
1989
1990    if (isNumberMatchingDesc(nationalNumber, metadata.getPremiumRate())) {
1991      return PhoneNumberType.PREMIUM_RATE;
1992    }
1993    if (isNumberMatchingDesc(nationalNumber, metadata.getTollFree())) {
1994      return PhoneNumberType.TOLL_FREE;
1995    }
1996    if (isNumberMatchingDesc(nationalNumber, metadata.getSharedCost())) {
1997      return PhoneNumberType.SHARED_COST;
1998    }
1999    if (isNumberMatchingDesc(nationalNumber, metadata.getVoip())) {
2000      return PhoneNumberType.VOIP;
2001    }
2002    if (isNumberMatchingDesc(nationalNumber, metadata.getPersonalNumber())) {
2003      return PhoneNumberType.PERSONAL_NUMBER;
2004    }
2005    if (isNumberMatchingDesc(nationalNumber, metadata.getPager())) {
2006      return PhoneNumberType.PAGER;
2007    }
2008    if (isNumberMatchingDesc(nationalNumber, metadata.getUan())) {
2009      return PhoneNumberType.UAN;
2010    }
2011    if (isNumberMatchingDesc(nationalNumber, metadata.getVoicemail())) {
2012      return PhoneNumberType.VOICEMAIL;
2013    }
2014
2015    boolean isFixedLine = isNumberMatchingDesc(nationalNumber, metadata.getFixedLine());
2016    if (isFixedLine) {
2017      if (metadata.isSameMobileAndFixedLinePattern()) {
2018        return PhoneNumberType.FIXED_LINE_OR_MOBILE;
2019      } else if (isNumberMatchingDesc(nationalNumber, metadata.getMobile())) {
2020        return PhoneNumberType.FIXED_LINE_OR_MOBILE;
2021      }
2022      return PhoneNumberType.FIXED_LINE;
2023    }
2024    // Otherwise, test to see if the number is mobile. Only do this if certain that the patterns for
2025    // mobile and fixed line aren't the same.
2026    if (!metadata.isSameMobileAndFixedLinePattern() &&
2027        isNumberMatchingDesc(nationalNumber, metadata.getMobile())) {
2028      return PhoneNumberType.MOBILE;
2029    }
2030    return PhoneNumberType.UNKNOWN;
2031  }
2032
2033  /**
2034   * Returns the metadata for the given region code or {@code null} if the region code is invalid
2035   * or unknown.
2036   */
2037  PhoneMetadata getMetadataForRegion(String regionCode) {
2038    if (!isValidRegionCode(regionCode)) {
2039      return null;
2040    }
2041    synchronized (regionToMetadataMap) {
2042      if (!regionToMetadataMap.containsKey(regionCode)) {
2043        // The regionCode here will be valid and won't be '001', so we don't need to worry about
2044        // what to pass in for the country calling code.
2045        loadMetadataFromFile(currentFilePrefix, regionCode, 0, metadataLoader);
2046      }
2047    }
2048    return regionToMetadataMap.get(regionCode);
2049  }
2050
2051  PhoneMetadata getMetadataForNonGeographicalRegion(int countryCallingCode) {
2052    synchronized (countryCodeToNonGeographicalMetadataMap) {
2053      if (!countryCallingCodeToRegionCodeMap.containsKey(countryCallingCode)) {
2054        return null;
2055      }
2056      if (!countryCodeToNonGeographicalMetadataMap.containsKey(countryCallingCode)) {
2057        loadMetadataFromFile(
2058            currentFilePrefix, REGION_CODE_FOR_NON_GEO_ENTITY, countryCallingCode, metadataLoader);
2059      }
2060    }
2061    return countryCodeToNonGeographicalMetadataMap.get(countryCallingCode);
2062  }
2063
2064  boolean isNumberPossibleForDesc(String nationalNumber, PhoneNumberDesc numberDesc) {
2065    Matcher possibleNumberPatternMatcher =
2066        regexCache.getPatternForRegex(numberDesc.getPossibleNumberPattern())
2067            .matcher(nationalNumber);
2068    return possibleNumberPatternMatcher.matches();
2069  }
2070
2071  boolean isNumberMatchingDesc(String nationalNumber, PhoneNumberDesc numberDesc) {
2072    Matcher nationalNumberPatternMatcher =
2073        regexCache.getPatternForRegex(numberDesc.getNationalNumberPattern())
2074            .matcher(nationalNumber);
2075    return isNumberPossibleForDesc(nationalNumber, numberDesc) &&
2076        nationalNumberPatternMatcher.matches();
2077  }
2078
2079  /**
2080   * Tests whether a phone number matches a valid pattern. Note this doesn't verify the number
2081   * is actually in use, which is impossible to tell by just looking at a number itself.
2082   *
2083   * @param number       the phone number that we want to validate
2084   * @return  a boolean that indicates whether the number is of a valid pattern
2085   */
2086  public boolean isValidNumber(PhoneNumber number) {
2087    String regionCode = getRegionCodeForNumber(number);
2088    return isValidNumberForRegion(number, regionCode);
2089  }
2090
2091  /**
2092   * Tests whether a phone number is valid for a certain region. Note this doesn't verify the number
2093   * is actually in use, which is impossible to tell by just looking at a number itself. If the
2094   * country calling code is not the same as the country calling code for the region, this
2095   * immediately exits with false. After this, the specific number pattern rules for the region are
2096   * examined. This is useful for determining for example whether a particular number is valid for
2097   * Canada, rather than just a valid NANPA number.
2098   * Warning: In most cases, you want to use {@link #isValidNumber} instead. For example, this
2099   * method will mark numbers from British Crown dependencies such as the Isle of Man as invalid for
2100   * the region "GB" (United Kingdom), since it has its own region code, "IM", which may be
2101   * undesirable.
2102   *
2103   * @param number       the phone number that we want to validate
2104   * @param regionCode   the region that we want to validate the phone number for
2105   * @return  a boolean that indicates whether the number is of a valid pattern
2106   */
2107  public boolean isValidNumberForRegion(PhoneNumber number, String regionCode) {
2108    int countryCode = number.getCountryCode();
2109    PhoneMetadata metadata = getMetadataForRegionOrCallingCode(countryCode, regionCode);
2110    if ((metadata == null) ||
2111        (!REGION_CODE_FOR_NON_GEO_ENTITY.equals(regionCode) &&
2112         countryCode != getCountryCodeForValidRegion(regionCode))) {
2113      // Either the region code was invalid, or the country calling code for this number does not
2114      // match that of the region code.
2115      return false;
2116    }
2117    PhoneNumberDesc generalNumDesc = metadata.getGeneralDesc();
2118    String nationalSignificantNumber = getNationalSignificantNumber(number);
2119
2120    // For regions where we don't have metadata for PhoneNumberDesc, we treat any number passed in
2121    // as a valid number if its national significant number is between the minimum and maximum
2122    // lengths defined by ITU for a national significant number.
2123    if (!generalNumDesc.hasNationalNumberPattern()) {
2124      int numberLength = nationalSignificantNumber.length();
2125      return numberLength > MIN_LENGTH_FOR_NSN && numberLength <= MAX_LENGTH_FOR_NSN;
2126    }
2127    return getNumberTypeHelper(nationalSignificantNumber, metadata) != PhoneNumberType.UNKNOWN;
2128  }
2129
2130  /**
2131   * Returns the region where a phone number is from. This could be used for geocoding at the region
2132   * level.
2133   *
2134   * @param number  the phone number whose origin we want to know
2135   * @return  the region where the phone number is from, or null if no region matches this calling
2136   *     code
2137   */
2138  public String getRegionCodeForNumber(PhoneNumber number) {
2139    int countryCode = number.getCountryCode();
2140    List<String> regions = countryCallingCodeToRegionCodeMap.get(countryCode);
2141    if (regions == null) {
2142      String numberString = getNationalSignificantNumber(number);
2143      logger.log(Level.WARNING,
2144                 "Missing/invalid country_code (" + countryCode + ") for number " + numberString);
2145      return null;
2146    }
2147    if (regions.size() == 1) {
2148      return regions.get(0);
2149    } else {
2150      return getRegionCodeForNumberFromRegionList(number, regions);
2151    }
2152  }
2153
2154  private String getRegionCodeForNumberFromRegionList(PhoneNumber number,
2155                                                      List<String> regionCodes) {
2156    String nationalNumber = getNationalSignificantNumber(number);
2157    for (String regionCode : regionCodes) {
2158      // If leadingDigits is present, use this. Otherwise, do full validation.
