feat: update regex, bug fixes, add tests

src/phonenumberutil/helper_constants/helper_constants.rs

@@ -50,11 +50,11 @@ pub const VALID_ALPHA_INCL_UPPERCASE: &'static str = "A-Za-z";
 // prefix. This can be overridden by region-specific preferences.
 pub const DEFAULT_EXTN_PREFIX: &'static str = " ext. ";
 
-pub const POSSIBLE_SEPARATORS_BETWEEN_NUMBER_AND_EXT_LABEL: &'static str = "0001";
+pub const POSSIBLE_SEPARATORS_BETWEEN_NUMBER_AND_EXT_LABEL: &'static str = "[ \u{00A0}\\t,]*";
 
 // Optional full stop (.) or colon, followed by zero or more
 // spaces/tabs/commas.
-pub const POSSIBLE_CHARS_AFTER_EXT_LABEL: &'static str = "[ \u{00A0}\\t,]*";
-pub const OPTIONAL_EXT_SUFFIX: &'static str = "[:\\.\u{FF0E}]?[ \u{00A0}\\t,-]*";
+pub const POSSIBLE_CHARS_AFTER_EXT_LABEL: &'static str = "[:\\.\u{FF0E}]?[ \u{00A0}\\t,-]*";
+pub const OPTIONAL_EXT_SUFFIX: &'static str = "#?";
 
 pub const NANPA_COUNTRY_CODE: i32 = 1;

src/phonenumberutil/phone_number_regexps_and_mappings.rs

@@ -307,13 +307,12 @@ impl PhoneNumberRegExpsAndMappings {
             separator_pattern: Regex::new(&format!("[{}]+", VALID_PUNCTUATION)).unwrap(),
             extn_patterns_for_matching: create_extn_pattern(false),
             extn_pattern: Regex::new(&format!("(?i)(?:{})$", &extn_patterns_for_parsing)).unwrap(),
-            valid_phone_number_pattern: Regex::new(&format!("(?i){}(?:{})?", 
+            valid_phone_number_pattern: Regex::new(&format!("(?i)(?:{})(?:{})?", 
                 &valid_phone_number,
-                extn_patterns_for_parsing
-            )).unwrap(),
-            valid_alpha_phone_pattern: Regex::new(&format!("(?i)(?:.*?[{}]){{3}}",
-                VALID_ALPHA
+                &extn_patterns_for_parsing
             )).unwrap(),
+            // from java
+            valid_alpha_phone_pattern: Regex::new("(?:.*?[A-Za-z]){3}.*").unwrap(),
             // The first_group_capturing_pattern was originally set to $1 but there
             // are some countries for which the first group is not used in the
             // national pattern (e.g. Argentina) so the $1 group does not match