feat: update regex, bug fixes, add tests

.vscode/settings.json

@@ -0,0 +1,3 @@
+{
+    "java.configuration.updateBuildConfiguration": "interactive"
+}

Cargo.toml

@@ -10,7 +10,7 @@ build = "build/rust_build.rs"
 log = "0.4.27"
 # helpful error package
 thiserror = "2.0.12"
-# google protobuf lib required to use .proto files from assets
+# protobuf lib required to use .proto files from assets
 protobuf = "3.7.2"
 # optimized concurrent map
 dashmap = "6.1.0"

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 POSSIBLE_CHARS_AFTER_EXT_LABEL: &'static str = "[:\\.\u{FF0E}]?[ \u{00A0}\\t,-]*";
-pub const OPTIONAL_EXT_SUFFIX: &'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
+                &extn_patterns_for_parsing
-            )).unwrap(),
-            valid_alpha_phone_pattern: Regex::new(&format!("(?i)(?:.*?[{}]){{3}}",
-                VALID_ALPHA
             )).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

src/tests/mod.rs

@@ -1 +1,2 @@
 mod tests;
+pub(self) mod region_code;

src/tests/region_code.rs

@@ -0,0 +1,160 @@
+pub struct RegionCode {}
+
+impl RegionCode {
+    pub fn ad() -> &'static str {
+        "AD"
+    }
+
+    pub fn ae() -> &'static str {
+        "AE"
+    }
+
+    pub fn am() -> &'static str {
+        "AM"
+    }
+
+    pub fn ao() -> &'static str {
+        "AO"
+    }
+
+    pub fn aq() -> &'static str {
+        "AQ"
+    }
+
+    pub fn ar() -> &'static str {
+        "AR"
+    }
+
+    pub fn au() -> &'static str {
+        "AU"
+    }
+
+    pub fn bb() -> &'static str {
+        "BB"
+    }
+
+    pub fn br() -> &'static str {
+        "BR"
+    }
+
+    pub fn bs() -> &'static str {
+        "BS"
+    }
+
+    pub fn by() -> &'static str {
+        "BY"
+    }
+
+    pub fn ca() -> &'static str {
+        "CA"
+    }
+
+    pub fn ch() -> &'static str {
+        "CH"
+    }
+
+    pub fn cl() -> &'static str {
+        "CL"
+    }
+
+    pub fn cn() -> &'static str {
+        "CN"
+    }
+
+    pub fn co() -> &'static str {
+        "CO"
+    }
+
+    pub fn cs() -> &'static str {
+        "CS"
+    }
+
+    pub fn cx() -> &'static str {
+        "CX"
+    }
+
+    pub fn de() -> &'static str {
+        "DE"
+    }
+
+    pub fn fr() -> &'static str {
+        "FR"
+    }
+
+    pub fn gb() -> &'static str {
+        "GB"
+    }
+
+    pub fn hu() -> &'static str {
+        "HU"
+    }
+
+    pub fn it() -> &'static str {
+        "IT"
+    }
+
+    pub fn jp() -> &'static str {
+        "JP"
+    }
+
+    pub fn kr() -> &'static str {
+        "KR"
+    }
+
+    pub fn mx() -> &'static str {
+        "MX"
+    }
+
+    pub fn nz() -> &'static str {
+        "NZ"
+    }
+
+    pub fn pl() -> &'static str {
+        "PL"
+    }
+
+    pub fn re() -> &'static str {
+        "RE"
+    }
+
+    pub fn ru() -> &'static str {
+        "RU"
+    }
+
+    pub fn se() -> &'static str {
+        "SE"
+    }
+
+    pub fn sg() -> &'static str {
+        "SG"
+    }
+
+    pub fn un001() -> &'static str {
+        "001"
+    }
+
+    pub fn us() -> &'static str {
+        "US"
+    }
+
+    pub fn uz() -> &'static str {
+        "UZ"
+    }
+
+    pub fn yt() -> &'static str {
+        "YT"
+    }
+
+    pub fn zw() -> &'static str {
+        "ZW"
+    }
+
+    /// s a region code string representing the "unknown" region.
+    pub fn get_unknown() -> &'static str {
+        Self::zz()
+    }
+
+    pub fn zz() -> &'static str {
+        "ZZ"
+    }
+}