Function sample_file

src/utils/estimation.rs:81–135

Sample a file to get an estimate of matching lines

(
    input_file_path: &str, 
    sample_size: u64,
    prefix: &str,
    suffix: &str,
    infix: Option<&str>
)

/// Sample a file to get an estimate of matching lines
async fn sample_file(
    input_file_path: &str, 
    sample_size: u64,
    prefix: &str,
    suffix: &str,
    infix: Option<&str>
) -> Result<(u64, u64, u64), Error> {
    let sample_file = File::open(input_file_path).await?;
    let sample_reader = BufReader::new(sample_file);
    let mut sample_lines = sample_reader.lines();
    
    let mut sample_count = 0;
    let mut sample_matched = 0;
    let mut sample_bytes_read = 0;
    let check_suffix = !suffix.is_empty();
    let check_prefix = !prefix.is_empty();
    let check_infix = infix.is_some();
    
    // Read the sample
    while let Some(line) = sample_lines.next_line().await? {
        if sample_count >= sample_size {
            break;
        }
        
        sample_bytes_read += line.len() as u64 + 1; // +1 for newline
        
        if !line.trim().is_empty() {
            sample_count += 1;
            
            let phone = line.trim();
            let suffix_match = !check_suffix || phone.ends_with(suffix);
            let prefix_match = !check_prefix || phone.starts_with(prefix);
            
            // Check infix if needed
            let infix_match = if check_infix {
                let infix_val = infix.unwrap();
                if phone.len() >= 6 {
                    // Extract the infix (6th and 5th characters from the end)
                    let potential_infix = &phone[phone.len() - 6..phone.len() - 4];
                    potential_infix == infix_val
                } else {
                    false // Phone number too short for infix
                }
            } else {
                true // No infix check needed
            };
            
            if suffix_match && prefix_match && infix_match {
                sample_matched += 1;
            }
        }
    }
    
    Ok((sample_matched, sample_count, sample_bytes_read))
}

/// Estimate workload for full scan (number generation) mode
fn estimate_fullscan_workload(