pandora_lib_promethion/src/helpers.rs

use anyhow::Context;
use bitcode::{Decode, Encode};
use glob::glob;
use log::{debug, warn};
use serde::{Deserialize, Serialize};
use std::{
    cmp::Ordering,
    collections::HashMap,
    fmt, fs,
    iter::Sum,
    ops::{Add, Div},
    path::{Path, PathBuf},
};

pub fn find_unique_file(dir_path: &str, suffix: &str) -> anyhow::Result<String> {
    let mut matching_files = Vec::new();

    for entry in
        fs::read_dir(dir_path).with_context(|| format!("Failed to read directory: {}", dir_path))?
    {
        let entry = entry.with_context(|| "Failed to read directory entry")?;
        let path = entry.path();

        if path.is_file()
            && path
                .file_name()
                .and_then(|name| name.to_str())
                .map(|name| name.ends_with(suffix))
                .unwrap_or(false)
        {
            matching_files.push(path);
        }
    }

    match matching_files.len() {
        0 => Err(anyhow::anyhow!("No file found ending with '{}'", suffix))
            .with_context(|| format!("In directory: {}", dir_path)),
        1 => Ok(matching_files[0].to_string_lossy().into_owned()),
        _ => Err(anyhow::anyhow!(
            "Multiple files found ending with '{}'",
            suffix
        ))
        .with_context(|| format!("In directory: {}", dir_path)),
    }
}

pub fn path_prefix(out: &str) -> anyhow::Result<String> {
    let out_path = Path::new(&out);

    let out_dir = out_path
        .parent()
        .ok_or_else(|| anyhow::anyhow!("Can't parse the dir of {}", out_path.display()))?;

    let name = out_path
        .file_name()
        .and_then(|name| name.to_str())
        .ok_or_else(|| anyhow::anyhow!("Can't parse the file name of {}", out_path.display()))?;

    let stem = name
        .split_once('.')
        .map(|(stem, _)| stem)
        .ok_or_else(|| anyhow::anyhow!("Can't parse the file stem of {}", name))?;

    Ok(format!("{}/{stem}", out_dir.display()))
}

pub fn force_or_not(_path: &str, _force: bool) -> anyhow::Result<()> {
    // let path = Path::new(path);
    // let mut output_exists = path.exists();
    // let dir = path
    //     .parent()
    //     .context(format!("Can't parse the parent dir of {}", path.display()))?;
    //
    // if force && output_exists {
    //     fs::remove_dir_all(dir)?;
    //     fs::create_dir_all(dir)?;
    //     output_exists = false;
    // }
    //
    // if output_exists {
    //     info!("{} already exists.", path.display())
    // }
    Ok(())
}

use rayon::prelude::*;
use std::cmp::Ord;

pub struct VectorIntersection<T> {
    pub common: Vec<T>,
    pub only_in_first: Vec<T>,
    pub only_in_second: Vec<T>,
}

impl<T> fmt::Display for VectorIntersection<T> {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let total_items = self.common.len() + self.only_in_first.len() + self.only_in_second.len();

        writeln!(f, "Total items: {}", total_items)?;
        writeln!(
            f,
            "Common: {} ({:.2}%)",
            self.common.len(),
            percentage(self.common.len(), total_items)
        )?;
        writeln!(
            f,
            "Only in first: {} ({:.2}%)",
            self.only_in_first.len(),
            percentage(self.only_in_first.len(), total_items)
        )?;
        writeln!(
            f,
            "Only in second: {} ({:.2}%)",
            self.only_in_second.len(),
            percentage(self.only_in_second.len(), total_items)
        )
    }
}

fn percentage(part: usize, total: usize) -> f64 {
    if total == 0 {
        0.0
    } else {
        (part as f64 / total as f64) * 100.0
    }
}

impl<T> Default for VectorIntersection<T> {
    fn default() -> Self {
        Self {
            common: Vec::new(),
            only_in_first: Vec::new(),
            only_in_second: Vec::new(),
        }
    }
}

impl<T: Ord + Clone> VectorIntersection<T> {
    fn merge(&mut self, other: &mut Self) {
        self.common.append(&mut other.common);
        self.only_in_first.append(&mut other.only_in_first);
        self.only_in_second.append(&mut other.only_in_second);
    }
}

fn intersection<T: Ord + Clone>(vec1: &[T], vec2: &[T]) -> VectorIntersection<T> {
    let mut result = VectorIntersection::default();
    let mut i = 0;
    let mut j = 0;

    while i < vec1.len() && j < vec2.len() {
        match vec1[i].cmp(&vec2[j]) {
            Ordering::Less => {
                result.only_in_first.push(vec1[i].clone());
                i += 1;
            }
            Ordering::Greater => {
                result.only_in_second.push(vec2[j].clone());
                j += 1;
            }
            Ordering::Equal => {
                let val = &vec1[i];
                let mut count1 = 1;
                let mut count2 = 1;

