Files
clean_file/src/parser/mod.rs
T
2026-07-13 11:57:24 -04:00

264 lines
8.2 KiB
Rust

use std::io::BufRead;
pub const DEFAULT_OUTPUT_FILE: &str = "numbers.json";
const LETTERS: &str = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
#[derive(Clone, Debug, Default)]
pub struct NumberParser {
pub filepath: String,
}
impl NumberParser {
pub async fn file_dump(
&self,
) -> Result<Vec<schedtxt_models::contact::Contact>, std::io::Error> {
println!("Dumping files");
let mut objs: Vec<schedtxt_models::contact::Contact> = Vec::new();
match self.read_lines(&self.filepath).await {
Ok(lines) => {
for line in lines {
let contains_letters = line.chars().any(|c| LETTERS.contains(c));
if contains_letters {
continue;
} else {
let my_obj = schedtxt_models::contact::Contact {
phone_number: line.clone(),
..Default::default()
};
if !my_obj.phone_number.is_empty() {
objs.push(my_obj);
}
}
}
let updated_contacts = self.update_contacts(&objs).await;
let removed_duplicates = self.remove_dups(&updated_contacts).await;
Ok(removed_duplicates)
}
Err(err) => Err(err),
}
}
pub async fn save_file(&self, numbers: Vec<schedtxt_models::contact::Contact>, filename: &str) {
match serde_json::to_string(&numbers) {
Ok(json_string) => match std::fs::write(filename, json_string) {
Ok(_) => {
println!("Saved to: {filename:?}");
}
Err(err) => {
eprintln!("Error: {err:?}");
}
},
Err(err) => {
eprintln!("Error: {err:?}");
}
};
}
pub fn print_phone_numbers(&self, contacts: &Vec<schedtxt_models::contact::Contact>) {
println!("Printing phone numbers");
for contact in contacts {
println!("Phone Number: {:?}", contact.phone_number);
}
println!("Total phone numbers: {:?}", contacts.len());
}
async fn read_lines(&self, filename: &String) -> Result<Vec<String>, std::io::Error> {
match std::fs::File::open(filename) {
Ok(file) => {
let reader = std::io::BufReader::new(file);
reader.lines().collect()
}
Err(err) => Err(err),
}
}
async fn remove_dups(
&self,
contacts: &Vec<schedtxt_models::contact::Contact>,
) -> Vec<schedtxt_models::contact::Contact> {
let mut unique_contacts: Vec<schedtxt_models::contact::Contact> = Vec::new();
let mut unique_items = std::collections::HashSet::new();
for val in contacts {
unique_items.insert(val.phone_number.clone());
}
for unique_item in unique_items {
let contact = schedtxt_models::contact::Contact {
phone_number: unique_item,
..Default::default()
};
unique_contacts.push(contact);
}
unique_contacts
}
async fn update_contacts(
&self,
contacts: &Vec<schedtxt_models::contact::Contact>,
) -> Vec<schedtxt_models::contact::Contact> {
let mut updated: Vec<schedtxt_models::contact::Contact> = Vec::new();
for ct in contacts {
let phone_number = ct.phone_number.clone();
if phone_number.len() < 2 {
println!("Invalid number: {phone_number:?}");
continue;
}
let pars = self.remove_some_data(&phone_number).await;
match self.add_prefix(&pars) {
Ok(updated_parsed) => {
let contact = schedtxt_models::contact::Contact {
phone_number: updated_parsed,
..Default::default()
};
updated.push(contact);
}
Err(err) => {
eprintln!("Error: {err:?}");
std::process::exit(-2);
}
};
}
updated
}
fn add_prefix(&self, raw: &String) -> Result<String, std::io::Error> {
if raw.is_empty() {
Err(std::io::Error::other("Empty"))
} else {
let mut parsed: String = String::new();
let first_char = raw.chars().nth(0).unwrap();
if first_char == '1' {
parsed = format!("+{raw}");
} else if first_char != '+' {
parsed = format!("+1{raw}");
}
Ok(parsed)
}
}
async fn remove_some_data(&self, unparsed: &str) -> String {
let mut parsed: String = String::new();
for ch in unparsed.chars() {
if ch == ' ' {
continue;
} else {
if ch.is_numeric() {
parsed.push(ch);
}
}
}
parsed
}
}
#[cfg(test)]
mod tests {
use schedtxt_models::contact::Contact;
const TEST_NUMBERS_0: &str = "tests/numbers-0.txt";
#[test]
fn test_file_dump() {
let file_exists = {
let path = std::path::Path::new(TEST_NUMBERS_0);
path.exists()
};
assert_eq!(true, file_exists, "File does not exists {TEST_NUMBERS_0:?}");
let filename = String::from(TEST_NUMBERS_0);
let prsr = super::NumberParser { filepath: filename };
let original: Vec<Contact> = vec![
Contact {
phone_number: String::from("+13055550112"),
..Default::default()
},
Contact {
phone_number: String::from("+13175550123"),
..Default::default()
},
Contact {
phone_number: String::from("+14155550135"),
..Default::default()
},
Contact {
phone_number: String::from("+16085550190"),
..Default::default()
},
Contact {
phone_number: String::from("+17135550167"),
..Default::default()
},
Contact {
phone_number: String::from("+16025550177"),
..Default::default()
},
Contact {
phone_number: String::from("+13605550158"),
..Default::default()
},
Contact {
phone_number: String::from("+12175550194"),
..Default::default()
},
];
match async_std::task::block_on(prsr.file_dump()) {
Ok(contacts) => {
assert!((contacts.len() > 0), "No contacts has been parsed");
let phone_numbers_1: std::collections::HashSet<_> =
contacts.iter().map(|p| p.phone_number.clone()).collect();
let phone_numbers_2: std::collections::HashSet<_> =
original.iter().map(|p| p.phone_number.clone()).collect();
assert_eq!(phone_numbers_1, phone_numbers_2, "No match");
}
Err(err) => {
assert!(false, "Error: {err:?}");
}
};
}
#[test]
fn test_save_file() {
let prsr = super::NumberParser {
filepath: String::from(TEST_NUMBERS_0),
};
let results = async_std::task::block_on(prsr.file_dump());
assert!(results.is_ok(), "File dump not successful");
let test_output_file: &str = "test-numbers.json";
let contacts = results.unwrap();
async_std::task::block_on(prsr.save_file(contacts, test_output_file));
let file_exists = {
let path = std::path::Path::new(test_output_file);
path.exists()
};
assert!(file_exists, "Contacts were not saved");
match std::fs::remove_file(test_output_file) {
Ok(_) => {
println!("File removed");
}
Err(err) => {
assert!(false, "Error: {err:?}");
}
}
}
}