JAVA: libphonenumber 5.1: AYTF changes and metadata updates.

git-svn-id: http://libphonenumber.googlecode.com/svn/trunk/resources@516 ee073f10-1060-11df-b6a4-87a95322a99c

2 files changed, 305 insertions, 83 deletions

diff --git a/PhoneNumberMetaData.xml b/PhoneNumberMetaData.xml
index fb53359..4528132 100644
--- a/PhoneNumberMetaData.xml
+++ b/PhoneNumberMetaData.xml
@@ -22,7 +22,7 @@
            official short names in English according to ISO 3166-1.
 
      For more information on what each element represents, see
-     resources/phonemetadata.proto
+     java/com/google/i18n/phonenumbers/phonemetadata.proto
 
      Note that if you want to add validation metadata, the generalDesc nationalNumberPattern and
      possibleNumberPattern must be provided. If this is missing, then the country will be
@@ -2471,13 +2471,14 @@
       <mobile>
         <!-- Including the whole range of 75 despite the document restricting it to only a few,
              since diallable numbers have been found outside the range that the document specifies.
-             Including 716 as well since many numbers seem to have this prefix. -->
+             Including 716 as well since many numbers seem to have this prefix. Including 65 as a
+             prefix for Airtel, since they have informed us they are using it. -->
         <nationalNumberPattern>
           (?:
             6(?:
-              0\d|
+              [05]\d|
               1[01]|
-              6[0-4]|
+              6[0-7]|
               8[0-2]
             )|
             7(?:
@@ -2503,27 +2504,30 @@
     <territory id="BG" countryCode="359" internationalPrefix="00"
                nationalPrefix="0" nationalPrefixFormattingRule="$NP$FG">
       <availableFormats>
+        <!-- Formatting rules follow the conventions seen in web-search results. A space has been
+             used to separate the area code from the rest of the number, based on sites like
+             http://www.goldenpages.bg. -->
         <numberFormat pattern="(2)(\d{5})">
           <leadingDigits>29</leadingDigits>
-          <format>$1/$2</format>
+          <format>$1 $2</format>
         </numberFormat>
         <numberFormat pattern="(2)(\d{3})(\d{3,4})">
           <leadingDigits>2</leadingDigits>
-          <format>$1/$2 $3</format>
+          <format>$1 $2 $3</format>
         </numberFormat>
         <numberFormat pattern="(\d{3})(\d{4})">
           <leadingDigits>
             43[124-7]|
             70[1-9]
           </leadingDigits>
-          <format>$1/$2</format>
+          <format>$1 $2</format>
         </numberFormat>
         <numberFormat pattern="(\d{3})(\d{3})(\d{2})">
           <leadingDigits>
             43[124-7]|
             70[1-9]
           </leadingDigits>
-          <format>$1/$2 $3</format>
+          <format>$1 $2 $3</format>
         </numberFormat>
         <numberFormat pattern="(\d{3})(\d{2})(\d{3})">
           <leadingDigits>[78]00</leadingDigits>
@@ -2536,7 +2540,7 @@
             8[1-6]|
             9[1-7]
           </leadingDigits>
-          <format>$1/$2 $3</format>
+          <format>$1 $2 $3</format>
         </numberFormat>
         <numberFormat pattern="(\d{2})(\d{3})(\d{3,4})">
           <leadingDigits>
@@ -2549,8 +2553,8 @@
       </availableFormats>
       <generalDesc>
         <nationalNumberPattern>
-            [23567]\d{5,7}|
-            [489]\d{6,8}
+          [23567]\d{5,7}|
+          [489]\d{6,8}
         </nationalNumberPattern>
         <possibleNumberPattern>\d{5,9}</possibleNumberPattern>
       </generalDesc>
@@ -5978,13 +5982,17 @@
         <exampleNumber>12345678</exampleNumber>
       </fixedLine>
       <mobile>
-        <!-- Adding 65 and 78 from numbers found online. -->
+        <!-- Adding 65 and 78 from numbers found online. Also, prefix 670 is added since the carrier
+             Mobilis Algeria provided it.-->
         <nationalNumberPattern>
           (?:
             5[56]|
-            6[569]|
             7[7-9]
-          )\d{7}
+          )\d{7}|
+          6(?:
+            [569]\d|
+            70
+          )\d{6}
         </nationalNumberPattern>
         <possibleNumberPattern>\d{9}</possibleNumberPattern>
         <exampleNumber>551234567</exampleNumber>
@@ -7152,26 +7160,32 @@
 
     <!-- Gabon -->
     <!-- http://www.itu.int/oth/T020200004E/en -->
-    <territory id="GA" countryCode="241" internationalPrefix="00"
-               nationalPrefix="0" nationalPrefixFormattingRule="$NP$FG">
+    <territory id="GA" countryCode="241" internationalPrefix="00" leading_zero_possible="true"
+               nationalPrefix="0">
       <availableFormats>
-        <numberFormat pattern="(\d)(\d{2})(\d{2})(\d{2})">
+        <numberFormat pattern="(1)(\d{2})(\d{2})(\d{2})" nationalPrefixFormattingRule="$NP$FG">
+          <leadingDigits>1</leadingDigits>
+          <format>$1 $2 $3 $4</format>
+        </numberFormat>
+        <numberFormat pattern="(0\d)(\d{2})(\d{2})(\d{2})">
+          <leadingDigits>0</leadingDigits>
           <format>$1 $2 $3 $4</format>
         </numberFormat>
       </availableFormats>
       <generalDesc>
-        <nationalNumberPattern>[1-7]\d{6}</nationalNumberPattern>
-        <possibleNumberPattern>\d{7}</possibleNumberPattern>
+        <nationalNumberPattern>[01]\d{6,7}</nationalNumberPattern>
+        <possibleNumberPattern>\d{7,8}</possibleNumberPattern>
       </generalDesc>
-      <!-- A 7-digit fixed-line plan was implemented as of June 17, 2012. (The doc refers to an
-           8-digit plan since it counts the initial 0.) -->
+      <!-- A 7-digit fixed-line plan was scheduled to be implemented on June 17, 2012 to unify fixed
+           line and mobile numbering. However the change to remove the leading '0' from mobile NDCs
+           has not gone through. This is now scheduled to happen on October 28, 2012. -->
       <fixedLine>
         <nationalNumberPattern>1\d{6}</nationalNumberPattern>
         <exampleNumber>1441234</exampleNumber>
       </fixedLine>
       <mobile>
-        <nationalNumberPattern>[2-7]\d{6}</nationalNumberPattern>
-        <exampleNumber>6031234</exampleNumber>
+        <nationalNumberPattern>0[2-7]\d{6}</nationalNumberPattern>
+        <exampleNumber>06031234</exampleNumber>
       </mobile>
       <emergency>
         <nationalNumberPattern>
@@ -11234,8 +11248,11 @@
       <availableFormats>
         <!-- The leading zero for fixed numbers will be prepended before the matching of these
              regular expressions. -->
-        <numberFormat pattern="(0[26])(\d{3,4})(\d{4})">
-          <leadingDigits>0[26]</leadingDigits>
+        <numberFormat pattern="(\d{2})(\d{3,4})(\d{4})">
+          <leadingDigits>
+            0[26]|
+            55
+          </leadingDigits>
           <format>$1 $2 $3</format>
         </numberFormat>
         <numberFormat pattern="(0[26])(\d{4})(\d{5})">
@@ -11250,15 +11267,33 @@
           <leadingDigits>0[13-57-9][0159]</leadingDigits>
           <format>$1 $2 $3</format>
         </numberFormat>
-        <numberFormat pattern="(0\d{2})(\d{4,6})">
-          <leadingDigits>0[13-57-9][0159]</leadingDigits>
+        <numberFormat pattern="(\d{3})(\d{3,6})">
+          <leadingDigits>
+            0[13-57-9][0159]|
+            8(?:
+              03|
+              4[17]|
+              9[245]
+            )
+          </leadingDigits>
+          <leadingDigits>
+            0[13-57-9][0159]|
+            8(?:
+              03|
+              4[17]|
+              9(?:
+                2|
+                [45][0-4]
+              )
+            )
+          </leadingDigits>
           <format>$1 $2</format>
         </numberFormat>
         <numberFormat pattern="(0\d{3})(\d{3})(\d{4})">
           <leadingDigits>0[13-57-9][2-46-8]</leadingDigits>
           <format>$1 $2 $3</format>
         </numberFormat>
-        <numberFormat pattern="(0\d{3})(\d{4,6})">
+        <numberFormat pattern="(0\d{3})(\d{2,6})">
           <leadingDigits>0[13-57-9][2-46-8]</leadingDigits>
           <format>$1 $2</format>
         </numberFormat>
@@ -11267,29 +11302,36 @@
