added more functions from original code, added macro

src/i18n/mod.rs

@@ -0,0 +1,3 @@
+mod region_code;
+
+pub use region_code::RegionCode;

src/i18n/region_code.rs

@@ -0,0 +1,13 @@
+pub struct RegionCode {
+}
+
+impl RegionCode {
+    /// Returns a region code string representing the "unknown" region.
+    pub fn get_unknown() -> &'static str {
+        return Self::zz();
+    }
+
+    pub fn zz() -> &'static str {
+        return "ZZ";    
+    }
+}

src/interfaces/mod.rs

@@ -4,7 +4,7 @@ use crate::proto_gen::phonemetadata::PhoneNumberDesc;
 /// implementation of the matcher and allow different implementations to be
 /// swapped in easily.
 
-pub(crate) trait MatcherApi {
+pub(crate) trait MatcherApi: Send + Sync {
   /// Returns whether the given national number (a string containing only decimal
   /// digits) matches the national number pattern defined in the given
   /// PhoneNumberDesc message.

src/lib.rs

@@ -5,3 +5,11 @@ mod proto_gen;
 mod phonenumberutil;
 mod regexp_cache;
 mod regex_based_matcher;
+pub mod i18n;
+pub mod regex_util;
+
+/// I decided to create this module because there are many 
+/// boilerplate places in the code that can be replaced with macros, 
+/// the name of which will describe what is happening more 
+/// clearly than a few lines of code.
+mod macros;

src/macros/mod.rs

@@ -0,0 +1,21 @@
+// std::borrow::Cow
+// std::option::Option
+
+/// This macro extracts owned value from cow
+/// but if cow is borrowed it returns default given value
+/// 
+/// it's helpful when function returns `Cow<'_, T>` as result,
+/// where `Cow::Borrowed` option marks that value was not modified
+/// and we can use owned original instead of copying it.
+macro_rules! owned_from_cow_or {
+    ($getcow:expr, $default:expr) => {{
+        if let std::borrow::Cow::Owned(s) = $getcow {
+            s
+        } else {
+            $default
+        }
+    }};
+}
+
+pub(crate) use owned_from_cow_or;
+

src/phonenumberutil/errors.rs

@@ -0,0 +1,9 @@
+use thiserror::Error;
+
+use crate::regexp_cache::ErrorInvalidRegex;
+
+#[derive(Debug, PartialEq, Error)]
+pub enum PhoneNumberUtilError {
+    #[error("{0}")]
+    InvalidRegexError(#[from] ErrorInvalidRegex)
+}

src/phonenumberutil/helper_functions.rs

@@ -253,11 +253,13 @@ pub(super) fn create_extn_pattern(for_parsing: bool) -> String {
 /// * `remove_non_matches` - indicates whether characters that are not able to be
 ///   replaced should be stripped from the number. If this is false, they will be
 ///   left unchanged in the number.
+/// 
+/// Returns: normalized_string
 pub(super) fn normalize_helper(
     normalization_replacements: &HashMap<char, char>,
     remove_non_matches: bool,
-    phone_number: &mut String,
-) {
+    phone_number: &str
+) -> String {
     let mut normalized_number = String::with_capacity(phone_number.len());
     // Skip UTF checking because strings in rust are valid UTF-8 already
     for phone_char in phone_number.chars() {
@@ -269,7 +271,7 @@ pub(super) fn normalize_helper(
         // If neither of the above are true, we remove this character.
     }
 
-    *phone_number = normalized_number;
+    normalized_number
 }
 
 /// Returns `true` if there is any possible number data set for a particular
@@ -307,7 +309,7 @@ pub(super) fn desc_has_data(desc: &PhoneNumberDesc) -> bool {
 
 /// Returns the types we have metadata for based on the PhoneMetadata object
 /// passed in.
-pub(super) fn get_supported_types_for_metadata(
+pub(super) fn populate_supported_types_for_metadata(
     metadata: &PhoneMetadata,
     types: &mut HashSet<PhoneNumberType>,
 ) {
@@ -327,6 +329,13 @@ pub(super) fn get_supported_types_for_metadata(
         });
 }
 
+pub(super) fn get_supported_types_for_metadata(metadata: &PhoneMetadata) -> HashSet<PhoneNumberType> {
+    const EFFECTIVE_NUMBER_TYPES: usize = 11 /* count */ - 2 /* filter type or unknown */;
+    let mut types = HashSet::with_capacity(EFFECTIVE_NUMBER_TYPES);
+    populate_supported_types_for_metadata(metadata, &mut types);
+    types
+}
+
 /// Helper method to check a number against possible lengths for this number
 /// type, and determine whether it matches, or is too short or too long.
 pub(super) fn test_number_length(
@@ -444,7 +453,7 @@ pub(crate) fn copy_core_fields_only(from_number: &PhoneNumber, to_number: &mut P
 /// Determines whether the given number is a national number match for the given
 /// PhoneNumberDesc. Does not check against possible lengths!
 pub(super) fn is_match(
-    matcher_api: Box<dyn MatcherApi>,
+    matcher_api: &Box<dyn MatcherApi>,
     number: &str,
     number_desc: &PhoneNumberDesc,
 ) -> bool {

src/phonenumberutil/mod.rs

@@ -1,13 +1,17 @@
 mod helper_constants;
 pub mod helper_functions;
+mod errors;
 mod enums;
 mod phonenumberutil;
-mod regex_and_mappings;
 mod phone_number_regexps_and_mappings;
 
