feat: Population data (#31)

* doc: update

* feat: population data

README.md

@@ -10,12 +10,13 @@ use keshvar::{CurrencyCode, WeekDay, Region, SubdivisionType, find_by_name};
 let country_name_in_any_language = "estados unidos"; // The US in spanish!
 let country = find_by_name(country_name_in_any_language).unwrap();
 assert_eq!(country.iso_long_name(), "The United States of America");
-assert_eq!(country.nationality(), Some("American"));
+assert_eq!(country.maybe_nationality(), Some("American"));
 assert_eq!(country.currency_code(), CurrencyCode::USD);
 assert_eq!(country.start_of_week(), WeekDay::Sunday);
 assert_eq!(country.emoji(), "🇺🇸");
 assert_eq!(country.country_code(), 1);
-assert_eq!(country.region(), Some(Region::Americas));
+assert!(country.maybe_population().unwrap() > 330_000_000);
+assert_eq!(country.maybe_region(), Some(Region::Americas));
 assert!(country.unofficial_name_list().contains(&"United States"));
 assert!(country.spoken_language_list().contains(&"en"));
 assert!(country.distance_unit().is_mi()); // KM/MI
@@ -288,11 +289,11 @@ use keshvar::{
 };
 
 let country = find_by_iso_short_name("united states of america").unwrap();
-assert_eq!(Some("American"), country.nationality());
+assert_eq!(Some("American"), country.maybe_nationality());
 
 let country = find_by_iso_long_name("ukraine").unwrap();
 assert_eq!(Alpha2::UA, country.alpha2());
-assert_eq!(Some(SubRegion::EasternEurope), country.subregion());
+assert_eq!(Some(SubRegion::EasternEurope), country.maybe_subregion());
 
 let country = find_by_code(971).unwrap(); // The United Arab Emirates (Asia)
 assert_eq!(Alpha3::ARE, country.alpha3());
@@ -306,4 +307,4 @@ keshvar (/keʃvar/ or کِشوَر) simply means `country` in [persian/farsi lan
 See [**CONTRIBUTING.md** file](https://github.com/pouriya/keshvar/blob/master/CONTRIBUTING.md).
 
 # License
-`keshvar` source-code generator and the generated source are distributed under BSD-3-Clause license (See `LICENSE` file) but the data that is used to feed the generator is distributed under MIT License (See [`countries` license file](https://github.com/countries/countries/blob/master/LICENSE)).  
+`keshvar` source-code generator and the generated source are distributed under BSD-3-Clause license (See `LICENSE` file) but the data that is used to feed the generator is distributed under MIT License (See [`countries` license file](https://github.com/countries/countries/blob/master/LICENSE)). Additionally, The population data which is from [World Bank dataset](http://data.worldbank.org/indicator/SP.POP.TOTL) is licensed by its maintainers under the [Public Domain Dedication and License (PDDL)](http://opendatacommons.org/licenses/pddl/1.0/).  

keshvar-code-generator/Cargo.toml

@@ -8,3 +8,4 @@ edition = "2021"
 serde = { version = "1.0", features = ["derive"]}
 serde_yaml = "0.9"
 anyhow = "1.0"
+csv = "1.3.0"

