Refactor build script

.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"

build/rust_build.rs

@@ -1,8 +1,9 @@
+
+/** TODO: uncomment, move to macros and refactor
 /**
  * This file represents content of https://github.com/google/libphonenumber/tree/master/tools/cpp
  */
 
-use std::{collections::BTreeMap, fs::File, io::{BufRead, BufReader}, num::ParseIntError, path::Path};
 
 use thiserror::Error;
 
@@ -68,9 +69,9 @@ fn parse_prefixes(path: &str, prefixes: &mut BTreeMap<i32, String>) -> Result<()
     Ok(())
 }
 
+*/
 
-
+fn main() {
-fn main() -> Result<(), BuildError> {
     protobuf_codegen::Codegen::new()
         .pure()
         .includes(["resources"])
@@ -78,5 +79,4 @@ fn main() -> Result<(), BuildError> {
         .input("resources/phonenumber.proto")
         .cargo_out_dir("proto_gen")
         .run_from_script();
-    Ok(())
 }

src/generated/metadata/mod.rs

@@ -0,0 +1,6 @@
+mod metadata;
+mod test_metadata;
+
+pub use metadata::METADATA;
+pub use test_metadata::TEST_METADATA;
+

src/generated/mod.rs

@@ -0,0 +1,2 @@
+pub mod proto;
+pub mod metadata;

src/interfaces/mod.rs

@@ -1,5 +1,5 @@
 
-use crate::proto_gen::phonemetadata::PhoneNumberDesc;
+use crate::phonemetadata::PhoneNumberDesc;
 /// Internal phonenumber matching API used to isolate the underlying
 /// implementation of the matcher and allow different implementations to be
 /// swapped in easily.

src/lib.rs

@@ -1,7 +1,7 @@
 mod shortnumberinfo;
 mod interfaces;
 /// This module is automatically generated from /resources/*.proto
-mod proto_gen;
+mod generated;
 mod phonenumberutil;
 mod regexp_cache;
 mod regex_based_matcher;
@@ -30,6 +30,6 @@ pub use phonenumberutil::{
     errors,
     enums,
 };
-pub use proto_gen::phonemetadata;
+pub use generated::proto::phonemetadata;
-pub use proto_gen::phonenumber;
+pub use generated::proto::phonenumber;
 mod tests;

src/phonenumberutil/errors.rs

@@ -7,10 +7,10 @@ use crate::regexp_cache::ErrorInvalidRegex;
 #[derive(Debug, PartialEq, Error)]
 pub enum InternalLogicError {
     #[error("{0}")]
-    InvalidRegexError(#[from] ErrorInvalidRegex),
+    InvalidRegex(#[from] ErrorInvalidRegex),
     
     #[error("{0}")]
-    InvalidMetadataForValidRegionError(#[from] InvalidMetadataForValidRegionError)
+    InvalidMetadataForValidRegion(#[from] InvalidMetadataForValidRegionError)
 }   
 
 #[derive(Debug, PartialEq, Error)]
@@ -18,9 +18,9 @@ pub enum ParseError {
     // Removed as OK variant
     // NoParsingError,
     #[error("Invalid country code")]
-    InvalidCountryCodeError, // INVALID_COUNTRY_CODE in the java version.
+    InvalidCountryCode, // INVALID_COUNTRY_CODE in the java version.
-    #[error("Not a number")]
+    #[error("Not a number: {0}")]
-    NotANumber,
+    NotANumber(#[from] NotANumberError),
     #[error("Too short after idd")]
     TooShortAfterIdd,
     #[error("Too short Nsn")]
@@ -28,11 +28,19 @@ pub enum ParseError {
     #[error("Too long nsn")]
     TooLongNsn, // TOO_LONG in the java version.
     #[error("{0}")]
-    InvalidRegexError(#[from] ErrorInvalidRegex),
+    InvalidRegex(#[from] ErrorInvalidRegex),
+}
+
+#[derive(Debug, PartialEq, Error)]
+pub enum NotANumberError {
+    #[error("Number not matched a valid number pattern")]
+    NotMatchedValidNumberPattern,
+    #[error("Invalid phone context")]
+    InvalidPhoneContext,
     #[error("{0}")]
-    ParseNumberAsIntError(#[from] ParseIntError),
+    FailedToParseNumberAsInt(#[from] ParseIntError),
     #[error("{0}")]
-    ExtractNumberError(#[from] ExtractNumberError),
+    FailedToExtractNumber(#[from] ExtractNumberError),
 }
 