             [13]|
             8(?:
               00|
-              4[78]|
-              99
+              4[08]|
+              9[59]
             )
           </leadingDigits>
-          <format>$1 $2 $3</format>
-        </numberFormat>
-        <numberFormat pattern="(\d{3})(\d{4})(\d{4})">
-          <leadingDigits>3</leadingDigits>
-          <format>$1 $2 $3</format>
-        </numberFormat>
-        <numberFormat pattern="(\d{3})(\d{3,6})">
           <leadingDigits>
+            [13]|
             8(?:
-              03|
-              92
+              00|
+              4[08]|
+              9(?:
+                5[5-9]|
+                9
+              )
             )
           </leadingDigits>
+          <format>$1 $2 $3</format>
+        </numberFormat>
+        <numberFormat pattern="(\d{4})(\d{4})">
+          <leadingDigits>894</leadingDigits>
+          <leadingDigits>894[5-9]</leadingDigits>
           <format>$1 $2</format>
         </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{4})(\d{4})">
+          <leadingDigits>3</leadingDigits>
+          <format>$1 $2 $3</format>
+        </numberFormat>
       </availableFormats>
       <generalDesc>
         <nationalNumberPattern>
-          [0189]\d{5,10}|
+          [01589]\d{5,10}|
           3(?:
             [12457-9]\d{8}|
             [36]\d{7,9}
@@ -11298,18 +11340,75 @@
         <possibleNumberPattern>\d{6,11}</possibleNumberPattern>
       </generalDesc>
       <noInternationalDialling>
-        <nationalNumberPattern>848\d{6,7}</nationalNumberPattern>
-        <possibleNumberPattern>\d{9,10}</possibleNumberPattern>
-        <exampleNumber>8481234567</exampleNumber>
+        <nationalNumberPattern>848\d{6}</nationalNumberPattern>
+        <possibleNumberPattern>\d{9}</possibleNumberPattern>
+        <exampleNumber>848123456</exampleNumber>
       </noInternationalDialling>
       <fixedLine>
+        <!-- Maximum lengths in the ITU document are 10, but it states above that for numbers
+             starting with 01, the maximum length is in fact 11. In fact, online, numbers can be
+             found with other prefixes that are 11 digits long as well, so we allow it for all the
+             three and four digit area codes. -->
         <nationalNumberPattern>
           0(?:
             [26]\d{4,9}|
-            [13-57-9](?:
-              [0159]\d{4,8}|
-              [2-46-8]\d{5,8}
-            )
+            (?:
+              1(?:
+                [0159]\d|
+                [27][1-5]|
+                31|
+                4[1-4]|
+                6[1356]|
+                8[2-57]
+              )|
+              3(?:
+                [0159]\d|
+                2[1-4]|
+                3[12]|
+                [48][1-6]|
+                6[2-59]|
+                7[1-7]
+              )|
+              4(?:
+                [0159]\d|
+                [23][1-9]|
+                4[245]|
+                6[1-5]|
+                7[1-4]|
+                81
+              )|
+              5(?:
+                [0159]\d|
+                2[1-5]|
+                3[2-6]|
+                4[1-79]|
+                6[4-6]|
+                7[1-578]|
+                8[3-8]
+              )|
+              7(?:
+                [0159]\d|
+                2[12]|
+                3[1-7]|
+                4[2346]|
+                6[13569]|
+                7[13-6]|
+                8[1-59]
+              )|
+              8(?:
+                [0159]\d|
+                2[34578]|
+                3[1-356]|
+                [6-8][1-5]
+              )|
+              9(?:
+                [0159]\d|
+                [238][1-5]|
+                4[12]|
+                6[1-8]|
+                7[1-6]
+              )
+            )\d{2,7}
           )
         </nationalNumberPattern>
         <possibleNumberPattern>\d{6,11}</possibleNumberPattern>
@@ -11352,26 +11451,80 @@
           )\d{6}|
           89(?:
             2\d{3}|
+            4(?:
+              [0-4]\d{2}|
+              [5-9]\d{4}
