Update libphonenumber to v4.3

Review URL: http://codereview.appspot.com/5434054

java/geocoder/src/com/google/i18n/phonenumbers/geocoding/PhoneNumberOfflineGeocoder.java

@@ -135,10 +135,12 @@ public class PhoneNumberOfflineGeocoder {
   }
 
   /**
-   * Returns a text description for the given language code for the given phone number. The
-   * description might consist of the name of the country where the phone number is from and/or the
-   * name of the geographical area the phone number is from. This method assumes the validity of the
-   * number passed in has already been checked.
+   * Returns a text description for the given phone number, in the language provided. The
+   * description might consist of the name of the country where the phone number is from, or the
+   * name of the geographical area the phone number is from if more detailed information is
+   * available.
+   *
+   * <p>This method assumes the validity of the number passed in has already been checked.
    *
    * @param number  a valid phone number for which we want to get a text description
    * @param languageCode  the language code for which the description should be written
@@ -156,10 +158,44 @@ public class PhoneNumberOfflineGeocoder {
   }
 
   /**
-   * Returns a text description for the given language code for the given phone number. The
-   * description might consist of the name of the country where the phone number is from and/or the
-   * name of the geographical area the phone number is from. This method explictly checkes the
-   * validity of the number passed in.
+   * As per {@link #getDescriptionForValidNumber(PhoneNumber, Locale)} but also considers the
+   * region of the user. If the phone number is from the same region as the user, only a lower-level
+   * description will be returned, if one exists. Otherwise, the phone number's region will be
+   * returned, with optionally some more detailed information.
+   *
+   * <p>For example, for a user from the region "US" (United States), we would show "Mountain View,
+   * CA" for a particular number, omitting the United States from the description. For a user from
+   * the United Kingdom (region "GB"), for the same number we may show "Mountain View, CA, United
+   * States" or even just "United States".
+   *
+   * <p>This method assumes the validity of the number passed in has already been checked.
+   *
+   * @param number  the phone number for which we want to get a text description
+   * @param languageCode  the language code for which the description should be written
+   * @param userRegion  the region code for a given user. This region will be omitted from the
+   *     description if the phone number comes from this region. It is a two-letter uppercase ISO
+   *     country code as defined by ISO 3166-1.
+   * @return  a text description for the given language code for the given phone number, or empty
+   *     string if the number passed in is invalid
+   */
+  public String getDescriptionForValidNumber(PhoneNumber number, Locale languageCode,
+                                             String userRegion) {
+    // If the user region matches the number's region, then we just show the lower-level
+    // description, if one exists - if no description exists, we will show the region(country) name
+    // for the number.
+    String regionCode = phoneUtil.getRegionCodeForNumber(number);
+    if (userRegion.equals(regionCode)) {
+      return getDescriptionForValidNumber(number, languageCode);
+    }
+    // Otherwise, we just show the region(country) name for now.
+    return getCountryNameForNumber(number, languageCode);
+    // TODO: Concatenate the lower-level and country-name information in an appropriate
+    // way for each language.
+  }
+
+  /**
+   * As per {@link #getDescriptionForValidNumber(PhoneNumber, Locale)} but explicitly checks
+   * the validity of the number passed in.
    *
    * @param number  the phone number for which we want to get a text description
    * @param languageCode  the language code for which the description should be written
@@ -173,6 +209,26 @@ public class PhoneNumberOfflineGeocoder {
     return getDescriptionForValidNumber(number, languageCode);
   }
 
+  /**
+   * As per {@link #getDescriptionForValidNumber(PhoneNumber, Locale, String)} but
+   * explicitly checks the validity of the number passed in.
+   *
+   * @param number  the phone number for which we want to get a text description
+   * @param languageCode  the language code for which the description should be written
+   * @param userRegion  the region code for a given user. This region will be omitted from the
+   *     description if the phone number comes from this region. It is a two-letter uppercase ISO
+   *     country code as defined by ISO 3166-1.
+   * @return  a text description for the given language code for the given phone number, or empty
+   *     string if the number passed in is invalid
+   */
+  public String getDescriptionForNumber(PhoneNumber number, Locale languageCode,
+                                        String userRegion) {
+    if (!phoneUtil.isValidNumber(number)) {
+      return "";
+    }
+    return getDescriptionForValidNumber(number, languageCode, userRegion);
+  }
+
   /**
    * Returns an area-level text description in the given language for the given phone number.
    *

java/geocoder/test/com/google/i18n/phonenumbers/geocoding/PhoneNumberOfflineGeocoderTest.java

@@ -116,6 +116,26 @@ public class PhoneNumberOfflineGeocoderTest extends TestCase {
         geocoder.getDescriptionForNumber(KO_NUMBER3, Locale.KOREAN));
   }
 