2159      // Metadata cannot be null because the region codes come from the country calling code map.
2160      PhoneMetadata metadata = getMetadataForRegion(regionCode);
2161      if (metadata.hasLeadingDigits()) {
2162        if (regexCache.getPatternForRegex(metadata.getLeadingDigits())
2163                .matcher(nationalNumber).lookingAt()) {
2164          return regionCode;
2165        }
2166      } else if (getNumberTypeHelper(nationalNumber, metadata) != PhoneNumberType.UNKNOWN) {
2167        return regionCode;
2168      }
2169    }
2170    return null;
2171  }
2172
2173  /**
2174   * Returns the region code that matches the specific country calling code. In the case of no
2175   * region code being found, ZZ will be returned. In the case of multiple regions, the one
2176   * designated in the metadata as the "main" region for this calling code will be returned. If the
2177   * countryCallingCode entered is valid but doesn't match a specific region (such as in the case of
2178   * non-geographical calling codes like 800) the value "001" will be returned (corresponding to
2179   * the value for World in the UN M.49 schema).
2180   */
2181  public String getRegionCodeForCountryCode(int countryCallingCode) {
2182    List<String> regionCodes = countryCallingCodeToRegionCodeMap.get(countryCallingCode);
2183    return regionCodes == null ? UNKNOWN_REGION : regionCodes.get(0);
2184  }
2185
2186  /**
2187   * Returns a list with the region codes that match the specific country calling code. For
2188   * non-geographical country calling codes, the region code 001 is returned. Also, in the case
2189   * of no region code being found, an empty list is returned.
2190   */
2191  public List<String> getRegionCodesForCountryCode(int countryCallingCode) {
2192    List<String> regionCodes = countryCallingCodeToRegionCodeMap.get(countryCallingCode);
2193    return Collections.unmodifiableList(regionCodes == null ? new ArrayList<String>(0)
2194                                                            : regionCodes);
2195  }
2196
2197  /**
2198   * Returns the country calling code for a specific region. For example, this would be 1 for the
2199   * United States, and 64 for New Zealand.
2200   *
2201   * @param regionCode  the region that we want to get the country calling code for
2202   * @return  the country calling code for the region denoted by regionCode
2203   */
2204  public int getCountryCodeForRegion(String regionCode) {
2205    if (!isValidRegionCode(regionCode)) {
2206      logger.log(Level.WARNING,
2207                 "Invalid or missing region code ("
2208                  + ((regionCode == null) ? "null" : regionCode)
2209                  + ") provided.");
2210      return 0;
2211    }
2212    return getCountryCodeForValidRegion(regionCode);
2213  }
2214
2215  /**
2216   * Returns the country calling code for a specific region. For example, this would be 1 for the
2217   * United States, and 64 for New Zealand. Assumes the region is already valid.
2218   *
2219   * @param regionCode  the region that we want to get the country calling code for
2220   * @return  the country calling code for the region denoted by regionCode
2221   * @throws IllegalArgumentException if the region is invalid
2222   */
2223  private int getCountryCodeForValidRegion(String regionCode) {
2224    PhoneMetadata metadata = getMetadataForRegion(regionCode);
2225    if (metadata == null) {
2226      throw new IllegalArgumentException("Invalid region code: " + regionCode);
2227    }
2228    return metadata.getCountryCode();
2229  }
2230
2231  /**
2232   * Returns the national dialling prefix for a specific region. For example, this would be 1 for
2233   * the United States, and 0 for New Zealand. Set stripNonDigits to true to strip symbols like "~"
2234   * (which indicates a wait for a dialling tone) from the prefix returned. If no national prefix is
2235   * present, we return null.
2236   *
2237   * <p>Warning: Do not use this method for do-your-own formatting - for some regions, the
2238   * national dialling prefix is used only for certain types of numbers. Use the library's
2239   * formatting functions to prefix the national prefix when required.
2240   *
2241   * @param regionCode  the region that we want to get the dialling prefix for
2242   * @param stripNonDigits  true to strip non-digits from the national dialling prefix
2243   * @return  the dialling prefix for the region denoted by regionCode
2244   */
2245  public String getNddPrefixForRegion(String regionCode, boolean stripNonDigits) {
2246    PhoneMetadata metadata = getMetadataForRegion(regionCode);
2247    if (metadata == null) {
2248      logger.log(Level.WARNING,
2249                 "Invalid or missing region code ("
2250                  + ((regionCode == null) ? "null" : regionCode)
2251                  + ") provided.");
2252      return null;
2253    }
2254    String nationalPrefix = metadata.getNationalPrefix();
2255    // If no national prefix was found, we return null.
2256    if (nationalPrefix.length() == 0) {
2257      return null;
2258    }
2259    if (stripNonDigits) {
2260      // Note: if any other non-numeric symbols are ever used in national prefixes, these would have
2261      // to be removed here as well.
2262      nationalPrefix = nationalPrefix.replace("~", "");
2263    }
2264    return nationalPrefix;
2265  }
2266
2267  /**
2268   * Checks if this is a region under the North American Numbering Plan Administration (NANPA).
2269   *
2270   * @return  true if regionCode is one of the regions under NANPA
2271   */
2272  public boolean isNANPACountry(String regionCode) {
2273    return nanpaRegions.contains(regionCode);
2274  }
2275
2276  /**
2277   * Checks whether the country calling code is from a region whose national significant number
2278   * could contain a leading zero. An example of such a region is Italy. Returns false if no
2279   * metadata for the country is found.
2280   */
2281  boolean isLeadingZeroPossible(int countryCallingCode) {
2282    PhoneMetadata mainMetadataForCallingCode =
2283        getMetadataForRegionOrCallingCode(countryCallingCode,
2284                                          getRegionCodeForCountryCode(countryCallingCode));
2285    if (mainMetadataForCallingCode == null) {
2286      return false;
2287    }
2288    return mainMetadataForCallingCode.isLeadingZeroPossible();
2289  }
2290
2291  /**
2292   * Checks if the number is a valid vanity (alpha) number such as 800 MICROSOFT. A valid vanity
2293   * number will start with at least 3 digits and will have three or more alpha characters. This
2294   * does not do region-specific checks - to work out if this number is actually valid for a region,
2295   * it should be parsed and methods such as {@link #isPossibleNumberWithReason} and
2296   * {@link #isValidNumber} should be used.
2297   *
2298   * @param number  the number that needs to be checked
2299   * @return  true if the number is a valid vanity number
2300   */
2301  public boolean isAlphaNumber(String number) {
2302    if (!isViablePhoneNumber(number)) {
2303      // Number is too short, or doesn't match the basic phone number pattern.
2304      return false;
2305    }
2306    StringBuilder strippedNumber = new StringBuilder(number);
2307    maybeStripExtension(strippedNumber);
2308    return VALID_ALPHA_PHONE_PATTERN.matcher(strippedNumber).matches();
2309  }
2310
2311  /**
2312   * Convenience wrapper around {@link #isPossibleNumberWithReason}. Instead of returning the reason
2313   * for failure, this method returns a boolean value.
2314   * @param number  the number that needs to be checked
2315   * @return  true if the number is possible
2316   */
2317  public boolean isPossibleNumber(PhoneNumber number) {
2318    return isPossibleNumberWithReason(number) == ValidationResult.IS_POSSIBLE;
2319  }
2320
2321  /**
2322   * Helper method to check a number against a particular pattern and determine whether it matches,
2323   * or is too short or too long. Currently, if a number pattern suggests that numbers of length 7
2324   * and 10 are possible, and a number in between these possible lengths is entered, such as of
2325   * length 8, this will return TOO_LONG.
