Support semicolon as extension character while parsing phone numbers. (#1458)

* Support semicolon as extension character while parsing

* Add notes to pending_code_changes.txt

* JS port: Support semicolon as extension character while parsing

* Update comments in phonenumberutil.js

cpp/src/phonenumbers/phonenumberutil.cc

@@ -208,9 +208,9 @@ string CreateExtnPattern(const string& single_extn_symbols) {
   // The first regular expression covers RFC 3966 format, where the extension is
   // added using ";ext=". The second more generic one starts with optional white
   // space and ends with an optional full stop (.), followed by zero or more
-  // spaces/tabs and then the numbers themselves. The third one covers the
-  // special case of American numbers where the extension is written with a hash
-  // at the end, such as "- 503#".
+  // spaces/tabs/commas and then the numbers themselves. The third one covers
+  // the special case of American numbers where the extension is written with a
+  // hash at the end, such as "- 503#".
   // Note that the only capturing groups should be around the digits that you
   // want to capture as part of the extension, or else parsing will fail!
   // Canonical-equivalence doesn't seem to be an option with RE2, so we allow
@@ -451,8 +451,8 @@ class PhoneNumberRegExpsAndMappings {
   // will be run as a case-insensitive regexp match. Wide character versions are
   // also provided after each ASCII version.
   // For parsing, we are slightly more lenient in our interpretation than for
-  // matching. Here we allow a "comma" as a possible extension indicator. When
-  // matching, this is hardly ever used to indicate this.
+  // matching. Here we allow "comma" and "semicolon" as possible extension
+  // indicators. When matching, these are hardly ever used to indicate this.
   const string extn_patterns_for_parsing_;
 
  public:
@@ -570,7 +570,7 @@ class PhoneNumberRegExpsAndMappings {
                    punctuation_and_star_sign_, kDigits,
                    "]*")),
         extn_patterns_for_parsing_(
-            CreateExtnPattern(StrCat(",", kSingleExtnSymbolsForMatching))),
+            CreateExtnPattern(StrCat(",;", kSingleExtnSymbolsForMatching))),
         regexp_factory_(new RegExpFactory()),
         regexp_cache_(new RegExpCache(*regexp_factory_.get(), 128)),
         diallable_char_mappings_(),

cpp/test/phonenumbers/phonenumbermatcher_test.cc

@@ -253,6 +253,10 @@ class PhoneNumberMatcherTest : public testing::Test {
     // With trailing numbers after a comma. The 45 should not be considered an
     // extension.
     context_pairs.push_back(NumberContext("", ", 45 days a year"));
+    // When matching we don't consider semicolon along with legitimate extension
+    // symbol to indicate an extension. The 7246433 should not be considered an
+    // extension.
+    context_pairs.push_back(NumberContext("", ";x 7246433"));
     // With a postfix stripped off as it looks like the start of another number.
     context_pairs.push_back(NumberContext("Call ", "/x12 more"));
 

cpp/test/phonenumbers/phonenumberutil_test.cc

@@ -2824,6 +2824,9 @@ TEST_F(PhoneNumberUtilTest, IsNumberMatchMatches) {
   EXPECT_EQ(PhoneNumberUtil::EXACT_MATCH,
             phone_util_.IsNumberMatchWithTwoStrings("+64 3 331-6005 extn 1234",
                                                     "+6433316005#1234"));
+  EXPECT_EQ(PhoneNumberUtil::EXACT_MATCH,
+            phone_util_.IsNumberMatchWithTwoStrings("+64 3 331-6005 extn 1234",
+                                                    "+6433316005;1234"));
   // Test proto buffers.
   PhoneNumber nz_number;
   nz_number.set_country_code(64);
@@ -3792,6 +3795,15 @@ TEST_F(PhoneNumberUtilTest, ParseExtensions) {
             phone_util_.Parse("(800) 901-3355 , ext 7246433", RegionCode::US(),
                               &test_number));
   EXPECT_EQ(us_with_extension, test_number);
+  EXPECT_EQ(PhoneNumberUtil::NO_PARSING_ERROR,
+            phone_util_.Parse("(800) 901-3355 ; 7246433", RegionCode::US(),
+                              &test_number));
+  EXPECT_EQ(us_with_extension, test_number);
+  // To test an extension character without surrounding spaces.
+  EXPECT_EQ(PhoneNumberUtil::NO_PARSING_ERROR,
+            phone_util_.Parse("(800) 901-3355;7246433", RegionCode::US(),
+                              &test_number));
+  EXPECT_EQ(us_with_extension, test_number);
   EXPECT_EQ(PhoneNumberUtil::NO_PARSING_ERROR,
             phone_util_.Parse("(800) 901-3355 ,extension 7246433",
                               RegionCode::US(),