+            )|
+            5(?:
+              [0-4]\d{2}|
+              [5-9]\d{6}
+            )|
             9\d{6}
           )
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{6,9}</possibleNumberPattern>
+        <possibleNumberPattern>\d{6,10}</possibleNumberPattern>
         <exampleNumber>899123456</exampleNumber>
       </premiumRate>
       <sharedCost>
-        <nationalNumberPattern>84[78]\d{6,7}</nationalNumberPattern>
-        <possibleNumberPattern>\d{9,10}</possibleNumberPattern>
-        <exampleNumber>8481234567</exampleNumber>
+        <nationalNumberPattern>
+          84(?:
+            [08]\d{6}|
+            [17]\d{3}
+          )
+        </nationalNumberPattern>
+        <possibleNumberPattern>\d{6,9}</possibleNumberPattern>
+        <exampleNumber>848123456</exampleNumber>
       </sharedCost>
-      <!-- The plan says these should be 6 digits long, but when you go to telephone companies in
-           Italy, such as http://www.gnetwork.it/EmailServizi/Numerazioni178/tabid/91/Default.aspx
-           and when you search for 178 numbers, they seem to all be 7 digits, so we cover both
-           lengths here. -->
       <personalNumber>
-        <nationalNumberPattern>178\d{6,7}</nationalNumberPattern>
+        <nationalNumberPattern>
+          1(?:
+            78\d|
+            99
+          )\d{6}
+        </nationalNumberPattern>
         <possibleNumberPattern>\d{9,10}</possibleNumberPattern>
         <exampleNumber>1781234567</exampleNumber>
       </personalNumber>
+      <voip>
+        <nationalNumberPattern>55\d{8}</nationalNumberPattern>
+        <possibleNumberPattern>\d{10}</possibleNumberPattern>
+        <exampleNumber>5512345678</exampleNumber>
+      </voip>
+      <shortCode>
+        <!-- Later, when we have better support for short-codes, we could add the codes
+             used for MMS and SMS services here too. -->
+        <nationalNumberPattern>
+          1(?:
+            1(?:
+              [47]|
+              6\d{3}
+            )|
+            2\d{2}|
+            4(?:
+              82|
+              9\d{1,3}
+            )|
+            5(?:
+              00|
+              1[58]|
+              2[25]|
+              3[03]|
+              44
+            )|
+            86|
+            9(?:
+              2(?:
+                [01]\d{2}|
+                [2-9]\d
+              )|
+              4\d|
+              696
+            )
+          )
+        </nationalNumberPattern>
+        <possibleNumberPattern>\d{3,6}</possibleNumberPattern>
+        <exampleNumber>114</exampleNumber>
+      </shortCode>
       <emergency>
         <nationalNumberPattern>11[2358]</nationalNumberPattern>
         <possibleNumberPattern>\d{3}</possibleNumberPattern>
@@ -11909,8 +12062,40 @@
           <leadingDigits>088</leadingDigits>
           <format>$1-$2-$3</format>
         </numberFormat>
-        <numberFormat pattern="(\d{3})(\d{4})(\d{4})">
-          <leadingDigits>037</leadingDigits>
+        <numberFormat pattern="(\d{3})(\d{3})(\d{3,4})">
+          <leadingDigits>
+            0(?:
+              37|
+              66
+            )
+          </leadingDigits>
+          <format>$1-$2-$3</format>
+        </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{4})(\d{4,5})">
+          <leadingDigits>
+            0(?:
+              37|
+              66
+            )
+          </leadingDigits>
+          <format>$1-$2-$3</format>
+        </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{5})(\d{5,6})">
+          <leadingDigits>
+            0(?:
+              37|
+              66
+            )
+          </leadingDigits>
+          <format>$1-$2-$3</format>
+        </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{6})(\d{6,7})">
+          <leadingDigits>
+            0(?:
+              37|
+              66
+            )
+          </leadingDigits>
           <format>$1-$2-$3</format>
         </numberFormat>
         <!-- Some leading digits are explicitly reserved for a particular purpose.