+  public void testGetDescriptionForNumberWithUserRegion() {
+    // User in Italy, American number. We should just show United States, in German, and not more
+    // detailed information.
+    assertEquals("Vereinigte Staaten von Amerika",
+        geocoder.getDescriptionForNumber(US_NUMBER1, Locale.GERMAN, "IT"));
+    // Unknown region - should just show country name.
+    assertEquals("Vereinigte Staaten von Amerika",
+        geocoder.getDescriptionForNumber(US_NUMBER1, Locale.GERMAN, "ZZ"));
+    // User in the States, language German, should show detailed data.
+    assertEquals("Kalifornien",
+        geocoder.getDescriptionForNumber(US_NUMBER1, Locale.GERMAN, "US"));
+    // User in the States, language French, no data for French, so we fallback to English detailed
+    // data.
+    assertEquals("CA",
+        geocoder.getDescriptionForNumber(US_NUMBER1, Locale.FRENCH, "US"));
+    // Invalid number - return an empty string.
+    assertEquals("", geocoder.getDescriptionForNumber(US_INVALID_NUMBER, Locale.ENGLISH,
+                                                      "US"));
+  }
+
   public void testGetDescriptionForInvalidNumber() {
     assertEquals("", geocoder.getDescriptionForNumber(KO_INVALID_NUMBER, Locale.ENGLISH));
     assertEquals("", geocoder.getDescriptionForNumber(US_INVALID_NUMBER, Locale.ENGLISH));

java/libphonenumber/src/com/google/i18n/phonenumbers/AsYouTypeFormatter.java

@@ -356,7 +356,7 @@ public class AsYouTypeFormatter {
           }
           return ableToFormat
              ? prefixBeforeNationalNumber + tempNationalNumber
-             : tempNationalNumber;
+             : accruedInput.toString();
         } else {
           return attemptToChooseFormattingPattern();
         }
@@ -428,8 +428,7 @@ public class AsYouTypeFormatter {
     // number (excluding national prefix) have been entered.
     if (nationalNumber.length() >= MIN_LEADING_DIGITS_LENGTH) {
       getAvailableFormats(nationalNumber.substring(0, MIN_LEADING_DIGITS_LENGTH));
-      maybeCreateNewTemplate();
-      return inputAccruedNationalNumber();
+      return maybeCreateNewTemplate() ? inputAccruedNationalNumber() : accruedInput.toString();
     } else {
       return prefixBeforeNationalNumber + nationalNumber.toString();
     }
@@ -446,7 +445,7 @@ public class AsYouTypeFormatter {
       }
       return ableToFormat
           ? prefixBeforeNationalNumber + tempNationalNumber
-          : tempNationalNumber;
+          : accruedInput.toString();
     } else {
       return prefixBeforeNationalNumber.toString();
     }

java/libphonenumber/src/com/google/i18n/phonenumbers/PhoneNumberMatcher.java

@@ -294,8 +294,8 @@ final class PhoneNumberMatcher implements Iterator<PhoneNumberMatch> {
         block.equals(UnicodeBlock.COMBINING_DIACRITICAL_MARKS);
   }
 
-  private static boolean isCurrencySymbol(char character) {
-    return Character.getType(character) == Character.CURRENCY_SYMBOL;
+  private static boolean isInvalidPunctuationSymbol(char character) {
+    return character == '%' || Character.getType(character) == Character.CURRENCY_SYMBOL;
   }
 
   /**
@@ -407,15 +407,15 @@ final class PhoneNumberMatcher implements Iterator<PhoneNumberMatch> {
         // punctuation, check the previous character.
         if (offset > 0 && !LEAD_CLASS.matcher(candidate).lookingAt()) {
           char previousChar = text.charAt(offset - 1);
-          // We return null if it is a latin letter or a currency symbol.
-          if (isCurrencySymbol(previousChar) || isLatinLetter(previousChar)) {
+          // We return null if it is a latin letter or an invalid punctuation symbol.
+          if (isInvalidPunctuationSymbol(previousChar) || isLatinLetter(previousChar)) {
             return null;
           }
         }
         int lastCharIndex = offset + candidate.length();
         if (lastCharIndex < text.length()) {
           char nextChar = text.charAt(lastCharIndex);
-          if (isCurrencySymbol(nextChar) || isLatinLetter(nextChar)) {
+          if (isInvalidPunctuationSymbol(nextChar) || isLatinLetter(nextChar)) {
             return null;
           }
         }

java/libphonenumber/src/com/google/i18n/phonenumbers/PhoneNumberUtil.java

@@ -97,6 +97,11 @@ public class PhoneNumberUtil {
 
   private static final String RFC3966_EXTN_PREFIX = ";ext=";
 
+  // A map that contains characters that are essential when dialling. That means any of the
+  // characters in this map must not be removed from a number when dialing, otherwise the call will
+  // not reach the intended destination.
+  private static final Map<Character, Character> DIALLABLE_CHAR_MAPPINGS;
+
   // Only upper-case variants of alpha characters are stored.
   private static final Map<Character, Character> ALPHA_MAPPINGS;
 
@@ -156,6 +161,12 @@ public class PhoneNumberUtil {
     combinedMap.putAll(asciiDigitMappings);
     ALPHA_PHONE_MAPPINGS = Collections.unmodifiableMap(combinedMap);
 