2326   */
2327  private ValidationResult testNumberLengthAgainstPattern(Pattern numberPattern, String number) {
2328    Matcher numberMatcher = numberPattern.matcher(number);
2329    if (numberMatcher.matches()) {
2330      return ValidationResult.IS_POSSIBLE;
2331    }
2332    if (numberMatcher.lookingAt()) {
2333      return ValidationResult.TOO_LONG;
2334    } else {
2335      return ValidationResult.TOO_SHORT;
2336    }
2337  }
2338
2339  /**
2340   * Helper method to check whether a number is too short to be a regular length phone number in a
2341   * region.
2342   */
2343  private boolean isShorterThanPossibleNormalNumber(PhoneMetadata regionMetadata, String number) {
2344    Pattern possibleNumberPattern = regexCache.getPatternForRegex(
2345        regionMetadata.getGeneralDesc().getPossibleNumberPattern());
2346    return testNumberLengthAgainstPattern(possibleNumberPattern, number) ==
2347        ValidationResult.TOO_SHORT;
2348  }
2349
2350  /**
2351   * Check whether a phone number is a possible number. It provides a more lenient check than
2352   * {@link #isValidNumber} in the following sense:
2353   *<ol>
2354   * <li> It only checks the length of phone numbers. In particular, it doesn't check starting
2355   *      digits of the number.
2356   * <li> It doesn't attempt to figure out the type of the number, but uses general rules which
2357   *      applies to all types of phone numbers in a region. Therefore, it is much faster than
2358   *      isValidNumber.
2359   * <li> For fixed line numbers, many regions have the concept of area code, which together with
2360   *      subscriber number constitute the national significant number. It is sometimes okay to dial
2361   *      the subscriber number only when dialing in the same area. This function will return
2362   *      true if the subscriber-number-only version is passed in. On the other hand, because
2363   *      isValidNumber validates using information on both starting digits (for fixed line
2364   *      numbers, that would most likely be area codes) and length (obviously includes the
2365   *      length of area codes for fixed line numbers), it will return false for the
2366   *      subscriber-number-only version.
2367   * </ol>
2368   * @param number  the number that needs to be checked
2369   * @return  a ValidationResult object which indicates whether the number is possible
2370   */
2371  public ValidationResult isPossibleNumberWithReason(PhoneNumber number) {
2372    String nationalNumber = getNationalSignificantNumber(number);
2373    int countryCode = number.getCountryCode();
2374    // Note: For Russian Fed and NANPA numbers, we just use the rules from the default region (US or
2375    // Russia) since the getRegionCodeForNumber will not work if the number is possible but not
2376    // valid. This would need to be revisited if the possible number pattern ever differed between
2377    // various regions within those plans.
2378    if (!hasValidCountryCallingCode(countryCode)) {
2379      return ValidationResult.INVALID_COUNTRY_CODE;
2380    }
2381    String regionCode = getRegionCodeForCountryCode(countryCode);
2382    // Metadata cannot be null because the country calling code is valid.
2383    PhoneMetadata metadata = getMetadataForRegionOrCallingCode(countryCode, regionCode);
2384    PhoneNumberDesc generalNumDesc = metadata.getGeneralDesc();
2385    // Handling case of numbers with no metadata.
2386    if (!generalNumDesc.hasNationalNumberPattern()) {
2387      logger.log(Level.FINER, "Checking if number is possible with incomplete metadata.");
2388      int numberLength = nationalNumber.length();
2389      if (numberLength < MIN_LENGTH_FOR_NSN) {
2390        return ValidationResult.TOO_SHORT;
2391      } else if (numberLength > MAX_LENGTH_FOR_NSN) {
2392        return ValidationResult.TOO_LONG;
2393      } else {
2394        return ValidationResult.IS_POSSIBLE;
2395      }
2396    }
2397    Pattern possibleNumberPattern =
2398        regexCache.getPatternForRegex(generalNumDesc.getPossibleNumberPattern());
2399    return testNumberLengthAgainstPattern(possibleNumberPattern, nationalNumber);
2400  }
2401
2402  /**
2403   * Check whether a phone number is a possible number given a number in the form of a string, and
2404   * the region where the number could be dialed from. It provides a more lenient check than
2405   * {@link #isValidNumber}. See {@link #isPossibleNumber(PhoneNumber)} for details.
2406   *
2407   * <p>This method first parses the number, then invokes {@link #isPossibleNumber(PhoneNumber)}
2408   * with the resultant PhoneNumber object.
2409   *
2410   * @param number  the number that needs to be checked, in the form of a string
2411   * @param regionDialingFrom  the region that we are expecting the number to be dialed from.
2412   *     Note this is different from the region where the number belongs.  For example, the number
2413   *     +1 650 253 0000 is a number that belongs to US. When written in this form, it can be
2414   *     dialed from any region. When it is written as 00 1 650 253 0000, it can be dialed from any
2415   *     region which uses an international dialling prefix of 00. When it is written as
2416   *     650 253 0000, it can only be dialed from within the US, and when written as 253 0000, it
2417   *     can only be dialed from within a smaller area in the US (Mountain View, CA, to be more
2418   *     specific).
2419   * @return  true if the number is possible
2420   */
2421  public boolean isPossibleNumber(String number, String regionDialingFrom) {
2422    try {
2423      return isPossibleNumber(parse(number, regionDialingFrom));
2424    } catch (NumberParseException e) {
2425      return false;
2426    }
2427  }
2428
2429  /**
2430   * Attempts to extract a valid number from a phone number that is too long to be valid, and resets
2431   * the PhoneNumber object passed in to that valid version. If no valid number could be extracted,
2432   * the PhoneNumber object passed in will not be modified.
2433   * @param number a PhoneNumber object which contains a number that is too long to be valid.
2434   * @return  true if a valid phone number can be successfully extracted.
2435   */
2436  public boolean truncateTooLongNumber(PhoneNumber number) {
2437    if (isValidNumber(number)) {
2438      return true;
2439    }
2440    PhoneNumber numberCopy = new PhoneNumber();
2441    numberCopy.mergeFrom(number);
2442    long nationalNumber = number.getNationalNumber();
2443    do {
2444      nationalNumber /= 10;
2445      numberCopy.setNationalNumber(nationalNumber);
2446      if (isPossibleNumberWithReason(numberCopy) == ValidationResult.TOO_SHORT ||
2447          nationalNumber == 0) {
2448        return false;
2449      }
2450    } while (!isValidNumber(numberCopy));
2451    number.setNationalNumber(nationalNumber);
2452    return true;
2453  }
2454
2455  /**
2456   * Gets an {@link com.android.i18n.phonenumbers.AsYouTypeFormatter} for the specific region.
2457   *
2458   * @param regionCode  the region where the phone number is being entered
2459   * @return  an {@link com.android.i18n.phonenumbers.AsYouTypeFormatter} object, which can be used
2460   *     to format phone numbers in the specific region "as you type"
2461   */
2462  public AsYouTypeFormatter getAsYouTypeFormatter(String regionCode) {
2463    return new AsYouTypeFormatter(regionCode);
2464  }
2465
2466  // Extracts country calling code from fullNumber, returns it and places the remaining number in
2467  // nationalNumber. It assumes that the leading plus sign or IDD has already been removed. Returns
2468  // 0 if fullNumber doesn't start with a valid country calling code, and leaves nationalNumber
2469  // unmodified.
2470  int extractCountryCode(StringBuilder fullNumber, StringBuilder nationalNumber) {
2471    if ((fullNumber.length() == 0) || (fullNumber.charAt(0) == '0')) {
2472      // Country codes do not begin with a '0'.