keshvar-code-generator/src/countries.rs

@@ -230,6 +230,15 @@ pub fn generate_country(destination_file: &PathBuf, info: &CountryInfo) -> Resul
             "DistanceUnit",
             format!("DistanceUnit::{:?}", info.distance_unit),
         ),
+        (
+            "const POPULATION",
+            "Option<u64>",
+            if info.population != 0 {
+                format!("Some({})", info.population)
+            } else {
+                "None".to_string()
+            },
+        ),
     ] {
         file.write_all(format!("    pub {}: {} = {};\n", name, _type, value).as_bytes())?;
     }
@@ -423,25 +432,25 @@ pub fn new() -> Country {{
         continent: Continent::{},
         country_code: {},
         currency_code: CurrencyCode::{},
-        gec: {},
+        maybe_gec: {},
         #[cfg(feature = "geo")]
         geo: geo::new(),
         international_prefix: {:?},
-        ioc: {},
+        maybe_ioc: {},
         iso_long_name: {:?},
         iso_short_name: {:?},
         official_language_list: {:?}.to_vec(),
         spoken_language_list: {:?}.to_vec(),
         national_destination_code_length_list: {:?}.to_vec(),
         national_number_length_list: {:?}.to_vec(),
         national_prefix: {:?},
-        nationality: {},
+        maybe_nationality: {},
         number: {:?},
         postal_code: {:?},
         postal_code_format: {},
-        region: {},
+        maybe_region: {},
         start_of_week: WeekDay::{},
-        subregion: {},
+        maybe_subregion: {},
         un_locode: {:?},
         unofficial_name_list: {:?}.to_vec(),
         world_region: WorldRegion::{},
@@ -455,8 +464,9 @@ pub fn new() -> Country {{
         g20_member: {},
         eu_member: {},
         eea_member: {},
-        vat_rates: {},
+        maybe_vat_rates: {},
         distance_unit: {},
+        maybe_population: {},
     }}
 }}
             "#,
@@ -520,6 +530,11 @@ pub fn new() -> Country {{
                 "None".to_string()
             },
             format!("DistanceUnit::{:?}", info.distance_unit),
+            if info.population != 0 {
+                format!("Some({})", info.population)
+            } else {
+                "None".to_string()
+            },
         )
         .as_bytes(),
     )?;

keshvar-code-generator/src/main.rs

@@ -6,6 +6,7 @@ mod countries;
 mod currency_code;
 mod gec;
 mod ioc;
+mod population;
 mod region;
 mod search;
 mod structs;
@@ -226,6 +227,7 @@ fn code_gen_countries(data_directory: PathBuf, output_directory: PathBuf) -> Res
             }
         }
     }
+    population::read_from_file("population.csv", &mut countries_info_list)?;
     log!("Region features: {:?}", region_features);
     log!("Continent features: {:?}", continent_features);
     log!("SubRegion features: {:?}", subregion_features);

keshvar-code-generator/src/population.rs

@@ -0,0 +1,39 @@
+use crate::structs::CountryInfo;
+use anyhow::{Context, Result};
+use std::fs;
+use std::io::BufReader;
+use std::path::Path;
+
+pub fn read_from_file<F: AsRef<Path>>(
+    filename: F,
+    countries_info_list: &mut Vec<(String, CountryInfo)>,
+) -> Result<()> {
+    let file = fs::File::open(&filename).context(format!(
+        "Could not open population file {:?}",
+        filename.as_ref()
+    ))?;
+    let file_reader = BufReader::new(file);
+    let mut csv_reader = csv::ReaderBuilder::new()
+        .flexible(true)
+        .from_reader(file_reader);
+    for record_result in csv_reader.records() {
+        let record = record_result?;
+        if record.len() < 3 {
+            continue;
+        }
+        let alpha3 = &record[1];
+        if let Some((_, info)) = countries_info_list
+            .iter_mut()
+            .find(|(_, info)| info.alpha3 == alpha3)
+        {
+            info.population = if record.iter().last().unwrap().trim().is_empty() {
+                &record[record.len() - 2]
+            } else {
+                record.iter().last().unwrap()
+            }
+            .parse()
+            .unwrap()
+        }
+    }
+    Ok(())
+}

keshvar-code-generator/src/structs.rs

@@ -66,6 +66,8 @@ pub struct CountryInfo {
     pub eu_member: Option<bool>,
     pub vat_rates: Option<VatRates>,
     pub distance_unit: DistanceUnit,
+    #[serde(default)]
+    pub population: u64,
 }
 
 #[derive(Clone, Debug, Copy, Serialize, Deserialize)]