@@ -12078,7 +12263,7 @@
               9[1-578]
             )|
             2(?:
-              2[03-9]|
+              2[03-689]|
               3[3-58]|
               4[0-468]|
               5[04-8]|
@@ -12147,7 +12332,7 @@
             )|
             2(?:
               2(?:
-                [04-9]|
+                [04-689]|
                 3[23]
               )|
               3[3-58]|
@@ -12265,7 +12450,7 @@
             )|
             2(?:
               2(?:
-                [04-9]|
+                [04-689]|
                 3[23]
               )|
               3[3-58]|
@@ -12404,7 +12589,7 @@
             )|
             2(?:
               2(?:
-                [04-9]|
+                [04-689]|
                 3[23]
               )|
               3[3-58]|
@@ -12537,7 +12722,7 @@
           <leadingDigits>
             1|
             2(?:
-              23|
+              2[37]|
               5[5-9]|
               64|
               78|
@@ -12567,7 +12752,7 @@
           <leadingDigits>
             1|
             2(?:
-              23|
+              2[37]|
               5(?:
                 [57]|
                 [68]0|
@@ -12605,7 +12790,7 @@
           <leadingDigits>
             1|
             2(?:
-              23|
+              2[37]|
               5(?:
                 [57]|
                 [68]0|
@@ -12681,18 +12866,19 @@
         <nationalNumberPattern>
           [1-9]\d{8,9}|
           0(?:
-            3\d{9}|
+            [36]\d{7,14}|
             7\d{5,7}|
             8\d{7}
           )
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{7,11}</possibleNumberPattern>
+        <possibleNumberPattern>\d{7,16}</possibleNumberPattern>
       </generalDesc>
       <noInternationalDialling>
         <!-- Toll-free numbers with a leading "00" cannot be dialled internationally. -->
         <nationalNumberPattern>
           0(?:
-            37\d{8}|
+            37\d{6,13}|
+            66\d{6,13}|
             777(?:
               [01]\d{2}|
               5\d{3}|
@@ -12701,7 +12887,7 @@
             882[1245]\d{4}
           )
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{7,11}</possibleNumberPattern>
+        <possibleNumberPattern>\d{7,16}</possibleNumberPattern>
         <exampleNumber>0777012</exampleNumber>
       </noInternationalDialling>
       <fixedLine>
@@ -12767,11 +12953,20 @@
       </pager>
       <tollFree>
         <!-- http://www.kddi.com/english/business/free_call_dx/number.html and
-             http://tm.softbank.jp/english/business/phone_service/freecall_sp/index.html -->
+             http://tm.softbank.jp/english/business/phone_service/freecall_sp/index.html
+             http://eonet.jp/home/denwa/service/access.html
+             http://ci.fusioncom.co.jp/feature/
+             http://www.auhikari.jp/service/tel/connection/index.html -->
+        <!-- Note that in fact, the number length for 0037 and 0066 numbers should extend to 21
+             digits, confirmed with Yahoo JP. However, this extends well beyond the maximum number
+             length allowed by ITU and hence our library, so we only allow numbers up to 17 digits
+             for now (including both leading 00s). -->
         <nationalNumberPattern>
           120\d{6}|
           800\d{7}|
           0(?:
+            37\d{6,13}|
+            66\d{6,13}|
             777(?:
               [01]\d{2}|
               5\d{3}|
@@ -12780,7 +12975,7 @@
             882[1245]\d{4}
           )
         </nationalNumberPattern>
-        <possibleNumberPattern>\d{7,10}</possibleNumberPattern>
+        <possibleNumberPattern>\d{7,16}</possibleNumberPattern>