                // Count occurrences in vec1
                while i + 1 < vec1.len() && &vec1[i + 1] == val {
                    i += 1;
                    count1 += 1;
                }

                // Count occurrences in vec2
                while j + 1 < vec2.len() && &vec2[j + 1] == val {
                    j += 1;
                    count2 += 1;
                }

                // Add to common
                result
                    .common
                    .extend(std::iter::repeat(val.clone()).take(count1.min(count2)));

                // Add excess to only_in_first or only_in_second
                match count1.cmp(&count2) {
                    Ordering::Greater => {
                        result
                            .only_in_first
                            .extend(std::iter::repeat(val.clone()).take(count1 - count2));
                    }
                    Ordering::Less => {
                        result
                            .only_in_second
                            .extend(std::iter::repeat(val.clone()).take(count2 - count1));
                    }
                    Ordering::Equal => {
                        // No excess elements, do nothing
                    }
                }
                i += 1;
                j += 1;
            }
        }
    }

    result.only_in_first.extend(vec1[i..].iter().cloned());
    result.only_in_second.extend(vec2[j..].iter().cloned());

    result
}

pub fn par_intersection<T: Ord + Send + Sync + Clone>(
    vec1: &[T],
    vec2: &[T],
) -> VectorIntersection<T> {
    let chunk_size = (vec1.len() / rayon::current_num_threads()).max(1);

    vec1.par_chunks(chunk_size)
        .map(|chunk| {
            let start = vec2.partition_point(|x| x < &chunk[0]);
            let end = vec2.partition_point(|x| x <= &chunk[chunk.len() - 1]);

            // Ensure start is not greater than end
            if start <= end {
                intersection(chunk, &vec2[start..end])
            } else {
                // If start > end, there's no intersection for this chunk
                VectorIntersection::default()
            }
        })
        .reduce(VectorIntersection::default, |mut acc, mut x| {
            acc.merge(&mut x);
            acc
        })
}

pub fn temp_file_path(suffix: &str) -> std::io::Result<PathBuf> {
    let temp_path = tempfile::Builder::new()
        .prefix("pandora-temp-")
        .suffix(suffix)
        .rand_bytes(5)
        .tempfile()?
        .into_temp_path();

    Ok(temp_path.to_path_buf())
}

pub fn estimate_shannon_entropy(dna_sequence: &str) -> f64 {
    let m = dna_sequence.len() as f64;

    // Early return for empty sequences
    if m == 0.0 {
        return 0.0;
    }

    // Count occurrences of each base
    let mut bases = HashMap::<char, usize>::new();
    for base in dna_sequence.chars() {
        *bases.entry(base).or_insert(0) += 1;
    }

    // Calculate Shannon entropy
    let shannon_entropy_value: f64 = bases
        .values()
        .map(|&n_i| {
            let p_i = n_i as f64 / m;
            if p_i > 0.0 {
                -p_i * p_i.log2()
            } else {
                0.0 // Avoid log2(0)
            }
        })
        .sum();

    shannon_entropy_value
}

pub fn mean<T>(values: &[T]) -> f64
where
    T: Copy + Add<Output = T> + Div<Output = T> + Sum + Into<f64>,
{
    let count = values.len();

    if count == 0 {
        return 0.0;
    }

    let sum: T = values.iter().copied().sum();
    sum.into() / count as f64
}

pub fn bin_data<V>(data: Vec<V>, bin_size: V) -> Vec<(V, usize)>
where
    V: Copy + PartialOrd + std::ops::AddAssign + std::ops::Add<Output = V>,
{
    // Sort the data
    let mut sorted_data = data.clone();
    sorted_data.sort_by(|a, b| a.partial_cmp(b).unwrap());

    // Initialize bins
    let mut bins: Vec<(V, usize)> = Vec::new();
    let mut current_bin_start = sorted_data[0];
    let mut count = 0;

    for &value in &sorted_data {
        if value < current_bin_start + bin_size {
            count += 1;
        } else {
            bins.push((current_bin_start, count));
            current_bin_start += bin_size;
            count = 1;
        }
    }