2473      return 0;
2474    }
2475    int potentialCountryCode;
2476    int numberLength = fullNumber.length();
2477    for (int i = 1; i <= MAX_LENGTH_COUNTRY_CODE && i <= numberLength; i++) {
2478      potentialCountryCode = Integer.parseInt(fullNumber.substring(0, i));
2479      if (countryCallingCodeToRegionCodeMap.containsKey(potentialCountryCode)) {
2480        nationalNumber.append(fullNumber.substring(i));
2481        return potentialCountryCode;
2482      }
2483    }
2484    return 0;
2485  }
2486
2487  /**
2488   * Tries to extract a country calling code from a number. This method will return zero if no
2489   * country calling code is considered to be present. Country calling codes are extracted in the
2490   * following ways:
2491   * <ul>
2492   *  <li> by stripping the international dialing prefix of the region the person is dialing from,
2493   *       if this is present in the number, and looking at the next digits
2494   *  <li> by stripping the '+' sign if present and then looking at the next digits
2495   *  <li> by comparing the start of the number and the country calling code of the default region.
2496   *       If the number is not considered possible for the numbering plan of the default region
2497   *       initially, but starts with the country calling code of this region, validation will be
2498   *       reattempted after stripping this country calling code. If this number is considered a
2499   *       possible number, then the first digits will be considered the country calling code and
2500   *       removed as such.
2501   * </ul>
2502   * It will throw a NumberParseException if the number starts with a '+' but the country calling
2503   * code supplied after this does not match that of any known region.
2504   *
2505   * @param number  non-normalized telephone number that we wish to extract a country calling
2506   *     code from - may begin with '+'
2507   * @param defaultRegionMetadata  metadata about the region this number may be from
2508   * @param nationalNumber  a string buffer to store the national significant number in, in the case
2509   *     that a country calling code was extracted. The number is appended to any existing contents.
2510   *     If no country calling code was extracted, this will be left unchanged.
2511   * @param keepRawInput  true if the country_code_source and preferred_carrier_code fields of
2512   *     phoneNumber should be populated.
2513   * @param phoneNumber  the PhoneNumber object where the country_code and country_code_source need
2514   *     to be populated. Note the country_code is always populated, whereas country_code_source is
2515   *     only populated when keepCountryCodeSource is true.
2516   * @return  the country calling code extracted or 0 if none could be extracted
2517   */
2518  // @VisibleForTesting
2519  int maybeExtractCountryCode(String number, PhoneMetadata defaultRegionMetadata,
2520                              StringBuilder nationalNumber, boolean keepRawInput,
2521                              PhoneNumber phoneNumber)
2522      throws NumberParseException {
2523    if (number.length() == 0) {
2524      return 0;
2525    }
2526    StringBuilder fullNumber = new StringBuilder(number);
2527    // Set the default prefix to be something that will never match.
2528    String possibleCountryIddPrefix = "NonMatch";
2529    if (defaultRegionMetadata != null) {
2530      possibleCountryIddPrefix = defaultRegionMetadata.getInternationalPrefix();
2531    }
2532
2533    CountryCodeSource countryCodeSource =
2534        maybeStripInternationalPrefixAndNormalize(fullNumber, possibleCountryIddPrefix);
2535    if (keepRawInput) {
2536      phoneNumber.setCountryCodeSource(countryCodeSource);
2537    }
2538    if (countryCodeSource != CountryCodeSource.FROM_DEFAULT_COUNTRY) {
2539      if (fullNumber.length() <= MIN_LENGTH_FOR_NSN) {
2540        throw new NumberParseException(NumberParseException.ErrorType.TOO_SHORT_AFTER_IDD,
2541                                       "Phone number had an IDD, but after this was not "
2542                                       + "long enough to be a viable phone number.");
2543      }
2544      int potentialCountryCode = extractCountryCode(fullNumber, nationalNumber);
2545      if (potentialCountryCode != 0) {
2546        phoneNumber.setCountryCode(potentialCountryCode);
2547        return potentialCountryCode;
2548      }
2549
2550      // If this fails, they must be using a strange country calling code that we don't recognize,
2551      // or that doesn't exist.
2552      throw new NumberParseException(NumberParseException.ErrorType.INVALID_COUNTRY_CODE,
2553                                     "Country calling code supplied was not recognised.");
2554    } else if (defaultRegionMetadata != null) {
2555      // Check to see if the number starts with the country calling code for the default region. If
2556      // so, we remove the country calling code, and do some checks on the validity of the number
2557      // before and after.
2558      int defaultCountryCode = defaultRegionMetadata.getCountryCode();
2559      String defaultCountryCodeString = String.valueOf(defaultCountryCode);
2560      String normalizedNumber = fullNumber.toString();
2561      if (normalizedNumber.startsWith(defaultCountryCodeString)) {
2562        StringBuilder potentialNationalNumber =
2563            new StringBuilder(normalizedNumber.substring(defaultCountryCodeString.length()));
2564        PhoneNumberDesc generalDesc = defaultRegionMetadata.getGeneralDesc();
2565        Pattern validNumberPattern =
2566            regexCache.getPatternForRegex(generalDesc.getNationalNumberPattern());
2567        maybeStripNationalPrefixAndCarrierCode(
2568            potentialNationalNumber, defaultRegionMetadata, null /* Don't need the carrier code */);
2569        Pattern possibleNumberPattern =
2570            regexCache.getPatternForRegex(generalDesc.getPossibleNumberPattern());
2571        // If the number was not valid before but is valid now, or if it was too long before, we
2572        // consider the number with the country calling code stripped to be a better result and
2573        // keep that instead.
2574        if ((!validNumberPattern.matcher(fullNumber).matches() &&
2575             validNumberPattern.matcher(potentialNationalNumber).matches()) ||
2576             testNumberLengthAgainstPattern(possibleNumberPattern, fullNumber.toString())
2577                  == ValidationResult.TOO_LONG) {
2578          nationalNumber.append(potentialNationalNumber);
2579          if (keepRawInput) {
2580            phoneNumber.setCountryCodeSource(CountryCodeSource.FROM_NUMBER_WITHOUT_PLUS_SIGN);
2581          }
2582          phoneNumber.setCountryCode(defaultCountryCode);
2583          return defaultCountryCode;
2584        }
2585      }
2586    }
2587    // No country calling code present.
2588    phoneNumber.setCountryCode(0);
2589    return 0;
2590  }
2591
2592  /**
2593   * Strips the IDD from the start of the number if present. Helper function used by
2594   * maybeStripInternationalPrefixAndNormalize.
2595   */
2596  private boolean parsePrefixAsIdd(Pattern iddPattern, StringBuilder number) {
2597    Matcher m = iddPattern.matcher(number);
2598    if (m.lookingAt()) {
2599      int matchEnd = m.end();
2600      // Only strip this if the first digit after the match is not a 0, since country calling codes
2601      // cannot begin with 0.
2602      Matcher digitMatcher = CAPTURING_DIGIT_PATTERN.matcher(number.substring(matchEnd));
2603      if (digitMatcher.find()) {
2604        String normalizedGroup = normalizeDigitsOnly(digitMatcher.group(1));
2605        if (normalizedGroup.equals("0")) {
2606          return false;
2607        }
2608      }
2609      number.delete(0, matchEnd);
2610      return true;
2611    }
2612    return false;
2613  }
2614
2615  /**
2616   * Strips any international prefix (such as +, 00, 011) present in the number provided, normalizes
2617   * the resulting number, and indicates if an international prefix was present.
2618   *
2619   * @param number  the non-normalized telephone number that we wish to strip any international
2620   *     dialing prefix from.
2621   * @param possibleIddPrefix  the international direct dialing prefix from the region we
2622   *     think this number may be dialed in
2623   * @return  the corresponding CountryCodeSource if an international dialing prefix could be
2624   *     removed from the number, otherwise CountryCodeSource.FROM_DEFAULT_COUNTRY if the number did
2625   *     not seem to be in international format.
2626   */
2627  // @VisibleForTesting
2628  CountryCodeSource maybeStripInternationalPrefixAndNormalize(
2629      StringBuilder number,
2630      String possibleIddPrefix) {
2631    if (number.length() == 0) {
2632      return CountryCodeSource.FROM_DEFAULT_COUNTRY;
2633    }
2634    // Check to see if the number begins with one or more plus signs.
2635    Matcher m = PLUS_CHARS_PATTERN.matcher(number);
2636    if (m.lookingAt()) {
2637      number.delete(0, m.end());
2638      // Can now normalize the rest of the number since we've consumed the "+" sign at the start.