+    HashMap<Character, Character> diallableCharMap = new HashMap<Character, Character>();
+    diallableCharMap.put('+', '+');
+    diallableCharMap.put('*', '*');
+    diallableCharMap.putAll(asciiDigitMappings);
+    DIALLABLE_CHAR_MAPPINGS = Collections.unmodifiableMap(diallableCharMap);
+
     HashMap<Character, Character> allPlusNumberGroupings = new HashMap<Character, Character>();
     // Put (lower letter -> upper letter) and (upper letter -> upper letter) mappings.
     for (char c : ALPHA_MAPPINGS.keySet()) {
@@ -407,9 +418,8 @@ public class PhoneNumberUtil {
    */
   public enum Leniency {
     /**
-     * Phone numbers accepted are
-     * {@linkplain PhoneNumberUtil#isPossibleNumber(Phonenumber.PhoneNumber) possible}, but not
-     * necessarily {@linkplain PhoneNumberUtil#isValidNumber(Phonenumber.PhoneNumber) valid}.
+     * Phone numbers accepted are {@linkplain PhoneNumberUtil#isPossibleNumber(PhoneNumber)
+     * possible}, but not necessarily {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid}.
      */
     POSSIBLE {
       @Override
@@ -418,9 +428,8 @@ public class PhoneNumberUtil {
       }
     },
     /**
-     * Phone numbers accepted are
-     * {@linkplain PhoneNumberUtil#isPossibleNumber(Phonenumber.PhoneNumber) possible} and
-     * {@linkplain PhoneNumberUtil#isValidNumber(Phonenumber.PhoneNumber) valid}. Numbers written
+     * Phone numbers accepted are {@linkplain PhoneNumberUtil#isPossibleNumber(PhoneNumber)
+     * possible} and {@linkplain PhoneNumberUtil#isValidNumber(PhoneNumber) valid}. Numbers written
      * in national format must have their national-prefix present if it is usually written for a
      * number of this type.
      */
@@ -1059,9 +1068,8 @@ public class PhoneNumberUtil {
   }
 
   /**
-   * Same as {@link #format(Phonenumber.PhoneNumber, PhoneNumberUtil.PhoneNumberFormat)}, but
-   * accepts a mutable StringBuilder as a parameter to decrease object creation when invoked many
-   * times.
+   * Same as {@link #format(PhoneNumber, PhoneNumberFormat)}, but accepts a mutable StringBuilder as
+   * a parameter to decrease object creation when invoked many times.
    */
   public void format(PhoneNumber number, PhoneNumberFormat numberFormat,
                      StringBuilder formattedNumber) {
@@ -1222,7 +1230,7 @@ public class PhoneNumberUtil {
    */
   public String formatNumberForMobileDialing(PhoneNumber number, String regionCallingFrom,
                                              boolean withFormatting) {
-    String regionCode = getRegionCodeForNumber(number);
+    String regionCode = getRegionCodeForCountryCode(number.getCountryCode());
     if (!isValidRegionCode(regionCode)) {
       return number.hasRawInput() ? number.getRawInput() : "";
     }
@@ -1231,11 +1239,19 @@ public class PhoneNumberUtil {
     // Clear the extension, as that part cannot normally be dialed together with the main number.
     PhoneNumber numberNoExt = new PhoneNumber().mergeFrom(number).clearExtension();
     PhoneNumberType numberType = getNumberType(numberNoExt);
-    if ((regionCode.equals("CO")) && (regionCallingFrom.equals("CO")) &&
-        (numberType == PhoneNumberType.FIXED_LINE)) {
-      formattedNumber =
-          formatNationalNumberWithCarrierCode(numberNoExt, COLOMBIA_MOBILE_TO_FIXED_LINE_PREFIX);
-    } else if ((regionCode.equals("BR")) && (regionCallingFrom.equals("BR")) &&
+    if (regionCode.equals("CO") && regionCallingFrom.equals("CO")) {
+      if (numberType == PhoneNumberType.FIXED_LINE) {
+        formattedNumber =
+            formatNationalNumberWithCarrierCode(numberNoExt, COLOMBIA_MOBILE_TO_FIXED_LINE_PREFIX);
+      } else {
+        // E164 doesn't work at all when dialing within Colombia.
+        formattedNumber = format(numberNoExt, PhoneNumberFormat.NATIONAL);
+      }
+    } else if (regionCode.equals("PE") && regionCallingFrom.equals("PE")) {
+      // In Peru, numbers cannot be dialled using E164 format from a mobile phone for Movistar.
+      // Instead they must be dialled in national format.
+      formattedNumber = format(numberNoExt, PhoneNumberFormat.NATIONAL);
+    } else if (regionCode.equals("BR") && regionCallingFrom.equals("BR") &&
         ((numberType == PhoneNumberType.FIXED_LINE) || (numberType == PhoneNumberType.MOBILE) ||
          (numberType == PhoneNumberType.FIXED_LINE_OR_MOBILE))) {
       formattedNumber = numberNoExt.hasPreferredDomesticCarrierCode()
@@ -1251,7 +1267,9 @@ public class PhoneNumberUtil {
       formattedNumber = (regionCallingFrom.equals(regionCode))
           ? format(numberNoExt, PhoneNumberFormat.NATIONAL) : "";
     }
-    return withFormatting ? formattedNumber : normalizeDigitsOnly(formattedNumber);
+    return withFormatting ? formattedNumber
+                          : normalizeHelper(formattedNumber, DIALLABLE_CHAR_MAPPINGS,
+                                            true /* remove non matches */);
   }
 
   /**
@@ -1339,7 +1357,11 @@ public class PhoneNumberUtil {
    * @return  the formatted phone number in its original number format
    */
   public String formatInOriginalFormat(PhoneNumber number, String regionCallingFrom) {
-    if (number.hasRawInput() && !isValidNumber(number)) {
+    if (number.hasRawInput() &&
+        (!hasFormattingPatternForNumber(number) || !isValidNumber(number))) {
+      // We check if we have the formatting pattern because without that, we might format the number
+      // as a group without national prefix. We also want to check the validity of the number
+      // because we don't want to risk formatting the number if we don't really understand it.
       return number.getRawInput();
     }
     if (!number.hasCountryCodeSource()) {
@@ -1358,6 +1380,18 @@ public class PhoneNumberUtil {
     }
   }
 