    // Push the last bin
    bins.push((current_bin_start, count));

    bins
}

// fn aggregate_data(data: &[(u128, u32)], num_bins: usize) -> Vec<(u32, u32)> {
//     if data.is_empty() || num_bins == 0 {
//         return vec![];
//     }
//
//     let bin_size = ((data.len() as f64) / (num_bins as f64)).ceil() as usize;
//
//     (0..num_bins)
//         .map(|i| {
//             let start = i * bin_size;
//             let end = (start + bin_size).min(data.len()); // Ensure `end` does not exceed `data.len()`
//
//             // If `start` is out of bounds, return (0, 0)
//             if start >= data.len() {
//                 return (0, 0);
//             }
//
//             let bin = &data[start..end];
//
//             let sum_x: u128 = bin.iter().map(|&(x, _)| x).sum();
//             let count = bin.len() as u128;
//             let mean_x = (sum_x / count) as u32; // Rounded down to nearest u32
//
//             let sum_n: u32 = bin.iter().map(|&(_, n)| n).sum();
//
//             (mean_x, sum_n)
//         })
//         .collect()
// }
//

pub fn app_storage_dir() -> anyhow::Result<PathBuf> {
    let app_name = env!("CARGO_PKG_NAME");
    let app_dir = dirs::data_dir()
        .context("Failed to get data directory")?
        .join(app_name);

    if !app_dir.exists() {
        fs::create_dir_all(&app_dir).context("Failed to create application directory")?;
    }

    Ok(app_dir)
}
use blake3::Hasher as Blake3Hasher;
use std::hash::{BuildHasher, Hasher};

pub struct Blake3Hash(Blake3Hasher);

impl Hasher for Blake3Hash {
    fn finish(&self) -> u64 {
        let hash = self.0.finalize();
        u64::from_le_bytes(hash.as_bytes()[..8].try_into().unwrap())
    }

    fn write(&mut self, bytes: &[u8]) {
        self.0.update(bytes);
    }
}

/// Default Hasher
#[derive(Default, Clone)]
pub struct Blake3BuildHasher;

impl BuildHasher for Blake3BuildHasher {
    type Hasher = Blake3Hash;

    fn build_hasher(&self) -> Self::Hasher {
        Blake3Hash(Blake3Hasher::new())
    }
}
// Custom 128-bit hash type
#[derive(PartialEq, Eq, Clone, Copy, Serialize, Deserialize, Debug, Encode, Decode)]
pub struct Hash128([u8; 16]);

impl std::hash::Hash for Hash128 {
    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
        state.write(&self.0);
    }
}

impl Hash128 {
    pub fn new(bytes: [u8; 16]) -> Self {
        Hash128(bytes)
    }
    pub fn to_bytes(&self) -> [u8; 16] {
        self.0
    }
}

pub fn get_dir_size(path: &Path) -> std::io::Result<u64> {
    let mut total_size = 0;
    if path.is_dir() {
        for entry in fs::read_dir(path)? {
            let entry = entry?;
            let path = entry.path();
            if path.is_file() {
                total_size += path.metadata()?.len();
            } else if path.is_dir() {
                total_size += get_dir_size(&path)?;
            }
        }
    }
    Ok(total_size)
}

/// Finds all files matching the given glob pattern.
///
/// # Arguments
///
/// * `pattern` - A glob pattern string (e.g., `"data/**/*.txt"`).
///
/// # Returns
///
/// * `Ok(Vec<PathBuf>)` with all successfully matched file paths.
/// * `Err` if the glob pattern is invalid or any matched path fails to resolve.
///
/// # Errors
///
/// Returns an error if:
/// - The glob pattern is invalid.
/// - A file path matching the pattern cannot be resolved.
///
/// # Examples
///
/// ```rust
/// let files = find_files("src/**/*.rs")?;
/// for file in files {
///     println!("{:?}", file);
/// }
/// ```
pub fn find_files(pattern: &str) -> anyhow::Result<Vec<PathBuf>> {
    let mut result = Vec::new();

    let entries = glob(pattern)
        .with_context(|| format!("Invalid glob pattern: '{}'", pattern))?;

    for entry in entries {
        let path = entry.with_context(|| format!("Failed to resolve path for pattern '{}'", pattern))?;
        result.push(path);
    }

    Ok(result)
}

// fn system_time_to_utc(system_time: SystemTime) -> Option<DateTime<Utc>> {
//     system_time
//         .duration_since(UNIX_EPOCH)
//         .ok()
//         .map(|duration| Utc.timestamp(duration.as_secs() as i64, duration.subsec_nanos()))
// }

pub fn list_directories(dir_path: &str) -> std::io::Result<Vec<String>> {
    let mut directories = Vec::new();

    for entry in fs::read_dir(dir_path)? {
        let entry = entry?;
        let path = entry.path();

        // Check if the path is a directory
        if path.is_dir() {
            if let Some(dir_name) = path.file_name().and_then(|name| name.to_str()) {
                directories.push(dir_name.to_string());
            }
        }
    }