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

@@ -313,9 +313,9 @@ public class PhoneNumberUtil {
     // One-character symbols that can be used to indicate an extension.
     String singleExtnSymbolsForMatching = "x\uFF58#\uFF03~\uFF5E";
     // For parsing, we are slightly more lenient in our interpretation than for matching. Here we
-    // allow a "comma" as a possible extension indicator. When matching, this is hardly ever used to
-    // indicate this.
-    String singleExtnSymbolsForParsing = "," + singleExtnSymbolsForMatching;
+    // allow "comma" and "semicolon" as possible extension indicators. When matching, these are
+    // hardly ever used to indicate this.
+    String singleExtnSymbolsForParsing = ",;" + singleExtnSymbolsForMatching;
 
     EXTN_PATTERNS_FOR_PARSING = createExtnPattern(singleExtnSymbolsForParsing);
     EXTN_PATTERNS_FOR_MATCHING = createExtnPattern(singleExtnSymbolsForMatching);
@@ -328,9 +328,9 @@ public class PhoneNumberUtil {
   private static String createExtnPattern(String singleExtnSymbols) {
     // There are three regular expressions here. The first covers RFC 3966 format, where the
     // extension is added using ";ext=". The second more generic one starts with optional white
-    // space and ends with an optional full stop (.), followed by zero or more spaces/tabs and then
-    // the numbers themselves. The other one covers the special case of American numbers where the
-    // extension is written with a hash at the end, such as "- 503#".
+    // space and ends with an optional full stop (.), followed by zero or more spaces/tabs/commas
+    // and then the numbers themselves. The other one covers the special case of American numbers
+    // where the extension is written with a hash at the end, such as "- 503#"
     // Note that the only capturing groups should be around the digits that you want to capture as
     // part of the extension, or else parsing will fail!
     // Canonical-equivalence doesn't seem to be an option with Android java, so we allow two options

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

@@ -981,6 +981,9 @@ public class PhoneNumberMatcherTest extends TestMetadataTestCase {
         "As I said on 03/10/2011, you may call me at ", ""));
     // With trailing numbers after a comma. The 45 should not be considered an extension.
     contextPairs.add(new NumberContext("", ", 45 days a year"));
+    // When matching we don't consider semicolon along with legitimate extension symbol to indicate
+    // an extension. The 7246433 should not be considered an extension.
+    contextPairs.add(new NumberContext("", ";x 7246433"));
      // With a postfix stripped off as it looks like the start of another number.
     contextPairs.add(new NumberContext("Call ", "/x12 more"));
 

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

@@ -2262,6 +2262,9 @@ public class PhoneNumberUtilTest extends TestMetadataTestCase {
     usWithExtension.setCountryCode(1).setNationalNumber(8009013355L).setExtension("7246433");
     assertEquals(usWithExtension, phoneUtil.parse("(800) 901-3355 x 7246433", RegionCode.US));
     assertEquals(usWithExtension, phoneUtil.parse("(800) 901-3355 , ext 7246433", RegionCode.US));
+    assertEquals(usWithExtension, phoneUtil.parse("(800) 901-3355 ; 7246433", RegionCode.US));
+    // To test an extension character without surrounding spaces.
+    assertEquals(usWithExtension, phoneUtil.parse("(800) 901-3355;7246433", RegionCode.US));
     assertEquals(usWithExtension,
                  phoneUtil.parse("(800) 901-3355 ,extension 7246433", RegionCode.US));
     assertEquals(usWithExtension,
@@ -2405,6 +2408,8 @@ public class PhoneNumberUtilTest extends TestMetadataTestCase {
     // Test numbers with extensions.
     assertEquals(PhoneNumberUtil.MatchType.EXACT_MATCH,
                  phoneUtil.isNumberMatch("+64 3 331-6005 extn 1234", "+6433316005#1234"));
+    assertEquals(PhoneNumberUtil.MatchType.EXACT_MATCH,
+                 phoneUtil.isNumberMatch("+64 3 331-6005 ext. 1234", "+6433316005;1234"));
     // Test proto buffers.
     assertEquals(PhoneNumberUtil.MatchType.EXACT_MATCH,
                  phoneUtil.isNumberMatch(NZ_NUMBER, "+6403 331 6005"));

java/pending_code_changes.txt

@@ -3,3 +3,5 @@ Code changes:
    of phone number objects. These have been marked deprecated for months. Any
    users of these methods should call PhoneNumberUtil.parse first to create a
    PhoneNumber object, and pass this in.
+ - Support semicolon as extension character while parsing phone numbers. This
+   is not applicable when you are trying to find the phone numbers.