+  private boolean hasFormattingPatternForNumber(PhoneNumber number) {
+    String phoneNumberRegion = getRegionCodeForCountryCode(number.getCountryCode());
+    PhoneMetadata metadata = getMetadataForRegion(phoneNumberRegion);
+    if (metadata == null) {
+      return false;
+    }
+    String nationalNumber = getNationalSignificantNumber(number);
+    NumberFormat formatRule =
+        chooseFormattingPatternForNumber(metadata.numberFormats(), nationalNumber);
+    return formatRule != null;
+  }
+
   /**
    * Formats a phone number for out-of-country dialing purposes.
    *
@@ -2028,10 +2062,10 @@ public class PhoneNumberUtil {
   /**
    * Check whether a phone number is a possible number given a number in the form of a string, and
    * the region where the number could be dialed from. It provides a more lenient check than
-   * {@link #isValidNumber}. See {@link #isPossibleNumber(Phonenumber.PhoneNumber)} for details.
+   * {@link #isValidNumber}. See {@link #isPossibleNumber(PhoneNumber)} for details.
    *
-   * <p>This method first parses the number, then invokes
-   * {@link #isPossibleNumber(Phonenumber.PhoneNumber)} with the resultant PhoneNumber object.
+   * <p>This method first parses the number, then invokes {@link #isPossibleNumber(PhoneNumber)}
+   * with the resultant PhoneNumber object.
    *
    * @param number  the number that needs to be checked, in the form of a string
    * @param regionDialingFrom  the region that we are expecting the number to be dialed from.
@@ -2683,15 +2717,14 @@ public class PhoneNumberUtil {
 
   /**
    * Takes two phone numbers as strings and compares them for equality. This is a convenience
-   * wrapper for {@link #isNumberMatch(Phonenumber.PhoneNumber, Phonenumber.PhoneNumber)}. No
-   * default region is known.
+   * wrapper for {@link #isNumberMatch(PhoneNumber, PhoneNumber)}. No default region is known.
    *
    * @param firstNumber  first number to compare. Can contain formatting, and can have country
    *     calling code specified with + at the start.
    * @param secondNumber  second number to compare. Can contain formatting, and can have country
    *     calling code specified with + at the start.
    * @return  NOT_A_NUMBER, NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH, EXACT_MATCH. See
-   *     {@link #isNumberMatch(Phonenumber.PhoneNumber, Phonenumber.PhoneNumber)} for more details.
+   *     {@link #isNumberMatch(PhoneNumber, PhoneNumber)} for more details.
    */
   public MatchType isNumberMatch(String firstNumber, String secondNumber) {
     try {
@@ -2723,14 +2756,13 @@ public class PhoneNumberUtil {
 
   /**
    * Takes two phone numbers and compares them for equality. This is a convenience wrapper for
-   * {@link #isNumberMatch(Phonenumber.PhoneNumber, Phonenumber.PhoneNumber)}. No default region is
-   * known.
+   * {@link #isNumberMatch(PhoneNumber, PhoneNumber)}. No default region is known.
    *
    * @param firstNumber  first number to compare in proto buffer format.
    * @param secondNumber  second number to compare. Can contain formatting, and can have country
    *     calling code specified with + at the start.
    * @return  NOT_A_NUMBER, NO_MATCH, SHORT_NSN_MATCH, NSN_MATCH, EXACT_MATCH. See
-   *     {@link #isNumberMatch(Phonenumber.PhoneNumber, Phonenumber.PhoneNumber)} for more details.
+   *     {@link #isNumberMatch(PhoneNumber, PhoneNumber)} for more details.
    */
   public MatchType isNumberMatch(PhoneNumber firstNumber, String secondNumber) {
     // First see if the second number has an implicit country calling code, by attempting to parse

java/libphonenumber/test/com/google/i18n/phonenumbers/AsYouTypeFormatterTest.java

@@ -93,6 +93,7 @@ public class AsYouTypeFormatterTest extends TestCase {
     assertEquals("+81 90 1234 5678", formatter.inputDigit('8'));
     assertEquals("+81 90 12 345 6789", formatter.inputDigit('9'));
     assertEquals("+81901234567890", formatter.inputDigit('0'));
+    assertEquals("+819012345678901", formatter.inputDigit('1'));
   }
 
   public void testAYTFUS() {
@@ -716,6 +717,15 @@ public class AsYouTypeFormatterTest extends TestCase {
     assertEquals("+81 222 12 567", formatter.inputDigit('7'));
     assertEquals("+81 222 12 5678", formatter.inputDigit('8'));
 
+    // 011113
+    formatter.clear();
+    assertEquals("0", formatter.inputDigit('0'));
+    assertEquals("01", formatter.inputDigit('1'));
+    assertEquals("011", formatter.inputDigit('1'));
+    assertEquals("011 1", formatter.inputDigit('1'));
+    assertEquals("011 11", formatter.inputDigit('1'));
+    assertEquals("011113", formatter.inputDigit('3'));
+
     // +81 3332 2 5678
     formatter.clear();
     assertEquals("+", formatter.inputDigit('+'));

java/libphonenumber/test/com/google/i18n/phonenumbers/PhoneNumberMatcherTest.java

@@ -266,6 +266,13 @@ public class PhoneNumberMatcherTest extends TestCase {
     findMatchesInContexts(possibleOnlyContexts, false, true);
   }
 