+use std::sync::LazyLock;
+
 pub use enums::{MatchType, PhoneNumberFormat, PhoneNumberType, ValidationResultErr, ValidNumberLenType};
 use thiserror::Error;
 
+use crate::phonenumberutil::phonenumberutil::PhoneNumberUtil;
+
 #[derive(Debug, Error)]
 pub enum ErrorType {
     #[error("No parsing")]
@@ -23,3 +27,7 @@ pub enum ErrorType {
     #[error("Too long nsn")]
     TooLongNsn, // TOO_LONG in the java version.
 }
+
+static PHONE_NUMBER_UTIL: LazyLock<PhoneNumberUtil> = LazyLock::new(|| {
+    PhoneNumberUtil::new()
+});

src/phonenumberutil/phone_number_regexps_and_mappings.rs

@@ -2,7 +2,11 @@ use std::collections::{HashMap, HashSet};
 
 use regex::Regex;
 
-use crate::{phonenumberutil::{helper_constants::{self, CAPTURE_UP_TO_SECOND_NUMBER_START, DIGITS, MIN_LENGTH_FOR_NSN, PLUS_CHARS, PLUS_SIGN, RFC3966_VISUAL_SEPARATOR, STAR_SIGN, VALID_ALPHA, VALID_ALPHA_INCL_UPPERCASE, VALID_PUNCTUATION}, helper_functions::create_extn_pattern}, regexp_cache::RegexCache};
+use crate::{phonenumberutil::{helper_constants::{
+    CAPTURE_UP_TO_SECOND_NUMBER_START, DIGITS, MIN_LENGTH_FOR_NSN, PLUS_CHARS, 
+    PLUS_SIGN, RFC3966_VISUAL_SEPARATOR, STAR_SIGN, VALID_ALPHA, VALID_ALPHA_INCL_UPPERCASE, 
+    VALID_PUNCTUATION
+}, helper_functions::create_extn_pattern}, regexp_cache::RegexCache};
 
 pub(super) struct PhoneNumberRegExpsAndMappings {
     /// Regular expression of viable phone numbers. This is location independent.
@@ -150,6 +154,17 @@ pub(super) struct PhoneNumberRegExpsAndMappings {
     /// Regular expression of valid domainname for the phone-context parameter,
     /// following the syntax defined in RFC3966.
     pub rfc3966_domainname_pattern: Regex,
+
+    /// *Rust note*: It's for some reason calculated inside function in C++,
+    /// so, we move it here
+    /// 
+    /// A pattern that is used to determine if a numberFormat under
+    /// availableFormats is eligible to be used by the AYTF. It is eligible when
+    /// the format element under numberFormat contains groups of the dollar sign
+    /// followed by a single digit, separated by valid phone number punctuation.
+    /// This prevents invalid punctuation (such as the star sign in Israeli star
+    /// numbers) getting into the output of the AYTF. 
+    pub is_format_eligible_as_you_type_formatting_regex: Regex
 }
 
 impl PhoneNumberRegExpsAndMappings {
@@ -200,6 +215,8 @@ impl PhoneNumberRegExpsAndMappings {
         alpha_map.insert('X', '9');
         alpha_map.insert('Y', '9');
         alpha_map.insert('Z', '9');
+        // IMPORTANT: only uppercase letters like in Java version
+
         self.alpha_mappings = alpha_map;
 
         let mut combined_map = HashMap::with_capacity(100);
@@ -241,7 +258,7 @@ impl PhoneNumberRegExpsAndMappings {
         self.all_plus_number_grouping_symbols = all_plus_number_groupings;
     }
 
-    fn new() -> Self {
+    pub fn new() -> Self {
         let alphanum = fast_cat::concat_str!(VALID_ALPHA_INCL_UPPERCASE, DIGITS);
         let extn_patterns_for_parsing = create_extn_pattern(true);
         let valid_phone_number = format!(
@@ -303,6 +320,9 @@ impl PhoneNumberRegExpsAndMappings {
             rfc3966_domainname_pattern: Regex::new(
                 &format!("^({}\\.)*{}\\.?$", rfc3966_domainlabel, rfc3966_toplabel)
             ).unwrap(),
+            is_format_eligible_as_you_type_formatting_regex: Regex::new(
+                &format!("[{}]*\\$1[{}]*(\\$\\d[{}]*)*",VALID_PUNCTUATION, VALID_PUNCTUATION, VALID_PUNCTUATION)
+            ).unwrap(),
         };
         instance.initialize_regexp_mappings();
         instance

src/phonenumberutil/regex_and_mappings.rs

@@ -1,3 +0,0 @@
-pub struct PhoneNumberRegExpsAndMappings{
-    
-}

src/regex_based_matcher.rs

@@ -1,6 +1,7 @@
 use log::{error};
+use super::regex_util::{RegexFullMatch, RegexConsume};
 