javascript/i18n/phonenumbers/phonenumberutil.js

@@ -763,13 +763,14 @@ i18n.phonenumbers.PhoneNumberUtil.CAPTURING_EXTN_DIGITS_ =
  * also provided after each ASCII version. There are three regular expressions
  * here. The first covers RFC 3966 format, where the extension is added using
  * ';ext='. The second more generic one starts with optional white space and
- * ends with an optional full stop (.), followed by zero or more spaces/tabs and
- * then the numbers themselves. The other one covers the special case of
- * American numbers where the extension is written with a hash at the end, such
- * as '- 503#'. Note that the only capturing groups should be around the digits
- * that you want to capture as part of the extension, or else parsing will fail!
- * We allow two options for representing the accented o - the character itself,
- * and one in the unicode decomposed form with the combining acute accent.
+ * ends with an optional full stop (.), followed by zero or more spaces/tabs
+ * /commas and then the numbers themselves. The other one covers the special
+ * case of American numbers where the extension is written with a hash at the
+ * end, such as '- 503#'. Note that the only capturing groups should be around
+ * the digits that you want to capture as part of the extension, or else parsing
+ * will fail! We allow two options for representing the accented o - the
+ * character itself, and one in the unicode decomposed form with the combining
+ * acute accent.
  *
  * @const
  * @type {string}
@@ -780,7 +781,7 @@ i18n.phonenumbers.PhoneNumberUtil.EXTN_PATTERNS_FOR_PARSING_ =
     i18n.phonenumbers.PhoneNumberUtil.CAPTURING_EXTN_DIGITS_ + '|' +
     '[ \u00A0\\t,]*' +
     '(?:e?xt(?:ensi(?:o\u0301?|\u00F3))?n?|\uFF45?\uFF58\uFF54\uFF4E?|' +
-    '[,x\uFF58#\uFF03~\uFF5E]|int|anexo|\uFF49\uFF4E\uFF54)' +
+    '[;,x\uFF58#\uFF03~\uFF5E]|int|anexo|\uFF49\uFF4E\uFF54)' +
     '[:\\.\uFF0E]?[ \u00A0\\t,-]*' +
     i18n.phonenumbers.PhoneNumberUtil.CAPTURING_EXTN_DIGITS_ + '#?|' +
     '[- ]+([' + i18n.phonenumbers.PhoneNumberUtil.VALID_DIGITS_ + ']{1,5})#';

javascript/i18n/phonenumbers/phonenumberutil_test.js

@@ -2928,6 +2928,11 @@ function testParseExtensions() {
       phoneUtil.parse('(800) 901-3355 x 7246433', RegionCode.US)));
   assertTrue(usWithExtension.equals(
       phoneUtil.parse('(800) 901-3355 , ext 7246433', RegionCode.US)));
+  assertTrue(usWithExtension.equals(
+      phoneUtil.parse('(800) 901-3355 ; 7246433', RegionCode.US)));
+  // To test an extension character without surrounding spaces.
+  assertTrue(usWithExtension.equals(
+      phoneUtil.parse('(800) 901-3355;7246433', RegionCode.US)));
   assertTrue(usWithExtension.equals(
       phoneUtil.parse('(800) 901-3355 ,extension 7246433', RegionCode.US)));
   assertTrue(usWithExtension.equals(
@@ -3120,6 +3125,9 @@ function testIsNumberMatchMatches() {
   assertEquals(i18n.phonenumbers.PhoneNumberUtil.MatchType.EXACT_MATCH,
                phoneUtil.isNumberMatch('+64 3 331-6005 extn 1234',
                                        '+6433316005#1234'));
+  assertEquals(i18n.phonenumbers.PhoneNumberUtil.MatchType.EXACT_MATCH,
+               phoneUtil.isNumberMatch('+64 3 331-6005 ext. 1234',
+                                       '+6433316005;1234'));
   // Test proto buffers.
   assertEquals(i18n.phonenumbers.PhoneNumberUtil.MatchType.EXACT_MATCH,
                phoneUtil.isNumberMatch(NZ_NUMBER, '+6403 331 6005'));