+  public void testPercentageNotSeenAsPhoneNumber() throws Exception {
+    ArrayList<NumberContext> possibleOnlyContexts = new ArrayList<NumberContext>();
+    possibleOnlyContexts.add(new NumberContext("", "%"));
+    // Numbers followed by % should be dropped.
+    findMatchesInContexts(possibleOnlyContexts, false, true);
+  }
+
   public void testPhoneNumberWithLeadingOrTrailingMoneyMatches() throws Exception {
     // Because of the space after the 20 (or before the 100) these dollar amounts should not stop
     // the actual number from being found.

java/libphonenumber/test/com/google/i18n/phonenumbers/PhoneNumberUtilTest.java

@@ -71,6 +71,8 @@ public class PhoneNumberUtilTest extends TestCase {
   private static final PhoneNumber IT_NUMBER =
       new PhoneNumber().setCountryCode(39).setNationalNumber(236618300L).
       setItalianLeadingZero(true);
+  private static final PhoneNumber JP_STAR_NUMBER =
+      new PhoneNumber().setCountryCode(81).setNationalNumber(2345);
   // Numbers to test the formatting rules from Mexico.
   private static final PhoneNumber MX_MOBILE1 =
       new PhoneNumber().setCountryCode(52).setNationalNumber(12345678900L);
@@ -614,7 +616,8 @@ public class PhoneNumberUtilTest extends TestCase {
     // US toll free numbers are marked as noInternationalDialling in the test metadata for testing
     // purposes.
     assertEquals("800 253 0000",
-        phoneUtil.formatNumberForMobileDialing(US_TOLLFREE, RegionCode.US, true));
+        phoneUtil.formatNumberForMobileDialing(US_TOLLFREE, RegionCode.US,
+                                               true /*  keep formatting */));
     assertEquals("", phoneUtil.formatNumberForMobileDialing(US_TOLLFREE, RegionCode.CN, true));
     assertEquals("+1 650 253 0000",
         phoneUtil.formatNumberForMobileDialing(US_NUMBER, RegionCode.US, true));
@@ -623,12 +626,26 @@ public class PhoneNumberUtilTest extends TestCase {
         phoneUtil.formatNumberForMobileDialing(usNumberWithExtn, RegionCode.US, true));
 
     assertEquals("8002530000",
-        phoneUtil.formatNumberForMobileDialing(US_TOLLFREE, RegionCode.US, false));
+        phoneUtil.formatNumberForMobileDialing(US_TOLLFREE, RegionCode.US,
+                                               false /* remove formatting */));
     assertEquals("", phoneUtil.formatNumberForMobileDialing(US_TOLLFREE, RegionCode.CN, false));
     assertEquals("+16502530000",
         phoneUtil.formatNumberForMobileDialing(US_NUMBER, RegionCode.US, false));
     assertEquals("+16502530000",
         phoneUtil.formatNumberForMobileDialing(usNumberWithExtn, RegionCode.US, false));
+
+    // An invalid US number, which is one digit too long.
+    assertEquals("+165025300001",
+        phoneUtil.formatNumberForMobileDialing(US_LONG_NUMBER, RegionCode.US, false));
+    assertEquals("+1 65025300001",
+        phoneUtil.formatNumberForMobileDialing(US_LONG_NUMBER, RegionCode.US, true));
+
+    // Star numbers. In real life they appear in Israel, but we have them in JP in our test
+    // metadata.
+    assertEquals("*2345",
+        phoneUtil.formatNumberForMobileDialing(JP_STAR_NUMBER, RegionCode.JP, false));
+    assertEquals("*2345",
+        phoneUtil.formatNumberForMobileDialing(JP_STAR_NUMBER, RegionCode.JP, true));
   }
 
   public void testFormatByPattern() {
@@ -730,6 +747,11 @@ public class PhoneNumberUtilTest extends TestCase {
     // When the raw input is unavailable, format as usual.
     PhoneNumber number7 = phoneUtil.parse("7345678901", RegionCode.US);
     assertEquals("734 567 8901", phoneUtil.formatInOriginalFormat(number7, RegionCode.US));
+
+    // This number is valid, but we don't have a formatting pattern for it. Fall back to the raw
+    // input.
+    PhoneNumber number8 = phoneUtil.parseAndKeepRawInput("02-4567-8900", RegionCode.KR);
+    assertEquals("02-4567-8900", phoneUtil.formatInOriginalFormat(number8, RegionCode.KR));
   }
 
   public void testIsPremiumRate() {

java/release_notes.txt

@@ -1,3 +1,16 @@
+November 24th, 2011: libphonenumber-4.3
+* Code changes
+ - Fix the problems with AYTF crashing for longer numbers entered with +CountryCode, and incorrectly
+   removing national prefix for some numbers.
+ - Improve PhoneNumberMatcher to not match numbers ending with '%'.
+ - Fix formatNumberForMobileDialing to handle Israeli star numbers, Peruvian and Colombian numbers.
+ - Modify formatInOriginalFormat to use the raw input if we don't have a formatting pattern for a
+   number.
+ - Simple offline geocoding function which takes into account the user's region.
+
+* Metadata changes
+ - Updates for CR, GN, JP, KE, PL, SG, SR, SX, TH, TK
+
 November 10th, 2011: libphonenumber-4.2
 * Code changes
  - Providing an "exact match" isEmergencyNumber method