         <exampleNumber>120123456</exampleNumber>
       </tollFree>
       <premiumRate>
@@ -12794,13 +12989,8 @@
         <exampleNumber>601234567</exampleNumber>
       </personalNumber>
       <voip>
-        <nationalNumberPattern>
-          (?:
-            037|
-            50
-          )\d{8}
-        </nationalNumberPattern>
-        <possibleNumberPattern>\d{10,11}</possibleNumberPattern>
+        <nationalNumberPattern>50\d{8}</nationalNumberPattern>
+        <possibleNumberPattern>\d{10}</possibleNumberPattern>
         <exampleNumber>5012345678</exampleNumber>
       </voip>
       <!-- Storing "unified number service" as UAN. -->
@@ -13634,7 +13824,7 @@
         <nationalNumberPattern>
           (?:
             5(?:
-              0[0-2568]|
+              0[0-25-9]|
               11|
               5\d
             )|
@@ -18338,7 +18528,7 @@
         <nationalNumberPattern>
           (?:
             [27]\d{2}|
-            3[0-59]\d|
+            3[0-79]\d|
             411
           )\d{3}
         </nationalNumberPattern>
diff --git a/PhoneNumberMetaDataForTesting.xml b/PhoneNumberMetaDataForTesting.xml
index 588a2d7..73440cf 100644
--- a/PhoneNumberMetaDataForTesting.xml
+++ b/PhoneNumberMetaDataForTesting.xml
@@ -177,6 +177,39 @@
       </premiumRate>
     </territory>
 
+    <!-- Belarus -->
+    <!-- This country has been coopted to test the case of a national prefix formatting rule with a
+         space in it. -->
+    <territory id="BY" countryCode="375" internationalPrefix="810" nationalPrefix="8"
+               nationalPrefixForParsing="80?|99999" nationalPrefixFormattingRule="$NP $FG">
+      <availableFormats>
+        <!-- We make some bogus formatting templates that differ only in length to check that
+             switching formatting patterns works as expected.-->
+        <numberFormat pattern="(\d{4})">
+          <leadingDigits>[1-8]</leadingDigits>
+          <format>$1</format>
+        </numberFormat>
+        <!-- This pattern has an override to remove the space after the national prefix. -->
+        <numberFormat pattern="(\d{2})(\d{3})" nationalPrefixFormattingRule="$NP$FG">
+          <leadingDigits>[1-8]</leadingDigits>
+          <format>$1 $2</format>
+        </numberFormat>
+        <numberFormat pattern="(\d{3})(\d{3})">
+          <leadingDigits>[1-8]</leadingDigits>
+          <format>$1 $2</format>
+        </numberFormat>
+      </availableFormats>
+      <generalDesc>
+        <!-- This numbering plan is completely bogus, but is used to test the AYTF logic. For that
+             reason, it is kept very simple. -->
+        <nationalNumberPattern>[1-9]\d{5}</nationalNumberPattern>
+        <possibleNumberPattern>\d{6}</possibleNumberPattern>
+      </generalDesc>
+      <fixedLine>
+        <exampleNumber>112345</exampleNumber>
+      </fixedLine>
+    </territory>
+
     <!-- Germany -->
     <territory id="DE" countryCode="49" internationalPrefix="00"
                nationalPrefix="0" nationalPrefixFormattingRule="$NP$FG">
@@ -701,11 +734,10 @@
 
     <!-- United States -->
     <!-- http://www.nanpa.com/reports/reports_npa.html -->
-    <!-- For testing purposes, numbers starting with 24 are not considered US
-    numbers.-->
+    <!-- For testing purposes, numbers starting with 24 are not considered US numbers.-->
     <territory id="US" countryCode="1" internationalPrefix="011"
                preferredExtnPrefix=" extn. " nationalPrefix="1"
-               mainCountryForCode="true" >
+               mainCountryForCode="true" nationalPrefixOptionalWhenFormatting="true">
       <availableFormats>
         <numberFormat pattern="(\d{3})(\d{4})">
           <format>$1 $2</format>