src/countries/ad.rs

@@ -44,6 +44,7 @@ pub mod consts {
     pub const EU_MEMBER: bool = false;
     pub const EEA_MEMBER: bool = false;
     pub const DISTANCE_UNIT: DistanceUnit = DistanceUnit::Km;
+    pub const POPULATION: Option<u64> = Some(79824);
     #[cfg(feature = "emojis")]
     pub const EMOJI: &str = "🇦🇩";
     #[cfg(feature = "translations")]
@@ -423,25 +424,25 @@ pub fn new() -> Country {
         continent: Continent::Europe,
         country_code: 376,
         currency_code: CurrencyCode::EUR,
-        gec: Some(GEC::AN),
+        maybe_gec: Some(GEC::AN),
         #[cfg(feature = "geo")]
         geo: geo::new(),
         international_prefix: "00",
-        ioc: Some(IOC::AND),
+        maybe_ioc: Some(IOC::AND),
         iso_long_name: "The Principality of Andorra",
         iso_short_name: "Andorra",
         official_language_list: ["ca"].to_vec(),
         spoken_language_list: ["ca"].to_vec(),
         national_destination_code_length_list: [2].to_vec(),
         national_number_length_list: [6, 7, 8, 9].to_vec(),
         national_prefix: "None",
-        nationality: Some("Andorran"),
+        maybe_nationality: Some("Andorran"),
         number: "020",
         postal_code: true,
         postal_code_format: Some("AD[1-7]0\\d"),
-        region: Some(Region::Europe),
+        maybe_region: Some(Region::Europe),
         start_of_week: WeekDay::Monday,
-        subregion: Some(SubRegion::SouthernEurope),
+        maybe_subregion: Some(SubRegion::SouthernEurope),
         un_locode: "AD",
         unofficial_name_list: ["Andorre", "Andorra", "アンドラ"].to_vec(),
         world_region: WorldRegion::EMEA,
@@ -590,7 +591,8 @@ pub fn new() -> Country {
         g20_member: false,
         eu_member: false,
         eea_member: false,
-        vat_rates: None,
+        maybe_vat_rates: None,
         distance_unit: DistanceUnit::Km,
+        maybe_population: Some(79824),
     }
 }

src/countries/ae.rs

@@ -53,6 +53,7 @@ pub mod consts {
     pub const EU_MEMBER: bool = false;
     pub const EEA_MEMBER: bool = false;
     pub const DISTANCE_UNIT: DistanceUnit = DistanceUnit::Km;
+    pub const POPULATION: Option<u64> = Some(9441129);
     #[cfg(feature = "emojis")]
     pub const EMOJI: &str = "🇦🇪";
     #[cfg(feature = "translations")]
@@ -895,25 +896,25 @@ pub fn new() -> Country {
         continent: Continent::Asia,
         country_code: 971,
         currency_code: CurrencyCode::AED,
-        gec: Some(GEC::AE),
+        maybe_gec: Some(GEC::AE),
         #[cfg(feature = "geo")]
         geo: geo::new(),
         international_prefix: "00",
-        ioc: Some(IOC::UAE),
+        maybe_ioc: Some(IOC::UAE),
         iso_long_name: "The United Arab Emirates",
         iso_short_name: "United Arab Emirates",
         official_language_list: ["ar"].to_vec(),
         spoken_language_list: ["ar"].to_vec(),
         national_destination_code_length_list: [2].to_vec(),
         national_number_length_list: [7, 8, 9].to_vec(),
         national_prefix: "0",
-        nationality: Some("Emirian"),
+        maybe_nationality: Some("Emirian"),
         number: "784",
         postal_code: false,
         postal_code_format: None,
-        region: Some(Region::Asia),
+        maybe_region: Some(Region::Asia),
         start_of_week: WeekDay::Monday,
-        subregion: Some(SubRegion::WesternAsia),
+        maybe_subregion: Some(SubRegion::WesternAsia),
         un_locode: "AE",
         unofficial_name_list: ["United Arab Emirates", "الإمارات العربية المتحدة", "Vereinigte Arabische Emirate", "Émirats Arabes Unis", "Emiratos Árabes Unidos", "アラブ首長国連邦", "Verenigde Arabische Emiraten"].to_vec(),
         world_region: WorldRegion::EMEA,
@@ -927,7 +928,8 @@ pub fn new() -> Country {
         g20_member: false,
         eu_member: false,
         eea_member: false,
-        vat_rates: Some(VatRates{standard: 5.0, reduced: Vec::from([]), super_reduced: None, parking: None}),
+        maybe_vat_rates: Some(VatRates{standard: 5.0, reduced: Vec::from([]), super_reduced: None, parking: None}),
         distance_unit: DistanceUnit::Km,
+        maybe_population: Some(9441129),
     }
 }