    Ok(directories)
}

/// Checks whether the modification time of `file1` is older than `file2`.
///
/// If `rm` is `true` and `file1` is older, attempts to remove the directory containing `file1`.
///
/// # Arguments
///
/// * `file1` - Path to the first file.
/// * `file2` - Path to the second file.
/// * `rm` - If true, and `file1` is older, attempts to remove its parent directory.
///
/// # Returns
///
/// * `Ok(true)` if `file1` is older than `file2`.
/// * `Ok(false)` otherwise.
///
/// # Errors
///
/// Returns an [`anyhow::Error`] if:
/// - Either `file1` or `file2` does not exist.
/// - File metadata cannot be read.
/// - File modification times cannot be retrieved.
/// - (if `rm == true`) Directory removal fails (if uncommented).
///
pub fn is_file_older(file1: &str, file2: &str, rm: bool) -> anyhow::Result<bool> {
    let mtime1 = fs::metadata(file1)
        .with_context(|| format!("Failed to read metadata for '{}'", file1))?
        .modified()
        .with_context(|| format!("Failed to get modified time for '{}'", file1))?;

    let mtime2 = fs::metadata(file2)
        .with_context(|| format!("Failed to read metadata for '{}'", file2))?
        .modified()
        .with_context(|| format!("Failed to get modified time for '{}'", file2))?;

    if mtime1 < mtime2 && rm {
        if let Some(file1_dir) = Path::new(file1).parent() {
            warn!("Removing old directory: {}", file1_dir.display());
            // fs::remove_dir(file1_dir)?;
        }
    }

    Ok(mtime1 < mtime2)
}

pub fn remove_dir_if_exists(dir: &str) -> anyhow::Result<()> {
    debug!("Trying to remove: {dir}");
    // match fs::remove_dir_all(dir) {
    //     Ok(_) => {}
    //     Err(e) if e.kind() == std::io::ErrorKind::NotFound => {}
    //     Err(e) => {
    //         anyhow::bail!("Failed to remove directory '{}': {}", dir, e);
    //     }
    // };
    Ok(())
}

/// Searches a directory for the first file matching a given name pattern.
///
/// This function looks for a file in the given directory whose filename
/// starts with the provided `starts_with` prefix and ends with the provided
/// `ends_with` suffix. It returns the full path to the first match found.
///
/// # Arguments
///
/// * `dir` - A reference to the directory path where the search will occur.
/// * `starts_with` - The required prefix of the filename (e.g., `"throughput_"`).
/// * `ends_with` - The required suffix of the filename (e.g., `".csv"`).
///
/// # Returns
///
/// * `Some(PathBuf)` - Path to the first file matching the pattern.
/// * `None` - If no matching file is found or the directory can't be read.
///
/// # Example
///
/// ```rust
/// let dir = std::path::Path::new("/path/to/data");
/// if let Some(path) = find_matching_file(dir, "throughput_", ".csv") {
///     println!("Found file: {}", path.display());
/// } else {
///     eprintln!("No matching file found.");
/// }
/// ```
pub fn find_matching_file(dir: &Path, starts_with: &str, ends_with: &str) -> Option<PathBuf> {
    fs::read_dir(dir)
        .ok()?
        .filter_map(Result::ok)
        .map(|entry| entry.path())
        .find(|path| {
            path.is_file()
                && path
                    .file_name()
                    .and_then(|name| name.to_str())
                    .map(|name| name.starts_with(starts_with) && name.ends_with(ends_with))
                    .unwrap_or(false)
        })
}

/// Searches for the first file in the given directory whose file name
/// satisfies the provided condition.
///
/// # Arguments
///
/// * `dir` - Path to the directory to search.
/// * `condition` - A closure that takes a file name (`&str`) and returns `true`
///   if the file matches the desired condition.
///
/// # Returns
///
/// An `Option<PathBuf>` containing the path to the first matching file,
/// or `None` if no file matches or the directory can't be read.
///
/// # Examples
///
/// ```
/// use std::path::Path;
/// let result = find_file(Path::new("."), |name| name.ends_with(".rs"));
/// ```
pub fn find_file<F>(dir: &Path, condition: F) -> Option<PathBuf>
where
    F: Fn(&str) -> bool,
{
    fs::read_dir(dir).ok()?.find_map(|entry| {
        let path = entry.ok()?.path();

        if path.is_file()
            && path
                .file_name()
                .and_then(|name| name.to_str())
                .map(&condition)
                .unwrap_or(false)
        {
            Some(path)
        } else {
            None
        }
    })
}