resources/PhoneNumberMetaData.xml

@@ -4436,7 +4436,7 @@
     <!-- Costa Rica -->
     <!-- http://www.itu.int/oth/T0202000030/en -->
     <territory id="CR" countryCode="506" internationalPrefix="00"
-      nationalPrefixForParsing="(1900)" carrierCodeFormattingRule="$CC $FG">
+      nationalPrefixForParsing="(19(?:0[0-2]|19|77))" carrierCodeFormattingRule="$CC $FG">
       <availableFormats>
         <numberFormat pattern="(\d{4})(\d{4})">
           <leadingDigits>
@@ -4463,8 +4463,11 @@
         <!-- Adding 8[67] prefix after numbers were found online where these prefixes have been
              assigned. -->
         <nationalNumberPattern>
-          57[01]\d[01]\d{3}|
-          8[36789]\d{6}
+          5(?:
+            0[0-4]\d{5}|
+            7[01]\d[01]\d{3}
+          )|
+          8[36-9]\d{6}
         </nationalNumberPattern>
         <possibleNumberPattern>\d{8}</possibleNumberPattern>
         <exampleNumber>83123456</exampleNumber>
@@ -4488,7 +4491,8 @@
             2(?:
               00\d|
               900
-            )\d{2}
+            )\d{2}|
+            5\d{5}
           )
         </nationalNumberPattern>
         <possibleNumberPattern>\d{8}</possibleNumberPattern>
@@ -4512,14 +4516,17 @@
             )|
             2(?:
               12|
+              22|
               34
             )|
+            333|
             400|
+            55[15]|
             7(?:
               00|
               1[78]|
               77
-            )|
+            )
           )
         </nationalNumberPattern>
         <possibleNumberPattern>\d{4}</possibleNumberPattern>
@@ -7900,7 +7907,7 @@
           55\d{6}|
           6(?:
             0(?:
-              2\d|
+              2[0-35-9]|
               3[3467]|
               5[2457-9]
             )|
@@ -7913,6 +7920,10 @@
         </nationalNumberPattern>
         <exampleNumber>60201234</exampleNumber>
       </mobile>
+      <voip>
+        <nationalNumberPattern>78\d{6}</nationalNumberPattern>
+        <exampleNumber>78123456</exampleNumber>
+      </voip>
       <!-- No national emergency numbers were located for Guinea. -->
     </territory>
 