2639      normalize(number);
2640      return CountryCodeSource.FROM_NUMBER_WITH_PLUS_SIGN;
2641    }
2642    // Attempt to parse the first digits as an international prefix.
2643    Pattern iddPattern = regexCache.getPatternForRegex(possibleIddPrefix);
2644    normalize(number);
2645    return parsePrefixAsIdd(iddPattern, number)
2646           ? CountryCodeSource.FROM_NUMBER_WITH_IDD
2647           : CountryCodeSource.FROM_DEFAULT_COUNTRY;
2648  }
2649
2650  /**
2651   * Strips any national prefix (such as 0, 1) present in the number provided.
2652   *
2653   * @param number  the normalized telephone number that we wish to strip any national
2654   *     dialing prefix from
2655   * @param metadata  the metadata for the region that we think this number is from
2656   * @param carrierCode  a place to insert the carrier code if one is extracted
2657   * @return true if a national prefix or carrier code (or both) could be extracted.
2658   */
2659  // @VisibleForTesting
2660  boolean maybeStripNationalPrefixAndCarrierCode(
2661      StringBuilder number, PhoneMetadata metadata, StringBuilder carrierCode) {
2662    int numberLength = number.length();
2663    String possibleNationalPrefix = metadata.getNationalPrefixForParsing();
2664    if (numberLength == 0 || possibleNationalPrefix.length() == 0) {
2665      // Early return for numbers of zero length.
2666      return false;
2667    }
2668    // Attempt to parse the first digits as a national prefix.
2669    Matcher prefixMatcher = regexCache.getPatternForRegex(possibleNationalPrefix).matcher(number);
2670    if (prefixMatcher.lookingAt()) {
2671      Pattern nationalNumberRule =
2672          regexCache.getPatternForRegex(metadata.getGeneralDesc().getNationalNumberPattern());
2673      // Check if the original number is viable.
2674      boolean isViableOriginalNumber = nationalNumberRule.matcher(number).matches();
2675      // prefixMatcher.group(numOfGroups) == null implies nothing was captured by the capturing
2676      // groups in possibleNationalPrefix; therefore, no transformation is necessary, and we just
2677      // remove the national prefix.
2678      int numOfGroups = prefixMatcher.groupCount();
2679      String transformRule = metadata.getNationalPrefixTransformRule();
2680      if (transformRule == null || transformRule.length() == 0 ||
2681          prefixMatcher.group(numOfGroups) == null) {
2682        // If the original number was viable, and the resultant number is not, we return.
2683        if (isViableOriginalNumber &&
2684            !nationalNumberRule.matcher(number.substring(prefixMatcher.end())).matches()) {
2685          return false;
2686        }
2687        if (carrierCode != null && numOfGroups > 0 && prefixMatcher.group(numOfGroups) != null) {
2688          carrierCode.append(prefixMatcher.group(1));
2689        }
2690        number.delete(0, prefixMatcher.end());
2691        return true;
2692      } else {
2693        // Check that the resultant number is still viable. If not, return. Check this by copying
2694        // the string buffer and making the transformation on the copy first.
2695        StringBuilder transformedNumber = new StringBuilder(number);
2696        transformedNumber.replace(0, numberLength, prefixMatcher.replaceFirst(transformRule));
2697        if (isViableOriginalNumber &&
2698            !nationalNumberRule.matcher(transformedNumber.toString()).matches()) {
2699          return false;
2700        }
2701        if (carrierCode != null && numOfGroups > 1) {
2702          carrierCode.append(prefixMatcher.group(1));
2703        }
2704        number.replace(0, number.length(), transformedNumber.toString());
2705        return true;
2706      }
2707    }
2708    return false;
2709  }
2710
2711  /**
2712   * Strips any extension (as in, the part of the number dialled after the call is connected,
2713   * usually indicated with extn, ext, x or similar) from the end of the number, and returns it.
2714   *
2715   * @param number  the non-normalized telephone number that we wish to strip the extension from
2716   * @return        the phone extension
2717   */
2718  // @VisibleForTesting
2719  String maybeStripExtension(StringBuilder number) {
2720    Matcher m = EXTN_PATTERN.matcher(number);
2721    // If we find a potential extension, and the number preceding this is a viable number, we assume
2722    // it is an extension.
2723    if (m.find() && isViablePhoneNumber(number.substring(0, m.start()))) {
2724      // The numbers are captured into groups in the regular expression.
2725      for (int i = 1, length = m.groupCount(); i <= length; i++) {
2726        if (m.group(i) != null) {
2727          // We go through the capturing groups until we find one that captured some digits. If none
2728          // did, then we will return the empty string.
2729          String extension = m.group(i);
2730          number.delete(m.start(), number.length());
2731          return extension;
2732        }
2733      }
2734    }
2735    return "";
2736  }
2737
2738  /**
2739   * Checks to see that the region code used is valid, or if it is not valid, that the number to
2740   * parse starts with a + symbol so that we can attempt to infer the region from the number.
2741   * Returns false if it cannot use the region provided and the region cannot be inferred.
2742   */
2743  private boolean checkRegionForParsing(String numberToParse, String defaultRegion) {
2744    if (!isValidRegionCode(defaultRegion)) {
2745      // If the number is null or empty, we can't infer the region.
2746      if ((numberToParse == null) || (numberToParse.length() == 0) ||
2747          !PLUS_CHARS_PATTERN.matcher(numberToParse).lookingAt()) {
2748        return false;
2749      }
2750    }
2751    return true;
2752  }
2753
2754  /**
2755   * Parses a string and returns it in proto buffer format. This method will throw a
2756   * {@link com.android.i18n.phonenumbers.NumberParseException} if the number is not considered to be
2757   * a possible number. Note that validation of whether the number is actually a valid number for a
2758   * particular region is not performed. This can be done separately with {@link #isValidNumber}.
2759   *
2760   * @param numberToParse     number that we are attempting to parse. This can contain formatting
2761   *                          such as +, ( and -, as well as a phone number extension. It can also
2762   *                          be provided in RFC3966 format.
2763   * @param defaultRegion     region that we are expecting the number to be from. This is only used
2764   *                          if the number being parsed is not written in international format.
2765   *                          The country_code for the number in this case would be stored as that
2766   *                          of the default region supplied. If the number is guaranteed to
2767   *                          start with a '+' followed by the country calling code, then
2768   *                          "ZZ" or null can be supplied.
2769   * @return                  a phone number proto buffer filled with the parsed number
2770   * @throws NumberParseException  if the string is not considered to be a viable phone number or if
2771   *                               no default region was supplied and the number is not in
2772   *                               international format (does not start with +)
2773   */
2774  public PhoneNumber parse(String numberToParse, String defaultRegion)
2775      throws NumberParseException {
2776    PhoneNumber phoneNumber = new PhoneNumber();
2777    parse(numberToParse, defaultRegion, phoneNumber);
2778    return phoneNumber;
2779  }
2780
2781  /**
2782   * Same as {@link #parse(String, String)}, but accepts mutable PhoneNumber as a parameter to
2783   * decrease object creation when invoked many times.
2784   */
2785  public void parse(String numberToParse, String defaultRegion, PhoneNumber phoneNumber)
2786      throws NumberParseException {
2787    parseHelper(numberToParse, defaultRegion, false, true, phoneNumber);
2788  }
2789
2790  /**
2791   * Parses a string and returns it in proto buffer format. This method differs from {@link #parse}
2792   * in that it always populates the raw_input field of the protocol buffer with numberToParse as
2793   * well as the country_code_source field.
2794   *
2795   * @param numberToParse     number that we are attempting to parse. This can contain formatting
2796   *                          such as +, ( and -, as well as a phone number extension.
2797   * @param defaultRegion     region that we are expecting the number to be from. This is only used
2798   *                          if the number being parsed is not written in international format.
2799   *                          The country calling code for the number in this case would be stored
2800   *                          as that of the default region supplied.
