pandora_lib_promethion/src/io/dict.rs

use std::{fs::File, io::BufReader};

use anyhow::Context;
use log::debug;

use crate::io::tsv::TsvLine;

/// Read a sequence dictionary (`.dict`) file and return `(name, length)` pairs.
///
/// Only `@SQ` lines are processed; other lines are skipped. Each `@SQ` line
/// must contain `SN:<name>` and `LN:<length>` tab-separated fields.
///
/// # Errors
///
/// Returns an error if the file cannot be opened, an `@SQ` line is missing
/// `SN:` or `LN:`, or `LN:` cannot be parsed as `u32`.
pub fn read_dict(path: &str) -> anyhow::Result<Vec<(String, u32)>> {
    debug!("Parsing {path}.");

    let mut reader =
        BufReader::new(File::open(path).with_context(|| format!("cannot open dict: {path}"))?);
    let mut line = TsvLine::new();
    let mut line_no = 0usize;
    let mut res = Vec::new();

    while line
        .read(&mut reader)
        .with_context(|| format!("failed reading dict line after {path}:{line_no}"))?
    {
        line_no += 1;
        let fields = line.split_fields();
        if fields.first().copied() != Some("@SQ") {
            continue;
        }

        let sn = fields
            .iter()
            .find_map(|f| f.strip_prefix("SN:"))
            .with_context(|| format!("Missing SN: in @SQ line at {path}:{line_no}"))?
            .to_string();

        let ln: u32 = fields
            .iter()
            .find_map(|f| f.strip_prefix("LN:"))
            .with_context(|| format!("Missing LN: in @SQ line at {path}:{line_no}"))?
            .parse()
            .with_context(|| format!("Invalid LN: value at {path}:{line_no}"))?;

        res.push((sn, ln));
    }

    Ok(res)
}