 #[derive(Debug, PartialEq, Error)]
@@ -43,18 +51,24 @@ pub enum ExtractNumberError {
     NotANumber,
 }
 
+impl From<ExtractNumberError> for ParseError {
+    fn from(value: ExtractNumberError) -> Self {
+        NotANumberError::FailedToExtractNumber(value).into()
+    }
+}
+
 #[derive(Debug, PartialEq, Error)]
 pub enum GetExampleNumberError {
     #[error("Parse error: {0}")]
-    ParseError(#[from] ParseError),
+    FailedToParse(#[from] ParseError),
     #[error("{0}")]
-    InternalLogicError(#[from] InternalLogicError),
+    Internal(#[from] InternalLogicError),
     #[error("No example number")]
-    NoExampleNumberError,
+    NoExampleNumber,
     #[error("Could not get number")]
-    CouldNotGetNumberError,
+    CouldNotGetNumber,
     #[error("Invalid metadata")]
-    InvalidMetadataError
+    InvalidMetadata
 }
 
 

src/phonenumberutil/generated/mod.rs

@@ -1,3 +0,0 @@
-pub mod metadata;
-pub mod test_metadata;
-

src/phonenumberutil/helper_constants.rs

@@ -28,7 +28,7 @@ pub const VALID_PUNCTUATION: &'static str = "-x\
 pub const CAPTURE_UP_TO_SECOND_NUMBER_START: &'static str = r"(.*)[\\/] *x";
 
 
-pub const REGION_CODE_FOR_NON_GEO_ENTITY: &'static str = "0001";
+pub const REGION_CODE_FOR_NON_GEO_ENTITY: &'static str = "001";
 
 pub const PLUS_SIGN: &'static str = "+";
 pub const STAR_SIGN: &'static str = "*";
@@ -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/helper_constants/mod.rs

@@ -1,3 +0,0 @@
-mod helper_constants;
-
-pub(super) use helper_constants::{*};

src/phonenumberutil/helper_functions.rs

@@ -4,10 +4,10 @@ use protobuf::Message;
 use strum::IntoEnumIterator;
 
 use crate::{
-    interfaces::MatcherApi, phonenumberutil::generated::metadata::METADATA, proto_gen::{
+    interfaces::MatcherApi, generated::metadata::METADATA, 
         phonemetadata::{PhoneMetadata, PhoneMetadataCollection, PhoneNumberDesc},
         phonenumber::PhoneNumber,
-    }
+    
 };
 
 use super::{

src/phonenumberutil/helper_types.rs

@@ -1,6 +1,6 @@
 use std::borrow::Cow;
 
-use crate::proto_gen::phonenumber::phone_number::CountryCodeSource;
+use crate::phonenumber::phone_number::CountryCodeSource;
 
 #[derive(Debug)]
 pub struct PhoneNumberWithCountryCodeSource<'a> {

src/phonenumberutil/mod.rs

@@ -6,7 +6,6 @@ pub mod phonenumberutil;
 mod phone_number_regexps_and_mappings;
 pub(self) mod helper_types;
 pub(self) mod comparisons;
-pub(crate) mod generated;
 
 use std::sync::LazyLock;
 

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/regex_based_matcher.rs

@@ -1,7 +1,7 @@
 use log::{error};
 use super::regex_util::{RegexFullMatch, RegexConsume};
 
-use crate::{interfaces, proto_gen::phonemetadata::PhoneNumberDesc, regexp_cache::{ErrorInvalidRegex, RegexCache}};
+use crate::{interfaces, phonemetadata::PhoneNumberDesc, regexp_cache::{ErrorInvalidRegex, RegexCache}};
 
 pub struct RegexBasedMatcher {
     cache: RegexCache,   

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"
+    }
+}

tools/scripts/generate_metadata.sh

@@ -3,7 +3,7 @@
 filedir="./$(dirname "$0")"
 javadir="$filedir/../java"
 project_home="$filedir/../.."
-generated_dir="$project_home/src/phonenumberutil/generated"
+generated_dir="$project_home/src/generated/metadata"
 echo $generated_dir
 
 resources_dir="$project_home/resources"
@@ -28,6 +28,9 @@ generate "PhoneNumberMetadata.xml" "metadata" "metadata" "METADATA"
 generate "PhoneNumberMetadataForTesting.xml" "test_metadata" "metadata" "TEST_METADATA"
 
 echo "\
-pub mod metadata;
+mod metadata;
-pub mod test_metadata;
+mod test_metadata;
+
+pub use metadata::METADATA;
+pub use test_metadata::TEST_METADATA;
 " > "$generated_dir/mod.rs"