2801   * @return                  a phone number proto buffer filled with the parsed number
2802   * @throws NumberParseException  if the string is not considered to be a viable phone number or if
2803   *                               no default region was supplied
2804   */
2805  public PhoneNumber parseAndKeepRawInput(String numberToParse, String defaultRegion)
2806      throws NumberParseException {
2807    PhoneNumber phoneNumber = new PhoneNumber();
2808    parseAndKeepRawInput(numberToParse, defaultRegion, phoneNumber);
2809    return phoneNumber;
2810  }
2811
2812  /**
2813   * Same as{@link #parseAndKeepRawInput(String, String)}, but accepts a mutable PhoneNumber as
2814   * a parameter to decrease object creation when invoked many times.
2815   */
2816  public void parseAndKeepRawInput(String numberToParse, String defaultRegion,
2817                                   PhoneNumber phoneNumber)
2818      throws NumberParseException {
2819    parseHelper(numberToParse, defaultRegion, true, true, phoneNumber);
2820  }
2821
2822  /**
2823   * Returns an iterable over all {@link PhoneNumberMatch PhoneNumberMatches} in {@code text}. This
2824   * is a shortcut for {@link #findNumbers(CharSequence, String, Leniency, long)
2825   * getMatcher(text, defaultRegion, Leniency.VALID, Long.MAX_VALUE)}.
2826   *
2827   * @param text              the text to search for phone numbers, null for no text
2828   * @param defaultRegion     region that we are expecting the number to be from. This is only used
2829   *                          if the number being parsed is not written in international format. The
2830   *                          country_code for the number in this case would be stored as that of
2831   *                          the default region supplied. May be null if only international
2832   *                          numbers are expected.
2833   */
2834  public Iterable<PhoneNumberMatch> findNumbers(CharSequence text, String defaultRegion) {
2835    return findNumbers(text, defaultRegion, Leniency.VALID, Long.MAX_VALUE);
2836  }
2837
2838  /**
2839   * Returns an iterable over all {@link PhoneNumberMatch PhoneNumberMatches} in {@code text}.
2840   *
2841   * @param text              the text to search for phone numbers, null for no text
2842   * @param defaultRegion     region that we are expecting the number to be from. This is only used
2843   *                          if the number being parsed is not written in international format. The
2844   *                          country_code for the number in this case would be stored as that of
2845   *                          the default region supplied. May be null if only international
2846   *                          numbers are expected.
2847   * @param leniency          the leniency to use when evaluating candidate phone numbers
2848   * @param maxTries          the maximum number of invalid numbers to try before giving up on the
2849   *                          text. This is to cover degenerate cases where the text has a lot of
2850   *                          false positives in it. Must be {@code >= 0}.
2851   */
2852  public Iterable<PhoneNumberMatch> findNumbers(
2853      final CharSequence text, final String defaultRegion, final Leniency leniency,
2854      final long maxTries) {
2855
2856    return new Iterable<PhoneNumberMatch>() {
2857      public Iterator<PhoneNumberMatch> iterator() {
2858        return new PhoneNumberMatcher(
2859            PhoneNumberUtil.this, text, defaultRegion, leniency, maxTries);
2860      }
2861    };
2862  }
2863
2864  /**
2865   * A helper function to set the values related to leading zeros in a PhoneNumber.
2866   */
2867  static void setItalianLeadingZerosForPhoneNumber(String nationalNumber, PhoneNumber phoneNumber) {
2868    if (nationalNumber.length() > 1 && nationalNumber.charAt(0) == '0') {
2869      phoneNumber.setItalianLeadingZero(true);
2870      int numberOfLeadingZeros = 1;
2871      // Note that if the national number is all "0"s, the last "0" is not counted as a leading
2872      // zero.
2873      while (numberOfLeadingZeros < nationalNumber.length() - 1 &&
2874             nationalNumber.charAt(numberOfLeadingZeros) == '0') {
2875        numberOfLeadingZeros++;
2876      }
2877      if (numberOfLeadingZeros != 1) {
2878        phoneNumber.setNumberOfLeadingZeros(numberOfLeadingZeros);
2879      }
2880    }
2881  }
2882
2883  /**
2884   * Parses a string and fills up the phoneNumber. This method is the same as the public
2885   * parse() method, with the exception that it allows the default region to be null, for use by
2886   * isNumberMatch(). checkRegion should be set to false if it is permitted for the default region
2887   * to be null or unknown ("ZZ").
2888   */
2889  private void parseHelper(String numberToParse, String defaultRegion, boolean keepRawInput,
2890                           boolean checkRegion, PhoneNumber phoneNumber)
2891      throws NumberParseException {
2892    if (numberToParse == null) {
2893      throw new NumberParseException(NumberParseException.ErrorType.NOT_A_NUMBER,
2894                                     "The phone number supplied was null.");
2895    } else if (numberToParse.length() > MAX_INPUT_STRING_LENGTH) {
2896      throw new NumberParseException(NumberParseException.ErrorType.TOO_LONG,
2897                                     "The string supplied was too long to parse.");
2898    }
2899
2900    StringBuilder nationalNumber = new StringBuilder();
2901    buildNationalNumberForParsing(numberToParse, nationalNumber);
2902
2903    if (!isViablePhoneNumber(nationalNumber.toString())) {
2904      throw new NumberParseException(NumberParseException.ErrorType.NOT_A_NUMBER,
2905                                     "The string supplied did not seem to be a phone number.");
2906    }
2907
2908    // Check the region supplied is valid, or that the extracted number starts with some sort of +
2909    // sign so the number's region can be determined.
2910    if (checkRegion && !checkRegionForParsing(nationalNumber.toString(), defaultRegion)) {
2911      throw new NumberParseException(NumberParseException.ErrorType.INVALID_COUNTRY_CODE,
2912                                     "Missing or invalid default region.");
2913    }
2914
2915    if (keepRawInput) {
2916      phoneNumber.setRawInput(numberToParse);
2917    }
2918    // Attempt to parse extension first, since it doesn't require region-specific data and we want
2919    // to have the non-normalised number here.
2920    String extension = maybeStripExtension(nationalNumber);
2921    if (extension.length() > 0) {
2922      phoneNumber.setExtension(extension);
2923    }
2924
2925    PhoneMetadata regionMetadata = getMetadataForRegion(defaultRegion);
2926    // Check to see if the number is given in international format so we know whether this number is
2927    // from the default region or not.
2928    StringBuilder normalizedNationalNumber = new StringBuilder();
2929    int countryCode = 0;
2930    try {
2931      // TODO: This method should really just take in the string buffer that has already
2932      // been created, and just remove the prefix, rather than taking in a string and then
2933      // outputting a string buffer.
2934      countryCode = maybeExtractCountryCode(nationalNumber.toString(), regionMetadata,
2935                                            normalizedNationalNumber, keepRawInput, phoneNumber);
2936    } catch (NumberParseException e) {
2937      Matcher matcher = PLUS_CHARS_PATTERN.matcher(nationalNumber.toString());
2938      if (e.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE &&
2939          matcher.lookingAt()) {
2940        // Strip the plus-char, and try again.
2941        countryCode = maybeExtractCountryCode(nationalNumber.substring(matcher.end()),
2942                                              regionMetadata, normalizedNationalNumber,
2943                                              keepRawInput, phoneNumber);
2944        if (countryCode == 0) {
2945          throw new NumberParseException(NumberParseException.ErrorType.INVALID_COUNTRY_CODE,
2946                                         "Could not interpret numbers after plus-sign.");
2947        }
2948      } else {
2949        throw new NumberParseException(e.getErrorType(), e.getMessage());
2950      }
2951    }
2952    if (countryCode != 0) {
2953      String phoneNumberRegion = getRegionCodeForCountryCode(countryCode);
2954      if (!phoneNumberRegion.equals(defaultRegion)) {
2955        // Metadata cannot be null because the country calling code is valid.
2956        regionMetadata = getMetadataForRegionOrCallingCode(countryCode, phoneNumberRegion);
2957      }
2958    } else {
2959      // If no extracted country calling code, use the region supplied instead. The national number
2960      // is just the normalized version of the number we were given to parse.