@@ -11033,10 +11044,14 @@
         </numberFormat>
         <!-- National-only toll-free numbers (0077 and 0088, where the leading "0" is considered the
              national prefix) -->
-        <numberFormat pattern="(\d{3})(\d{4,5})">
+        <numberFormat pattern="(\d{3})(\d{4})">
           <leadingDigits>077</leadingDigits>
           <format>$1-$2</format>
         </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{2})(\d{3,4})">
+          <leadingDigits>077</leadingDigits>
+          <format>$1-$2-$3</format>
+        </numberFormat>
         <numberFormat pattern="(\d{3})(\d{2})(\d{4})">
           <leadingDigits>088</leadingDigits>
           <format>$1-$2-$3</format>
@@ -11809,7 +11824,7 @@
         <nationalNumberPattern>
           [1-9]\d{8,9}|
           0(?:
-            7\d{5,6}|
+            7\d{5,7}|
             8\d{7}
           )
         </nationalNumberPattern>
@@ -11821,12 +11836,12 @@
           0(?:
             777(?:
               [01]\d{2}|
-              5\d{3}
+              5\d{3}|
+              8\d{4}
             )|
             882[1245]\d{4}
           )
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{7,9}</possibleNumberPattern>
         <exampleNumber>0777012</exampleNumber>
       </noInternationalDialling>
       <fixedLine>
@@ -11899,12 +11914,12 @@
           0(?:
             777(?:
               [01]\d{2}|
-              5\d{3}
+              5\d{3}|
+              8\d{4}
             )|
             882[1245]\d{4}
           )
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{7,10}</possibleNumberPattern>
         <exampleNumber>120123456</exampleNumber>
       </tollFree>
       <premiumRate>
@@ -11941,22 +11956,23 @@
                nationalPrefix="0" nationalPrefixFormattingRule="$NP$FG">
       <availableFormats>
         <numberFormat pattern="(\d{2})(\d{4,7})">
-          <leadingDigits>
-            [2-6]|
-            91
-          </leadingDigits>
+          <leadingDigits>[24-6]</leadingDigits>
           <format>$1 $2</format>
         </numberFormat>
         <numberFormat pattern="(\d{3})(\d{6,7})">
-          <leadingDigits>
-            [78]|
-            90
-          </leadingDigits>
+          <leadingDigits>7</leadingDigits>
           <format>$1 $2</format>
         </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{3})(\d{3,4})">
+          <leadingDigits>[89]</leadingDigits>
+          <format>$1 $2 $3</format>
+        </numberFormat>
       </availableFormats>
       <generalDesc>
-        <nationalNumberPattern>\d{6,10}</nationalNumberPattern>
+        <nationalNumberPattern>
+          20\d{4,7}|
+          [4-9]\d{5,9}
+        </nationalNumberPattern>
         <possibleNumberPattern>\d{4,10}</possibleNumberPattern>
       </generalDesc>
       <fixedLine>
@@ -11972,41 +11988,52 @@
         <exampleNumber>202012345</exampleNumber>
       </fixedLine>
       <mobile>
-        <!-- Adding 70[0-5], 71[6-9], 73[019], 753 and 77[145] after successful delivery of SMSs.
-             -->
+        <!-- Adding 70[3-5], and 775 after successful delivery of SMSs. -->
         <nationalNumberPattern>
           7(?:
             0[0-5]|
             [123]\d|
             5[0-3]|
-            7[0-5]
+            7[0-5]|
+            8[6-9]
           )\d{6}
         </nationalNumberPattern>
         <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>712123456</exampleNumber>
       </mobile>
-      <!-- There is no document that has the information on the actual length of premium rates and
-           tollfree numbers. The information below comes from research on existing numbers. -->
       <tollFree>
-        <nationalNumberPattern>
-          8(?:
-            00|
-            88
-          )\d{6,7}
-        </nationalNumberPattern>
+        <!-- Longer numbers have been found than the plan suggests, so we support them here too. -->
+        <nationalNumberPattern>800[245-8]\d{5,6}</nationalNumberPattern>
         <possibleNumberPattern>\d{9,10}</possibleNumberPattern>
-        <exampleNumber>800123456</exampleNumber>
+        <exampleNumber>800223456</exampleNumber>
       </tollFree>
       <premiumRate>
-         <nationalNumberPattern>
-           9(?:
-             00|
-             1
-           )\d{6,7}
-         </nationalNumberPattern>
-         <possibleNumberPattern>\d{8,10}</possibleNumberPattern>
-         <exampleNumber>900123456</exampleNumber>
+        <nationalNumberPattern>
+          9(?:
+            00[2-578]|
+            11\d
+          )\d{5}
+        </nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
+        <exampleNumber>900223456</exampleNumber>
       </premiumRate>
+      <shortCode>
+        <nationalNumberPattern>
+          1(?:
+            0[089]|
+            1(?:
+              [06]|
+              99
+            )|
+            2[123]|
+            3[013]|
+            4[14]|
+            501
+          )
+        </nationalNumberPattern>
+        <possibleNumberPattern>\d{3,4}</possibleNumberPattern>
+        <exampleNumber>116</exampleNumber>
+      </shortCode>
       <emergency>
         <!-- http://www.kenyapolice.go.ke/report_a_crime.asp -->
         <nationalNumberPattern>
@@ -17750,24 +17777,50 @@
           </leadingDigits>
           <format>$1 $2 $3 $4</format>
         </numberFormat>
+        <numberFormat pattern="(\d{2})(\d{4,6})">
+          <!-- Pattern for shorter fixed-line numbers. -->
+          <leadingDigits>
+            [124]|
+            3[2-4]|
+            5[24-689]|
+            6[1-3578]|
+            7[14-7]|
+            8[1-7]
+          </leadingDigits>
+          <format>$1 $2</format>
+        </numberFormat>
         <!-- We are formatting 70 numbers as per mobile numbers, based on information from some
              Poles that this is more usual. -->
         <numberFormat pattern="(\d{3})(\d{3})(\d{3})">
           <leadingDigits>
             39|
             5[013]|
-            6[069]|
+            6[0469]|
             7[0289]|
             8[08]
           </leadingDigits>
           <format>$1 $2 $3</format>
         </numberFormat>
+        <!-- Additional patterns for shorter pager numbers. -->
+        <numberFormat pattern="(\d{3})(\d{2})(\d{2,3})">
+          <leadingDigits>64</leadingDigits>
+          <format>$1 $2 $3</format>
+        </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{3})">
+          <leadingDigits>64</leadingDigits>
+          <format>$1 $2</format>
+        </numberFormat>
       </availableFormats>
       <generalDesc>
-        <nationalNumberPattern>[1-9]\d{8}</nationalNumberPattern>
-        <possibleNumberPattern>\d{9}</possibleNumberPattern>
+        <nationalNumberPattern>
+          [1-58]\d{6,8}|
+          9\d{8}|
+          [67]\d{5,8}
+        </nationalNumberPattern>
+        <possibleNumberPattern>\d{6,9}</possibleNumberPattern>
       </generalDesc>
       <fixedLine>
+        <!-- Grouped by prefix-length. -->
         <nationalNumberPattern>
           (?:
             1[2-8]|
@@ -17776,8 +17829,12 @@
             4[1-468]|
             5[24-689]|
             6[1-3578]|
-            7[14-7]|
-            8[1-79]|
+            7[14-6]|
+            8[1-7]
+          )\d{5,7}|
+          77\d{4,6}|
+          (?:
+            89|
             9[145]
           )\d{7}
         </nationalNumberPattern>
@@ -17792,22 +17849,31 @@
             88
           )\d{7}
         </nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>512345678</exampleNumber>
       </mobile>
+      <pager>
+        <nationalNumberPattern>642\d{3,6}</nationalNumberPattern>
+        <exampleNumber>642123456</exampleNumber>
+      </pager>
       <tollFree>
         <nationalNumberPattern>800\d{6}</nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>800123456</exampleNumber>
       </tollFree>
       <premiumRate>
         <nationalNumberPattern>70\d{7}</nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>701234567</exampleNumber>
       </premiumRate>
       <sharedCost>
         <nationalNumberPattern>801\d{6}</nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>801234567</exampleNumber>
       </sharedCost>
       <voip>
         <nationalNumberPattern>39\d{7}</nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>391234567</exampleNumber>
       </voip>
       <emergency>
@@ -19312,13 +19378,7 @@
         <exampleNumber>1312</exampleNumber>
       </shortCode>
       <emergency>
-        <nationalNumberPattern>
-          112|
-          9(?:
-            11|
-            9[59]
-          )
-        </nationalNumberPattern>
+        <nationalNumberPattern>99[59]</nationalNumberPattern>
         <possibleNumberPattern>\d{3}</possibleNumberPattern>
         <exampleNumber>999</exampleNumber>
       </emergency>
@@ -19914,7 +19974,7 @@
         <!-- Adding 74 from numbers found online. -->
         <nationalNumberPattern>
           (?:
-            7[1245]|
+            7[1-5]|
             8[1-9]
           )\d{5}
         </nationalNumberPattern>
@@ -20044,7 +20104,12 @@
         <possibleNumberPattern>\d{7}(?:\d{3})?</possibleNumberPattern>
       </generalDesc>
       <fixedLine>
-        <nationalNumberPattern>72154[2-8]\d{4}</nationalNumberPattern>
+        <nationalNumberPattern>
+          7215(?:
+            4[2-8]|
+            8[239]
+          )\d{4}
+        </nationalNumberPattern>
         <exampleNumber>7215425678</exampleNumber>
       </fixedLine>
       <mobile>
@@ -20054,7 +20119,7 @@
           7215(?:
             1[02]|
             2\d|
-            5[0346]|
+            5[03469]|
             8[01678]
           )\d{4}
         </nationalNumberPattern>
@@ -20411,6 +20476,7 @@
 