src/countries/af.rs

@@ -44,6 +44,7 @@ pub mod consts {
     pub const EU_MEMBER: bool = false;
     pub const EEA_MEMBER: bool = false;
     pub const DISTANCE_UNIT: DistanceUnit = DistanceUnit::Km;
+    pub const POPULATION: Option<u64> = Some(41128771);
     #[cfg(feature = "emojis")]
     pub const EMOJI: &str = "🇦🇫";
     #[cfg(feature = "translations")]
@@ -777,25 +778,25 @@ pub fn new() -> Country {
         continent: Continent::Asia,
         country_code: 93,
         currency_code: CurrencyCode::AFN,
-        gec: Some(GEC::AF),
+        maybe_gec: Some(GEC::AF),
         #[cfg(feature = "geo")]
         geo: geo::new(),
         international_prefix: "00",
-        ioc: Some(IOC::AFG),
+        maybe_ioc: Some(IOC::AFG),
         iso_long_name: "The Islamic Republic of Afghanistan",
         iso_short_name: "Afghanistan",
         official_language_list: ["ps", "tk", "uz"].to_vec(),
         spoken_language_list: ["ps", "tk", "uz"].to_vec(),
         national_destination_code_length_list: [2].to_vec(),
         national_number_length_list: [8, 9].to_vec(),
         national_prefix: "0",
-        nationality: Some("Afghan"),
+        maybe_nationality: Some("Afghan"),
         number: "004",
         postal_code: true,
         postal_code_format: Some("\\d{4}"),
-        region: Some(Region::Asia),
+        maybe_region: Some(Region::Asia),
         start_of_week: WeekDay::Monday,
-        subregion: Some(SubRegion::SouthernAsia),
+        maybe_subregion: Some(SubRegion::SouthernAsia),
         un_locode: "AF",
         unofficial_name_list: ["Afghanistan", "Afganistán", "アフガニスタン"].to_vec(),
         world_region: WorldRegion::APAC,
@@ -947,7 +948,8 @@ pub fn new() -> Country {
         g20_member: false,
         eu_member: false,
         eea_member: false,
-        vat_rates: None,
+        maybe_vat_rates: None,
         distance_unit: DistanceUnit::Km,
+        maybe_population: Some(41128771),
     }
 }

src/countries/ag.rs

@@ -51,6 +51,7 @@ pub mod consts {
     pub const EU_MEMBER: bool = false;
     pub const EEA_MEMBER: bool = false;
     pub const DISTANCE_UNIT: DistanceUnit = DistanceUnit::Km;
+    pub const POPULATION: Option<u64> = Some(93763);
     #[cfg(feature = "emojis")]
     pub const EMOJI: &str = "🇦🇬";
     #[cfg(feature = "translations")]
@@ -388,25 +389,25 @@ pub fn new() -> Country {
         continent: Continent::NorthAmerica,
         country_code: 1,
         currency_code: CurrencyCode::XCD,
-        gec: Some(GEC::AC),
+        maybe_gec: Some(GEC::AC),
         #[cfg(feature = "geo")]
         geo: geo::new(),
         international_prefix: "011",
-        ioc: Some(IOC::ANT),
+        maybe_ioc: Some(IOC::ANT),
         iso_long_name: "Antigua and Barbuda",
         iso_short_name: "Antigua and Barbuda",
         official_language_list: ["en"].to_vec(),
         spoken_language_list: ["en"].to_vec(),
         national_destination_code_length_list: [3].to_vec(),
         national_number_length_list: [10].to_vec(),
         national_prefix: "1",
-        nationality: Some("Antiguan, Barbudan"),
+        maybe_nationality: Some("Antiguan, Barbudan"),
         number: "028",
         postal_code: false,
         postal_code_format: None,
-        region: Some(Region::Americas),
+        maybe_region: Some(Region::Americas),
         start_of_week: WeekDay::Monday,
-        subregion: Some(SubRegion::Caribbean),
+        maybe_subregion: Some(SubRegion::Caribbean),
         un_locode: "AG",
         unofficial_name_list: ["Antigua and Barbuda", "Antigua und Barbuda", "Antigua et Barbuda", "Antigua y Barbuda", "アンティグア・バーブーダ", "Antigua en Barbuda"].to_vec(),
         world_region: WorldRegion::AMER,
@@ -420,7 +421,8 @@ pub fn new() -> Country {
         g20_member: false,
         eu_member: false,
         eea_member: false,
-        vat_rates: None,
+        maybe_vat_rates: None,
         distance_unit: DistanceUnit::Km,
+        maybe_population: Some(93763),
     }
 }