2961      normalize(nationalNumber);
2962      normalizedNationalNumber.append(nationalNumber);
2963      if (defaultRegion != null) {
2964        countryCode = regionMetadata.getCountryCode();
2965        phoneNumber.setCountryCode(countryCode);
2966      } else if (keepRawInput) {
2967        phoneNumber.clearCountryCodeSource();
2968      }
2969    }
2970    if (normalizedNationalNumber.length() < MIN_LENGTH_FOR_NSN) {
2971      throw new NumberParseException(NumberParseException.ErrorType.TOO_SHORT_NSN,
2972                                     "The string supplied is too short to be a phone number.");
2973    }
2974    if (regionMetadata != null) {
2975      StringBuilder carrierCode = new StringBuilder();
2976      StringBuilder potentialNationalNumber = new StringBuilder(normalizedNationalNumber);
2977      maybeStripNationalPrefixAndCarrierCode(potentialNationalNumber, regionMetadata, carrierCode);
2978      // We require that the NSN remaining after stripping the national prefix and carrier code be
2979      // of a possible length for the region. Otherwise, we don't do the stripping, since the
2980      // original number could be a valid short number.
2981      if (!isShorterThanPossibleNormalNumber(regionMetadata, potentialNationalNumber.toString())) {
2982        normalizedNationalNumber = potentialNationalNumber;
2983        if (keepRawInput) {
2984          phoneNumber.setPreferredDomesticCarrierCode(carrierCode.toString());
2985        }
2986      }
2987    }
2988    int lengthOfNationalNumber = normalizedNationalNumber.length();
2989    if (lengthOfNationalNumber < MIN_LENGTH_FOR_NSN) {
2990      throw new NumberParseException(NumberParseException.ErrorType.TOO_SHORT_NSN,
2991                                     "The string supplied is too short to be a phone number.");
2992    }
2993    if (lengthOfNationalNumber > MAX_LENGTH_FOR_NSN) {
2994      throw new NumberParseException(NumberParseException.ErrorType.TOO_LONG,
2995                                     "The string supplied is too long to be a phone number.");
2996    }
2997    setItalianLeadingZerosForPhoneNumber(normalizedNationalNumber.toString(), phoneNumber);
2998    phoneNumber.setNationalNumber(Long.parseLong(normalizedNationalNumber.toString()));
2999  }
3000
3001  /**
3002   * Converts numberToParse to a form that we can parse and write it to nationalNumber if it is
3003   * written in RFC3966; otherwise extract a possible number out of it and write to nationalNumber.
3004   */
3005  private void buildNationalNumberForParsing(String numberToParse, StringBuilder nationalNumber) {
3006    int indexOfPhoneContext = numberToParse.indexOf(RFC3966_PHONE_CONTEXT);
3007    if (indexOfPhoneContext > 0) {
3008      int phoneContextStart = indexOfPhoneContext + RFC3966_PHONE_CONTEXT.length();
3009      // If the phone context contains a phone number prefix, we need to capture it, whereas domains
3010      // will be ignored.
3011      if (numberToParse.charAt(phoneContextStart) == PLUS_SIGN) {
3012        // Additional parameters might follow the phone context. If so, we will remove them here
3013        // because the parameters after phone context are not important for parsing the
3014        // phone number.
3015        int phoneContextEnd = numberToParse.indexOf(';', phoneContextStart);
3016        if (phoneContextEnd > 0) {
3017          nationalNumber.append(numberToParse.substring(phoneContextStart, phoneContextEnd));
3018        } else {
3019          nationalNumber.append(numberToParse.substring(phoneContextStart));
3020        }
3021      }
3022
3023      // Now append everything between the "tel:" prefix and the phone-context. This should include
3024      // the national number, an optional extension or isdn-subaddress component. Note we also
3025      // handle the case when "tel:" is missing, as we have seen in some of the phone number inputs.
3026      // In that case, we append everything from the beginning.
3027      int indexOfRfc3966Prefix = numberToParse.indexOf(RFC3966_PREFIX);
3028      int indexOfNationalNumber = (indexOfRfc3966Prefix >= 0) ?
3029          indexOfRfc3966Prefix + RFC3966_PREFIX.length() : 0;
3030      nationalNumber.append(numberToParse.substring(indexOfNationalNumber, indexOfPhoneContext));
3031    } else {
3032      // Extract a possible number from the string passed in (this strips leading characters that
3033      // could not be the start of a phone number.)
3034      nationalNumber.append(extractPossibleNumber(numberToParse));
3035    }
3036
3037    // Delete the isdn-subaddress and everything after it if it is present. Note extension won't
3038    // appear at the same time with isdn-subaddress according to paragraph 5.3 of the RFC3966 spec,
3039    int indexOfIsdn = nationalNumber.indexOf(RFC3966_ISDN_SUBADDRESS);
3040    if (indexOfIsdn > 0) {
3041      nationalNumber.delete(indexOfIsdn, nationalNumber.length());
3042    }
3043    // If both phone context and isdn-subaddress are absent but other parameters are present, the
3044    // parameters are left in nationalNumber. This is because we are concerned about deleting
3045    // content from a potential number string when there is no strong evidence that the number is
3046    // actually written in RFC3966.
3047  }
3048
3049  /**
3050   * Takes two phone numbers and compares them for equality.
3051   *
3052   * <p>Returns EXACT_MATCH if the country_code, NSN, presence of a leading zero for Italian numbers
3053   * and any extension present are the same.
3054   * Returns NSN_MATCH if either or both has no region specified, and the NSNs and extensions are
3055   * the same.
3056   * Returns SHORT_NSN_MATCH if either or both has no region specified, or the region specified is
3057   * the same, and one NSN could be a shorter version of the other number. This includes the case
3058   * where one has an extension specified, and the other does not.
3059   * Returns NO_MATCH otherwise.
3060   * For example, the numbers +1 345 657 1234 and 657 1234 are a SHORT_NSN_MATCH.
3061   * The numbers +1 345 657 1234 and 345 657 are a NO_MATCH.
3062   *
3063   * @param firstNumberIn  first number to compare
3064   * @param secondNumberIn  second number to compare
3065   *
3066   * @return  NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH or EXACT_MATCH depending on the level of equality
3067   *     of the two numbers, described in the method definition.
3068   */
3069  public MatchType isNumberMatch(PhoneNumber firstNumberIn, PhoneNumber secondNumberIn) {
3070    // Make copies of the phone number so that the numbers passed in are not edited.
3071    PhoneNumber firstNumber = new PhoneNumber();
3072    firstNumber.mergeFrom(firstNumberIn);
3073    PhoneNumber secondNumber = new PhoneNumber();
3074    secondNumber.mergeFrom(secondNumberIn);
3075    // First clear raw_input, country_code_source and preferred_domestic_carrier_code fields and any
3076    // empty-string extensions so that we can use the proto-buffer equality method.
3077    firstNumber.clearRawInput();
3078    firstNumber.clearCountryCodeSource();
3079    firstNumber.clearPreferredDomesticCarrierCode();
3080    secondNumber.clearRawInput();
3081    secondNumber.clearCountryCodeSource();
3082    secondNumber.clearPreferredDomesticCarrierCode();
3083    if (firstNumber.hasExtension() &&
3084        firstNumber.getExtension().length() == 0) {
3085        firstNumber.clearExtension();
3086    }
3087    if (secondNumber.hasExtension() &&
3088        secondNumber.getExtension().length() == 0) {
3089        secondNumber.clearExtension();
3090    }
3091    // Early exit if both had extensions and these are different.
3092    if (firstNumber.hasExtension() && secondNumber.hasExtension() &&
3093        !firstNumber.getExtension().equals(secondNumber.getExtension())) {
3094      return MatchType.NO_MATCH;
3095    }
3096    int firstNumberCountryCode = firstNumber.getCountryCode();
3097    int secondNumberCountryCode = secondNumber.getCountryCode();
3098    // Both had country_code specified.