     <!-- Thailand -->
     <!-- http://www.itu.int/oth/T02020000CD/en -->
+    <!-- http://www.barascientific.com/bscnews/variety/emergency/Tel-4Digi.pdf -->
     <territory id="TH" countryCode="66" internationalPrefix="00"
                nationalPrefix="0" nationalPrefixFormattingRule="$NP$FG">
       <availableFormats>
@@ -20437,10 +20503,15 @@
       <generalDesc>
         <nationalNumberPattern>
           [2-9]\d{7,8}|
-          1\d{9}
+          1\d{3}(?:\d{6})?
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{8,10}</possibleNumberPattern>
+        <possibleNumberPattern>\d{4}|\d{8,10}</possibleNumberPattern>
       </generalDesc>
+      <noInternationalDialling>
+        <nationalNumberPattern>1\d{3}</nationalNumberPattern>
+        <possibleNumberPattern>\d{4}</possibleNumberPattern>
+        <exampleNumber>1100</exampleNumber>
+      </noInternationalDialling>
       <fixedLine>
         <nationalNumberPattern>
           (?:
@@ -20477,6 +20548,13 @@
         <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>601234567</exampleNumber>
       </voip>
+      <uan>
+        <!-- 4-digit numbers are classified under UAN. Most of these are commercial numbers, and
+             their costs range from free to 6 Baht/min. -->
+        <nationalNumberPattern>1\d{3}</nationalNumberPattern>
+        <possibleNumberPattern>\d{4}</possibleNumberPattern>
+        <exampleNumber>1100</exampleNumber>
+      </uan>
       <emergency>
         <nationalNumberPattern>
           1(?:
@@ -20587,14 +20665,19 @@
     <!-- Tokelau -->
     <!-- http://www.itu.int/oth/T02020000D2/en -->
     <territory id="TK" countryCode="690" internationalPrefix="00">
-      <!-- No more information beyond leading digit and number length can be found. Numbers are
-           formatted as a block. The example number is the test number provided in the ITU document.
-           -->
       <generalDesc>
-        <nationalNumberPattern>[1-9]\d{3}</nationalNumberPattern>
+        <nationalNumberPattern>[2-5]\d{3}</nationalNumberPattern>
         <possibleNumberPattern>\d{4}</possibleNumberPattern>
-        <exampleNumber>3190</exampleNumber>
       </generalDesc>
+      <fixedLine>
+        <nationalNumberPattern>[2-4]\d{3}</nationalNumberPattern>
+        <!-- The example number is the contact number from the ITU document. -->
+        <exampleNumber>3010</exampleNumber>
+      </fixedLine>
+      <mobile>
+        <nationalNumberPattern>5\d{3}</nationalNumberPattern>
+        <exampleNumber>5190</exampleNumber>
+      </mobile>
       <!-- No evidence that any emergency numbers exist for this tiny island has been found. -->
     </territory>
 

resources/PhoneNumberMetaDataForTesting.xml

@@ -346,9 +346,9 @@
           <format>$1 $2 $3 $4</format>
         </numberFormat>
         <numberFormat pattern="(\d{2})(\d{3})(\d{4})">
-          <leadingDigits>222|333</leadingDigits>
-          <leadingDigits>(?:222|333)1</leadingDigits>
-          <leadingDigits>(?:222|333)11</leadingDigits>
+          <leadingDigits>111|222|333</leadingDigits>
+          <leadingDigits>(?:111|222|333)1</leadingDigits>
+          <leadingDigits>(?:111|222|333)11</leadingDigits>
           <format>$1 $2 $3</format>
         </numberFormat>
         <numberFormat pattern="(\d{4})(\d)(\d{4})">
@@ -364,11 +364,15 @@
         </numberFormat>
         <!-- The following numberFormat is added to test that the format containing the star sign is
              not used by the AYTF. -->
-        <numberFormat pattern="(\d{4})">
+        <numberFormat nationalPrefixFormattingRule="$FG" pattern="(\d{4})">
           <leadingDigits>[23]</leadingDigits>
           <format>*$1</format>
         </numberFormat>
       </availableFormats>
+      <noInternationalDialling>
+        <nationalNumberPattern>[23]\d{3}</nationalNumberPattern>
+        <possibleNumberPattern>\d{4}</possibleNumberPattern>
+      </noInternationalDialling>
     </territory>
 
     <!-- Korea (Rep. of) -->