-use crate::{interfaces, proto_gen::phonemetadata::PhoneNumberDesc, regexp_cache::{self, RegexCache}};
+use crate::{interfaces, proto_gen::phonemetadata::PhoneNumberDesc, regexp_cache::{ErrorInvalidRegex, RegexCache}};
 
 pub struct RegexBasedMatcher {
     cache: RegexCache,   
@@ -15,23 +16,15 @@ impl RegexBasedMatcher {
         &self, phone_number: &str, 
         number_pattern: &str,
         allow_prefix_match: bool
-    ) -> Result<bool, regexp_cache::ErrorInvalidRegex> {
+    ) -> Result<bool, ErrorInvalidRegex> {
         let regexp = self.cache.get_regex(number_pattern)?;
 
         // find first occurrence
-        if let Some(mat) = regexp.find(phone_number) {
-            // if first position is not matched none of scenarios are  possible
-            if mat.start() != 0 {
-                return Ok(false);
-            }
-            // full match
-            if mat.end() == phone_number.len() {
-                return Ok(true);
-            } else if allow_prefix_match {
-                return Ok(true);
-            }
+        if allow_prefix_match {
+            Ok(regexp.consume_start(phone_number).is_some())
+        } else {
+            Ok(regexp.full_match(phone_number))
         }
-        Ok(false)
     }
 }
 

src/regex_util.rs

@@ -0,0 +1,91 @@
+use std::borrow::Cow;
+
+use regex::Regex;
+
+pub trait RegexFullMatch {
+    /// Eq of C fullMatch
+    fn full_match(&self, s: &str) -> bool;
+}
+
+pub trait RegexConsume {
+    /// Eq of C Consume
+    fn consume_start<'a>(&self, s: &'a str) -> Option<Cow<'a, str>> {
+        self.consume_start_capturing(s, &mut [])
+    }
+
+    fn consume_start_capturing<'a, 'b>(&self, s: &'a str, groups: &mut [&'b str]) -> Option<Cow<'a, str>>
+        where 'a: 'b;
+    
+    fn find_and_consume<'a>(&self, s: &'a str) -> Option<Cow<'a, str>> {
+        self.find_and_consume_capturing(s, &mut [])
+    }
+    
+    fn find_and_consume_capturing<'a, 'b>(&self, s: &'a str, groups: &mut [&'b str]) -> Option<Cow<'a, str>>
+        where 'a: 'b;
+}
+
+trait RegexMatchStart {
+    // Eq of looking_at
+    fn match_start(&self, s: &str) -> bool;
+}
+
+impl RegexFullMatch for Regex {
+    fn full_match(&self, s: &str) -> bool {
+        let found = self.find(s);
+        if let Some(matched) = found {
+            return matched.start() == 0 && matched.end() == s.len();
+        }
+        false
+    }
+}
+
+impl RegexMatchStart for Regex {
+    fn match_start(&self, s: &str) -> bool {
+        let found = self.find(s);
+        if let Some(matched) = found {
+            return matched.start() == 0;
+        }
+        false
+    }
+}
+
+impl RegexConsume for Regex {
+    fn consume_start_capturing<'a, 'b>(&self, s: &'a str, groups: &mut [&'b str]) -> Option<Cow<'a, str>>
+        where 'a: 'b {
+        _consume(self, s, groups, true)
+    }
+
+    fn find_and_consume_capturing<'a, 'b>(&self, s: &'a str, groups: &mut [&'b str]) -> Option<Cow<'a, str>>
+        where 'a: 'b {
+        _consume(self, s, groups, false)
+    }
+}
+
+fn _consume<'a, 'b>(
+    r: &Regex, input: &'a str, 
+    groups: &mut [&'b str], anchor_at_start: bool
+) -> Option<Cow<'a, str>>
+    where 'a: 'b {
+    let captures = r.captures(input)?;
+    let full_capture = captures.get(0)?;
+    if anchor_at_start && full_capture.start() != 0 {
+        return None
+    }
+    // Check if expected groups count is leq
+    // captures.len includes full group (0), so take captures.len() - 1
+    if groups.len() > captures.len() - 1 {
+        return None;
+    }
+
+    // If less matches than expected - fail.
+    for i in 1..=groups.len() {
+        // Groups are counted from 1 rather than 0.
+        if let Some(capture) = captures.get(i) {
+            groups[i-1] = capture.as_str();
+        } else {
+            // should never happen
+            return None
+        }
+    }
+    Some(Cow::Borrowed(&input[full_capture.end()..]))
+}

src/regexp_cache.rs

@@ -3,7 +3,7 @@ use std::sync::Arc;
 use dashmap::DashMap;
 use thiserror::Error;
 
-#[derive(Debug, Error)]
+#[derive(Debug, PartialEq, Error)]
 #[error("An error occurred while trying to create regex: {0}")]
 pub struct ErrorInvalidRegex(#[from] regex::Error);