3099    if (firstNumberCountryCode != 0 && secondNumberCountryCode != 0) {
3100      if (firstNumber.exactlySameAs(secondNumber)) {
3101        return MatchType.EXACT_MATCH;
3102      } else if (firstNumberCountryCode == secondNumberCountryCode &&
3103                 isNationalNumberSuffixOfTheOther(firstNumber, secondNumber)) {
3104        // A SHORT_NSN_MATCH occurs if there is a difference because of the presence or absence of
3105        // an 'Italian leading zero', the presence or absence of an extension, or one NSN being a
3106        // shorter variant of the other.
3107        return MatchType.SHORT_NSN_MATCH;
3108      }
3109      // This is not a match.
3110      return MatchType.NO_MATCH;
3111    }
3112    // Checks cases where one or both country_code fields were not specified. To make equality
3113    // checks easier, we first set the country_code fields to be equal.
3114    firstNumber.setCountryCode(secondNumberCountryCode);
3115    // If all else was the same, then this is an NSN_MATCH.
3116    if (firstNumber.exactlySameAs(secondNumber)) {
3117      return MatchType.NSN_MATCH;
3118    }
3119    if (isNationalNumberSuffixOfTheOther(firstNumber, secondNumber)) {
3120      return MatchType.SHORT_NSN_MATCH;
3121    }
3122    return MatchType.NO_MATCH;
3123  }
3124
3125  // Returns true when one national number is the suffix of the other or both are the same.
3126  private boolean isNationalNumberSuffixOfTheOther(PhoneNumber firstNumber,
3127                                                   PhoneNumber secondNumber) {
3128    String firstNumberNationalNumber = String.valueOf(firstNumber.getNationalNumber());
3129    String secondNumberNationalNumber = String.valueOf(secondNumber.getNationalNumber());
3130    // Note that endsWith returns true if the numbers are equal.
3131    return firstNumberNationalNumber.endsWith(secondNumberNationalNumber) ||
3132           secondNumberNationalNumber.endsWith(firstNumberNationalNumber);
3133  }
3134
3135  /**
3136   * Takes two phone numbers as strings and compares them for equality. This is a convenience
3137   * wrapper for {@link #isNumberMatch(PhoneNumber, PhoneNumber)}. No default region is known.
3138   *
3139   * @param firstNumber  first number to compare. Can contain formatting, and can have country
3140   *     calling code specified with + at the start.
3141   * @param secondNumber  second number to compare. Can contain formatting, and can have country
3142   *     calling code specified with + at the start.
3143   * @return  NOT_A_NUMBER, NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH, EXACT_MATCH. See
3144   *     {@link #isNumberMatch(PhoneNumber, PhoneNumber)} for more details.
3145   */
3146  public MatchType isNumberMatch(String firstNumber, String secondNumber) {
3147    try {
3148      PhoneNumber firstNumberAsProto = parse(firstNumber, UNKNOWN_REGION);
3149      return isNumberMatch(firstNumberAsProto, secondNumber);
3150    } catch (NumberParseException e) {
3151      if (e.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE) {
3152        try {
3153          PhoneNumber secondNumberAsProto = parse(secondNumber, UNKNOWN_REGION);
3154          return isNumberMatch(secondNumberAsProto, firstNumber);
3155        } catch (NumberParseException e2) {
3156          if (e2.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE) {
3157            try {
3158              PhoneNumber firstNumberProto = new PhoneNumber();
3159              PhoneNumber secondNumberProto = new PhoneNumber();
3160              parseHelper(firstNumber, null, false, false, firstNumberProto);
3161              parseHelper(secondNumber, null, false, false, secondNumberProto);
3162              return isNumberMatch(firstNumberProto, secondNumberProto);
3163            } catch (NumberParseException e3) {
3164              // Fall through and return MatchType.NOT_A_NUMBER.
3165            }
3166          }
3167        }
3168      }
3169    }
3170    // One or more of the phone numbers we are trying to match is not a viable phone number.
3171    return MatchType.NOT_A_NUMBER;
3172  }
3173
3174  /**
3175   * Takes two phone numbers and compares them for equality. This is a convenience wrapper for
3176   * {@link #isNumberMatch(PhoneNumber, PhoneNumber)}. No default region is known.
3177   *
3178   * @param firstNumber  first number to compare in proto buffer format.
3179   * @param secondNumber  second number to compare. Can contain formatting, and can have country
3180   *     calling code specified with + at the start.
3181   * @return  NOT_A_NUMBER, NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH, EXACT_MATCH. See
3182   *     {@link #isNumberMatch(PhoneNumber, PhoneNumber)} for more details.
3183   */
3184  public MatchType isNumberMatch(PhoneNumber firstNumber, String secondNumber) {
3185    // First see if the second number has an implicit country calling code, by attempting to parse
3186    // it.
3187    try {
3188      PhoneNumber secondNumberAsProto = parse(secondNumber, UNKNOWN_REGION);
3189      return isNumberMatch(firstNumber, secondNumberAsProto);
3190    } catch (NumberParseException e) {
3191      if (e.getErrorType() == NumberParseException.ErrorType.INVALID_COUNTRY_CODE) {
3192        // The second number has no country calling code. EXACT_MATCH is no longer possible.
3193        // We parse it as if the region was the same as that for the first number, and if
3194        // EXACT_MATCH is returned, we replace this with NSN_MATCH.
3195        String firstNumberRegion = getRegionCodeForCountryCode(firstNumber.getCountryCode());
3196        try {
3197          if (!firstNumberRegion.equals(UNKNOWN_REGION)) {
3198            PhoneNumber secondNumberWithFirstNumberRegion = parse(secondNumber, firstNumberRegion);
3199            MatchType match = isNumberMatch(firstNumber, secondNumberWithFirstNumberRegion);
3200            if (match == MatchType.EXACT_MATCH) {
3201              return MatchType.NSN_MATCH;
3202            }
3203            return match;
3204          } else {
3205            // If the first number didn't have a valid country calling code, then we parse the
3206            // second number without one as well.
3207            PhoneNumber secondNumberProto = new PhoneNumber();
3208            parseHelper(secondNumber, null, false, false, secondNumberProto);
3209            return isNumberMatch(firstNumber, secondNumberProto);
3210          }
3211        } catch (NumberParseException e2) {
3212          // Fall-through to return NOT_A_NUMBER.
3213        }
3214      }
3215    }
3216    // One or more of the phone numbers we are trying to match is not a viable phone number.
3217    return MatchType.NOT_A_NUMBER;
3218  }
3219
3220  /**
3221   * Returns true if the number can be dialled from outside the region, or unknown. If the number
3222   * can only be dialled from within the region, returns false. Does not check the number is a valid
3223   * number. Note that, at the moment, this method does not handle short numbers.
3224   * TODO: Make this method public when we have enough metadata to make it worthwhile.
3225   *
3226   * @param number  the phone-number for which we want to know whether it is diallable from
3227   *     outside the region
3228   */
3229  // @VisibleForTesting
3230  boolean canBeInternationallyDialled(PhoneNumber number) {
3231    PhoneMetadata metadata = getMetadataForRegion(getRegionCodeForNumber(number));
3232    if (metadata == null) {
3233      // Note numbers belonging to non-geographical entities (e.g. +800 numbers) are always
3234      // internationally diallable, and will be caught here.
3235      return true;
3236    }
3237    String nationalSignificantNumber = getNationalSignificantNumber(number);
3238    return !isNumberMatchingDesc(nationalSignificantNumber, metadata.getNoInternationalDialling());
3239  }
3240
3241  /**
3242   * Returns true if the supplied region supports mobile number portability. Returns false for
3243   * invalid, unknown or regions that don't support mobile number portability.
3244   *
3245   * @param regionCode  the region for which we want to know whether it supports mobile number
3246   *                    portability or not.
3247   */
3248  public boolean isMobileNumberPortableRegion(String regionCode) {
3249    PhoneMetadata metadata = getMetadataForRegion(regionCode);
3250    if (metadata == null) {
3251      logger.log(Level.WARNING, "Invalid or unknown region code provided: " + regionCode);
3252      return false;
3253    }
3254    return metadata.isMobileNumberPortableRegion();
3255  }
3256}
3257