pandora_lib_bindings/src/progs/cramino.rs

use std::{
    collections::HashMap,
    env,
    fs::{self, File},
    io::{self, BufRead},
    path::PathBuf,
};

use anyhow::anyhow;
use log::info;
use uuid::Uuid;

use crate::utils::{exec_bin_to_file, RunBin};

#[derive(Debug)]
pub struct Cramino {
    pub prog_path: String,
    pub args: Vec<String>,
    pub bam_file: PathBuf,
    pub result_path: PathBuf,
    pub results: Option<CraminoRes>,
    pub remove_result_file: bool,
}

impl Default for Cramino {
    fn default() -> Self {
        let mut result_path = env::temp_dir();
        result_path.push(format!("{}_cramino.txt", Uuid::new_v4()));

        Self {
            prog_path: "cramino".to_string(),
            args: vec![
                "--hist".to_string(),
                "--checksum".to_string(),
                "--karyotype".to_string(),
            ],
            result_path,
            results: None,
            remove_result_file: false,
            bam_file: PathBuf::new(),
        }
    }
}

impl Cramino {
    pub fn with_result_path(mut self, result_path: &str) -> Self {
        self.result_path = PathBuf::from(result_path);
        self
    }

    pub fn with_threads(mut self, n_threads: usize) -> Self {
        self.args.push("-t".to_string());
        self.args.push(n_threads.to_string());
        self
    }

    pub fn with_bam(mut self, bam_file: &str) -> anyhow::Result<Self> {
        let bam_file = PathBuf::from(bam_file);
        if !bam_file.exists() {
            Err(anyhow!("BAM file doesn't exists."))
        } else {
            self.bam_file = bam_file;
            Ok(self)
        }
    }
}

impl RunBin for Cramino {
    fn exec(&self) -> anyhow::Result<()> {
        if self.result_path.exists() {
            info!("Result file already exists.");
            Ok(())
        } else {
            let mut args = self.args.clone();
            args.push(self.bam_file.to_string_lossy().to_string());
            exec_bin_to_file(&self.prog_path, &args, &self.result_path)
        }
    }

    fn parse_results(&mut self) -> anyhow::Result<()> {
        let file = File::open(&self.result_path)?;
        let reader = io::BufReader::new(file);
        let mut lines = reader.lines();

        let mut cramino_res = CraminoRes {
            file_name: String::new(),
            number_of_reads: 0,
            yield_gb: 0.0,
            mean_coverage: 0.0,
            n50: 0,
            median_length: 0.0,
            mean_length: 0.0,
            median_identity: 0.0,
            mean_identity: 0.0,
            path: String::new(),
            creation_time: String::new(),
            checksum: String::new(),
            normalized_read_count_per_chromosome: HashMap::new(),
        };

        while let Some(Ok(line)) = lines.next() {
            if line.starts_with('#') {
                continue; // Skip the normalized read count header line
            }

            let parts: Vec<&str> = line.split_whitespace().collect();
            if parts.len() < 2 {
                continue; // Skip any malformed lines
            }

            if !parts[0].starts_with("chr") {
                match parts[0] {
                    "File" => cramino_res.file_name = parts[2].to_string(),
                    "Number" => cramino_res.number_of_reads = parts[3].parse().unwrap_or(0),
                    "Yield" => cramino_res.yield_gb = parts[2].parse().unwrap_or(0.0),
                    "N50" => cramino_res.n50 = parts[1].parse().unwrap_or(0),
                    "Median" => match parts[1] {
                        "length" => cramino_res.median_length = parts[2].parse().unwrap_or(0.0),
                        "identity" => cramino_res.median_identity = parts[2].parse().unwrap_or(0.0),
                        _ => {}
                    },
                    "Mean" => match parts[1] {
                        "coverage" => cramino_res.mean_coverage = parts[2].parse().unwrap_or(0.0),
                        "length" => cramino_res.mean_length = parts[2].parse().unwrap_or(0.0),
                        "identity" => cramino_res.mean_identity = parts[2].parse().unwrap_or(0.0),
                        _ => {}
                    },
                    "Path" => cramino_res.path = parts[1].to_string(),
                    "Creation" => cramino_res.creation_time = parts[2..].join(" "),
                    "Checksum" => cramino_res.checksum = parts[1].to_string(),
                    _ => {}
                }
            } else {
                if parts.len() == 2 {
                    let chromosome = parts[0].to_string();
                    let count = parts[1].parse().unwrap_or(0.0);
                    cramino_res
                        .normalized_read_count_per_chromosome
                        .insert(chromosome, count);
                }
            }
        }

        self.results = Some(cramino_res);
        Ok(())
    }

    fn clean(&self) -> anyhow::Result<()> {
        if self.remove_result_file {
            fs::remove_file(&self.result_path)?;
        }
        Ok(())
    }
}

#[derive(Debug)]
pub struct CraminoRes {
    pub file_name: String,
    pub number_of_reads: u64,
    pub yield_gb: f64,
    pub mean_coverage: f64,
    pub n50: u32,
    pub median_length: f64,
    pub mean_length: f64,
    pub median_identity: f64,
    pub mean_identity: f64,
    pub path: String,
    pub creation_time: String,
    pub checksum: String,
    pub normalized_read_count_per_chromosome: HashMap<String, f64>,
}

impl CraminoRes {
    pub fn is_woman(&self) -> anyhow::Result<bool> {
        if let Some(v) = self.normalized_read_count_per_chromosome.get("chrY") {
            Ok(*v < 0.1)
        } else {
            Err(anyhow::anyhow!("Can't get chrY normalized read count."))
        }
    }
}