log_ingest/src/files.rs

use anyhow::{Result, anyhow};
use chrono::NaiveDate;
use flate2::read::GzDecoder;
use std::fs::File;
use std::io::{BufRead, BufReader, Read};
use std::path::PathBuf;

/// Enum-based reader to avoid Box<dyn BufRead> heap allocation and dynamic dispatch
pub enum LogReader {
    Plain(BufReader<File>),
    Gzip(BufReader<GzDecoder<File>>),
}

impl Read for LogReader {
    fn read(&mut self, buf: &mut [u8]) -> std::io::Result<usize> {
        match self {
            LogReader::Plain(r) => r.read(buf),
            LogReader::Gzip(r) => r.read(buf),
        }
    }
}

impl BufRead for LogReader {
    fn fill_buf(&mut self) -> std::io::Result<&[u8]> {
        match self {
            LogReader::Plain(r) => r.fill_buf(),
            LogReader::Gzip(r) => r.fill_buf(),
        }
    }

    fn consume(&mut self, amt: usize) {
        match self {
            LogReader::Plain(r) => r.consume(amt),
            LogReader::Gzip(r) => r.consume(amt),
        }
    }
}

/// Discovers log files for a given date range
pub struct LogFileDiscovery {
    base_dir: PathBuf,
    filename: String,
}

impl LogFileDiscovery {
    pub fn new(base_dir: PathBuf, filename: String) -> Self {
        Self { base_dir, filename }
    }

    /// Returns an iterator over all log files in the date range
    pub fn discover(&self, from: NaiveDate, to: NaiveDate) -> Result<Vec<LogFile>> {
        let mut files = Vec::new();

        let mut current = from;
        while current <= to {
            if let Some(log_file) = self.find_log_for_date(current)? {
                files.push(log_file);
            }
            current = current.succ_opt().ok_or_else(|| anyhow!("Date overflow"))?;
        }

        Ok(files)
    }

    fn find_log_for_date(&self, date: NaiveDate) -> Result<Option<LogFile>> {
        // Build path: <base_dir>/yyyy/mm/dd/<filename>.gz or <filename>
        let date_path = self
            .base_dir
            .join(date.format("%Y").to_string())
            .join(date.format("%m").to_string())
            .join(date.format("%d").to_string());

        // Try gzipped first
        let gz_path = date_path.join(format!("{}.gz", self.filename));
        if gz_path.exists() {
            return Ok(Some(LogFile {
                path: gz_path,
                compressed: true,
            }));
        }

        // Try uncompressed
        let plain_path = date_path.join(&self.filename);
        if plain_path.exists() {
            return Ok(Some(LogFile {
                path: plain_path,
                compressed: false,
            }));
        }

        // No file found for this date
        Ok(None)
    }
}

#[derive(Debug)]
pub struct LogFile {
    pub path: PathBuf,
    pub compressed: bool,
}

impl LogFile {
    /// Returns a buffered reader for this log file, handling compression transparently
    pub fn reader(&self) -> Result<LogReader> {
        let file = File::open(&self.path)?;

        if self.compressed {
            let decoder = GzDecoder::new(file);
            Ok(LogReader::Gzip(BufReader::new(decoder)))
        } else {
            Ok(LogReader::Plain(BufReader::new(file)))
        }
    }
}

/// For reading a single file directly (e.g., for testing)
pub fn read_log_file(path: &str) -> Result<LogReader> {
    let file = File::open(path)?;

    if path.ends_with(".gz") {
        let decoder = GzDecoder::new(file);
        Ok(LogReader::Gzip(BufReader::new(decoder)))
    } else {
        Ok(LogReader::Plain(BufReader::new(file)))
    }
}